Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Proceedings Day 1 April 18, 2007 WELCOME, INTRODUCTIONS, AND WORKSHOP OBJECTIVES: Dr. Alan Leshner Dr. Leshner: Good morning everyone. I am Alan Leshner. I am the CEO (Chief Executive Officer) of the American Association for the Advancement of Science and Executive Publisher of Science magazine, but I am here in my role as chair of the Institute of Medicineâs (IOMâs) Forum on Neuroscience and Nervous System Disorders. I am delighted to welcome everyone. This is the workshop Autism and the Environment: Challenges and Opportunities for Research. The major purpose of this workshop is to work together to try to figure out how we can do a better job to bring the full power of science to bear on a public problem of tremendous magnitude and tremendous import. The IOMâs Forum on Neuroscience and Nervous System Disorders has the purpose of building partnerships and discussions to further understand the brain and the nervous system, to understand disorders and their structure and function, as well as clinical prevention and treatment strategies. What the forum does is to bring together leaders from the public and private sectors, including federal agencies, the pharmaceutical industry, advocacy organizations, and the academic community to have conversa- tions about these general critical issues. 5
6 AUTISM AND THE ENVIRONMENT In addition, we try to serve an educational function, educating the press, the public, and policy makers about neuroscience and nervous system disorders. One of the mechanisms through which we operate is workshops like the one today that provide a venue for discussion about key challenges and opportunities in the field. The Forum was asked to host this workshop by the U.S. Secretary of Health and Human Services, and William Raub will speak in a moment to help explain its origins, but it came about as a result of a series of discussions among members of the autism community, the Office of the Secretary, and Dr. Harvey Fineberg, president of the Institute of Medicine. I do want to specifically thank Kelli Ann Davis, Jim Moody, and Mark BlaxillâMark has also been a member on our planning committeeâwho were instrumental members of the advocacy community in all of these discussions. Let me say a few words about the format for todayâs meeting. You all have a copy of the agenda and I wonât read it to you, but let me just reiterate that the workshop objectives are to look to the future, to look for and try to identify the most promising scientific opportunities for improving the understanding of potential environmental factors in autism, to talk about what infrastructure, what tools and technologies are available and what is needed, what kinds of interdisciplinary approaches are needed and other kinds of infrastructure, investments and then to talk about exploring potential partnerships that are needed to support and conduct autism research. The format that we are using, if you look at the agenda (Appendix B), is to have a series of speakers in each of numerous settings. There actually are way too many speakers for a normal workshop, but we were unable to figure out how to keep this in proportion and make sure that we covered this very complex issue fully. So, we are going to be rather ruthless in maintaining the organization of the workshop. Each speaker has been given 15 minutes. We will have one or two minutes for what I will call critical questions of clarification right after each talk, but really we mean critical clarification, not a discussion and not a discourse. Then at the end of each session, we have allocated actually a substantial amount of time for discussion among the session participants and then we have allocated time at the end of the day for continued discussion, but also an opportunity for members of the audience to participate at that time as well.
PROCEEDINGS 7 I would ask those people who are planning to ask questions or develop questions over the course of the dayâs events or the 2 daysâ events, please, no statements, no long harangues. This is about questions. This is a scientific meeting and we are looking for scientific opportunities and ways that we can move the science forward. So, I remind you of that, but I will surely remind you of that again. Again, the purpose of the workshop is to stimulate discussion about how best to move research on autism and the environment forward and I need to make some, I apologize, bureaucratic announcements on behalf of the IOM. One, this is not a consensus conference. We are not, in fact, expecting to come to a consensus by the end of the meeting. What we are expecting to do is to hear an array of opportunities, an array of needs, an array of challenges that will be identified. A proceeding of the workshop will be written. (NOTE: This is the pub- lished document to which Dr. Leshner refers in his remarks.) It will not, again, be an official consensus statement or consensus report of the Institute of Medicine. It is outside our authority as a forum to produce those kinds of reports, but, hopefully, what it will do is set the stage for a research agenda moving forward, and that is really what our goal is today. I do want to thank the planning committee, which has been so instrumental in this workshop. It could not have been done without the joint activity of people from many different sectors with interest in this. One request of the speakers before we move on and that isâand I apologize for doing this late, but we are a little bit concerned that with all the discussion and all the talk, it may be a bit difficult to capture each speakerâs view of what a major opportunity or a major gap to be filled might be. Therefore, if it is not too late to do that organizationally in your head, if at the end of your talk you could articulate at least one, just so the recorders can write down, here is one potential gap in scientific knowledge or potential scientific opportunity that needs to be filled. If we can focus in that way toward the future, I think we can make a larger contribution than we could otherwise. I donât want to take too much time. We are already a bit constrained and I, again, want to thank the members of the planning committee, who did such a wonderful job of pulling this together. I want to thank all the speakers, who will be with us today and tomorrow, and all of you who are participating in this.
8 AUTISM AND THE ENVIRONMENT Let me turn now to Dr. William Raub, science advisor to the Secretary of the Department of Health and Human Services (DHHS), who just will make some additional comments about the workshop and its origins and its particular charge. CHARGE TO WORKSHOP PARTICIPANTS Dr. William Raub Dr. Raub: Good morning everyone. I add my welcome to Dr. Leshnerâs. I am delighted at the wonderful turnout. The journey that brought us here began almost 2 years ago with a protest action directed against the Executive Branch. Parents of autistic children and other advocates were mindful of a reelection campaign promise to eliminate mercury-based compounds from vaccines. Staff of the Domestic Policy Council asked me to host a meeting whereby representatives of the advocates could state their concerns in person. I did so and had the privilege of meeting a group of impressive individu- als, several of whom are here today. I would like to be able to say that the protest event, the follow-up meeting at Health and Human Services, and the subsequent communica- tion from the Council back to the advocates left everyone satisfied. But that was not the case, and deep divisions remain over the matter of vaccine safety. Nevertheless, several of the advocates asked if I would be amenable to hosting further meetings, whereby they could lay out additional concerns about how the institutions of science have approached the problem of autism. I quickly agreed, having been moved deeply by the quality of the advocatesâ preparedness, the sincerity of their representa- tions, and the power of their testimony regarding the crushing burden that autism places on not only the affected children, but also the entire family. That led to a series of meetings with various combinations of repre- sentatives from the autism advocacy communityâbut always focused on what science has or has not done and what more it can or should do. Last October, at one of those sessions, Dr. Harvey Fineberg, president of the Institute of Medicine, joined Mark Blaxill, Kelli Ann Davis, Jim Moody, and me to discuss the IOMâs studies of vaccine safety and related
PROCEEDINGS 9 activities. Out of that meeting arose the notion that some sort of IOM- anchored, autism-oriented event could make a uniquely important contribution to shaping the research agenda against this dreaded disease. During the weeks that followed, Dr. Fineberg and I posed this generic concept to Dr. Thomas Insel, director of the National Institute of Mental Health (NIMH) at the National Institutes of Health (NIH) and chairman of the HHS Interagency Autism Coordinating Committee. Then we widened the circle to include two other NIH leaders: Dr. David Schwartz, director of the National Institute of Environmental Health Sciences (NIEHS), and Dr. Duane Alexander, director of the National Institute of Child Health and Human Development. This team quickly determined that an IOM-hosted workshop focused on potential environmental factors contributing to the etiology or pathogenesis of autism would make for a highly desirable and value-added contribution to ongoing NIH-based efforts to develop a strategic plan for autism research. Without equivocation, Drs. Fineberg and Leshner affirmed that the IOM neuroscience forum would host such a workshop. Drs. Bruce Altevogt and Andrew Pope and their staff recruited and facilitated the deliberations of a first-class planning committee. The three institutes that I have already mentioned, plus the National Institute of Neurological Disorders and Stroke and the Centers for Disease Control and Prevention (CDC), agreed to provide the requisite funding. As indicated by the agenda and the advanced materials on the IOM website, the planning committee tried to ensure that no potentially important environmental contributor to autism has been overlooked or excluded. Although the workshop is not intended to reprise the analysis of the epidemiological evidence related to vaccine safety, the planning committee recognized that vaccine constituents, especially organic chemicals used as preservatives or adjuvants, obviously qualify as environmental agents that warrant attention. In other words, our research agenda should include studies of any and all environmental agents that plausibly might contribute to causing or exacerbating autism, irrespective of the medium of exposure. I am hopeful that the next 2 days will prove to be an important milestone for autism researchânot only because this workshop is addressing vitally important questions about the cause or causes of the disease, but also because the agenda is the product of collaboration between advocates for autistic children and their families and the scientific community. To be sure, other aspects of the autism challenge deserve similar
10 AUTISM AND THE ENVIRONMENT attention, especially the paucity of effective treatments, and autism advocates and the scientific community have much further to go to achieve the full measure of mutual understanding and trust. But our challenge here and now is clear: to step together onto the path to a better day, to set the stage for other important steps to come, and to make other advocates and scientists want to be part of that advancing throng. Thank you for being here. Dr. Leshner: Thank you, Bill, and thank you very much for your important efforts in getting this meeting organized, getting it stimulated, and setting the appropriate stage for it. I do want to, again, thank the planning committee. I neglected to mention that the importance of this meeting from a scientific and a public health perspective is reflected by the very large number of members of our forum, who came today, in spite of it not being an official regular meeting of the forum. I really very much appreciate the help and support. Almost the entire forum has come today from many different sectors. I think that is an important statement. I also want to reiterate William Raubâs thanks to Bruce Altevogt, Sarah Hanson, and their colleagues from the IOM, who have done a phenomenal amount of work putting this all together and making sure that it happens. Let me not take more time, but rather turn to Laura Bono, who has been a member of our workshop planning committee, is a board member of the National Autism Association, and has agreed to bring to the group the perspectives of the advocacy community. PERSPECTIVES OF THE ADVOCACY COMMUNITY1 Ms. Laura Bono Ms. Bono: I am Laura Bono, founding board member and past chair of the National Autism Association. I have been asked to talk about the perspectives of the advocacy community. My time is short, so I will get right to the point of what many in the advocacy community want and think. Declare autism a national health emergency under the Public Health 1 Throughout Ms. Bonoâs presentation, she may refer to slides that can be found online at http://www.iom.edu/?id=42455.
PROCEEDINGS 11 Act and treat it with urgency. Thirty-six thousand children, who should be living normal lives, will succumb to the diagnosis this year alone, affecting the trajectory of their lives and that of their parents forever. Autism is estimated to cost $3.2 million per child over a lifetime. Using the conservative estimate in the United States of 500,000 children means this epidemic will cost society close to $2 trillion. Autism is both economically and emotionally devastating to the children and their families. Many families are on the brink of bankruptcy as they struggle to get insurance and the medical attention their children need. Murder/suicides of parents and their autistic children are on the rise. I canât discuss the perspectives of the advocacy community without citing the failings of the CDC. We believe the CDC has a performance and credibility problem. Their failure to declare an epidemic beginning with the 1989 birth cohort to study the time trend data or to examine the toxic and viral body burdens of children are why we are here today, over 15 years too late. Julie Gerberding, director of the CDC, said recently in a February 8, 2007, CDC press release when they announced the 1-in-150 rate that she wasnât sure if the rates are truly rising or if they are getting better at studies. âOur estimates are becoming better and more consistent, though we canât tell yet if there is a true increase in ASDs [autism spectrum disorders] or if the changes are the result of our better studies.â This denial thwarts research into environmental factors and just isnât acceptable. How many autistic individuals did you know when you were under the age of 21? Since it is impossible to have a genetic epidemic, literally hundreds of millions of taxpayer dollars could have been appropriately directed to geneâenvironment and other susceptibility initiatives. Even more could have been spent on learning about the critical mechanisms involved in response to environmental neurotoxicants. We could have been focusing on what changed in the environment and when. We could have been investigating the environmental trigger for years and successfully helping suffering children. We urgently need to begin these initiatives now. Many in the advocacy community are thankful because starting today, the government is finally going to make environmental research a priority, which will lead to better treatments and recovery. Because if autism is environmental, then it is treatable and preventable. It is no longer hopeless, or lifelong. It is hopeful, with a possible cure.
12 AUTISM AND THE ENVIRONMENT Recent clinical investigations have identified numerous comorbid disease states in children with autism. These include immune system abnormalities; inflammatory bowel disease; oxidative stress; disordered urine and serum chemistries, including elevated porphyrins; methylation disturbances; increased body burdens of metals, including mercury and lead; chronic viral, fungal, and bacterial infections; and microglial activation in the brain. Studies must be initiated as soon as possible to increase the focus on the identification of these comorbid disease states. Parents and clinicians alike are reporting that when these problems are acknowledged and treated, it can result in marked improvement in childrenâs learning and behavior. Some children recover completely. This should be a wakeup call to us all. The research paradigm needs to shift from autistic children are genetically defective to autistic children are sick and treatable. We should only grant money to genetic vulnerability and epidemiology studies that have a clear environmental hypothesis. Research detoxification treatments; identify and validate biomarkers; study biomedical imbalances and treatments that are working; investigate the role of vaccines, including thimerosal, aluminum, and live viruses; research the role of the immune, gastrointestinal, and endocrine systems; and study the recovered childrenâs pre- and post-diagnosis medical files for clues. Because it is the environment, we need to leave no stone unturned. There is a growing body of evidence implicating vaccine overload, mercury and aluminum from vaccines. Thousands of parents agree with this research. They watch their children regress after being vaccinated. Their autistic children have been diagnosed with heavy metal poisoning and immune system dysfunction and when treated, get better. Regardless of controversy surrounding any theory, we must research and produce successful antioxidant, methylation, and blood-brain barrier chelation treatments, as well as immune system, detoxification, and inflammation interventions. I want to remind you that you are tasked with setting in motion the crucial environmental research that hundreds of thousands of children are silently waiting for now. The guiding principles should be to pursue research and treatments that will impact the most lives as quickly as possible and follow clues provided by treatments currently working in children. Such an agenda would best be served by a translational research
PROCEEDINGS 13 protocol where clinicians who care for children with autism advise research into the most promising areas of intervention. It is imperative that the working group proceed with urgency and follow the truth wherever it leads. Recovery happens every day to some, but our goal should be for all. We need to accelerate environmental research; demand even more money to address the problem; issue RFAs [request for applications] and have research proposals scored according to autism matrix goals; get answers; interpret them expediently; and continue to work the problem until we beat it. We can do this. And our hope starts with you. Thank you. Dr. Leshner: Thank you very much for your very powerful statement setting the stage. I would like to respond that we share the sense of urgency. We share the sense that science and research will be the hope, and we of course share the goal of bringing the full power of science to bear on this public health problem of great urgency in tremendous proportion. I hope that we will live up to the charge that you have just given us to look at the full array of environmental factors and the ways in which they can cause this disorder, affect its progression and then, of course, the variety of ways in which we can approach it. I think that your point about the need to both inform clinical practice, but also to listen to clinical practice is a very important charge and I assure you that we are planning to take advantage of that and listen to that carefully. So, I really tremendously appreciate the statement you have just made and I promise you that we will do our best to respond to it.
Session I AutismâThe Clinical Problem: âWhat Do We Know? What Do We Need?â Dr. Leshner: I would like to turn to a discussion of the clinical problem and introduce Sarah Spence, who will serve as the session chair. I would like to point out that we are on time. So, donât feel any pressure, Dr. Spence, or speakers in this session. Dr. Spence is staff clinician at NIMH, where she works in the Pediatrics and Developmental Neuropsychiatry Branch. Thank you. Dr. Spence: Thank you, Dr. Leshner. I think we may have one of the most difficult sessions to do, which is to introduce the clinical problem and do it in an hour and 20 minutes and no more. So, I am not going to spend a lot of time on the introduction. I think the most important thing to keep in mind during this session is that it is about what we know and what we need. It is about introducing the main issues to set the stage for a productive discussion later on today and getting a diverse audience, kind of onto a level playing field about what the issues are. So, to start with the clinical problem, I am going to introduce my boss, Dr. Susan Swedo, the chief of the branch that I work in at the NIMH. CLINICAL OVERVIEW: HOW CAN THE CLINICAL MANIFESTATIONS OF AUTISM SHED LIGHT ON POTENTIAL ENVIRONMENTAL ETIOLOGIES?2 Dr. Susan Swedo Dr. Swedo: Since I only have 15 minutes today to describe all of autism to you and why we believe that the environment plays such a crucial role in this disorder, Iâm going to be using videos to show you in a few seconds what it would take me a very long time to try to explain. 2 Throughout Dr. Swedoâs presentation, she may refer to slides that can be found online at http://www.iom.edu/?id=42456. 15
16 AUTISM AND THE ENVIRONMENT Autism is characterized by two areas of deficit, deficits in social interactions and communication deficits. It is also defined by an excess of repetitive behaviors or fixated interests. Now, these fixated interests and repetitive behaviors are not usually present during the very earliest stages of the illness and increase in time as the child becomes older. As we know, autism is a developmental disorder. By definition, symptoms must appear before age 3 years and affect development. The crucial thing is that the development affects the symptom expression and the symptom expression also affects development. Since this is a disorder of social communication, which is essential to all development-al interactions, autism can quickly take you very far off of your expected trajectory. Autism is one of several pervasive developmental disorders (PDDs), which are now commonly called the autism spectrum disorders. I think that expanding the continuum to include all pervasive developmental disorders as âautismâ is a bit confusing and dilutes the meaning of the term, so I am asking that we keep our focus today on those children who meet full criteria for autism. Rett disorder is caused by a genetic mutation, which leads to symptoms very similar to autism and, in fact, until we knew what the gene was, girls with Rett disorder were included in the autism group. Since the gene has been identified for Rett disorder, it is now considered to be separate from other autistic disorders. Similarly, childhood disin- tergrative disorder presents with symptoms of autism, except that the children donât begin to regress and lose their skills until after age 3. Here is an example of the social deficits in autism. One of the crucial components of social interactions is joint attentionâbeing able to pay attention to things that are of interest to others. (Video shown of a child performing a task of joint attention.) Here you see a normal volunteer from our lab. His reward is a bunny and he is very clearly excited and he tries to share that excitement with the examiner. Here is a 4-year-old girl with autism performing the same task. The bunny is behind the examiner again. You see the examiner saying âLook! Is it a bunny?â but the child is oblivious, preoccupied with other thoughts. I am going to replay that section of the video and ask you to also watch the repetitive behaviors that she exhibits. Notice that the child pulls her hands into her sides. Then when she gets excited, there is a repetitive motion.
PROCEEDINGS 17 Another common social interaction is shared enjoyment. (Video shown.) The young child with typical development says, âWow! Look at that!â when shown the bubble gun. He invites his mother to share in his enjoyment of the new toy before asking if he can have a turn operating it. It is easy to see how excited he is by the toy. He uses gestures to make his needs known, as well as his verbal comments. (Next video is shown.) Here is XXXX,3 a little boy of about the same age, with fairly severe autistic symptoms, presented with the same bubble task. He clearly sees the bubbles, is interested in them. The examiner gives him every cue she can to get him to ask for more bubbles, but he doesnât. He just seems terribly confused and somewhat upset. (Video segment.) Here is an example of an autistic childâs perseverative behaviors. You probably have heard about the autistic children who spin the wheels of the bus, rather than playing with the bus as it is intended. Here the pop-up toy has become an instant area of fixated interest for him. He isnât playing with it as intended, but rather, chooses to repetitively open and shut one of the doors. The examiner is trying to get him to look over at the bunny. But he is not willing to attend to anything but the pop-up toy. Even when she gently takes the toy away, he remains fixated on the spot where it was sitting. So, this child demonstrates both deficits in social communications and an excess of repetitive behaviors. The causes of autism that are known are mainly genetic. About 10 percent of children diagnosed with autism have been found to have a genetic cause. Less than 1 percent have been attributed to teratogens, such as valproic acid or thalidomide. That leaves about 90 percent of the kids, or 9 out of 10, for whom the cause is idiopathic, meaning we just donât know. That does not mean that there is no known cause. It just means that the cause is not known. When autism is related to a genetic defect, the pathogenesis is relatively âsimple.â Even then, there is a great deal that happens between the genetic mutation and the manifestation of neuronal dysfunction and/or damage. But when something in the environment is causing the symptoms, it is even harder to make a direct link. But the working model is that environmental factors, in a genetically susceptible population, lead to neuronal dysfunction and/or damage and the symptoms of autism. The tricky thing about that pathogenic model is the fact that it has so 3 Out of respect to privacy for the family, the name of this individual has been replaced with âXXXX.â
18 AUTISM AND THE ENVIRONMENT many stages, each of which is actually broken down into many, many more steps. So, for the purposes of the conference, you are going to be hearing a lot about genetic mechanisms that might create vulnerabilities, about the environmental factors that trigger the symptom onset, and even though weâll be addressing individual parts of the diagram, we need to keep the larger picture in mind at all times. Potential environmental triggers that have been suggested are numerous. They include the toxicants, which will be discussed by Isaac Pessah; the infectious agents, which Ian Lipkin will be speaking to in a later session; and household exposures, such as household chemicals and cleaning products. The household exposures are one of the areas of study for the NIEHS-sponsored CHARGE (Childhood Autism Risks from Genetics and the Environment) study and the CDC-sponsored CADDRE studies (Centers for Autism and Developmental Disabilities Research and Epidemiology). Food, dietary supplements, and vitamins and minerals may also be involved in autism. If you think about how we eat today, compared to how we ate in the 1950s, it is mind boggling how many changes there have been. Of particular interest have been changes in folic acid supplements, and the utilization of aspartame, because both have been associated with other neurologic conditions. Additional environmental factors include drugs, medications, and herbal remedies. For example, as a pediatrician, I know that there was a dramatic change in the treatment of children with fever following the Reyeâs syndrome epidemic. And practice guidelines required a switch from giving children aspirin following vaccinations to prescribing Tylenol and/or ibuprofen. We donât know what the effect of that might have been, but it is certainly an area for investigation. Other medical interventions that might play an etiologic role include the use of ultrasounds during pregnancy, and the administration of vaccinesânot just the contents of those vaccines, but also the increasing number and the immunologic challenges that are faced by our children today, in comparison with previous generations. Technological advances include the ultrasounds, but also microwave ovens, cell phones, and everything else. So, you really end up with an overwhelming array of environmental factors to consider because in essence, everything encountered by the mom, the dad, and the child could be a potential environmental trigger. There are some clinical clues that suggest that the environment is
PROCEEDINGS 19 playing a role in the etiology of autism: first, the association with the teratogenic agents is a direct cause-and-effect relationship; second, the reported prevalence of autism is increasing at dramatic rates; and third, the fact that the symptoms frequently have their onset between 12 and 18 months of age (not at birth). I think this is the thing that the parents see as the most compelling evidence that there must have been an environmental trigger. They tell us, âMy child was healthy and then he wasnâtââsomething must have happened in between. The change from typical development to autism certainly may have been the result of an environmental exposure, but we have to keep in mind the fact that many disorders that are genetically based do not present in the first year or even 2 years of life. Sickle cell disease is a prime example. In addition, there are disorders like Rett syndrome in which the girls are developing normally until about 12 to 15 months of age and then have a regression and lose their skills. So, I think that the age at onset of symptoms in autism is an important clue, but it isnât evidence on its own. Medical comorbidities may also provide information about environmental factors in autism. For example, within the past few years, there has been increasing attention to the link between autism and immune dysfunction that suggests a common environmental exposure is increasing prevalence rates for both autism and autoimmune disorders. We will hear more about that during this workshop as well. A request has been made that we start paying attention to the response to treatments that are being given to these children in order to find clues to the original etiology of symptoms. Many parents and practitioners are finding that symptoms can be dramatically improved or eliminated by a variety of biological and dietary interventions. At the NIH, we are attempting to do systematic studies of some of the more commonly used treatments, because open-label trials and anecdotal reports of benefit can be very difficult to assess because the child is developing naturally during that same period of time. The regressive subtype of autism is one of the most clinically compelling pieces of evidence for environmental triggers. The regressive subtype of autism is actually regressive âsubtypes,â just as there are multiple autisms. For most children with regressive autism, they develop normally until about 12 to 30 months of age, when they begin to lose the language they have acquired and stop interacting socially. However, 12 to 30 months of age is a tremendous span in development, and suggests
20 AUTISM AND THE ENVIRONMENT that even within the regressive group, there is likely to be a significant amount of heterogeneity. Fifteen to 50 percent of children with autism will have regressive features, depending on how narrowly you define âregression.â If you take the strictest definition, which requires that the child has at least 10 words and loses those, then the proportion is closer to 15 percent. To date, the prognosis for the regression group is reported to be particularly poor. Of note is the fact that regression can be very acute. We have already seen children at the NIMH clinic who were developing normally, became ill, and within a few weeks had lost all of their verbal and social skills. For most children, the process is slower and subtler; it is a painstaking process to find out how they were developing at each developmental phase and to begin to pinpoint the area at which the regression occurred. The final caveat in consideration of the regressive autism subtype is work from Dr. Geraldine Dawson and her colleagues at the University of Washington which shows that for many of the children, development wasnât completely normal before the regression occurred, but there is still a very obvious loss of acquired skills. Here is an example of a little girl who had a clear regression. She is the one that you saw with the self- stimulatory behaviors and the lack of attention. Here she is at 6 months of age. Her dad calls her name, and she gets a huge smile. Here she is at her 1-year birthday party. Again, her father calls her name, and see if you can tell when he says it. You canât, can you? So, she had already lost attention to her name. By the time she is a year and a half, he is shouting her name repeatedly, and she is completely oblivious to his presence. She had also lost words during this period. As you can see in the videos, the regression is profound. The family describes it as having their daughter âstolenâ from them by the autism. I think that is a superb description to keep in mind of the regressive subtype. The child is developing on an expected trajectory and then falls off completely. Certainly in regressive autism, the hunt for the environmental trigger should take prominence, but how do we trace back from the clinical picture to that environmental trigger? As I said earlier, it is complex. Each of these cartoon boxes has multiple stages, multiple phases, and multiple levels to be investigatedâit is a huge task, but it isnât hopeless. I was asked to tell you what I think we need to do to find these environmental factors. First, we need a standardized definition of autism and related disorders. We really need to be dealing with as clinically
PROCEEDINGS 21 homogeneous a group as possible, because within that homogeneous group, we are going to find biological heterogeneity. We already know this from all of the other medical disorders of childhood, and particularly from Type 1 diabetes and leukemia, where knowing exactly what the clinical picture looked like helped us to get to the pathophysiology. We need brain pathology. As we had our planning conference calls, it became very clear that until we know what is happening in the brain, there is not much point in trying to figure out when or where the trigger occurred. It would be helpful to have incidence data from populations with disparate risk factors. If we could look at developing nations and their rates of autism, we might be able to find clues to environmental triggers here in the United States and elsewhere in the industrial world. In order for such studies to be meaningful, however, we need to use the same diagnostic criteria for each time and place. It is very clear from work being done by international epidemiologists that if you change the diagnostic cut-off scores by just one point, the prevalence rates change dramatically. Obviously, the same thing would be true for the incidence data and would complicate any international comparisons. We need systematic evaluation of anecdotal case reports as we already know from genetic disorders that it is the exception that ends up proving the rule. So, we need to start looking for those exceptions and studying them in depth. At the same time, we need to be doing randomized control led trials of novel therapeutics, using reliable, valid, developmentally appropriate and change-sensitive outcome measuresâ such measures still need to be developed. And finally, we need identification of clinically meaningful subtypes, perhaps by identifying unique ages of onset, similarities of clinical presentations or associated symptoms, or by identifying a group with similar developmental or clinical trajectories. Since I am out of time, I will stop and take questions. Dr. Spence: So, the next 2 or 3 minutes we can use for questions directly related to Dr. Swedoâs talk or else we can move on. Dr. Swedo: Since there donât appear to be any questions, I am going to spend the next 3 minutes talking about PANDAS (Pediatric Autoim- mune Neuropsychiatric Disorders Associated with Streptococcal infect- ions) and how we at the NIMH were able to use clinically meaningful subtypes of obsessive-compulsive disorder (OCD) to go from the unique clinical presentation to the environmental trigger, and meaningful
22 AUTISM AND THE ENVIRONMENT treatment and prevention strategies in a relatively short period of time. Our hope is that we will be able to find a similarly informative subgroup of children with autism. The PANDAS subgroup differs from other children with OCD in that it has a very abrupt onset and an episodic course, in which there are periods of both relapse and remission. Boys predominate in this young population of children. When the children are acutely ill, they have developmental regression, social isolation and aggressiveness, emotional lability, sensory defensiveness, sleep difficulties, and choreiform movements. The symptoms are found in many children with autism, as well as in Sydenhamâs chorea, which is the neurologic manifestation of rheumatic fever. The association between obsessive-compulsive disorder and Sydenhamâs chorea is what led us to suspect that strep bacteria might be the environmental trigger for the abrupt-onset form of obsessive- compulsive disorder. A decade of research suggested that the presence of untreated strep bacteria in a genetically susceptible host could cause an abnormal immune response and lead to clinical manifestations of obsessive-compulsive disorders and tics. We already knew that only a few of the 120 strains of strep were capable of producing rheumatic fever, and that not all children were susceptible to the poststrep complications. In fact, only about 1 in 20 families was susceptible to rheumatic fever. It seemed like a difficult model to investigateânot all strep infections could cause symptoms and not all children would be affected, so there would be many false starts and dead ends. However, by starting with this model, we were able to borrow from the experience with rheumatic fever eradication, and conducted a controlled trial of antibiotic prophylaxis that showed beautifully that preventing strep infections was capable of preventing neuropsychiatric symptom exacerbations. By giving antibiotics to prevent strep, we were preventing episodes of OCD and tics. The slide shows the results of the trial for the first 10 patientsâon the left side of the red line is the year prior to study entry and on the right side is the year of antibiotic administration; just visually scanning the data, you can see that there are fewer symptomatic months (represented by the bars) during the year of antibiotics administration. The summary data showed that the children went from having two strep infections on average per year to zero strep infections, and that they went from having 2.4 to 0.7 neuropsychiatric exacerbations during that same period. What isnât shown here are the follow-up data demonstrating that continued antibiotic prophylaxis has
PROCEEDINGS 23 rendered over 75 percent of these children asymptomatic. The genetically susceptible host allows us to develop trait markers and susceptibility markers. We had great hope for a short period of time that the D8/17 marker would serve as a susceptibility marker for the PANDAS subgroup. Unfortunately, the original monoclonal antibody clone was lost and we havenât found one that has equal sensitivity and specificity, but the hunt goes on. The postulated abnormal immune response led to two lines of investigation. First, the search for a disease marker which would reliably distinguish children in the PANDAS subgroup from others with OCD, and the development of immunomodulatory treatments for severely affected children. Dr. Madeline Cunningham and Christine Kirvan have been the heroes in the search for disease markers. They have demonstrated that cross-reactive antibodies recognizing the strep cell walls also recognize neurons within the basal ganglia and that the titers in the Sydenhamâs chorea group (shown on the left side of the graph in the red squares) are much higher than those in the PANDAS subgroup, but the PANDAS children are significantly higher than the normal controls, and most importantly, acute and convalescent titers are dramatically different in both Sydenhamâs chorea and PANDAS. Thus, the antibody titers may be useful not only in identifying PANDAS versus non-PANDAS cases, but also in following disease progression and response to treatment. We also conducted a placebo-controlled trial of immunomodulatory treatments in which plasma exchange and IVIG (intravenous immu- noglobulin) therapy both were effective in reducing symptom severity by more than 50 percent in the first month following treatment, whereas placebo had no discernible effects. In conjunction with relieving symptoms, the immunomodulatory therapies also reduced the size of the abnormally enlarged caudate nucleus, as seen in this individual. So, for PANDAS, we were able to identify a medical model for disease etiology and to use that model to prevent symptom onset by preventing strep infections. We were also able to identify the genetically susceptible host and develop markers of disease activity, and even develop treatments that were effective in eradicating the neuropsychiat- ric symptoms. I would challenge us to try and do the same thing in autism. It is not going to be easy, but if we start with clinically meaningful subtypes of autism, we will be able to identify the etiologic triggers and keep them from doing harm. We will also be able to identify
24 AUTISM AND THE ENVIRONMENT biomarkers of genetic susceptibility and develop diagnostic tests that will identify vulnerable populations. And, of course, our ultimate goal is to move from the clinical observations to developing new methods for prevention and cure. Thank you. Dr. Spence: Thank you, Dr. Swedo. Next we have Dr. Pat Levitt, who is a professor of pharmacology at Vanderbilt. GENES AND THE ENVIRONMENT: HOW MAY GENETICS BE USED TO INFORM RESEARCH SEARCHING FOR POTENTIAL ENVIRONMENTAL TRIGGERS?4 Dr. Patrick Levitt Dr. Levitt: I am going to provide for you a neurobiologist perspective on where we are in terms of genetics and what some of the opportunities are in terms of genetics and designing the kinds of research we might be doing to understand geneâenvironment interactions. The first slide basically depicts the fact that we all understandâcomplex genetic disorders are complex. Complex genetic disorders are complex and what we are trying to understand are the combination of risk alleles, variations in gene sequences or in copy number of specific genes which, in combination, end up underlying risk or, in fact, directly perturb brain development that ends up generating the three core symptoms that are diagnostic of autism spectrum disorders. You can see in the diagram that for any disorder, a combination of risk alleles may be correct, but there may be an intermediate phenotype rather than the features of the full disorder. We know that of the three major core symptoms that are used for an autism diagnosis, dysfunction in any one of these domains can run in families. There have been large twin studies to look at heritability independent of the autism diagnosis itself. The diagram also shows that the correct combination of risk alleles 4 Throughout Dr. Levittâs presentation, he may refer to slides that can be found online at http://www.iom.edu/?id=42457.
PROCEEDINGS 25 might require specific environmental factors in order for the full-blown disorder to be expressed. There also are issues of incomplete penetrance where you may have the correct combination of genetic risk, but for reasons unknown, an individual has modifier factors that reduce the impact of the risk alleles. This means that one does not express the disorder. So, I want you to keep something in mind. I take this from Daniel Weinberger, who studies schizophrenia at the NIMH and he makes this point, I think, very well. Genes are involved in the assembly of specialized cells to perform specific functions. Thus, there are no âsocial behaviorâ or âcommunicationâ genes. If we are looking for those, you might as well stop now because they donât exist. Genes donât know about social behavior. They donât know about communication. What they know about are assembling tissues and cells to perform specific functions, and when there are mutations or changes in the sequence of those genes that affect function or expression levels, or differences in the copy number of those genes, we see alterations in the assembly of cells and the specific functions that they underlie. So, what do we know from a genetics perspective? Well, there have been three approaches used: (1) linkage studies that look for excess sharing of genomic regions on chromosomes that track with the disease; (2) allelic association studies, where we look for excess sharing of alleles; this is accomplished by studying single nucleotide polymer- phisms (SNPs), in which a single nucleotide is changed, or differences in microsatellite sequences at a single locus; and (3) a defined copy number variation (CNV), where we look for submicroscopic changes (thousands of bases, rather than macroscopically identified millions). CNVs thus are not obvious changes, such as chromosomal rearrange- ments, but submicroscopic changes that alter chromosome structure, which could be either deletions or duplications. Keep in mind that most of the chromosome is not occupied by sequences of bases that encode the transcript that will be translated into protein, but rather encode regions whose functions we really donât understand, but we think may be involved in regulation of gene expression. I want to mention here some of the previous and current caveats to what we know in terms of ASD and genetics. You need to keep these in mind as you read genetic studies to determine the degree to which you can rely on the findings and conclusions. First, there may and are likely ascertainment biases. This means that the subject population that has
26 AUTISM AND THE ENVIRONMENT been studied genetically may not necessarily be broadly representative, or perhaps they are broadly represented, but they represent one small domain of the spectrum. Second, until recently the sample sizes used in studies typically were small and underpowered. Why is this a problem? Well, as my friend, Ted Slotkin, tells me, if you do enough comparisons, you will find something. For genetics, this means that if you try to find an association between many different SNPs and a disorder, eventually you will identify some relationship statisticallyâbut one needs to correct for what we call Type 1 error, that is, false-positive results. There are debates regarding the best ways to correct, and many earlier studies may not have corrected at all, leaving us with nonreplicable findings. Third, the accuracy and completeness of the diagnosis and characterization of the phenotypes are essential to understand who you are studying in terms of a cohort to be used in a genetic study. If this is not done at a high standard, an already heterogeneous disorder like autism becomes even more difficult to study genetically because the study population may be diluted with poorly defined subjects. Fourth, in the past there have been issues with technical quality control; that is, the quality of the assays used to identify SNPs and other changes. This is becoming a nonissue as technology advances. Finally, and perhaps most important, there is for the most part a lack of assigned gene/variant function, in which the polymorphism does something to gene expression or gene function. Keep in mind that this concern comes from a biologist. One may identify a variant associated with autism, but if it is not a coding variation that would clearly change the coding of an amino acid, what does it do? You are stuck with that finding in terms of translating that to a biological substrate for the brain changes that may underlie the disorder. So, what do we know about linkage? Well, the most recent autism genome project consortium identified a modest signal on chromosome 11p, and this is being followed up. In addition, by doing some data filtering, a few other loci seen in previous studies were seen, including regions on 2q and 7q. From previous work, there are in the literature dozens of other reports of linkage, but the bottom line is that with disease heterogeneity, as we have seen in schizophrenia, for example, and other disorders, the linkage signals are generally relatively small and there may be difficulty in replication from study to study. This is telling us something about the disorder, that there is locus heterogeneity. Thus, there are likely to be many different genes or combinations of risk alleles
PROCEEDINGS 27 that may underlie ASD. Regarding CNV, there has been a lot of discussion over the last month about two studies, the AGP (Autism Genome Project) study and a study out of the Cold Spring Harbor group, essentially identifying somewhere between 8 and 10 percent of the individuals in their study having CNVs. The findings are exciting, as CNVs have been implicated in other disorders (e.g., certain cancers), but the findings are not without issues. There is no overlap in terms of chromosomal sites, as far as I could tell, between what was found in the Cold Spring Harbor study and in the AGP study and one does not know the biological significance yet. We are going to talk about that later perhaps, I think, in terms of what CNV might contribute to this disorder. Regarding rare mutations, we know that there are loss of function mutations that have been identified in a single individual that was part of a genetic study, or even in individual families in which there is an autism diagnosis. I have listed some of those genes up there. The reason I list those is because it turns out that a number of those mutations are found in genes that at least biologically have some things in common; they are involved in synapse formation and function. Keep in mind, however, that rare mutations generally do not translate into genetic variations across large segments of the affected population. They are important in trying to understand the genetic contribution to the neurobiological disruptions. Genetic syndromes with co-occurring ASD diagnosis have often been overlooked in the past. My friend Art Beaudet talks about these all the time. Disorders such as Fragile X, Rett, Angelman, and Timothy have a relatively high prevalence of co-occurring autism diagnosis. In addition, there are some common themes in terms of the neurobiological changes known to occur in each of these disorders, related to the changes in neural development. Keep in mind that genetically the causes are quite distinct from each other, but the high co-occurrence suggests that there may be many genetic routes to impact negatively the three core functions used to diagnose ASD. The literature is also replete with reports of association of common risk alleles with ASD; that is, gene variants have been identified from standard association studies that give us some clues regarding the impact of common variants on genetic risk. I have listed some of those on this slide: (1) nonfunctional risk alleles, meaning that there has been a change in the sequence of the gene, but we donât know what that sequence change means. I have listed some of those genes up there. The neurexins,
28 AUTISM AND THE ENVIRONMENT the GABA beta 3 subunits and Gral-2 and some risk alleles that have been identified, but not necessarily replicated in every study; and (2) functional risk alleles, that is, variants that have been identified that either change the function of the gene product or change how much of the gene product is actually produced. The promoter region of the reelin gene is one example, and I have placed a red circle around Met, a finding from my laboratory that I will tell you about in a moment. I have posed some questions related to the influence of genetics on autism expression that might also be retitled âGaps in Knowledgeâ (1) How much of genetic risk is due to direct impact of mutations on brain development? (2) How much of what we are talking about in terms of genetic influence is actually the combination of genetic mutations changing the trajectory or course of neural development in wiring the brain up? (3) How much of the risk is due to direct impact of mutations on peripheral functions, that is, other organ systems that influence brain development? I raise this as a possibility because we know that peripheral organ development and brain development are linked physiologically. (4) How much of the risk is due to genetically established sensitivities to environmental perturbations? We know this exists experimentally, but we really donât understand it in the clinical population. (5) How much of the phenotypic heterogeneity of individuals with ASD are influenced combinatorially through genetic and environmental factors? That can be viewed as my red herring question. So, here is my concept of where we are with understanding autism brain pathophysiology. I am being facetious, but that is a thimble in case you didnât recognize that blurry image. In essence, we know very little about the changes in brain development and brain organization that underlie ASD. That is a real problem in trying to understand the causes. Genes, environment, or both? How can you answer any of the questions I posed without knowing what exactly is disrupted in terms of brain architecture and development? Part of the problem, in my opinion, has been that the geneâenvironment debate has been held in isolated silos, that is, separated disciplines in which there is rare exchange of ideas. The silos, or disciplines, need to interconnect. This harkens back, and we talked about it on the conference call among presenters, to when developmental neurobiologists spent an enormous amount of time trashing each othersâ work because one was either in the ânature schoolâ or the ânurture schoolâ regarding brain development. Of course, that was silly because we know that the brain is built through a genetic blueprint
PROCEEDINGS 29 that takes information from the outside world and utilizes it to direct the developmental course to wire up circuits. This geneâenvironment interaction is one of the unique properties of the brain. So, of course, regarding ASD, it is not genetic versus environmental, irrespective of whether you think there is a principal cause that is genetic or environmental. Because ASDs have at their core disrupted brain development, in terms of etiology, both genetic and environmental influences must play roles because this is in the basis for brain development. So, here is a concept regarding what we might do to address mechanisms: Translational approaches that incorporate multiple technical strategies. There are a number of different strategies in which we are trying to link these domains experimentally. One approach is to focus on neurodevelopmental genes that have been characterized for altering the assembly of circuits that are likely to be disrupted in individuals who develop autism; it does allow investigators to move freely between animal models in which the biological functions of the genes are studied, and going back and working with human geneticists to try to determine whether there are meaningful relationships that would make sense in terms of variations of that gene that might underlie partial risk for ASD. One also can begin from human genetic research data and develop model systems that probe biological functions, trying to make sense in terms of what has been identified as a variant associated with the disorder that carries genetic risk. I would suggest that it doesnât necessarily help to knock out a gene in a mouse if the variant that has been identified in the human genetic research is not a complete null, but rather a variant that alters protein function or levels of gene expression. Genetic knockout studies may generate some very interesting biological findings, but these may not necessarily be relevant to the pathophysi- ology of ASD. Of course, with model systems, such as genetically engineered mice, you can do experiments. You can manipulate the system both genetically and environmentally at different developmental ages. I have diagrammed an example, in which one can expose genetically manipulated mice to different environmental factors that we know change the course of trajectory and development. The impact of exposure may be influenced by genetic variation and you can design experiments to do this in developing model systems. So, I just want to highlight for the last minute or two what we have
30 AUTISM AND THE ENVIRONMENT done in our laboratory. There is only one data slide, and it summarizes work published in 2006 in the Proceedings of the National Academy of Sciences. We took the approach of studying the role of a gene in brain development and then extending these to human genetic studies for several reasons. We were examining the role of a tyrosine receptor, Met, in cerebral cortical development. Met actually has been the focus of thousands of scientific studies because its dysregulation is implicated in certain kinds of cancers. It turns out that this gene is expressed in the brain during development and is important for a number of different processes, including cell migration, development of excitatory and inhibitory neurons, synapse formation, and myelination. We were studying Met in an animal model. The brain architecture changes we found when we manipulated levels of Met expression, together with long-term changes behaviorally, paralleled changes in ASD. We also realized the Met is located under a linkage peak on chromosome 7 in humans, a region implicated multiple times in studies of ASD. The major finding is that we identified an SNP in the 5' region of the gene that controls how much of the gene is expressed. We showed experimentally that it reduced how much of the Met gene is expressed, and we believe the mechanism for this is due to the 5' SNP associated with ASD reducing the ability of two transcription factors to bind to this region of the gene. Transcription factors are proteins that control how much of a gene is turned on in specific locations and at any particular time during development. Thus, the Met variant that is strongly associated with ASD actually had a functional outcome. It changed how much of the gene was actually produced. Met is involved in brain development, but we also thought more broadly about this when we were debating about doing the human genetics studies. Met is also involved in gastrointestinal repair, in immune response regulation, and some other peripheral functions that are consistent with the co-occurring medical issues that are described clinically for individuals with ASD. We spent a lot of time with clinicians to talk to them about whether this made sense because it is not a small number of children who have co-occurring medical conditions. Though still unsettled, it may be a relatively large number. These detailed delineations of the population are telling us about disorder etiology and perhaps even the biology as well.
PROCEEDINGS 31 It turns out that the transcription factor impacted by the ASD- associated variant in Met is SP1, which happens to be a transcription factor whose binding to DNA is disrupted by a number of environmental toxins. So, here one can see that the possibility of combining environmental toxin work with this variant in a humanized mouse model, for example, or introducing the humanized mutations in cells, opens up the possibility of studies that examine combined genetic and environmental influences. So, for one example, we have actually shown that if you expose cells that have either the G or C (ASD-associated) variants of this gene to BaP, which is a common environmental toxin, levels of gene expression are reduced quite dramatically for both the common and ASD-associated variants. Keep in mind that the common SNP (G) results in more than double the amount of gene transcription in the cells than the ASD- associated variant. Iâve added here a hypothetical threshold for when a disorder is expressed. If the toxin reduces levels of Met expression for both the G and C variants, but the C variant starts out lower, the environmental exposure will result in even lower levels of expression that reduce below the threshold. In this example, even with BaP exposure, expression of the gene with the G allele still does not drop below disorder threshold. Thus, BaP does not directly cause the disorder, but has differential effects due to genetic variation. So, what do we need to do in the future? Iâve listed some suggestions here. We need to increase subject ascertainment, character-izing populations in great detail, which will allow geneticists, psychologists, and neuroscientists to stratify groups more accurately to determine if certain phenotypes are associated with specific genetic variants, including SNPs, CNVs, and other genetic changes. Given that we all agree that we need to be very careful about how we phenotype in doing the genetic studies, it simply doesnât make sense to start out with a cohort of 1,000, because by the time you stratify based in different characteristics of ASD, or even life history, the study will be underpowered. Deep sequencing to identify more functional variants will be important to pursue. If we are going to translate the genetics to more than just associations or statistical arguments, we have to translate the findings to biologically relevant changes. Thus, functional characterization of the variants is a very high priority. There needs to be continued wise investments in model systems that
32 AUTISM AND THE ENVIRONMENT will allow us to pursue geneâenvironment influences more rigorously than can be done in human populations. Finally, if we are going to understand functional etiology of ASD, if we are trying to identify the genes that underlie risk, and we are searching for environmental factors that cause changes in brain development, we need to know where these candidate genes are expressed in the developing human brain, and where these environmental factors have their impact. There is a difference between mouse and human brains, and it is essential to keep in mind that one cannot always extrapolate findings between species because of fundamental differences, particularly related to brain areas that simply are not represented in the mouse, but which may be at the heart of ASD. For this type of information, there is an enormous gap in terms of understanding where key genes might be playing a role in neurodevelopment, and how their perturbation may impact the core features of ASD. Thank you. Dr. Spence: Thank you, Dr. Levitt. Now, we will move right on to Dr. Isaac Pessah, who is the director of the Childrenâs Center for Environmental Health and Disease Prevention at the University of CaliforniaâDavis, at the M.I.N.D. Institute. He will give a toxicology talk. HOW MAY ENVIRONMENTAL FACTORS IMPACT POTENTIAL MECHANISMS IN HUMANS?5 Dr. Isaac Pessah Dr. Pessah: Thank you. Pat Levitt really summed up very nicely the complexity of the heritability of autism and autism spectrum disorder. What the last 25, 30 years have taught us is that any chromosome in the genome has multiple linkage sites, whether they are replicable or not, but the fact is that many, many genes may be involved in conferring susceptibility to autism, and to a toxicologist, one would say if there are that many genes involved and more than one gene in any individual that is susceptible to autism, environment must play a factor. So, what do we know about the scope of the problem in terms of how 5 Throughout Dr. Pessahâs presentation, he may refer to slides that can be found online at http://www.iom.edu/?id=42458.
PROCEEDINGS 33 environment may influence outcome? Well, here are some data that are in the broadest sense not necessarily hard-wired data and very plastic in how you interpret them, but of about 4 million births per year in the United States, about 120,000 show major birth defects and these include structural defects, growth retardation, functional deficits. We believe that this underestimates the problem because most neurological and behavioral problems are not diagnosed until early childhood or young adulthood. So, this is essentially an underestimate at birth. At present, the cost and the causes of the majority of the developmental defects are not understood, but it is believed that about 3 percent of all developmental defects may be attributed to exposure to toxic chemicals and 25 percent of all developmental defects may be due to a combination of genetic and environmental factors, where the personâs genome essentially confers increased susceptibility to the environmental hits that occur both during gestation and postnatally. Now, I wish I could tell you that the environmental issues, the chemicals, are simpler to address. In fact, they are much more compli- cated. These are very old data. They were released by a National Toxicology Program Report back in 1992 that was mediated through the National Academies. Essentially, this thorough review of the literature came to some rather startling conclusionsâthat at the time there were greater than 53,000 commercially important chemicals in use, and approximately 80 percent lacked adequate toxicity testing for risk assessment. This is especially true in the vein of neurodevelopmental toxicity testing. You would think that pesticides are more highly regulated and, therefore, we have more information on them and, in effect, we do, but still they concluded that 64 percent lack adequate data for risk assess- ment. Cosmetics are not any better, and food additivesâhere we are talking about intentional food additives that may have unintentional consequences. One example that I like to use is the fact that high-fructose corn sweetener is processed through reagents that are generated by chloralka- lide plants. Chloralkalide plants use a mercury cell process and now NIST (National Institute of Standards and Technology) and some other labs, including ours at Davis, have identified very low levels of mercury within high-fructose corn sweetener, very low levels. Yet, one doesnât know how to use this information for risk assessment because of volume and we actually donât know the form.
34 AUTISM AND THE ENVIRONMENT I have updated information that there is now at the Environmental Protection Agency (EPA) a very active program to try to identify which chemicals we should, in fact, be prioritizing for additional toxicity testing so we can improve our risk assessment. The little pie chart that probably doesnât appear on your monitor essentially shows that there are hundreds, if not thousands, of chemicals in various environmental samples, including foods, water, and so forth, that we really have begun to prioritize in terms of trying to understand risk assessment. What is really needed is information about the additivity of various combinations of these compounds that may cause actually synergism, where each compound doesnât cause an effect, but together they have a much greater effect. Some of these exposures may antagonize each other, and also very important to risk assessment is the relative timing of exposure. So, letâs get back to autism. Well, I would like to propose, and I think others have as well, that autism is really a multisystem disorder. We have focused on the developing nervous system, but because of the number of genes involved and the heritability pattern and the fact that greater than 90 percent are idiopathic, we have no clue how it is caused; we should assume that children who are susceptible to autism may actually be more adversely impacted by environmental exposures than the typical child. What are the possible mechanisms involved? This was a large task for me. I am going to generalize and then give you two examples. One is a simple example that will provide a framework for additional studies and one is more complicated. If we look at this kind of hypothetical curve here, where we are looking at the percentage of kids having adverse effects that we can actually measure as a function of a particular cadre of toxicant exposure, this could be a single chemical or a complex mixture, even highly inbred rats, even cell lines will show you variability about the meaning for the adverse outcome that you are measuring. So, you will have hypersensitive individuals and you will have resistant individuals. Why that is in a highly inbred strain like a B6 mouse, we actually donât know. In autism, one could hypothesize that we could have a shift to the left of the sensitivity curves because autism has genes that may impact susceptibility, either through altered metabolism processing or, in fact, more sensitive target sites. But it is more compli- cated than that because we really believe that there are autisms, not any
PROCEEDINGS 35 one type of autism. This is probably because different populations of kids susceptible to autism have a different pattern of gene expression and different susceptibility markers that contribute to their ultimate susceptibility to autism. So, this becomes a very complicated problem. One way to approach a very complicated gene environment issue is to acknowledge that genetic susceptibility will play off of environmental exposure, and if the timing of those exposures are correct or critically timed, then the prevalence and severity of the developmental disorder will be influenced. Letâs take an example. This is a very simple example. Timothy syndrome is a very rare disorder. It is not a genetic heritable disorder that wipes out hundreds of genes. It is not one that deletes a gene. It is one that actually inserts a missense mutation in the coding region of Cav1.2, changing a G at 406 to an R. The single missense mutation causes the calcium channel for which it codes to inactivate much more slowly than it should. This leads to an abnormal calcium signal. If you look at the kids that suffer from Timothy syndrome, you see that they have a 60 percent autism rate, mental retardation at about 25 percent. I am trying to follow the lines hereâ21 percent rate of seizure disorder. The reason that these kids were identified initially was they have long QT. They have an arrhythmatic heart, which contributes to the clinical presentation and, yet, 80 percent are on spectrum; 60 percent are autistic. This is a very rare disorder, but could it lead us to understand geneâenvironment interaction? Well, it is well known among toxicolo- gists that one of the major targets of Cav1.2 and some of the other L-type calcium channels is mercury, cadmium, and lanthanum. In fact, we use them routinely in the lab to block these channels. So, this provides a kind of homework, a simple model where you can go in, make a mouse, and then test the hypothesis of a surgical strike on a particular gene and how it might influence genetic susceptibility. We need to look at this as a system. We now know that PTEN highly regulates through its effects on PI-3 kinase-dependent phorylation of PKB, the inactivation kinetics of Cav1.2. PTEN has been another susceptibility marker for autism. This presents a framework from which we can learn from rare disorders a particular strategy to use to try to understand a complex disease, a rare mutation to understand a complex disease. Now, letâs get a little more complicated in the sense that there have
36 AUTISM AND THE ENVIRONMENT been several neurobiologists who have proposed that one of the sort of patterns you see in autism that may be more generally applied is an imbalance between excitation and inhibition within the developing nervous system. Here now we have several pieces of evidence. Some are more controversial than others, but, nevertheless, they converge on a common defect, which is a more general defect. It is a functional defect in autism, which is a deficiency in GABA-ergic signaling. GABA is the major inhibitory neurotransmitter in the nervous system, and we now have evidence for methylation problems involving the MeCP2 protein, which is associated with a host of genetic outcomes in terms of transcriptional outcomes, but one of the genes that is impacted is the GABA receptor beta 3 subunit. GABA receptors are also influenced throughâhave been identified to be involved or at least linked toâautism through linkage studies, and they suggest complex epistatic interactions between, letâs say, the GABA receptor alpha 4 and GABA receptor beta 1 genes. Finally, polymor- phisms between GABA receptor alpha 1 and GABA receptor beta 3 have also been suggested, as indicated by Pat Levitt. So, how might this imbalance in GABA-ergic transmission be influenced by pesticides? Well, toxicologists have known for years that one of the major targets of chlorinated hydrocarbons is, in fact, targeting GABA receptors. That is the way in which they work in insects and how they have been proposed to at least have acute toxicological effects in mammals. Then I want to touch on how PCBs might be modifying this. So, several polychlorinated hydrocarbons of historical importance, in the parentheses on this slide are the dates at whichâthe years in which they were banned in the United States Lindane is still used for head lice control and scabies. Heptachlor, chlordane, dieldrin, kepone, and toxaphene have all been discontinued, but, in fact, they are extremely stable structures and there are exposures that still occur. How do they work? They essentially block the pore of GABA receptors. So, they decrease inhibitory neurotransmitters in the central nervous system (CNS). You might say these are of historical importance. Why do we even worry about them? The risk factors that they might contribute are relatively small. Well, it doesnât seem like we have learned about GABA receptors in terms of rationally designed new insecticides. What we find here is that one of the major insecticides currently used in every home is one called fipronil. It is a 4-alkyl-1-phenylpyrazole. In 2000 about 800 tons of it were applied. It goes by several names and we
PROCEEDINGS 37 used to think and we used to teach veterinary students that this is an insect selective. It was birationally developed and, therefore, it affected insect receptors and mammalian receptors; a paper back in 2004 did a comparison between the beta 3 expressed GABA receptor and compared it to the insect receptor and essentially showed no selectivity. This again provides a framework of how a GABA-ergic deficiency in autism may play off of an environmental exposure. Then finally in the broadest sense, we are all exposed to very low levels of persistent organic pollutants. One chemical class is the polychlorinated biphenyls. In a paper that will probably appear next week, there is now evidence that low-level perinatal exposure to these chemicals can, in fact, shift the balance between excitatory and inhibitory currents within the auditory cortex. What I show here is work from Tal Kenet and Mike Merzenich, which shows the normal tonotopic map of a rat at postnatal days 35 through 50, very nice organization and nice gradation from high frequency to low frequency, from blue to red and with a perinatal exposure associated with this imbalance and excitatory over inhibitory current, you see that perinatal exposure to the PCB is called a disruption of the tonotopic map. Now, this is not seen as an overt toxicity in the rat, but certainly if one could imagine this occurs to even a minor extent in children, it would affect language development. So, I want to finish, that a framework for future studies would be to accept the fact that there are several very complicated genetic suscepti- bilities in autism and the number and timing of environmental exposures need to be better defined and need to be relevant to the condition. We also need to pay attention to repair mechanisms that may be impaired, such as the DNA methylation. Here I have given you some examples of how this might work in a real hypothesis-driven research proposal where you might look at competitive and noncompetitive GABA blockers, alterations and self- signaling that may be modified and play off of those mechanisms, such as PCBs, PBDEs, and PCDEs, and then in terms of the framework for future studies, I think we need case-controlled studies so that we can have better comparisons and define subsets, as has already been mentioned. We need to pay particular attention to immunological susceptibilities because the very genes that I mentioned here play a major role in immune regulation, including Cav1.2. We need molecular, cellular, and
38 AUTISM AND THE ENVIRONMENT in vivo models that address mechanistic and behavioral outcomes at low subtoxic exposure levels. I think this is extremely important. We need to better define endpoints and, as will be discussed later in the meeting, nutritional-based models are also extremely important. So, I thank you. Dr. Spence: Thank you, Dr. Pessah. Does anybody have any questions related to this talk? Dr. Insel. Dr. Insel: Just as a question, in much of the work that Pat Levitt described, the genetic discoveries are going much more toward the sort of discovery-driven approach without necessarily having a hypothesis about a specific candidate. Are we able to do that in the realm of toxicology? Is there some kind of non-hypothesis-driven exploratory approach that we can take? Dr. Pessah: I think there are several individuals, some I think will speak later, that one could use simple cell culture models to identify changes in signaling pathways. These are relatively rapid throughput and discovery sort of oriented type approaches where one could, in fact, use individual compounds that are thought to be a potential problem and complex mixtures and identify how specific signaling pathways are altered or some morphological changes are impacted. But that I think is really how do you relate that back to humans, and autism is really the million-dollarâit is a very tough thing to do. That is why trying to understand some of the susceptibility genes and seeing if, in fact, we already know that they are targets or the pathways are targets may prove to be a bit quicker to try and identify risk factors and interventions. Dr. Spence: Other questions? Dr. Schwartz: This is David Schwartz. I was wondering if I could respond to Dr. Inselâs question. I would like Larry Needham to potentially add some information here. There are panels of toxins that one could look at in the serum, blood, hair, and other specimens and apply that in a very broad way to population-based studies. It just hasnât been done. The concept of approaching environmental etiology in an agnostic way, looking at as many toxins and exposures as possible, makes enormous sense, especially in a disease like this. Larry is in charge of the unit in the CDC that has developed over 150 assays to look at chemicals and toxins in various human specimens.
PROCEEDINGS 39 Dr. Needham: One thing we have talked aboutâwhat we have done and I will talk more about it tomorrowâis our division, and Eric Sampson is the director of our division and Henry Falk is the directorâ what we have done with the NHANES [National Health and Nutrition Examination Survey] survey is look now at about 200-and-some-odd chemicals, but we looked at the general population and, of course, we have talked about folks being on selected populations and doing educated studies and looking for increased amounts or decreased amounts and increases and decreases over time of concentrations of various chemicals. But that is where we need your help in selecting these chemicals for autism. Dr. Insel: But we are talking about hundreds, not millions. Dr. Needham: And also should we be focusing on those chemicals that are only changing in terms of concentration in the environment. Dr. Schwartz: Dr. Isaac Pessah started off by saying there are roughly 49,000 compounds and 80 percent of them still need to be investigated. We have no wayâand the way that we do in genomics now where we have a comprehensive approach, there is nothing comparable hereâbut we are in the process of developing that. That is the thrust of the environmental biology program. Think about where genetics was 10 years ago in terms of looking at hundreds, not tens of thousands. Dr. Needham: We need clues like on structural activity relationships and so forth if we can get some clues there. Dr. Spence: Should we move on and move this discussion to the end of the session? The last speaker for this session is Dr. Martha Herbert, who is assistant professor of neurology at Harvard Medical School. She is going to give us a talk about the concept of biomarkers. DEFINING AUTISM: BIOMARKERS AND OTHER RESEARCH TOOLS6 Dr. Martha Herbert Dr. Herbert: I am happy to be here and I want to reiterate that our instructions have been to give a broad and general overview of what we 6 Throughout Dr. Herbertâs presentation, she may refer to slides that can be found online at http://www.iom.edu/?id=42459.
40 AUTISM AND THE ENVIRONMENT know and what we need to know. I want to comment that the issue of biomarkers is very pertinent for research and it is also very pertinent to those of us in clinical practice. Why biomarkers in autism? Right now we have no biomarkers for diagnosis. Biomarkers would help in identifying pathophysiological mechanisms. Different biomarkers in different subgroups would be very helpful in multiple regardsâidentification of treatable features and prediction and tracking of treatment response. Overall, we need to focus on biology and pathophysiology, and there are multiple levels of the biological hierarchy at which we need to measure, and we need to make a more concerted effort to coordinate measures across these levels of the biological hierarchy. I would like to propose from the point of view of learning more about autism pathophysiology, more than the thimble that Pat Levitt showed in his slide show, that we consider taking a middle-out approach. Bottom up could be genes up and top down could be behavior down. We have been using in autism research a geneâbrain behavior model and I think that what we are talking about here is, and everyone has been saying more than that, it is genetically influenced, not just genetically determined and it is a whole organism. It is a systems model where pathogenesis is now geneâenvironment and epigenetics, and the biology needs to be broken out in details of molecular and cellular mechanisms, tissue and metabolism, altered connectivity and processing to get to the phenotype. It is the mechanisms that yield the phenotype. They may be caused or triggered by the pathogenesis, but it is the mechanisms that lead to what we call autism. This involves again breaking out the biology in more detail and it is also at the level of pathophysiological mechanisms that will identify biomarkers and also that will identify biomedical treatment targets. We have been talking about involvement of more than the brain in autism and I think it is interesting to go back to the very first paper on autism by Leo Kanner. Kanner commented on somatic symptoms in almost all of his cases, and I have highlighted, in red, eating problems, tonsils, diarrhea and fever, more tonsils and adenoids, frequent vomiting, tonsils, tube feeding, avitaminosis, malnutrition, vomiting, feeding, bronchitis, colds, streptococcus, infection, impetigo. One child who didnât have any infections, which would be a question of overactive immune system, frequent hospitalizations because of feeding, colds and otitis media, hormonal problems. These were
PROCEEDINGS 41 commented on. They did not fit into a model of autism as a behavioral disorder, but it is interesting to note that they were there. They were not life-threatening problems for the most part. It raises the question of what our thresholds are in a complex disorder for taking symptoms into account in the model. So, I think that what we need is a move again on emphasizing pathophysiology in my area of brain imaging research. We have been looking at this from a cognitive neuroscience point of view, looking at behavior as it is modulated by regional and neurosystems alterations. I think what is happening now is a greater interest in the tissue changes of the brain and looking at the brain based on the physical properties, the receptors, the growth factors, and so forth that may be targeted by a gene and particularly by environmental factors. I think we need to tackle the intersection between pathophysiology and cognitive neuroscience and I think for this we need a programmatic brainâbody biomarker linkage. I went and did some countings of biomarker-pertinent published articles. I went back to the literature and looked at all of the articles that measured biomarkers, starting in the Journal of Autism and Developmental Disorders and every paper in that journal and in its predecessor, Journal of Autism and Childhood Schizophrenia. In 36 years, there were 78 articles that had to do with biomarkers. You can see in the color coding of the Excel chart that most of them only measured one biomarker and only measured it once in a small cohort without relationship to other kinds of finger typing, whether it be biological or biochemical. The serotonin is the only one, which was measured multiple times. I had the hypothesis that as genetics became more organized over the years, there would be fewer biomarker studies, and the graph on the bottom left shows the blue line is a decrease in the proportion of biomarker articles as a percentage of total articles that year, a modest increase in genetics, although it should be pointed out that people donât publish their genetic studies in this journal. A preliminary count of all the articles in PubMed that use autism or autistic in the title yielded 400-some-odd articles out of about 7,000 that talked about biomarkers, with not a lot of repetition. Now, one of the things that is distinctive about our current period is that we are entering into an epoch where it is possible to measure things in new ways. It is possible to measure large numbers of analytes in small quantities of samples and it is also possible due to informatics advances to link
42 AUTISM AND THE ENVIRONMENT measurements across the levels of biological hierarchy. This may provide fresh and unique opportunities to get a grip on what is going on in autism. Biomarker challenges in autismâcritical issues are involved in terms of recognizing what is going on with the phenotypes so that we understand and measure things that are appropriate to the challenge. Many autistic childrenânot all, but manyâ have striking variations in their severities, striking good hair days, bad hair days and I will get back to that in a minute. There are also chronic features, oxidative stress, inflammation, metabolic perturbations. These are ongoing problems. They raise the question of what environmental factors do on an ongoing basis and not just to perturb development. How do they affect neuronal functioning on an ongoing basis? Treatment responsivenessâI will get back to that in a minute. We are seeing some stable improvement following treatment, multisystem involvement. It has been mentioned and, again, whether this impacts the brain primarily parallel or downstream. The heterogeneity is enormous and that has been mentioned, autisms, leaving the question open of where are the commonalities and final common pathways and how can that question influence our research agenda. Finally, some of the chronic pathophysiological features, such as the inflammation and the oxidative stress, appear nonspecific so that insofar as these are potentially treatment responsive, it is important to remember that what may be treatable may not be specific to autism. A particular thing that is important to remember is to not characterize autism as a static encephalopathy. There is a paper in press in pediatrics by Andy Zimmerman and others reporting transient marked improvement with fever, children who will start making eye contact and talking during the course of a fever and then it goes away when the fever resolves or somewhat afterward. Some children will have spikes in function sometimes under conditions of stress or emotional stimulation or they will say something quite articulate when they are normally not verbal, demonstrating a neurological capability of performing at that level, but which for some reason is suppressed, which raises the question of whether treatment is removing inhibition or giving skills or both. Transient improvement on antibiotics has been reported and I will get back to that. Improvement on allergy medications, variability in function related to food, allergen and toxic exposures and also treatment responsiveness, including published reports of loss of diagnosis with recovery documentation studies in process. Overall, this raises the
PROCEEDINGS 43 question that this is not just a disorder of neurological development. It is also a disorder of ongoing neuromodulator impact on brain function. This is a slide from Sidney Finegold at UCLA (University of Californiaâ Los Angeles) and the veteranâs hospital in Los Angeles. He is one of the authors of the study that demonstrated that oral vancomycin could transiently improve symptoms in autism. This was a follow-up study showing nine variants of clostridial bacteria found only in autistic subjects and three variants of clostridia found only in controls. These abnormal variants of bacteria can deplete vital nutrients, alter metabolism of xenobiotics, and affect immune function. All of this can cause or worsen metabolic stress. This suggests that in order to characterize subgroups and treat these children, we need to go beyond human metabolites and human genome to look at an extended metabolome. This is a delectable slide of a child standing in his own diarrhea and I donât think the day of autism is complete without looking at this, but the point here is that if you send a stool sample from this child to a clinical laboratory, it would probably come back negative. The measures that allow these bacteria to be identified are done in research labs, and they are not available to help practicing physicians. This is something that needs to change, and this is not just a research question. There are two reasons why measurements need to be coordinated across levels. It is not just that there is a great deal of variability among results of genetic studies. There is variability in behavior and now that we know that the genetics are not the only thing, we need to confront the variability at multiple levels. We donât know where the commonality is, where it fans in and what is stable across different people with autism. Is it connectivity? Is it more at the tissue metabolism level? Do all children have inflammation in their brains or only some? We donât know this. We need to get systematic about looking for this across all levels. Also, toxins, infection/immune, genes, and other things function clinically in a vicious cycle. Genetic susceptibility sets up vulnerability to toxins, which impairs immune systems, sets up infection, which alters gene expression and it becomes a self-amplifying vicious-circle feedback loop. So, conclusions. First, I would propose that metabolism needs to be a core focus in autism. We know that environmental factors perturb metabolism, even at low levels of exposure. We know that some of the same mechanisms and pathways get hit in the metabolic disorders as in
44 AUTISM AND THE ENVIRONMENT inborn errors of metabolism. But the spectrum and intensity of effect differs. We need to learn about multisystem and multilevel impacts. Metabolism is a target for biomedical treatments and also metabolic changes are a final common pathway on which can converge multiple different genetic mutations. What we need to do is to study how environmental perturbation of metabolism, which is not a disease category that we are taught about in medical training at this time, has different patterns and thresholds than inborn errors of metabolism and we need to develop practice parameters around this. Also what is needed is to develop infrastructural support of the study of metabolism. It is daunting because of the state sensitivity and the sensitivity to handling of sampling. I think we need consensus meetings to identify measures that are less sensitive to these problems. We should consider âomicsâ and other profiles, develop standard operating procedures, and in particular have a special focus on environmentally responsive metabolism. As these questions get clarified, we should develop a repository for metabolic samples of many kinds as determined by consensus with multicenter participation and encourage, strongly encourage, participation with contributions from research projects with well- phenotyped subjects. With regard to brain and metabolism, we know that brain and metabolism are both abnormal in autism and we also know that this is not consistent in its details between subjects. What we need is to learn how metabolism modulates brain and vice-versa. This requires integration and integration requires infrastructure. I would particularly suggest that in our studies of the brain, we have a much more concerted focus on characterization of brain as a physical organ, characterization of brain tissue. I also propose that we use more high-temporal-resolution brain function measures since the abnormalities in temporal measures, EEG, MEG, at the millisecond level are closer to being indicators of synaptic dysfunction and particularly that since synaptic dysfunction can be metabolically modulated, this is important. We need, as I have said, systematic metabolic characterization and we need to have an extended metabolome and also extended genome looking at gut microecology and its disruptions. We also need to have better characterization of change and treatment, which we know are possible. We need better tools to track treatment and change biologically. We should be studying nâs of 1,
PROCEEDINGS 45 repeated measures to see in the same individuals see what can change. We do not have good measures of change. We can study individuals who are diagnosed over time. We can study children at risk for autism over time, children undergoing treatment over time, and the marked good hair, bad hair days, for example, a child with a fever who improves or other phenomena like that, children who are off fluids and function better. We need subgrouping to identify mechanisms and to predict treatment response. We do not have published studies showing the separation of groups, such as the illustration that I have shown, which is a separation of Lou Gehrigâs disease, Alzheimerâs, and Parkinsonâs, but this kind of work needs to be done in autism. There is no reason to think that there will be one biomarker for autism. What we need is profiles of vulnerability and treatability. Environmental perturbation of metabolism is widespread in the organism, but its thresholds are different and the reference ranges we use will not pick it up. Autismâs sensitive physiology also may mean trouble for the individual, even when labs are within the population normal ranges. So, clinical reference ranges need to be rethought for this kind of complex disorder as part of our process. Finally, it has been pointed out that the environmental influences on autism suggest treatability and prevention and even though the focus of todayâs discussion is not on treatment, I think it is very important to understand that what we are talking about is also very much about identification of treatment targets and of treating them. Thank you. Dr. Spence: Are there questions for Dr. Herbert? Dr. Levitt: I just wanted to comment that there are many patients with clear-cut primary genetic disorders like the Fragile X syndrome and Rett syndrome and many, many others, who have good hair days and bad hair days and you have a lot of this kind of variation, even though they have a primary underlying genetic condition. Dr. Herbert: I would also like to point out that many or most people with some of those primary conditions donât have autism and people with genetic conditions have high vulnerability to environmental perturbation. So, it doesnât exclude an environmental role, even when there is a known genetic factor. Dr. Levitt: I am not arguing there is no environmental role. I think there is an environmental role and that these people are suspect and subject to these kind of perturbations, but they are particularlyâthey are
46 AUTISM AND THE ENVIRONMENT profoundly susceptible to these. Dr. Herbert: That is very telling. That is very interesting. So, this is a question of using controlled clinical settings to study change using systematic systems, biology, and biomarkers. Dr. Insel: Martha, I really liked your presentation a lot, but could I get you to just expand a little bit for us so that we understand what you mean about some of the studies you would like to see done? For instance, in metabolomics, what would be the tissue and for the microecology, microbiomics, it sounds like you would focus on gut, but can you giveâ where would you go? What would be the targets for some of these things? Dr. Herbert: At the moment, I think we havenât explored even blood and urine samples. I am mentioning gut because I wanted to show that this is more than a human metabolism problem. It is also our pathogens or our commensals, but I think we should haveâI do not pretend to be the one who can unilaterally dictate what an appropriate profile should be, but I think there is a lot that could be done with blood and urine samples and with a protocol for suitable spinal fluid, even when you donât have a standard study for children who may for one reason or another get a lumbar puncture, they should have available to them a standard operating procedure that can be used in a clinical setting to send samples to a repository. I really propose consensus meetings to make that decision. I think this is a very complicated area. I have been asked singlehandedly to offer certain organizations the answer to this and I just donât think it is appropriate. I think planning procedures are what we really need to have happen here. Dr. Spence: Martha, I actually had a question. As a clinician we do get abnormal labs sometimes and I know in metabolism that the state of the child at the time that the lab is drawn is very important. So, do you want to just speak to the challenges of kind of the reliability of some of these biomarkers and standardization? Dr. Herbert: I think there are some measures that are more state dependent than others. If we are going to organize a repository, we need to get people whose day job and 24/7 specialty is to handle these things, to identify a set of measures, which are most stable. It may be a limited set of measures. There are some measures where if you donât freeze it within 15 minutes and so forth and if you have it at a different time of day, it is very different. But by no means all of them. So, I think the first
PROCEEDINGS 47 step would be to get specialists in multiple disciplines to identify measures, which are more and which are less state sensitive, and I think we owe it to the children to really put the effort into doing that. Dr. Schwartz: You talked about biomarkers as a way of linking pathogenesis to phenotype in terms of mechanisms, but biomarkers could also be used as a way of biologically phenotyping the disease, and in particular this disease strikes me as a disease that is made up of several subtypes. I just wanted your comments and thoughts in terms of using biomarkers as a way of biologically phenotyping autism. Dr. Herbert: Absolutely. That was on my first slide and I further would expand what I am sayingâwhat I said about that, which is that treatment response measured with biomarkers is an even further way of subtyping. Some of the nutritional treatments that are used in autism in certain settings are relatively low risk and a difference in response to those could be related to a genetically modulated environmentally sensitive set of differences in pathophysiology. That is a great opportunity to learn more about disease mechanisms as a research probe. Ms. Bono: I just wanted to make a comment that in the blue folder in front of everyone, I made copies of an algorithm that some of the DAN practitionersâthat is Defeat Autism Nowâuse when they get children into their offices and they start subtyping them by blood and urine markers and how they would treat a child based on gut, based on immune problems. It is in the packet. So, if anyone wants to look at that, it is a good start. It is based on about 4,000 or 5,000 children. Dr. Spence: Great. Thank you. DISCUSSION Dr. Spence: I think we should probably move on to the discussion section. I am going to take one minute and try to sum this up. I think Dr. Herbert did a beautiful job of talking about the complexity in the search for biomarkers, even to the point of extending the metabolome beyond humans. So, how complex does that get? Add to that, Dr. Swedoâs description of the heterogeneity of the clinical picture, where there are tens of autisms, maybe twenties, maybe hundreds. Dr. Levittâs description of the multiple ways in which our 30,000 genes can contribute to this disorder and then Dr. Pessahâs description about toxicology and 53,000 known toxicants. So, if you do the math, I think this is a real challenge for us, but I think that we have
48 AUTISM AND THE ENVIRONMENT laid out some of the issues in framing this discussion as to what we actually need to go forward. The complexity is daunting, but I think we need to talk about how to overcome that. One thing that I would start with is by posing a question both to our session speakers and also to the table: Do we really need to understand that middle approach, that pathophysiology? Do we need to get past Dr. Levittâs thimble in order to get going on this or can it be attacked in multiple levels? Can the geneticist be working on genetics and the metabolomics people working on metabolome? So, that is where I am going to start. Dr. Akil: I feel like we cannot have parallel play because of the very nature of the beast. We cannot have genetics working over here and toxicologists and immunologists and metabolomics. The main reason I see that is the heterogeneity that everybody talks about. What I am not getting out of the discussion is, what are the variables along which you would separate or segregate? Do you wait and let things be self- segregating? For example, do you do huge numbers of subjects with a huge number of toxicology screens and then let that self-aggregate and say in those tens of thousands of autistic kids, there are five patterns and then plus a whole bunchâwhere we canât figure out anything. Do you do the same in genetics or do you intersect or do you do anything a priori behaviorally, like so when you come and say there is this gene or that gene, this calcium channel or this tyrosine kinase receptor, is there anything we can hang our hat on from a clinical point of view? So, I think while it is good to talk about all these approaches, the question is exactly how do you intelligently bring them together with statistical analysis and multilevel analysis and so on. I mean, I think that is what we need to grapple with. Participant: We are portraying this as an intractable problem and I think there is some preliminary parsing that many people have already done. There is a difference between boys and girls. There are kids who have GI (gastrointestinal) disturbances and kids who donât. There are kids who look abnormal from the get-go and there are other children who clearly are very different, later in childhood. So, I think there are already some areas where you can begin to break this down to some extent. Dr. Swedo: I would just echo that and say that clinically I think one of the things that we have spent a lot of time on was coming to diagnostic agreement on the behavioral characteristics that make up autism. I mean, if you look at the research papers, it has been largely aimed at better and better and earlier and earlier diagnosis of the social
PROCEEDINGS 49 and communications deficits. There has been almost no further look at the mind or the brain within the body. One of the things that is already underway is major phenotyping efforts to begin to do what you have suggested. Can we find those few kids as well as going after the 10,000 at the same time? Dr. Akil: Does it map the phenotyping map onto anything else that you can identify? Dr. Swedo: Exactly, but I think the way we set it up so far is that we were looking just at the two characteristics that would put them into that diagnostic group and couldnât begin to stratify the way Pat was suggesting we would need to for better genetic power. Dr. Leshner: Can I ask a similar question? I had actually the same question that Huda Akil has just asked. That is, is there a group with some common credibility or standing, who, in fact, could do the level of detailed phenotyping that you would need toâor at least to be able to parse the symptomology into agreeable form so that you could look at individual clusters of symptoms or look at subtypes or whatever in a way that might get us off the dime a little bit more easily? But who does that? Dr. Swedo: I will speak for my colleagues at the NIH and the CDC, but they might want to chime in as well. The Autism Phenome Project, that phenotyping effort, was a short-term goal of the IACC (Interagency Autism Coordinating Committee) research matrix and efforts are already underway. Our intramural research group is collaborating with the M.I.N.D. Institute on the pilot project for in-depth phenotyping, biolog- ical metabolomics, neuroimaging, all of the variables that we canâthe purpose of the first hundred children to be evaluated is to determine what is feasible, how reliable are things among different sites. The autism centers of excellence RFA was actually written for an impressive number of common measures that will include medical history, environmental exposures, and other things. Every child evaluated within one of the new autism centers of excellence will have those common measures done and those data will be entered into the National Database for Autism Research, NDAR, in an effort to very rapidly get large enough populations to start doing the phenotyping. Dr. Levitt: I just want to point out one other thing. The domains that we utilize to both diagnoseâthe functional domains that we use to diagnose and then to phenotypically subcategorizeâare those neurobiological domains by definition that are so heterogeneous within the typical population. We are not talking about measuring grip strength
50 AUTISM AND THE ENVIRONMENT here. Social behavior is by definition broadly heterogeneous within that normal distribution, and communication and language development from model systems and birds, all the way up through humans depends upon social behavior, social communication, and social interaction. So, it is a problem in terms of division and definition because of the very components of brain function that we are focusing on, and you see these domains disrupted in the broad spectrum of psychiatric disorders and it is equally difficult to phenotype in those as it is in autism spectrum disorders. Dr. Leshner: Not to perseverate too much on this, but I am one of the people who believe that some progress has been delayed in other brain and mental disorder conditions by overlumping and an overfocus on diagnostic categories and, therefore, focusing on symptom clusters or particularly disabilities, however you want to cast it, may be another route. Dr. Levitt: I agree with you completely. So, the characterization and the divisions and the stratifications are extremely important. I am just saying that the challenge is because of what we are given, the domains that are most difficult in terms of human behavior and function to characterize preciselyâand from a practical perspective, if you go around centers around the country, if you want to have an integrated approach, whether you are going to do a metabolomic approach or to integrate that, you need people who are really good, as good and as precise as the quality control we now insist upon in gene sequencing to characterize the populations and it isâI donât know. Maybe I amâbut it is extremely difficult to find those individuals; you need clinical psychologists and others who can do this well or who want to do it because it is an enormous task. If you go around the country and say have you been able to hire a clinical psychologist to work with your geneticistâbecause I can tell you that the genetics studies from my perspective, the non-syndromic genetic studies, are tainted simply because the populations that have been included are not well characterized. So, this is a real gap. This is a people power gap, just like nurses are a people power gap in hospitals in terms of medical care. Mr. Blaxill: Alan used the word âstandingâ and I think one of the things that is interesting about the problem clinically is that there are a lot of clinicians out there working the problem on a different model, on an environmentally based model, on a gut model, on an inflammation
PROCEEDINGS 51 model and oxidative stress model, which are sort of the things that are beginning to come into the discussion more frequently. There are a lot of people out there clinically who have been working the problem, but they tend not to have standing because they have been outside the mainstream of where the science has been. So, I think one of the things we need is to break down some of the cultural barriers to, oh, gosh, those people are kookie. They are doing crazy stuff. Some of them are, but a lot of them are actually conscientious, good clinicians, trying to treat sick kids and we ought to take that community more seriously than we do. They have a lot of data and they donât have any resources and if we spent some time, you know, learning from recovery, learning from treatment response, working with some hypotheses about inflammation and toxicology. There is a lot that could accelerate in terms of getting data faster, testing hypotheses faster and would serve to break down a lot of the divisions that have emerged emotionally about the whole issue. So, I think there is a big group out there that doesnât have standing and we ought to reach out to them maybe a bit more than we do. Dr. Leshner: I would like to make one comment and then I will shut up, I promise. I think that the essence of real translational research, though, is to, in fact, be listening to the clinical experience, have data inform the scientific agenda and so I think you are right. The mechanisms for doing that, our institute directors have to figure that out, not I, are really complicated and difficult because sorting through the inappropriate stuff to get to the appropriate stuff could be very difficult. But your point is very well made. Ms. Redwood: I would like to follow up on what Mr. Blaxill said from an advocacy perspective and what Laura presented as well. There is a sense of urgency here. I am concerned or my belief is that we donât really have to understand mechanisms to be able to intervene in a meaningful way with these children. They are very sick. I think the slide that Martha Herbert put up with regard to their gut disorders, there are things that we can do now if we focus on that n of 1 that Martha mentioned and work on trying to help these children medically. They have several medical problems. I know my son, for example, had very low cholesterol levels. They were in the nineties. He wasnât digesting his food. We treated him with Creon, which is a prescription digestive enzyme. He gained 14 pounds in
52 AUTISM AND THE ENVIRONMENT one year and because his brain was able to finally get nutrition, he improved functionally. He had very elevated levels of serum B12, but he also had methamolonic acid in his urine. He had a functional B12 deficiency. Treating him with B12 resulted in marked improvement. So, there are things we can do now. I would sort of argue with Martha not to focus on the brain, but to actually focus on the child with an n of 1 and document these medical abnormalities and treat them. We can do that now without necessarily having to know all the mechanisms involved. Dr. Newschaffer: I just wanted to add something. I think that the emphasis on etiologic heterogeneity and subtyping autism cases is critical and a number of folks have touched on that. What I havenât touched on this morning and it is a little bit out of my wheelhouseâ maybe Dr. Levitt can comment on thisâis the idea that we might also want to look a little bit at continuous outcomes related to the autism spectrum, the idea of endophenotypes or looking at continuous measures of social cognition in populations that perhaps include individuals affected with autism spectrum disorders, but also broader samples. Continuous endpoints are favorable to study in a number of different study designs, looking atâthat even include examinations of geneâ environmental interaction. So, while the etiologic heterogeneity and subtyping I think is paramount and that became clear, too, we also might want to think a little bit about continuous endpoints broadly distributed in the population and how that can inform what we know about geneâenvironment interactions on related behaviors. Dr. Herbert: A couple of things. I didnât mean to say we have to start with the brain. If I said that, it was a complete misunderstanding. I mean, obviously, I work in brain, but I think this is a whole-body condition. We are trained as scientists to really like precision and definition and careful definition. We are dealing with a situationâthe figure I have heard of the number of chemicals that we donât know very much about is much higher than 53,000 now. That was a 1992 number. These come in different combinations at different times and they pass through a certain number of final common pathways in our bodies, which our physiological capability is to handle them. But beyond a certain point, it is not going to be that precise. There is going to be continuous distribution not only because things are normally
PROCEEDINGS 53 continuously distributed in the population, but because the injuries are continuously distributed and they are not necessarily going to parse out quite uniquely. So, I think that we are going to have to do some very existential reflection on what it means to design a study when it is never going to come out neat and that we need to proceed anyway with taking care of people with the parts that we do know how to handle even while we donât have a grip on many of the other parts. That is just the nature of the beast. That is the nature of geneâ environment interaction in a situation where we did not think through ahead of time the impact of the individual or combined exposures that we are dealing with. Dr. James: I think there may be a unique opportunity in treatment trials. So far the placebo-controlled, double-blind studies have been disappointing. We basically donât see differences between the control and the autistic group, but I think it is important to look beyond the mean. I think within the mean there are responders and they are not characterized. If we could look at those that do respond and characterize them, this would be a very productive way to look at subpopulations and be able to get perhaps to more individualized approaches to treatment because I do believe that within the mean, there are responders and we should not neglect them. I think there is a huge opportunity to look at subtypes and individualize approaches to their treatment. Dr. Swedo: That could be actually very useful in other studies where they have done that. For example, there is a classic study in obsessive- compulsive disorder where responders to behavioral therapy and responders to medicine have the same types of changes on their PET scans. So, I think the response is key. One of the questions, though, is how do you look at the responders to placebo. I mean, that is what I struggle with in our placebo-controlled trials. I agree with you completely that you want to break it down to responders and nonresponders and see what those differences are. But if all they have received is a sugar pill and they respond, how would you go after them? Have to look at those two subgroups very carefully and look at what is different about the autistic children who did respond and hopefully might find something there. Equally important, I think, are the negative responders because within the spectrum, you will have children who absolutely have clear responses and then others who go the other way, who regress. I think they are also equally important to characterize individually, again,
54 AUTISM AND THE ENVIRONMENT working toward more individualized approaches to their treatment. Dr. Insel: This is very helpful. I am kind of listening to this discussion to get ideas about where the next generation of studies might be and what I am hearing is that part of what makes it such a difficult problem is it is an equation with two variables and we donât really have a good handle on either one. One option people are saying is that you could do within subject designs so that you donât have to worry about genetic variation at all and just look at how changing environmental factors alters outcomes. There is the real attraction in doing that in kids who may have responded. Would it also be useful if we had a repository of twins, where you could look at discordant twins or, again, you would take genes out of the equation or at least genetic sequence out of the equation and then be able to ask why does one child get the disease or why does one kid become more severely affected and another one not? Has that been done? Is that an option? Is that something we should chase? Dr. Lipkin: Tom, when we started putting together the AGRE (Autism Genetic Resource Exchange) database, which goes back to the mid-nineties, mid- to late nineties and I was the chair of the first Scientific Advisory Board, I argued for collection of 14 monozygotic (MZ) twins discordant for disease. At that point it was a small fraction of what we found, but there was no interest in proceeding with that. I would imagine we can still find these kids and there must be many more at this point who have been identified. It would be a useful thing to try to do. Dr. Insel: I think Dr. Hu has actually done some of this, using transcriptomics toâdo you want to say something about that? Participant: My name is Valerie Hu and last year we published a small study on monozygotic twins. That is identical twins that were discordant for autism and we found differential expression. That means turning on or off of a number of genes, the majority of which or at least half of which had no neurological functions in lymphoblastoid cell lines from the AGRE repository. We have since continued the study with case controls, sib pairs, and we found additional genes that play a role. They are pointing toward cholesterol metabolism and androgen biosynthesis. This is all new. We have also just started a pilot study looking at the epigenetic effects in the same monozygotic twin pairs who we studied before and we are getting confirmatory results supporting the same genes, some of
PROCEEDINGS 55 the same genes that we have identified by expression analysis and others. If you do a pathway analysis and look at some of the canonical pathways that are established biochemical and signaling pathways that are implicated by both the genes that we have identified by gene expression analysis as well as by the preliminary methylation analysis, they really converge and they have pointed to some very interesting pathways, some of which are really surprising where they confirmed the involvement of the steroid receptor activity signaling, as well as what surprised me, Type 2 diabetes signaling and insulin signaling. So, that might be a tie-in to some of the additional systemic problems. Dr. Lipkin: This was done with cell lines rather than primary tissue? Participant: Lymphoblastoid cell lines. So, you have to consider the caveat in using those. Dr. Lipkin: The notion at the time that we started when we proposed this was to collect cells, so that the RNA would be there available in PAX gene or Tempest Tube or so forth. That would beâit would perhaps even give you cleaner data. But I am very excited about what you have just described. Dr. Insel: What I am trying to hear from this group is where are the opportunities like this where you could control one of the two variables, either hold onto the genetics and take that out of the equation or hold onto the environment and take that out of the equation. I think we need some sense of what that range of possibilities would be because you are not going to solve this equation with those variables flipping around on you. Participant: What we are trying to do is to take a systems biology approach and Dr. Herbert referred to this earlier in terms of the approach that she would recommend. We are trying to pull in not just the genomics, but also the metabolomics plus the epigenetics and we are trying to construct a neuronal cell model so that we can in vitroâwell, one of the processes that seems to be constantly implicated not just by my studies, but many other peopleâs studies, from the M.I.N.D. Institute by Daniel Geschwindâs group, many others, is that of axon guidance or neuroextension, neuronal cells going out and finding their targets. So, we want to establish a neuronal in vitro cell model so that we can test the impact of various stressors that are both environmental as well as biologic stressors, such as elevated testosterone, for example, which is implicated in our study. I think an integrated approach is really necessary.
56 AUTISM AND THE ENVIRONMENT Dr. Spence: Thank you, Dr. Hu. Dr. Susser. Dr. Susser: I wanted to agree but also underscore Tom Inselâs point. I think that when we talk about complexity, what we are really trying to do is talk about it so that we could simplify things and, you know, one approach is MZ twins, so it simplifies one part of it. The other approach is to look for situations where people have common environmental exposure, which you implied in your comment, too. There are many such situations that we can identify and we can exploit. My view on this is that it becomes infinitely complex if we say that we have to know everything about everything and the way that we will get to identify causes, at least, is by thinking about designs, exemplified by the MZ twin design, but that would just be the controlled in genetic practice. Think about also where you control environmental factors and look for genetic effects. There are many such and I will just throw a couple out so you know there are possibilities. You can look at groups that have been exposed to congenital rubella, to rubella in utero. You can look at groups that have been exposed to different kinds of toxic poisoning. You could look at groups that have very old fathers, for example. There are many possibilities in this range of looking for designs that simplify the problem. Dr. Akil: I think it was Susan earlier who said comparisons across different cultures, right? The Institute of Medicine has sort of a big interest in global outreach. I sort of donât want this idea to go away. That is going in like the very other, very big direction instead of controlling everything within like small number of pairs of identical twins, asking the very broad question, which I think is again a very different approach. Two other little points. One is that in these twin studies, it would still be wonderful if we had as much genetic information as possible, as well as gene expression profiling. If we could afford to sequence it, I would sequence it, but short of that, it is not that expensive to get, because different pairs of twins may turn out to be different and having that information would be very helpful. Finally, I hope we keep in mind protective as well as vulnerability factors in all of our thinking. Dr. Spence: That is an excellent point. I think we are going to finish up and we have one more question from Dr. Choi. Dr. Choi: These equation-solving efforts seem to me to be very important, but their ultimate impact to my way of thinking likely does
PROCEEDINGS 57 rest on the platform of understanding the phenotype. I mean, essentially there is potentially a whole series of equations out there. So, imagine a worst-case setting where you have 20 monozygotic twins and each one has a different autism. So that it is very difficult to pull the signal out of noise there. So, I go back to the need to really get at phenotyping and I am sort of struck as an outsider to this field by what sounds to me like a relative need for a bit of catch-up ball here in getting at the phenotype; understanding that clinical phenotype when you are dealing with cognitive and behavioral disorders is very challenging. It does seem like the biological phenotyping effort is really lagging behind that effort in several other fields. I am really struck when you say that even blood and urine havenât been thoroughly examined, given the rich tradition of looking at those fluids, particularly urine in other cognitive and medical disorders. So, it seems like that is something that ought to be a full-court press. I hope that is part of the national effort. Dr. Leshner: We will add that to the agenda. As keeper of the clock, I am going to call this session to a close. I want to thank Dr. Spence and the speakers, who have done a wonderful job. I do want to comment on what a wonderful array of people are in this room. This speaks, I think, to the commitment of many, many people. We have institute directors from NIH, the deputy director of the National Science Foundation, the basic science agency and an array of wonderful scientists, both clinical and basic. So, I think we are well on our way.
Session II Lessons Learned from Other Disorders: âStandards of Evidenceâ Dr. Leshner: I think the first session got us off to a wonderful start. Again, we appreciate the speakers tremendously. The audience isnât too bad either. I would now like to introduce Dr. David Schwartz, the director of the National Institute of Environmental Health Sciences, who will chair the next session. Dr. Schwartz: Thank you, Alan. It is a pleasure to be here. Good morning. Welcome to Session II. Lessons learned from other disorders: Standards of evidence. Over the past couple of years, it has been a real pleasure for me to get to know the autism community, both the advocates and the scientists in the community. What has become abundantly clear to me is that this is a real challenge to our institute to try to help understand what is causing this disease and also how the environment can contribute to the understanding of the pathogenesis of this disorder as well as the phenotyping of the disorder. This session is set up as a way of taking some environmental diseases, diseases where we have uncovered causes in the environment that are related to complex diseases, identified the causes, and identifying the causes has led to a much richer understanding of the pathogenesis, genetics, and also ways to prevent the disease processes. So, to start this off, Phil Landrigan is going to give the first talk on environmental toxicants and neurodevelopment. Phil Landrigan is a pediatrician at Mt. Sinai School of Medicine. He is also an epidemiologist with a long history in environmental sciences and he is chair of the Department of Community and Preventive Medicine. 59
60 AUTISM AND THE ENVIRONMENT ENVIRONMENTAL TOXICANTS AND NEURODEVELOPMENT7 Dr. Philip Landrigan Dr. Landrigan: Let me begin by thanking the organizers of this conference. This is a very important gathering that brings together people from the community of concerned parents, the autism research community, the genetics community, the public health community, and the clinical pediatric community. The only way we are going to make progress against a complex and multifactorial disease such as autism is to have people of these diverse backgrounds share their insights and talk across their boundaries, as we are doing here today. I salute the IOM organizers, and I thank them for the extraordinary preparatory work that I know they have selflessly undertaken to make this day possible. The two central questions before us today are: (1) how can we accelerate the discovery of new knowledge about the preventable environmental causes of autism; and (2) how can we effectively translate these discoveries to the clinic and to the community to improve treatments and to strengthen the prevention of autism. I will approach these two questions by presenting two case studiesâ the cases of lead and of the organophosphate (OP) pesticides. Our work over the past several decades on lead and OPs has taught us a great deal about how chemicals can injure the developing brain. And additionally this work has taught us much about how to translate science to treatment and to prevention. I suspect that there may be many parallels in these examples that are relevant to the case of autism. I shall start with lead. When the medical and scientific communities first came to recognize lead poisoning more than 2,000 years ago, the condition was thought to be an occupational disease that principally affected adults. Lead poisoning in ancient times was seen principally in miners, smeltermen, painters, and potters. Starting in the late Middle Ages and through the industrial revolu- tion, our species started spewing lead widely into the environment. That environmental dissemination accelerated sharply in the 20th century with the addition of lead to gasoline. As a result, many people in addition to workers came to be exposed to lead. 7 Throughout Dr. Landriganâs presentation, he may refer to slides that can be found online at http://www.iom.edu/?id=42461.
PROCEEDINGS 61 Beginning in the early 1900s, physicians began to realize that lead poisoning could affect children. Until then, kids were thought to be immune to lead. The initial discovery of childhood lead poisoning was made by a clinician and an epidemiologist working together in Australia, Dr. Gibson and Dr. Turner, who were confronted clinically with a group of children who had GI problems, headaches, coma, and convulsions. Some of these children died. The disease was thought initially to be infectious, some kind of encephalitis. It was only through patient detective work over 14 years that Gibson and Turner came to realize that these children were, in fact, lead poisoned and that their poisoning had resulted from ingesting lead paint chips and contaminated dust on their verandas. The medical community came to learn additionally over succeeding decades that lead poisoning is a disease that can cause damage to children even in the absence of clinical symptoms. Prior to the 1940s, lead poisoning was thought to be a disease that either killed a child or from which the child recovered. But in 1943, Dr. Randolph Byers, a pioneering pediatric neurologist in Boston, realized that children who had suffered from lead poisoning, and who were thought to have fully recovered, could come back into the hospital a number of years later with behavioral problems. The event that triggered this clinical observation was an episode in which one of Dr. Byersâ former lead patients, a boy who had previously been a docile child, stabbed a teacher with a pair of scissors. That dramatic story prompted a study which showed that 19 of 20 of previously lead-poisoned children, who were thought to have recovered, had persistent hyperaggressive behavior. Dr. Byersâ work paved the way for work that our group at CDC did in the 1970s and that Herb Needlemanâs group in Boston and then in Pittsburgh did in the late seventies and into the eighties. These studies showed that asymptomatic children who were exposed to leadâlead from a smelter in El Paso and lead from paint in Bostonâcould have a decreased IQ, shortening of attention span, and problems in school. These effects were dose related and were more severe in the more heavily exposed children. They occurred entirely in the absence of clinical symptoms of lead poisoning. These studies introduced the notion of subclinical toxicity, the concept that there is a continuum of toxicity in which the clinically apparent effects of lead or other toxins have their subclinical counterparts. We came to recognize that lead could have subclinical
62 AUTISM AND THE ENVIRONMENT manifestations, defined as damage that is not clinically obvious, but that is quite real and very readily demonstrable on special testing such as testing of intelligence, attention span, or impulsivity. We have come to realize further that lead affects many brain functions in addition to intelligence. Behavior is another target of lead, and children with subclinical lead poisoning fail in school. They fail at life. They are dyslexic. They drop out. They are incarcerated. We have been able to recognize these causal connections between lead and brain injury because we have reliable exposure measuresâ biological markersânamely, the measurement of lead in blood, and more recently the measurement of lead in bone. In undertaking epidemiologic studies of environmental exposures, it is incredibly important to have stable biomarkers of exposure. Three further lessons emerge from the case study of lead. The first is recognition that when a neurotoxic chemical is widely dispersed in society as was lead in the years when we allowed it to be added to gasoline, subclinical toxicity is likely also to be widespread and can affect entire societies. In the 1970s, we were putting more than 100,000 pounds of lead into gasoline each year and then spewing all of that lead out into the environment through the tailpipes of cars. The result is that the population mean blood-lead level among American children in the 1970s was almost 20 micrograms per deciliter, a level that today would be considered dangerously high. It is clear in retrospect that subclinical neurotoxicity was widespread and that IQ was diminished across virtually the entire U.S. population. Moreover, if the mean IQ in those years was reduced by just 5 percent, as it almost certainly was, the result would have been a reduction of more than 50 percent in the number of gifted children and a corresponding increase of more than 50 percent of the number of kids who are going to have problems. Second, we have learned that not all individuals are equally sensitive to neurotoxic chemicals such as lead. Genetic and physiological differences convey sharp differences in vulnerability. Accordingly, we have begun since the decoding of the human genome to explore geneâ environment interactions that influence susceptibility to lead. I suspect that geneâenvironment interactions may be important also in the genesis of autism, and that some of the lessons that we are learning about lead will be relevant to the understanding of autism. The third, critically important lesson that emerges from the lead case study is that neurotoxicity which is caused by a toxic chemical in the
PROCEEDINGS 63 environment can be prevented. In 1976 as a consequence of our epidemiologic studies, and following additional studies of lead neurotoxicity that were supported by NIEHS in Cincinnati, Australia, and the former Yugoslavia, EPA made the decision to phase lead out of gasoline over a multiyear period. It was predicted by EPA that there would be a very modest decline in the population average blood-lead level, perhaps a 1- or a 2-microgram decline. What happened in reality was that there was a 50 percent decline in the average blood-lead level that paralleled incredibly closely the decline of the content of lead in gasoline between 1976 and 1980. This decline has continued to the present day, so that the average blood-lead level now in the USA is less than 2 micrograms per deciliter. In other words, we have achieved a better than 90 percent reduction in blood-lead levels in this country as a consequence of our scientific discovery of the developmental neurotoxicity of lead. Good science has driven this process at every step. But science alone was not enough to achieve prevention. Prevention required partnerships among scientists, regulators, elected officials, pediatricians, concerned parents, and society in general. The lessons for autism are clear. So, the question for those of us who care about public health is what can we do to speed this up. I think my second case study, the study of the neurotoxicity of the OP pesticides, is illustrative of the acceleration in the pace of discovery that can be achieved when appropriate resources are directed at a problem. Studies of the developmental and pediatric neurotoxicity of OP pesticides was triggered by a 1993 report from the National Academy of Sciences, entitled Pesticides in the Diets of Infants and Children. This report highlighted the unique exposures and the special vulnerability of children to pesticides. Its findings paved the way for passage in 1996 of the Food Quality Protection Act (FQPA), the principal federal law that governs the use of pesticides in agriculture. That law spoke for the first time in this nation in the language of law and policy about the importance of affording children special protections in law and regulation against neurotoxic chemicals. The passage of FQPA led to an outpouring of investment in childrenâs environmental health. This research was needed to fulfill the requirements of FQPA, which called for child-focused studies. It led to the creation of a national network of Childrenâs Environmental Health and Disease Prevention
64 AUTISM AND THE ENVIRONMENT Research Centers, the National Childrenâs Study, Pediatric Environmen- tal Health Specialty Units supported by CDC, creation of the Office of Childrenâs Environmental Health Protection at EPA, and a whole outpouring of investment in a field that had previously been very seriously neglected. The next figure is a curve showing the increase in funding over the past decade in childrenâs environmental health. A specific consequence of this increased investment in childrenâs environmental health is that in the study of organophosphate pesticides and their effects on the developing brain, we have made great progress, and we have done so at a far more rapid pace than in the case of lead. This rapid progress in understanding the developmental neurotoxicity of the OP insecticides began with Ted Slotkinâs work at Duke in which he showed that exposure of newborn rodents to organophosphates could cause anatomical problems in the brain, reductions in the number of cells, and behavioral problems. These deficits were fixed and not reversible. This work was followed by studies showing that the genetic variation in the enzyme paraoxonase (PON), which is centrally involved in metabolism of OP pesticides, could profoundly influence susceptibil-ity, a clear example of geneâenvironment interaction. Those studies were followed by the development of exposure assessment strategies, in which scientists learned how to measure organophosphates in body fluids, especially in urine. We learned also how to measure organophosphates in air and dust in homes, where they had been applied to control cockroaches in urban apartments. We found that these allegedly short-lived chemicals actually had a residence time in apartments that could be measured in weeks or months. Then we moved into the realm of clinical epidemiology. In this effort, our group at the Mt. Sinai School of Medicine and our colleagues at Columbia University and at University of CaliforniaâBerkeley recruited populations of mothers and their children who were followed prospectively. We recruited the moms when they were still pregnant, and we assessed the momsâ exposure to OP pesticides during pregnancy by measuring levels of OP metabolites in maternal urine. A critically important finding was that babies who were exposed during pregnancy to OPs had small head circumference at birth, a measure of delayed brain growth during the 9 months of pregnancy. This effect was especially striking in babies born to mothers who had low expression levels of PON. These babies had developmental delays. They
PROCEEDINGS 65 had cognitive defects. They had increased risk of attention deficit hyperactivity disorder (ADHD). And most important to our discussion here today, the babies exposed to OP pesticides during pregnancy appear to have an increased incidence of pervasive developmental disorder, which of course is a component of the autistic spectrum. As a consequence of these findings, EPA banned residential use of the two most widely used OP pesticidesâchlorpyrifos and diazinon. Reduction in the frequency of impaired babies was documented within months to result from this ban. In summary, these two case studies teach us the following lessons: â¢ Chemicals in the environment can injure the human brain. â¢ Children are especially vulnerable to brain injury caused by chemicals, and this vulnerability is generally greatest during the 9 months of pregnancy and in the earliest years of life. â¢ The brain injury caused in children by chemicals is sometimes symptomatic, but more often produces a range of abnormalities that impair function and that can be detected only through special testing. â¢ Chemicals can cause syndromes in children, such as ADHD and PDD. â¢ Chemically induced injury to the developing brain can be prevented by the application of scientific discovery. â¢ The pace of scientific discovery can be dramatically accelerated by focused investment in research. Conclusion: Where do we go from here? How do we apply the lessons learned from study of the neurotoxicity of lead and OP insecticides to better understand, treat, and prevent autism? I will argue that an overarching need is to build on the investments our society has made in the past decade and to continue to support research in childrenâs environmental health. Without continuing support for research, there will be no discovery. And if there is no discovery, there will be no new treatments and no new prevention. I identify four specific needs: 1. We must as a nation continue to support Centers of Excellence in Childrenâs Environmental Health and Disease Prevention Research. There exists a strong and highly productive national centers program. It is under review. Review is a good thing. As a
66 AUTISM AND THE ENVIRONMENT result of this review, there will be some change in the composition of those centers. However, the essential need is to continue to sustain multidisciplinary centers in childrenâs environmental health, and most especially to sustain the prospective birth cohort studies within those centers, which are the jewel in the crown. 2. We must support a large national prospective birth cohort study of American children such as the National Childrenâs Study. The National Childrenâs Study will follow 100,000 children, a statistically representative sample of all children born in the United States over a 21-year period from conception to adulthood. The goal is to identify the preventable environmental causes of autism and other diseases of children and then to apply those scientific findings to create a national blueprint for treatment and prevention. Given the current prevalence of autism in the U.S., a study of 100,000 children will give us almost 1,000 children with autism and 99,000 controls. Moreover, because the study will collect data on hundreds of environmental exposures (to be measured by CDC) and on the individual genetic susceptibility of each child in the study, it will provide us an unparalleled opportunity to examine interactions between genome and environment in child development. There will be no better opportunity in our lifetimes to discover the preventable environmental causes of autism. 3. We need training programs that increase the national workforce in environmental pediatrics. Today far too few pediatricians have more than minimal understanding of the pervasive influence of the environment on child health. 4. We must improve the testing of chemicals for potential toxicity, especially developmental neurotoxicity. We must end the current situation of deliberate ignorance, in which we produce new chemicals, disseminate them into the environment, but fail to test them for potential toxicity. Today there are over 80,000 chemicals in commerce. Approximately 3,000 of these chemicals are classed as high-production-volume (HPV) chemicals. Fewer than 20 percent of HPV chemicals have been tested for their potential capacity to injure the developing brain. This is an
PROCEEDINGS 67 untenable situation. Thank you. Dr. Schwartz: Thank you very much. There is time for one or two questions if anyone around the table has a question. Mr. Blaxill: Phil, could you comment a little bit on the institutional response to the lead problems and some of the resistance that the science faced? Dr. Landrigan: Well, yes. There was huge resistance. The problem is that lead was a very profitable chemical in the 1970s and there was a huge lead lobby that did their best to discount every scientific finding that was made in those years in public forums and in private meetings. The lead industry did their best to pillory Herb Needleman, whose picture I showed in the middle of the slide. They had a couple of scientists who were in their pay, although who didnât acknowledge that they were in the industryâs pay until later, came forward and charged Dr. Needleman with scientific fraud. His case was hung up for 4 years at the NIH, while that terribly painful process was cranked through. He was eventually completely vindicated and has won a whole series of prestigious awards since that time, but there was great resistance to learning the results of the research or to translating those research results into public policy. I think some of the reasons today that we have 80,000 chemicals in commerce of which fewer than 20 percent have been properly tested reflect the same legacy of special interests not caring to know about the toxicity of chemicals. I honestly think as a society, we need to get beyond that. We are flying blind if we allow kids to continue to be exposed to chemicals of untested toxicity. It is not a political issue. It sometimes gets portrayed as one, but it is not. What it is, it is an issue of protecting kids and I think it is an issue that people all across the political spectrum in this town should get together and say we really need to do something about this. We need to test the chemicals. We need to be examining the children. We need to be doing good research that leads to prevention. Dr. Slotkin: Just to add onto that, the same thing that happened with the lead story is also true even today with the pesticides and particularly the organophosphates, where scientific papers are being fought through letter-writing campaigns and whispering campaigns done by scientists and others in the pay of the chemical companies.
68 AUTISM AND THE ENVIRONMENT Dr. Schwartz: Thank you, Phil. We should move on to the next speaker so we have enough time for the discussion. Ezra Susser is the next speaker. Dr. Susser is a psychiatrist and an epidemiologist. He is chair of the Department of Epidemiology at Columbia University in the Mailman School of Public Health. PRENATAL STARVATION AND SCHIZOPHRENIA8 Dr. Ezra Susser Dr. Susser: I am going to talk about an example which is less advanced. It is an example where we have established a connection, some kind of link between an early prenatal exposure, prenatal famine, and the emergence of a disorder decades later, which is schizophrenia. But we donât yet know what the causal pathway is that accounts for this link. Learning that would lead us toward prevention and intervention. So, that is where we stand. It was hard to get to this point. I am going to talk about how we got there and then what we do next. Just a few words first about what the challenges are of establishing a relation between early prenatal exposure of any kind and then neurodevelopment outcome, and many of these pertain to autism, too. The problem with the time between conception to birth is that development is very rapid and we only have indirect ways of assessing what is happening to the fetus. What we usually do actually is measure what is happening to the mother. We use that indirectly as a window on what is happening to the fetus. So, that is one of the challenges. Another challenge that we have in this area, which pertains to autism as well as schizophrenia, is that we need very large numbers because they are not common diseases and we need to do very labor-intensive assessments to establish good diagnoses. That is extremely difficult to do. It is hard to assess hundreds of thousands of people as to who has autism, who has schizophrenia, and so forth. It is even more difficult for schizophrenia because you may have to wait for 30 years or more after birth before you can make the diagnosis. It is quite challenging to do this. 8 Throughout Dr. Susserâs presentation, he may refer to slides that can be found online at http://www.iom.edu/?id=42462.
PROCEEDINGS 69 But there are a number of examples where we have been able to overcome this challenge. I am going to talk about them. This begins with a study that we did based on the Dutch hunger winter that was at the end of World War II in Holland. It was due to a Nazi blockade of occupied Holland in the last part of World War II. We focused on the people who were conceived at the very height of that famine. These were people who were born from October 15 to December 31 of 1945 in this region of Holland. The picture there is just a picture of the food ration that was received by the whole population around the time that these children were conceived. It was very meager. What you see in the two graphs there, in the top graph, it shows the birth rate in this part of Holland across a 3-year period and you see that there is a very dramatic drop in fertility, which follows the drop in food supply around the time of conception. The blue shaded area in the graph marks the group that we identified as exposed, people with pericon- ceptual exposure to the famine. Then in the bottom figure, you see the outcomes that we measured in this exposed group. I have laid it out so that the exposed group is right under the blue shaded area of the exposed group in the top graph. We looked at three outcomes. We looked at anomalies of the central nervous system at birth. These, in retrospect, are mostly neural tube defects and that is the black bar and you can see it peaks in the exposed group. We looked at schizoid personality disorder, measured at age 18 in military recruitsâall malesâthat came from this population. You see that also peaks in the same exposed cohort. Then finally we looked at schizophrenia in adulthood using the National Psychiatric Registries of Holland. That also peaks in the same exposed cohort. So, you have a very sharply defined exposure to famine, a periconceptual or early gestational exposure, which resulted in a marked peak in three neurodevelopmental outcomes at three different stages in the life course. Whether those three outcomes have anything to do with each other, I canât tell you yet, except that they follow the same exposure. That was one study. It met a fairly high standard of evidence taken by itself in the sense that it was based on a historical event so people couldnât choose whether or not they would be exposed to the event of the famine. So, it is much stronger than your average observational study. Also, because it took place in Holland, where there was very good documentation of the food ration and the health of the population, and where decades later there were national psychiatric registries and other kinds of information
70 AUTISM AND THE ENVIRONMENT that contributed to the strength of the design. But you can never know from one study. There are always so many alternative explanations. There could be toxic exposures from other things that people eat when they were starving, like tulip bulbs, just to mention one. So, for a long time, we just couldnât be sure that this was a real finding. That has changed recently due to the work of another group. This is a group led by David St. Clair from Scotland and Lin He in Shanghai. They set out to test the schizophrenia finding of the Dutch famine study to see if they could replicate it in a completely different setting. Their study was based on a massive famine that occurred in China in 1959 and 1960 after the initiation of the Great Leap Forward. They had much larger numbers than in the previous Dutch study, but they only had annual data, so in that sense the study was less precise. However, based on the previous Dutch study, they could specifically hypothesize that in years in which the birth rate dropped, the schizophrenia rate would go up, if periconceptual or early gestational exposure to famine was indeed linked to schizophrenia. What you see here is in 1960 and 1961 the birth rate drops dramatically, and then you see there is a twofold increase in the risk of schizophrenia in the same birth years. It is relatively stable across the rest of the period. It is a very similar finding to the Dutch famine study and they were able to do this because they identified a region of China, the Wuhu Region, which is in Anhui Province, where, again, they were able to identify all the cases that occurred of schizophrenia in that area over a long period of time. So, it was like having a psychiatric registry over a long period of time. Now we have a third study. Subsequent to the Wuhu study, David St. Clair and Lin He contacted me and we also brought in MaryClaire King, and the four of us are now working together to pursue this question in a joint effort. We have finished a third study in the Guangxi Region in China where we have the same result, but even stronger. With three studies of this kind, I think we are fairly sure now that early prenatal famine is linked to schizophrenia in adulthood. But we still donât know why. We have a study with a fairly precise exposure, timing from the Dutch famine. Then we have the Chinese studies with very, very large numbers of cases. Together they provide fairly strong evidence, and the Chinese samples give us the ability to follow up these findings because there are very large numbers of people that we could
PROCEEDINGS 71 now go and study and try to figure out what it is that explains this latent effect. How do you go from something that happens in early embryonic development to something that may not happen until age 30 or 40? We donât know, but we are using biological reasoning. We are using available clues to guide ourselves here. I am going to move now from facts to speculation, which I state up front to prevent any misunderstanding. Iâll talk about some of these clues. There are several clues which suggest that the folate pathway could be important. One has to do with a gene which codes for an enzyme in the folic pathway, an enzyme called MTHFR. There is a variant in that enzyme, which has been associated with schizophrenia in very large samples. That leads us to consider whether there is something about the folate pathway that could be important. Another line of evidence is that the risk of neural tube defects is known to be related to folate, that is, the risk is reduced by periconceptual folate supplements. The increase in neural tube defects was exactly coincident with the increase in schizophrenia in the Dutch study. There are also other reasons for us to look at the folate pathway. So, that is one of the places that we are looking. How could the folate pathway be involved here? There are two hypotheses that I would like to mention to illustrate the way we can think about genes and environment together here. One is based on our knowledge that the folate pathway is very important in DNA synthesis and repair. Folate deficiency is thought to be one of the causes of de novo mutations. One way in which famine could be related to latent schizophrenia is that it is actually an environmental cause of genetic mutations. So, it is not exactly geneâenvironment interaction, but it is environment to gene to disease. A second hypothesis is âepigenetic.â We know that the folate pathway is also important in DNA methylation, which is one of the key mechanisms for epigenetic effects. We know from animal studies, for example, that maternal folate supplements influence the methylation of the DNA of offspring in utero. Since many people today have speculated that epigenetic effects are important in schizophrenia and there is some evidence along that line, this is another way that folate deficiency could affect the risk of schizophrenia. You donât need to bother with the details of the diagram of the folate pathway there. I put that diagram there to point out these roles of the
72 AUTISM AND THE ENVIRONMENT folate pathway for those who arenât familiar with it. If you look in the top left, you see pyrimidine and purine, which points out that folate is involved in DNA synthesis. If you look at the bottom, you see methyl transferases, DNA, RNA, which points out that it is also important in methylation of DNA. So our general hypothesis is that folate deficiency might be a cause of de novo genetic or epigenetic events. It is just that, a hypothesis, but it is an interesting one. If it is true, it would have enormous implications for prevention and that is one of the reasons that we wanted to go after it. Here we can use the example of neural tube defects as the classic model, the sort of hope or the Holy Grail, because we do know that periconceptual supplements of folate do reduce the incidence of neural tube defects. We know that from randomized clinical trials. Our dream is we may be able to do that kind of thing for schizophrenia. I donât want you to think that only prenatal famine or only prenatal experiences are important in schizophrenia, even when we are just thinking about the environment. I am pointing this out because I think in autism also one ought to keep a broad view, and the environment can have important effects at different times in the life course. In schizophrenia we can begin thinking about the environment even before conception. We have good evidence now that people born to older fathersâin other words, the fathers are older at the time of conceptionâ have a higher risk of schizophrenia than other people. The age at which men have children in a particular society is partly an environmental, sociocultural phenomenon. But we hypothesize here too that the mechanism linking this phenomenon to schizophrenia is de novo genetic mutation or epigenetic events. If the former is true, that could be considered as another example of an environmental factor leading to a genetic mutation. But in this case, the mutation could occur in the germ line even before conception. After birth, we have evidence suggesting that social factors influence the risk of schizophrenia. There is strong evidence nowadays that certain immigrant groups in western Europe have very high rates of schizo- phrenia. Urban living has an effect on the risk of schizophrenia. Very probably, so does cannabis use. These things are not mutually exclusive. We probably will find that it is not only one point in the life course that is important for the cause of these disorders. What can we say that might be useful in terms of autism? David asked us each to draw from our experience with these other diseases and
PROCEEDINGS 73 say what would we suggest could be applicable to finding the causes of autism. There are three things that I would take from the experience with schizophrenia. One is I think that we should create and support autism registries. This is already being done in some places. There is a good registry in western Australia and I know people are talking about doing this in the Scandinavian countries and to some extent in the United States, but I think that is going to be key. Second, I think that we need to establish what I call pregnancy birth cohorts. These are very large populations of people on whom we have measured their prenatal exposures. So, we have archived prenatal and cord blood samples, for example, that can be used over a long period of time to measure environmental toxins, nutritional states, and infections, and we also have genetic information on these people and their mothers and fathers. You have to actually begin early in pregnancy in order to collect the information that you want to collect. You canât do it retrospectively. You have to do it prospectively and you need very large numbers. It is possible to do this. It is already being done in one large cohort of 100,000 in Norway. Within that cohort we have the Autism Birth Cohort or the ABC. Maybe Ian Lipkin and Allen Wilcox will talk more about this later. Then we are also starting, in the United States, the National Childrenâs Study, as Phil Landrigan mentioned. So, there are two examples where I think we have studies that are going to yield some answers to these questions for autism. Finally, along the lines of the example that I showed and the comment that I made earlier, I think we should look forâwe call them natural experiments. I am not sure if that is a misnomer. But we should look for historical events that result in people being exposed to harmful or protective factors and we should go to those places and study those people. There are so many opportunities to do that once you recognize the design as a useful approach to study these diseases. Dr. Schwartz: Thank you very much, Dr. Susser. Are there any questionsâI guess we can take just one question for Dr. Susser. Dr. Beaudet. Dr. Beaudet: Can I ask if there is any hint that the incidence of schizophrenia might be dropping as weâlooking at the parallel incidence of neural tube defect, but with a much later age of diagnosis? Dr. Susser: Well, it is a controversial subject. Some people think that the incidence of schizophrenia is dropping and others think it isnât.
74 AUTISM AND THE ENVIRONMENT We donât actually have an answer to that, but the time to look would be a little bit later, you know, because it is 20 or 30 years after we started with the fortification of folate and so forth. So, we need to wait another 10 years to really get that answer even. The other thing that I would say about that is that somebody should follow up the randomized trials that were done of folate supplementation to prevent NCDs and look both for adverse effects that we may not have known about and the good effects like reduction in schizophrenia in those studies. We do have these randomized trials and the people are already in their teens now, who are in them. Dr. Beaudet: I will show some data just to suggest the increase in folate was going on in the seventies and eighties and didnât only occur on into the nineties and so on. Dr. Susser: Yes, that is true, but the fourth vacation was introduced then. Dr. Wilcox: Ezra, in the context of natural experiment, it is interesting that in Norway and maybe other Scandinavian countries, they have relatively low natural levels of folate in the diet and a great resistance to taking something artificial like vitamins. So, there are national differences that could be exploited for this kind of question. Dr. Susser: Exactly right. We intend to do it. I know you have done it to some extent in very effective ways. Dr. Schwartz: Well, thanks again. Our next speaker is Fernando Martinez. Dr. Martinez is a true environmental scientist. He is a pulmonologist, a pediatrician, an epidemiologist, and a geneticist. He is an integrated investigator all unto himself. He is the director of the Arizona Respiratory Center at the University of Arizona as professor of pediatrics there, and he is going to discuss some lessons learned about environmental asthma. ASTHMA9 Dr. Fernando Martinez Dr. Martinez: I hope that the examples that I will produce today are 9 Throughout Dr. Martinezâs presentation, he may refer to slides that can be found online at http://www.iom.edu/?id=42463.
PROCEEDINGS 75 kind of intermediate between what Dr. Susser has shown and Dr. Landrigan has shown in terms of our degree of understanding. I would like to say that perhaps they have to do with two issues that have been raised, natural experiments and the potential role and interests that protective effects may have. So, what I am going to talk about is what I am interested in, which is asthma. You are going to see things here that may sound very familiar to you. Asthma is a heterogeneous set of related conditions in which recurrent, partially reversible airway obstruction is the final common pathway. The clinical expression of asthma can start at any age, but we have now found in the last 5, 10, 20 years, that the first manifestation of the disease usually occurs during the preschool years. That may sound also familiar. In our case we have well-defined intermediate phenotypes for asthma that are strongly related with the disease burden and therefore they can be studied separately and I will show you some examples of that. For example, aeroallergy, bronchial responsiveness, or total serum IgE. One thing we know about asthma is that in the last 40 years it has clearly increased in frequency and you can figure that out both through the diagnosis of asthma and through asthma symptoms as reported by parents. So, there is a strong hint that asthma is an environmental disease and, of course, we have an advantage, I think, with respect to autism in that asthma is a very variable disease, an extremely variable disease. We have known for years what the main triggers for the disease are, and, of course, some of them have been pursued as potential inceptors of the disease. In other words this is a concept that is very important. In asthma we know very well that there is a difference between what could cause the disease at its very beginning and what triggers the disease once the disease process has developed. Unfortunately, we have not been very able to show for any of the triggers that they are involved in inception of the disease. The one that I am involved with and that is the only reason why it doesnât have a question markâit should have also a question mark, but we are all biased, of courseâis the lack of certain protective effects. That is the one I am going to stress more today. It all started with a form of natural experiment. A researcher in Britain, David Strachan, working with one of the largest birth cohort studies as two of the previous speakers have talked about, the 1958 birth cohort in Britain, found a startling finding, which is that children who had older siblings at home were much less likely to have what could be
76 AUTISM AND THE ENVIRONMENT considered intermediate phenotype for asthma, which is hay fever, than those who did not. That was truer for those who had older siblings than for those who had younger siblings at home. This observation was ignored I think for years until we, in our own longitudinal study, which was started in the 1980s, tried to reproduce it. This, as you can see, is 11 years later. What we found was very interesting. Here are âin the trianglesâ the children who were exposed to other children, be it because they had older siblings or because they were taken to day care. As you can see here at the beginning of life, they tend to have, of course, more viral infections, which are strongly associated with wheezing with this age period. But very interestingly if you follow them enough by the sixth to seventh year, when the atopic form of the disease, the allergic form, starts to be more prevalent, these children are clearly and significantly protected. A series of other studies came showing that there were other protective exposures. For example, this has now been reproduced, replicated in 10 studies and not replicated in 2. If you have a dog in the home, you are less likely to develop asthma in the first years of life than if you do not have a dog. Now, what is common between having a dog in the home and being exposed to other children? Well, in several studies now, it has been shown that day care and homes with a lot of children have high concentrations of a marker of microbial exposure in the homes, which is endotoxin. This has been shown now repeatedly in many studies. This is true for pets, as it is true for day care and homes with heavy concentration of children. But perhaps the most interesting solid natural experiment is the one that you see here in this slide, which is a form of living, which still exists in Central Europe in which children and adults live in single-family farms as the ones you see here. In these single-family farms, an empirical observation was that there was really very little asthma. Researchers listened to local physicians who were telling them that there was very little asthma in this environment and went to study it. What they found only a year after we published that paper on day care and other siblings was that, lo and behold, both for subjective and for objective measures of asthma, the children who live in those farms that you saw there were between 5 and 10 times less likely to have asthma than those living in the same rural
PROCEEDINGS 77 communities, but away from those farms. They also studied endotoxin concentrations in the homes, which is here in the x axis in relation to the likelihood of having illnesses in this environment. As you can see here, immediately after the first publication about those farming environments, what they found was that there was a striking inverse relationship between endotoxin exposure and the likelihood of having asthma, particularly allergic asthma, but not nonallergic asthma, which is an important issue because the clinical expression is identical. It is impossible to distinguish them. Even more and I donât show this slide to take more time, but in nonatopic asthma the relation is inverse to this one. In other words the more endotoxin, the more nonatopic asthma. We will get back to this concept in a moment. Of course, this was a very extraordinary environment. So, it was necessary to try to reproduce this in a less extraordinary environment. If you want to consider Manchester in England a less extraordinary environment, here it is. As you can see, these researchers in Manchester clearly reproduced in an urban setting the findings that had been reported before for the extraordinary environment in rural communities in Europe, in Central Europe. Now, an explanation has been proposed for this association, which is very simply as you can see here, that endotoxin or LPS increases the expression of IL-12 and IL-18, which in turn has a downregulating effect on Th2 differentiation, T helper cell 2 differentiation, which is the central and most important determinant of having atopic diseases. So, it was proposed that if you have endotoxin exposure, you deviate your immune responsiveness away from the Th2 mediated response, which is responsible for atopic asthma. If you donât, as you see on the right side, you upregulate the likelihood of having an atopic response. Asthma is also an allergic disease, and here I have put the latest twin studies published by the same research group with respect to asthma and interestingly with respect to autism, just published this year. You can see there are many things in common between the genetics of asthma and autism. I think the reason why the validity of asthma appears to be greater is simply because this was done earlier in life and the twin studies of autism were done later in life or perhaps because it is true since I know very little about autism, I donât know the answer. Something very interesting and paradoxical, however, is that these twin studies have both shown no shared environmental influences
78 AUTISM AND THE ENVIRONMENT affecting the concordance of asthma between twins, which may be true, but may also be a complete artifact, due to the fact that the models used suppose that there are no geneâenvironment interactions. I will get back to that concept in a moment. But much like in the disease of your interest, in asthma what we have had is that no single study has shown strong statistical evidence of a single gene being responsible for the disease. We have 15 chromosomal regions in which there seem to be asthma genes. Sound familiar? And only three regions have been clearly reproduced in at least two studies, if not three. Same thing as for autism. Well, one of the reasons why people have been able to reproduce linkage with asthma in chromosome 5q is because there is, I think, a large array of potential genes that can be candidates. It is a problem of luck. One of those genes is CD14, which I showed before. CD14 was in the middle of this potential pathophysiologic explanation. Why? Because CD14 is one element that is a member of the receptor system for LPS. CD14 is a crucial member of the receptor system for the exposure and protection in the exceptional farming environments and also in the nonexceptional environments in Manchester. We sequenced that gene in populations and we found five main closely linked, single-nucleotide polymorphisms in the five prime regulatory regions, and for one of them we showed functionality. This morning, we were told that that was important and I am just showing one slide of probably 10 I could show about the functionality. You will have to believe me that transcription rates are increased in carriers of the T allele at position â159. We now know that this is a problem of balance in those particular 5' regions between SP1 and SP3 transcription factors. It was logical to suppose that if you had more CD14, you would be more sensitive to the environment and you would have less atopy. There was more atopy in our population among children who were CC and CT, who had low expression of CD14 and those who had TT. We thought we had to put ourselves in the hall of fame of geneticists, who had found something important until, of course, we fell into the same problem that every single other person working with complex diseases has fallen into, which is that three researchers were able to reproduce this and three researchers were unable to reproduce this result. The three researchers who reproduced us called us and congratulated us. The three researchers who were unable to reproduce us said you guys donât know what you are talking about. Of course, we immediately
PROCEEDINGS 79 thought that the right place to study this was where the exposure to endotoxin was the highest, the farming environments, because here perhaps if we could determine if these people who live this way and heavily expose their children to brown stuff that is hereâabout which I wonât talk before lunchâcould probably heavily expose the children to endotoxin and others could not be exposed. So, we could study geneâ environment interactions. Of course here just to show to you how these things work is the relation between the same polymorphism and atopy without considering the environment among farmers, there is nothing there with CD14. The trick was to put it in relation to the environment and when we did that, something very interesting happened, which is that the sensitivity to the exposure to endotoxin was completely different, depending on your genotype. The CCs were heavily sensitive to the environment. The CTs and TTs were not. Now we have shown in other functional experiments that baseline unstimulated production of CD14 is higher in TTs, but a stimulated production is higher in CCs and CTs. So, what happens is that you have a very flat line for TTs and CTs and a very steep line for CCs. That creates a very interesting paradoxical situation, which is that at lower levels of exposure, the CCs are at risk whereas at high levels of exposure the TTs are at risk and the CCs are protected. This is due to the fact that the genes donât act alone. If you donât believe that this is trueâand I would agree if you see only one study, which was done in the exceptional environmentâthis next slide shows the results of the same analysis done in Manchester and the result is exactly the same. The CCs show this very steep relation- ship between risk of being allergic and the exposure to endotoxin. TTs and CTs show much less response to the exposure to the point that CC risk is lower at high level of exposure. Among African American adults, the same thing has been shown by Williams and co-workers in Detroit. What are my proposed conclusions? From our experience, natural experiments are very important and they may be true for both risk exposures and protective exposures. In our case it was protective exposures and they provided significant cues or clues for us to understand. I think I would have to say the hygiene hypothesis as this is called is still very controversial, but I think it has focused us into an area of exposures and has allowed us to understand the disease much better. Not only that, it has inspired new treatment. Very recently in the New
80 AUTISM AND THE ENVIRONMENT England Journal of Medicine, a paper has been published in which ligands of TLR9 are used as adjuvants for allergic desensitization, with the idea that they wanted to reproduce a little bit what could be present in the environment in this particular condition. I believe that we have to understand biological systems as plastic with heterogeneous responses to the environment and I think the example of CD14 that I have presented to you is characteristic of what the complex related genetics are going to be. They are going to be nonlinear. They are going to be weakly linked and strongly context dependent. Suggested approaches: I think that following up on replicated enhancing or protective exposures may prove extremely rewarding. I am not an expert in the field, the very controversial but very interesting fact that Mexico-born mothers have children with less autism than those who are not, maybe because I am Hispanic, too, calls very much my attention. I do understand that this may well be due to bias because they may seek less access. They may recognize this less, but being a physician who works with a lot of Mexico-born mothers, I am quite aware of how worried they are about the health of their children. So, I am not very convinced about their argument. I think that we have now technologies, both at the genetic and epigenetic level that allow us to assess genomewide the potential for genetic and epigenetic factors to be present. That may be related to exposure. So, I think that in studies such as the National Childrenâs Study, we could determine if replicated exposures could be useful to determine the type of geneâ environment interactions that, only in an example, I have provided to you in the case of asthma. Thank you. Dr. Schwartz: Fernando, that was great. Let me just make a suggestion, then we will open it for group discussion and focus on the general topic. DISCUSSION Dr. Schwartz: Your talk, Fernando, brought up a really important point, which is that environment can be used to narrow the pathophysiologic phenotype in such a way that you can understand the genetics of and also potentially the biology that underlies a very complex disorder like asthma and consequently a complex disorder like autism.
PROCEEDINGS 81 But I wanted to ask the group in general about natural experiments and whether we can expand our concept of natural experiments to autism. What are the natural populations or the cohorts that might be available and amenable to further study? Can we follow as it relates to an autism endpoint or subclinical early condition that is along the pathogenic line or clinical line or development of autism? Phil Landrigan, you brought up and, Ezra Susser, you brought up the issue of the natural experiments. Are there populations? Clearly, the National Childrenâs Study is going to be an outstanding study that will allow us to follow kids over time through development, but it is also going to take a long time. Are there populations that have been exposed that we should be looking at more carefully for autism influence? Dr. Landrigan: David, a couple of responses to that. First of all, I think the National Childrenâs Study is very powerful, but it is probably not going to answer every question. When I think of natural experiments, maybe because I spent many years at CDC, I think of clusters. I think of Brick Township in New Jersey, for example; I think of the group of children now three or four decades, who were exposed in utero to thalidomide and I think there is need for highly focused studies, which look at children who suffered unique exposures. I also think it is terribly important when those kinds of studies are done that we do as Ezra Susser suggested in regard to big cohort studies. That is, that we take samples and we archive them because there is always the very high possibility that new diagnostic techniques or new genetic probes will be developed in future years that will enable the scientists that follow us to examine those specimens and ask questions that are not possible to ask today. I think with regard to the Childrenâs Study, I would say that it wonât be that long. It will certainly be in our lifetime that we have data on the relationship between the environment and autism now that federal funding has been made available by the Congress and in such a way that it doesnât destroy the budget of NICHD. We are going to be moving forward. The first recruitment will take place beginning in about 12 months. That means that we will have a large number of 3-year-old children in the study in about 5 or 6 years, something like that, 7 years at the most. So, I would argue we will begin to make data on geneâenvironment determinants of autism in that particular population available by 2010 or 2011, 2012, somewhere in that range, not tomorrow, but not 25 years either.
82 AUTISM AND THE ENVIRONMENT Dr. Lipkin: I am really not going to be talking much about the Norwegian cohort because I was asked to talk about infectious diseases. But, in fact, that study is well underway and it has been running now for several years and at present we are close to 80,000 children recruited. I would think that the time frame for having real data there is much shorter. It is in the neighborhood of 1 to 2 years. That study includes prenatal data on the child, on the mother. It includes genetic information on the father, on the mother, on the child, cord blood, urine, a wide variety of sample types. One of the problems and challenges that we face right now is that although there are resources that have been allocated for establishing the cohort and for collecting these samples, that funding will be expiring in the not-too-distant future. Furthermore, there is really no allocation as yet to do any sort of work to analyze environmental exposures or to look for biomarkers, anything of the like. Now, there are a number of people who are here who are working with that cohort. Ezra is involved with this. Alan Wilcox, Mady Hornig, and myself. We would encourage people to collaborate and begin using this resource as soon as possible. Dr. Schwartz: Is this a population that is large enough to look at autism as an endpoint? Dr. Lipkin: It is 100,000 children. Dr. Schendel: There is also a Danish cohort that was assembled beginning in the mid-1990s of 100,000 pregnant women followed up, including their children. That is a database that is available. It doesnât have the intensive clinical evaluations that I think are being funded for the Norwegian cohort, but it does have biologic samples collected at multiple time points in pregnancy and cord blood of the children and newborn blood spots and has baseline data of the features of the mother during pregnancy and postnatal development of the child, which is clearly a resource that could be used in tooling our de novo studies. Dr. Lipkin: Just to make one point of distinction, the Norwegian cohort, which followed the Danish cohort, actually has materials that have been collected specifically for proteomic analysis and transcript profiling. It was really connected primarily not to do only genetic studies but to really look at functional data, so that rather than having blood spots, we actually have materials that are stored at minus 70 degrees and really, I mean, the opportunity to do proteomics is really going to be unparalleled until such time as the National Child Study comes online.
PROCEEDINGS 83 Dr. Schwartz: So, it sounds like a terrific population. What are the limitations? Are there limitations within this population? Dr. Lipkin: Of course, one issue is whether this population is too isolated to tell us what is happening in the American population. That is one issue that has been raised. I donât know. Would anybody else like to speak to this? Dr. Schendel: I would like to throw out to the group a question that I wanted to raise in the first session this morning. It might also apply to Session II and certainly to Session III as another opportunity for comparisons which might serve as a natural experiment, which is the sex bias in autism, the fact that you have this extensive male bias, but the extent of the male bias varies depending on the phenotypic profile of the group, with girls obviously displaying autism much less frequently than boys. I am throwing this out to the speakers of these sessions. Is that an opportunity that we can use for an investigation of clues for protective or risk mechanisms for autism? Dr. Insel: One example of that which shows up in the recent literature is a point that Dr. Susser made about paternal age as a factor; much greater odds ratios when you look at girls with autism than boys, about 18-fold increase that your father will be over 40. In girls about a fivefold increase with boys. Dr. Schendel: My point is using that dichotomy between boys and girls as a field for identifying potential mechanisms might explain the susceptibility to autism. Dr. Slotkin: I actually think that is not going to be fruitful because basically that sex difference exists for almost all neurodevelopmental disorders and it is likely a reflection that you can have the same degree of initial impact, but the female brain is more plastic because estrogen receptors regulate plasticity. So, you guys have an advantage over us apparently. I donât think that is something that is going to be a fruitful etiological factor for autism because it is simply shared by everything from ADHD to physical trauma of the brain. Dr. Beaudet: I would like to just strongly disagree with that opinion. I think that there is a lot of evidence that genetic aspects of females and males with autism are very different and I will show a little bit of data and some speculation about this, but I think that the causes of autism in females and the causes of autism in males are very different, I believe. Dr. Schwartz: Art Beaudet, can you expand on that in terms of
84 AUTISM AND THE ENVIRONMENT environment? Dr. Beaudet: It might be better to wait until my presentation when I will show a slide which would be much better to address the question. I think the boys are much more likely to have major environmental factors. Dr. Alexander: We are really in a very fortunate position, I think, to have the potential for three large national studies for environmental influences on childrenâs health and development from different environments. The Scandinavian environment probably has significant differences from what we experience in the United States. That will be an interesting comparison in and of itself. Autism is clearly one of the major outcomes that we are going to be looking at in the National Childrenâs Study and as Phil Landrigan pointed out, it will take us about 4 years to recruit the hundred thousand sample as those kids ageâby age 3 we will have basically all the autism kids diagnosed and identified. So, we will be able to start looking at the analyses we intend to do with regard to autism for the whole cohort within 7 years. One thing that relates to some of the discussions this morning, where it was pointed out that many of the either biomarkers or the toxicants are looked at singly and at one point in time. The advantage we have here is we will be able to do comparisonsâanalyses of interactive effects potentially with multiple exposures at several different points in time and also in relationship to genotype. That is an additional factor that we are going to have going for us in the National Childrenâs Study that I think is going to help shed a lot of information on the questions we are asking today. Dr. Akil: This is probably silly and I am going to sound like a hippie, but that is okay. Old hippie. I was struck by the discussion today about how in a way we are getting an evolutionary message. It is an evolutionary experiment in that the environmental factors that we can handle well, like living with a cow, maybe because we evolved in some kind of selection so that we could cohabit with a cow. If anything, when it is protective, it has an advantage; whereas insecticides that somebody synthesizes we have no evolutionary advantage in protecting ourselves against them and they seem to be quite hurtful. I am thinking about these geneâenvironment interactions and how to group things, it might be helpful to kind of have this very general idea about whether it is something that humanity has coped with in the past
PROCEEDINGS 85 versus if it is newly introduced. So, I apologize for the very nonscientific take on all of this. Dr. Susser: I didnât want to leave your question hanging about opportunities for, quote, natural experiments and just mention a few examples. One I think that would be important would be to follow up on populations exposed prenatally or in early life to infectious diseases, which may be outside the United States. There were early findings that related rubella infection in utero to autism; no one knows if it is a true finding or not. We donât really have prenatal rubella in large numbers in the United States now, but you have massive epidemics in other countries. Another example would be populations exposed to toxins in industrial disasters, also common outside the United States, maybe in the United States. There are many examples one can find and the only other thing I wanted to say on that is that we should also look for positive things that happen to populations, and use those as experiments to see if they have a positive impact on autism. Dr. Schwartz: Let me ask very specifically, is the Agricultural Health Study a population that we could use to look at this carefully in? This study is a rather large population of, I believe, 60,000 to 100,000. Alan Wilcox, could you tell us something about this? Dr. Wilcox: This is a cohort that was set up at our institute of agricultural health workers, mostly men, but some women, men, and their spouses, farmers and their spouses, who apply pesticides and I think somebody may know better than me, but it is about 60,000 peopleâ90 were enrolled. Well, okay. So, the family members have been also enrolled, but in smaller numbers. So, I guess the question is whether anything is being done with following the neurodevelopment of those kids. You were on the advisory panel for that, werenât you. Do you know? Ms. Bono: I was on the panel, the advisory panel for about 6 years, but I havenât been for awhile. My understanding was that the number of children and wives was actually not as large and I donât know how much information. I know neurodevelopmental disorders at one point was on the panel of things we tried to look at. But I am not sure where that went. Dr. Schwartz: The question is could we expand it to that? It would take an investment, but is it worth the investment to expand that cohort so that we could find out whether pesticides in agricultural chemicals are important in the development of autism.
86 AUTISM AND THE ENVIRONMENT Ms. Singer: I have a question for Dr. Landrigan. Acknowledging that there is still work to be done with regard to lead, I think many of us would look at lead as a real success story. I was hoping that those of us who are parents, as Lynn said, feel a real sense of urgency. I was hoping that you could share with us some of the factors that will speed up your results and also that slow down your results to the extent that they apply to autism. Dr. Landrigan: First, I would caution you not to proclaim that lead poisoning is over. There are still many tens of thousands of kids in this country with elevated lead levels. Yes, we have knocked down the average by 90 percent, but there are still pockets of kids, principally minority kids living in inner city old housing, who have terrible exposures, not just immigrant families. So, just to clear the record on that. With regard to the factors, I think the biggest factor that sped up the discovery of the developmental neurotoxicity of the organophosphate pesticides was the decision that a number of the federal agencies made beginning in the mid- to late 1990s to substantially increase the investment in studying the impact of chemical toxins on childrenâs health, with a particular focus on brain development. I think in a lot of ways that is the take-home lesson I would like to give you from my little talk, that the lead studies, which languished for decades, were terribly slow to produce results and by contrast the pesticide studies, which were really very generously supported and spanned the gamut from the most basic science through clinical research to epidemiologic studies all the way through to intervention studies, yielded some dramatic results in less than a decade. Dr. Martinez: If I may add to that, given our own experience in asthma, researchers as a collective tend to be quite conservative with respect to knowledge. It is difficult to put into the collective brain of the research community completely new ideas. There is a difficulty with fundingâusing the very limited funds that exist for things that are of very high risk. For that reason, I think that the new approach that the NIH has taken to fund more risky research is extremely important. It is also a problem of our way of thinking as scientists and you have to justify that a little bit because there are a hundred ideasâI say a hundred, could be a thousand new ideasâthat come up in a single year and you know that two or three are going to be successful. You have to be very lucky to be working with
PROCEEDINGS 87 the one that is the successful one. I could have showed you many examples of unsuccessful ones in our field. So, it is just a problem of trying collectively to find those that are most promising and being willing to fund risky research that is at least in part solidly based on the knowledge we have today and at least replicate it. That is the other thing, not just one of theâor two, but replicate it. Participant: I am Wendy Harnisher. I am the parent of two boys with autism. I was in the overflow room, so I actually want to bring up something that I heard before the break. I liked something that Susan Swedo said about Tylenol and vaccines. It is my understanding that Tylenol reduces glutathione levels in our children and glutathione is responsible for pulling toxins, including heavy metals, out of the body. I think this is something we should seriously look at. Also, I wanted to comment on something that Dr. Pessah said about the calcium channel being a main target for mercury. I think we need to look at mercury as well as other heavy metals. I am currently chelating both of my boys and they are dumping a lot of lead and they are also dumping a lot of mercury and with each dump, they are getting better. I think we should look at populations that do not have autism, such as the Amish, but not just the Amish. There is a pediatrician in Chicago who claims that none of his patients have autism. I know we are not supposed to bring up vaccines here today, but he doesnât vaccinate his kids. Dr. Leshner: Thank you. Dr. Schwartz: Thanks for your comments. Really appreciate them. I do have a question for the group at large because Phil Landrigan brought up this issue of how subclinical early indicators of response to lead were helpful in identifying safe levels of lead and also moving the research forward. It made me wonder whether there are subclinical phenotypes or subclinical, preclinical biological responses in autism that would help us identify etiologic agents. It is a full-blown disease or there must be spectra of this disease and I guess the question is should we be looking at any of those less severe forms of the disease? Dr. Levitt: In the context of the several studies that have just come out on the examination of baby siblings, there are several studies that have come out where they have looked at what you might call intermediate phenotypes or whether there are clear indications that there
88 AUTISM AND THE ENVIRONMENT are atypical trajectory of social, behavioral, development, and communication. I think the answer from several studies is, yes, there are and you can identify those; that is, skilled individuals doing research in that area can identify that there is a typical development and trajectory in those domains. Now, how that relates to whether those children are going to end up on the spectrum or not, I think is an open question, but I thinkâwhen were the baby sibling initiatives started historically? How long has it been? Participant: About 5 years. Dr. Levitt: So, 5 years. So, the first studiesâthe Davis study, the study that came out of Vanderbilt, there is a third study. All three basically show very similar things with reasonable numbers so the statistics I think are okayâindicate that you can actually begin to think about doing that where you could identify individuals, children, young children before the full-blown diagnosis that then would need to be followed, but it has been 5 years and now we are just getting the first indications that this might be a fruitful way to go. Dr. Insel: David, I heard a slightly different message in the talks and if we could go back to Dr. Martinezâs comment about the two forms of asthma and how if I heard you right, you said they were clinically indistinguishable and yet you had a very different pathophysiology pathway. How did that happen? How did you get there, knowing that you couldnât do it just from the clinical phenomena? Dr. Martinez: What helped most is what you could call biomarkers in a very generic way. In other words we learned. It was tough because there was a period during which the lumpers had the prevalence or the splitters among the scientists. So, everything is asthma. Treat it with inhaled steroids and everybody is going to be okay. But with time we started learning that if you take, for example, responsiveness or you take being sensitized toâor not, those are very different kids or adults, who have the same symptoms. Of course, if you go into a lot of details once you know they are different, you start seeing that there may be differences, but it is very difficult to do that before you have the specific biomarker that allows you to do that. So, in a certain sense I was hearing before with respect to efforts of phenotyping, I think the efforts of phenotyping, somebody said that beforeâI donât remember whoâinclude the biomarkers. It is not that you can start by trying to squeeze your brain to distinguish clinical characteristics and that is the only way you are going to do it. You have
PROCEEDINGS 89 to include the biomarkers. We found out, for example, in these studies that children who are skin dispositive to allergies are very different from those who are not. They have a different prognosis. They are much less chronic. They tend to decrease with puberty. I wonât go into all the details. But we knew that once we developed the biomarkers to understand how to distinguish the different groups and that is something that could help a lot in this particular area, too. Dr. Schwartz: And the environment helped distinguish the different groups. So, there were several factors, biomarkers, genetics, and environment. Dr. Martinez: There is no doubt. It is an integrated process. It cannot be done one first and then the next. Dr. Hertz-Picciotto: I guess I wanted to sort of link some of the discussions that we had earlier with the discussions just now and raise the question, so, we have these large birth cohorts, which have a lot of promise in terms of being able to go back because there will be stored specimens that we can do a lot of varieties of biomarker testing on. On the other hand, the question of the actual characterization of the phenotypes in those very large cohort studies, I think, is a concern that needs to be raised because the diagnosisâwell, I donât know very much about the Norway situation and what is being done in terms of the phenotyping and just even in terms of the ascertainment of the diagnosis itself, but it does seem clear that from our study, from the CHARGE study, that there is a substantial percentage of children who donât meet criteria for the condition. Maybe those are the ones that fall into this category of having some of the markers and are still informative, but I think it is important that even in terms of knowing who actually has autism and who doesnât in these large studies, it is going to be a big challenge in the National Childrenâs Study having just spent the last 6 weeks writing one of those proposals and working with some of the counties that donât have academic centers and I suspect donât have a whole lot of people doingâwith the expertise to, in fact, do the diagnosis in those areas. So, I think that is another challenge and in the National Childrenâs Study, one of the things that I noticed was not part of the RFP and not part of the protocol is anything prior to 36 months in relation to autism. So, there is no screener that is happening at 12 months, 18 months, 24 months that is part of the current protocol. It is an area where I think some work is going to need to be done.
90 AUTISM AND THE ENVIRONMENT Maybe that will identify a large percentage of the ones who at 36 months will meet criteria and maybe it will identify some of the subclinical conditions that may or may not go on to, in fact, be autism after the 6 months and could be done on the 100,000 children, conceivably. Not to mention the subphenotyping issues that were brought up earlier in the first session this morning. Dr. Herbert: There are some papers by Dr. Deborah Feinâs group at the University of Connecticut on children losing their diagnoses and it is informative potentially to look at what they loseâwhat they have left after they lose their diagnosis. One group had specific language impairment, attention deficit. Does the way that the phenotype decomposes in the course of treatment tell you something about how it is stuck together? There has been work on the idea that the different behavioral traits transmit separately. But we really donât know what that means biologically. Is that purely a matter of genetics? Is it a matter of other biological issues? Is it a matter of gut bugs, a variety of things? In any case, I think if we had biological measurements in the course of tracking the progress of people in treatment, we could learn something. I want to make one last comment, which is that in order to gather this data, we are going to have to have some tolerance for exploratory measures, where we arenât exactly sure what it is going to show, but that this is a good time in history to take that on. Dr. Schwartz: Great comment. Sallie. Ms. Bernard: Couple of things quick. One is just to your point. It might be interesting to look and see how autism composes in addition to decomposes because with my son, he first had language and then attention deficit and then autism. So, PDD and then autism. So, that is sort of an interesting rule-out. Also, I want to go back to your question about what populations we can study and you know, while these big cohort studies, the Norwegian study and the National Childrenâs Study, are vitally important and we need to do it, we do need to keep the idea of urgency and there are populations that exist right now that we could be studying that focus on autism. The speaker from the other room brought up the Amish. We have talked about the baby sib studies and those are very specific populations that we could go in very quickly and study the rate of autism and look at exposure histories in those groups and seeâget some good information
PROCEEDINGS 91 right away without waiting 3 to 4 years and spending a huge amount of money. The last thing I would like to point out is that we focus very much on complex diseases with the idea that it is a foregone conclusion that autism really is a complex disease and there are multiple genes, 10 or 20 genes and there are a thousand exposures that could be possible. I would just like to remind us that historically there is a disease called acrodynia or Pink disease that was one of the number one childhood diseases in Australia about 50 to 75 years ago. It had one cause, and it was mercury. I just want us to think about and not rule out the possibility that the causes of autism could be more limited to what I have heard in the discussion today. Dr. Schwartz: Thanks very much. One last comment. Participant: Mary McKenna, University of Maryland School of Medicine. I am a neurochemist and I was very struck by Marthaâs talk earlier and also by listening to the other people talking. Many of the nutrient and alterations in metabolism that have been brought up, for instance, folate, B12, B6 are very important for brain development and it would seem to me that it would be extremely useful to come up with some sort of metabolic panel for testing for any kid suspected of autism, where you would look at all these vitamins that are possibly at risk, where you would look for heavy metals and you start screaming right when any sort of diagnosis is first suspected because you may come up with a lot of useful information if you start obtaining things like that. It doesnât seem like there is any sort of uniform consensus right now as to what the approach is and what biomarkers and what metabolic information and heavy metal information to get right from the beginning. Dr. Schwartz: One last comment, Laura. Make it short, though. Ms. Bono: I just want to mention something about what may be called the âhit and runâ and that as we are gathering urine and blood specimens on children, it is very hard to perhaps track the toxicant that hurt the child. Speaking to metals, which is the same general idea, before we started chelating my son in 2000, we ran a test at Duke and we got a 5 on his blood lead levels. But he has dumped more lead and mercury and aluminum and nickel and tin, which, of course, points to the synergistic effect of metals more than any kid I have ever seen. I mean, lead levels off the chart every single time and he started out with a blood level of 5, which leads to the theory currently of some doctors that these
92 AUTISM AND THE ENVIRONMENT kids are non-excretors. So, you wouldnât find that in the blood when you first start testing them. The blood is not the organ of toxicity. The mercury and lead and other things are going to other areas. Dr. Schwartz: Thank you. Dr. Leshner: Thank you, Dr. Schwartz and the speakers and all the participants. I think we have had a tremendous morning. I have had a tremendous morning. You can speak for yourselves. I think it has been very interesting. I like very much the spirit of leaving all the questions on the table; that is to say to take as broad a conceivable look at this as we can because as we go forward, although by necessity individual scientists and groups will develop their own priorities and their own specific projects. I think there is no question that as a field or at least as an outsider to the field, listening to it, sure does need to leave the field as open as possible and there is a tremendous amount of research yet to be done and to be discussed.
Session III Environment and Biology I: What Are the Tools for AutismâWhat Do We Have, What Do We Need? Dr. Leshner: As far as I can tell, there is general agreement that it has been a terrific morning, and therefore we are putting terrific pressure on the afternoon speakers, so donât let us down. People have been very well behaved and stayed on target, asked questions, didnât make long speeches, so everybody so far has behaved very well, identified gaps. Dr. Levitt: Good afternoon. My name is Pat Levitt. In this session, there is one content session and then the discussion; this is the session that has the unenviable task of putting together environment and biology. That is the title of it. The first speaker is Art Beaudet, who is professor and chair in the Department of Molecular and Human Genetics at the Baylor College of Medicine. HOW MAY ENVIRONMENTAL FACTORS IMPACT POTENTIAL MOLECULAR AND EPIGENETIC MECHANISMS?10 Dr. Arthur Beaudet Dr. Beaudet: I am going to maybe be a bit provocative and try to argue that there is a substantial chunk of autism where we now can predict what is going on. I made a diagram here. You heard earlier this morning the mention of maybe 10 percent of autism being genetic. Iâd like to argue that this is 40 or 50 percent. This is maybe an exaggeration, maybe it wonât be quite that high, but there is quite a group that we know where we are going to end up. These individuals have chromosomal defects and single gene defects. You have seen some of these mentioned. You have heard someone mention how we have better and better techniques for how to 10 Throughout Dr. Beaudetâs presentation, he may refer to slides that can be found online at http://www.iom.edu/?id=42981. 93
94 AUTISM AND THE ENVIRONMENT search for these. But we know the way we search for them today is like looking at icebergs and only looking above the water line. So we know that this group is much larger because of our inadequate ability to detect small changes. Then I would argue there is a second group over here that is much more unknown as far as what is going on, and much more likely to be a candidate for involvement with environmental interactions, epigenetics, you name it. I think we just donât know where we are there. This is a review from about a year ago that is very nice, indicating how many known definitive chromosomal abnormalities are seen in autism. These are mostly so-called de novo events in children. Their parents are normal. Duplications of chromosome 15 are by far the greatest, and there are quite a lot of deletions of chromosome 22 as well. So we know there are deletions and duplications that can involve every single chromosome that can give rise to autism. There are a couple of papers that have appeared recently that further emphasize this, using a methodology called array comparative genomic hybridization to detect larger events across the genome with greater efficiency. One paper recently reported detecting abnormalities that are presumed causative in children, with 27 and a half of the children with syndromic autism. These are children who are dysmorphic. They are likely more cognitively impaired, probably both mentally retarded and autistic in most cases, and unusual looking. If you see them in a grocery store, you will see that they have some physical abnormalities. The report by Jonathan Sebat has been mentioned, where he found in about 10 percent of simplex cases these kinds of abnormalities. We know that these methods being used will miss many, many kinds of genetic lesions which would give the same functional effect. We have heard about advanced paternal age. Dr. Susser has this publication here. I will just say that we have an ability to make a very good guess what the problem is with advanced paternal age. It is probably point mutations, so it is probably causing a de novo effect on a single gene. I think the fact that we are seeing it more in females than males will make sense in a minute. If you were to take away anything from my presentation, I would say this is the message. I would say there is a group of mutations that are identified, chromosomal, single gene, in autistic patients. We do see what their primary defect is. It is a strong genetic effect, it is a very
PROCEEDINGS 95 highly penetrant effect. This group tends to be dysmorphic and they tend to have cognitive impairment. From what we know about how we found these, we know that our ascertainment method is pretty terrible, so I am speculating that there has got to be more of these which are likely, more of the same mutations. That would give you then a residual, very small group of females who are the pink section up here, and this huge group of males who are less impaired, less dysmorphic, and more puzzling as to their etiology. The reason the paternal age effect makes sense is that we know that paternal age effects will be relevant to this group of mutations up here. For almost all of these we have equal maleâfemale distribution. So I think there is a big chunk of autism which we maybe would have said was 10 percent or 5 percent 5 years ago, that I think is going to be closer to 40 or 50 percent of the total. That leads to the second phenomenon down here, which seems very different. Geneticists think about things being heritable. You have heard comments about monozygous twins. I just want to make the point that de novo genetic events are highly heritable in genetic terminology. That is, if you take Down syndrome and you have identical twins, they will both have Down syndrome 100 percent of the time. So their phenotype of Down syndrome is determined by their phenotype, and we say the heritability is 100 percent. But their parents donât have Down syndrome. The abnormality is not inherited. I think this is the case for all of the autism genetic defects that we know about at present. They are by and large de novo genetic events. We would expect them to be highly concordant in monozygous twins and much less frequently concordant in dizygous twins, which is what the bulk of the twin data says about autism. The rest of it, I have to say, turns more to this leftover group that we understand less. I have worked with a couple of disorders, Prader-Willi syndrome and Angelman syndrome, that involve the phenomenon of genomic imprinting that I donât have time to go into here. On chromosome 15, if you have a deletion of a particular region, you have Prader-Willi syndrome, and if you have a deletion in the same region on the maternal chromosome you have Angelman syndrome. If you inherit two copies of chromosome 15 from your mother you have Prader-Willi syndrome, and two copies of chromosome 15 from your father you have Angelman syndrome.
96 AUTISM AND THE ENVIRONMENT The deletions are genetic. If you sequence the genome, you will find 5 or 6 million base pairs of DNA have been lost. These are epigenetic. If you sequence the gene in the epigenetic cases, the sequence is perfectly normal, but the fact that these genes behave differently whether they are of maternal origin or paternal origin explains the problem. Both of these events, which are the bulk of events that cause this kind of abnormality, are de novo. That is, the parents donât have the deletions and they donât have the two copies of a chromosome from a single parent. This is emphasized for us that a diagnosis could be quite hard to figure out. If you have some cases in the mix being epigenetic, some being genetic, the epigenetic or genetic events could be de novo or inherited. It creates quite a complicated model, and so we have tried to explore this mostly as it relates to the patients who we understand less. This de novo component fits very well with what I have been talking about, the genetic group. We donât really have definitive evidence as to whether the epigenetic component is going to be an important one or not. In this, I am very interested, particularly from the epigenetic status, about the environmental interaction, particularly folic acid. The genotype has to have a certain epigenetic state in order to give rise to the pheno- type, so I am very supportive of the idea that there will be environmental genetic interactions going on. This graph is widely talked about and looked at. I just want to make the point that some people think there is a substantially increased incidence of autism. I think it is clear this is partly artifactual by how children are diagnosed and ascertainment and so on. But if there is any component of this that is real, it is very, very important to detect for the reasons that have already been stated this morning in terms of under- standing the causation and trying to develop treatment. If there is something going on, what could be going on? I just want to mention two issues. Prenatal ultrasound, in the event that it might not get mentioned otherwise. Paternal age we have already talked about. I want to talk about folic acid a bit more. This is a paper from last year in PNAS looking at ultrasound exposure of mice and the effect on the neuronal migration in these developing mice. I think this is a good example of the kind of area that we need to be thinking about as far as any kind of environmental factor. These people made some recommendations that we shouldnât be doing
PROCEEDINGS 97 prenatal ultrasound as a recreational activity, and that we need more research in this area. This is how prenatal ultrasound has increased over the right year interval. You have seen this figure before about folic acid, and I will use it now to transition to folic acid, just to make a few points. When I have tried to express some concern that folic acid could be a problem that could be increasing the incidence of autism, people have said it was that the fortification came too late. But I think if you look at the data, that is not correct. In the NHANES studies in the 1970s, we had 23 percent of people reporting they took a daily vitamin. In the later 1970s and 1980s, 35 percent, and this went up with time. The FDA (Food and Drug Administration) prohibited putting much folic acid in vitamins until the mid-70s, and most vitamins had none, but a few had a tiny amount. But in 1973 they raised the limit to 0.4 milligrams of folic acid. So one-a-day vitamins went from none to 0.4 milligrams in 1976 and Vidaylin went from none to 0.4 milligrams in 1977. This is before the neural tube defect perspective. There are data from the Framingham study that people who reported that they took a daily vitamin or ate ready-to-eat cereal had a folic acid level roughly two to three times higher than people who reported they did not. This was in the 1990s before fortification. We had two groups in the population, those who were taking a folate supplement and those who were not. So I think these changes are reasonably compatible with the possibility that timewise, folic acid is a potential factor. Why have we been very interested in it? My laboratory has been interested in epigenetics, and it is known that using folic acid intake in mice and in humans, you can alter gene expression because of the way it contributes to DNA methylation and histone methylation. This is a publication from some time back, where coat color in these mice is under a particular genetic element which is responsive to DNA methylation. You can change the coat color of the mice by feeding the mother differing amounts of folic acid and other methylation-related compounds during the pregnancy. This is a study from humans. I wonât try to take you through the technology, but just to say it demonstrates that folic acid can change gene expression in humans as well. This is a gene which should have only one of the two bands here present in a normal situation. These are patients with renal failure in high homocysteines, and those with the
98 AUTISM AND THE ENVIRONMENT highest homocysteines are expressing both the maternal and paternal copy of the gene, so they have two bands; that is abnormal. But when you put them on folate supplementation they go back to expressing just one band, which is the normal state. So again, folic acid can influence gene expression in mice and in humans in certain situations, and often this involves this phenomenon of genomic imprinting, where the maternal copy of a gene and the paternal copy can differ. So folic acid definitely changes the action of some genes, probably especially imprinted genes. The laboratory acid intake of the population at large, and particularly reproductive-age women, has dramatically increased over the last three decades. Your folate level and maybe imprinted gene expression are different today than they were 15 years ago, and we need to know more about whether folic acid intake is increasing or decreasing the intake of any diseases. The following are suggestions for potential research areas. I think genomewide studies at the exxon level and single-gene level and single- nucleotide level will expand this group, which I propose will turn out to be genetic, but we donât have very good ability to detect them right now. This will separate out this strong mutation group from the other puzzling group that is left. I think that epigenetic approaches are very worthwhile for the idiopathic portion that has not got specific genetic lesions. Dr. Levitt: We have time for one or two clarifying questions. Dr. Pessah: When I started out in looking at autism many years ago, only about 10 years ago actually, the emphatic view was 90 to 95 percent heritable genetics. What has changed over the last 10 years to make it 50- 50? Dr. Beaudet: Well, I donât know. It is different opinions about what the heritability is. I think one question had been, why is it so concordant in monozygote twins and nonconcordant in dizygote twins? That is totally explained by de novo events. Whether it is advanced paternal age causing a point mutation or whether it is trisomy 21, these de novo copy number variants, they all will give you 100 percent concordance in monozygote twins and a much lower concordance in dizygote twins. I think also, this whole issue that these people have genetic conditions, their genotype determines their phenotype, but it is not inherited. So if you try to compare their genotype to their parentsâ genotype, you donât find the expected implications.
PROCEEDINGS 99 Dr. Herbert: You said that the known genetic mutations had a 100 percent concordance monozygotic and 5 percent dizygotic. Where are those data from? Dr. Beaudet: I would say on general principles, if you take any new mutation event, whether it is trisomy 21 or whether it is achondroplasia, Rett syndrome, any new mutation event happens prior to fertilization or prior to twinning. The monozygote twinning takes place later on, and the twins have the identical genotype, including the genotypic error that they have. Dr. Herbert: You are saying they are concordant for autism. We donât know whether the gene causes it or it is just a risk factor. If it is just a risk factor, then you canât assume that it is going to be 100 percent concordance. So if it is 100 percent concordance in these genetic errors, are there data that support what you said? Dr. Beaudet: The question is, when you find these kinds of errors, how convinced are you that they are the cause of that childâs abnormality? Dr. Herbert: Close to a risk factor, a high risk factor. Dr. Beaudet: Right, or totally irrelevant. I think that there is some of that. If you look for these de novo events in the control population, you do see some de novo events in the normal population. But statistically, most of these new events are almost certainly the cause of the childâs disability. They all have major effects, for the most part. There may be weaker effects that we havenât discovered yet, but the ones that we are looking at here, they mostly have physical abnormalities associated with them in terms of dysmorphic features and birth defects, they are mostly mentally retarded, and they meet the criteria for autism. Dr. Levitt: Thatâs it. We have a lot of time for discussion this afternoon, and the godfather is looking at me. Thanks very much, Art. Mark Nobel is our next speaker. He is going to talk about environ- mental factors impacting cell function. Mark is a professor of genetics at the University of Rochester Medical Center.
100 AUTISM AND THE ENVIRONMENT HOW MAY ENVIRONMENTAL FACTORS IMPACT POTENTIAL CELL-BASED MECHANISMS?11 Dr. Mark Noble Dr. Noble: Thank you so much for this opportunity to come and learn from you all. It is very exciting to me to have the opportunity to take part in this discussion. I am going to approach this talk from the perspective of our efforts to develop a comprehensive approach to the field of stem cell medicine. This work began with our initial isolation of CNS progenitor cells almost 25 years ago, and now extends to cover many components of stem cell medicine that are separate from the use of cell transplantation to repair damaged tissue. In order to discuss our work, I have to introduce you to some of the cellular players in the CNS. The only point that I want to make with this slide is that when people talk about development, they mostly talk about stem cells and they talk about differentiated cell types, neurons and myelin-forming oligodendrocytes and astrocytes. From our attempts to understand the cellular basis of developmental maladies, however, it seems the most interesting cells are the progenitor cells that lie in the middle. These lineage-restricted progenitor cells are the workhorses of building tissues. They are the ones that are responding to environmental signals. They are the ones that are building your whole nervous system during development, and these are the ones where we focus our attention. There are several such progenitors that we study. Our greatest interest, however, has been studying myelination because of damage to myelin being the largest category of neurological disorder, showing up in all traumatic injuries, most chronic degenerative conditions, and in respect to this meeting, with some very interesting findings in respect to autism. Through our studies of all these progenitors and what happens to them during development, we have come to realize that many develop- mental maladies are diseases of precursor cells. You have abnormalities in the generation of specific cells, with specific cell types being generated too early in some conditions, and not at all in others. Or you donât make enough of certain cell types. We have been trying to 11 Throughout Dr. Nobleâs presentation, he may refer to slides that can be found online at http://www.iom.edu/?id=42465.
PROCEEDINGS 101 understand how these abnormalities occur to identify the underlying principles at work when bad things happen to good cells. There are several insights from our work that I would like to share with you. Development is a progression, in which different cells appear at different time points. In some tissues, in some lineages, they go into a single end-stage cell, in some you are generating different cell types, such as different neurons, at different times. In the specific context of myelination you have a sequence of cells, from a stem cell to the first level of restricted progenitor, to the second level of restricted progenitor, and finally on to an oligodendrocyte. There are multiple insults that we have discovered, such as thyroid hormone deficiency and iron defi- ciency, that can have the same outcome of not generating enough myelin. Weâve been investigating the effects of different insults on aberrant myelination in a variety of ways, but the one that currently looks most interesting emerges from studies on the very ancient evolutionary problem of controlling the balance between self-renewal and differentia- tion. This is a central problem in understanding normal and abnormal development, and it is also an old problem, dating back to the first organisms that had multiple cell types. Our hope was that if we could solve this problem once, we might have solutions that apply to many different cell types. We started this work as many others have done, which is to discover molecules that impact on the balance between division and differentia- tion. In the years of this work, we and others discovered that thyroid hormone promotes differentiation. We discovered that platelet-derived growth factor is a basal mitogen that is needed and is sufficient to promote division of oligodendrocyte progenitor cells and allows oligodendrocyte generation. We discovered molecules that suppress differentiation, such as neurotropin-3 and FGF, which enhance self- renewal. We and others also identified other factors, such as bone morphogenetic protein, that enhance astrocyte generation. What we then wanted to know is how all this different information received by a cell becomes integrated. Lots of people study this integration question, and we see this diagram of intersecting signaling pathways in many meetings. But we wanted to ask a different question, which is, how does this information become integrated in the context of physiology?
102 AUTISM AND THE ENVIRONMENT The physiology that we have been most interested in is redox physiology. I want to explicitly say that I am not that interested in oxidative stress. What I am interested in is the normal use of redox balance in controlling development and cell function. This is a different question from that asking about stresses that kill cells. We study oxidative stress a bit, but what has turned out to be much more interest- ing is this area of normal physiological control, which is simply the balance between reducing and oxidizing equivalents in a cell. The first discovery we made in this area came from studies in cell death. What we found is that modifying the redox state by tiny amounts, by 15 percent, has tremendous effects on biological outcomes. For example, here we are looking at survival of spinal ganglion neurons, given suboptimal amounts of nerve growth factor in the purple bars here. If we add N-acetylcysteine, a cysteine pro drug that is taken up by cells and can be used to increase glutathione levels, we can give enough N-acetylcysteine to cause a 15 percent change in glutathione content, and that is sufficient to obtain a 1,300 percent increase in the number of surviving neurons. We wanted to understand how it was possible for small changes in redox state to have such large effects on cellular function, and we started taking this apart through our work on progenitor cells. Through our work on progenitor cells, we discovered a number of general principles. It now is clear that the redox state is a central regulator of precursor cell function, controlling whether a cell divides or differentiates as well as whether cells survive. For example, we found that all the classic signaling molecules that we study converge on regulation of redox state. Neurotropin-3, FGF, thyroid hormone, BMP, every signaling molecule we have examined converges on redox state, and the redox changes that they induce are necessary for them to exert their functions. We have also discovered that the organism uses developmental genetic or redox state to control precursor cell function. There are some general principles relevant to our data that I want to mention. If an oligodendrocyte progenitor cell is a little more reduced, it is more responsive to mitogens and survival factors. If it is a little more oxidized, it is more responsive to differentiation and death factors. Now let me tie it into some of the discussions that have occurred this morning. We decided to start working in the field of toxicology because we wanted to study redox perturbations that have real-world significance. In the laboratory we carry out genetic manipulations and pharmacologi-
PROCEEDINGS 103 cal manipulations, but it is toxicology that has real significance for all of us in the world. Many of these toxicants are pro-oxidants. So we started working on them. We started with methylmercury, for which there is an extensive literature. This literature says that astrocytes and neurons have an LD-50 for methylmercury of about 500 nanomolar, equivalent to 100 parts per billion. This is a level of methylmercury exposure that would occur only rarely. When we went to study progenitor cells, we found that the ideas of vulnerability of cells of the CNS were entirely off base, and that progenitor cells are vulnerable to methylmercury at concentrations of 20â30 nanomolar, that is, from 4 to 6 parts per billion. It currently appears that all the progenitor cells that we look at are vulnerable to things like methylmercury and thimerosal and other things at these exposure levels. What we found, which was recently published in PLoS Biology, is that environmentally relevant levels of toxicants make cells more oxidized precisely in the range of relevance to our work on development and on cell-extrinsic signaling molecules. And, just as we would predict, progenitor cells exposed to such pro-oxidants drop out of division and become more vulnerable to other physiological stressors. What we did next was to take apart signaling in the cell from the nucleus back to the receptor to understand the mechanistic basis by which toxicants disrupt normal progenitor cell function. What we discovered when we looked first at the PDGF pathway was that the effects were absolutely confusing, as our signaling colleagues told us they would be and the literature told us they would be, which was that everything was suppressedâNF kappa B-mediated transcription, serum response element-mediated transcription, Erk phosphorylation, AKT phosphorylation, and phosphorylation of PDGF receptor. We next were fortunate to choose the right control experiments to carry out, which was to look at the effects of methylmercury exposure on signaling pathway activation by other ligands. What we next asked was what happens to neurotropin-3 (NT-3) signaling in these cells. The answer was that environmental toxicants had no effect whatsoever. So that suggested that there is specificity in these changes and that specificity resides at the receptor level. By looking at the PDGF signaling pathway and the NT-3 signaling pathway, we found a novel regulatory pathway that appears to be a
104 AUTISM AND THE ENVIRONMENT convergent point for multiple chemically diverse toxicants. What happens when a cell becomes oxidized, apparently regardless of the oxidizing agent that is chosen, is that the cell activates Fyn kinase, an enzyme that is a member of the src family of kinases. Activated Fyn then activates a ubiquitin ligase called c-Cbl. c-Cbl is a negative regulator of receptor signaling for some receptor tyrosine kinases. What happens when you activate this pathway is that c-Cbl attaches ubiquitin to the receptor for platelet-derived growth factor, the EGF receptor, C-Met, and some other receptors of interest, and they become degraded more rapidly. As a consequence, you suppress all the downstream signaling because you donât have as many of the receptors anymore. Methylmercury has no effect on trkC (the receptor for NT-3) because this receptor is not a c-Cbl target. In fact, receptors that are not c-Cbl targets completely appear to be unaffected by activation of the re- dox/Fyn/c-Cbl pathway. Our current studies demonstrate that multiple substances with pro- oxidant activity converge on Fyn activation, including multiple toxicants, multiple chemotherapeutic agents, ethanol, thyroid hormone, and other agents. This convergence may be due to activation of Fyn kinase by oxidized glutathione. So if you increase oxidized glutathione, you activate this pathway. We invested many years in developing in vitro systems that mimic what happens in the animal, so that we could use in vitro studies to accurately predict in vivo outcomes. Indeed, that is the case for our studies on toxicology. If we expose developing rats to levels of methylmercury in ranges as low as 100 parts per billion in the maternal drinking water, levels that are 10 percent or less of what other people mostly study, what we see in vivo is that cerebellum levels of PDGF receptor are decreased, levels of epidermal growth factor receptor are decreased, and levels of neurotro- pin-3 receptor are not affected. The same is true in the hippocampus. Consistent with a lot of our other work and redox regulation in respect to development, the cortex is unaffected, just as we would predict. It is very interesting to us that the cerebellum is a target in so many toxicant exposure paradigms, and is of interest of course in autism. We next want to understand why do some people have an outcome that is bad and other people have no outcome at all that you can see when they are exposed to similar levels of environmental toxicants. So we got very interested in strain differences. The literature taught us that SJL
PROCEEDINGS 105 animals are responsive to organic mercurials and CBA animals are not. The literature also says that this is because of immune system problems. We were hoping that maybe that wasnât the case, so we purified progenitor cells and asked at the single cell level what happens. It turns out that progenitor cells from the corpus callosum of SJL animals are much more vulnerable to anything that we throw at them. For example, here we are looking at cell division in the corpus callosum following a clinical exposure of thimerosal. You have reduced division in the corpus callosum and subventricular zones of SJL animals, but not in CBA animals. We have similar effects in tissue cultures of progenitors from these animals, where cells are more vulnerable in a strain- dependent manner. Our current hypothesis is that SJL cells are more vulnerable because they are more oxidized, and indeed they are more oxidized. As we look at other strains of mice, what we are finding is that if the cells are derived from the animals that are more oxidized, they are more vulnerable, and if they are derived from animals that are more reduced, they are more resistant. We next asked whether, if you have a genetically more oxidized animal, whether this Fyn/Cbl pathway itself is more activated just because you are genetically more oxidized. That would be the outcome of our predictions, and indeed, it is what happens. If we look in the central nervous system of SJL mice, they have lower levels of PDGF receptor than we find in CBAs. And they have lower levels of the C-Met receptor. So it looks like being genetically more oxidized also activates this pathway. Thus, we are currently thinking that being more oxidized, which occurs frequently in children with autism spectrum disorders, is a marker (and potentially a mechanism) of vulnerability that needs to be more closely studied. Finally, because I mentioned thimerosal, I want to be explicit about what I think our thimerosal results say at this point in time. It is actually quite remarkable to find how little data exists on thimerosal toxicity. In our studies, we do not see any difference between thimerosal and methylmercury. Thimerosal is as toxic as methylmercury, but it also is not more toxic than methylmercury. Based upon these biological discoveries, it is appears that it is not wrong to be concerned about the possible contributions of thimerosal to neurological syndromes. But I would say that such a concern needs to be studied in the context of the idea that there is a subpopulation of children more vulnerable to the
106 AUTISM AND THE ENVIRONMENT effects of exposure to potentially toxic substances, thus, if there is something particularly interesting about thimerosal. I think what it may be is that you have susceptible individuals, and the probability of a susceptible individual being exposed to a high level of thimerosal would have been higher than if he or she might be exposed to methylmercury or lead simply because it was used in so many vaccines in the 1990s. That is what would make thimerosal unique, rather than its chemistry. At a cellular level, it is important to stress that all the environmental toxicants that make cells more oxidized are something to be concerned about as potential disruptors of normal nervous system development. Dr. Levitt: We have time for a few clarifying questions if anyone has any. Dr. Hertz-Picciotto: I am just wondering if you might comment on how these results might apply to the very consistent finding of Purkinje cell loss that has been seen in so many studies of autism. Dr. Noble: We have worked out the effects of early mercury exposure in SJL versus CBA animals. In CBAs we see no effect on Purkinje cells; in SJLs it is a disaster. Cell membrane is reduced, they are out of position, the carburization is compromised. So yes, we see a very strong correlation. Dr. Levitt: Any other clarifying questions? We have a lot of general discussion time. Dr. Pessah: I just have a question about the last 30 years of neuroscience and the dish. Iâm not sure that even 10 percent of those that study cells in the dish, including progenitor cell progression, are not using 20 percent oxygen in their incubator. So essentially everything is underoxidized when you make the measurement. Dr. Noble: Not everyone, and not us. The question, so that everybody knows what we are talking about, when most people do tissue culture experiments they are growing cells in atmospheric oxygen. The baby is exposed to an oxygen concentration, depending upon which papers you believe, of somewhere between 3 percent and 5 percent. Atmospheric oxygen is 21 percent. That would oxidize cells and make them do strange things. We have looked at this extensively. What we find is that for most parameters that we study, the cells donât actually change their behavior between 5 percent and 21 percent. We continue to run them at 5 percent because we believe they should change their behavior, but it looks as thoughâif one wants to talk about this on an evolutionary basis, the idea
PROCEEDINGS 107 would be that there is no selective pressure in the brain to respond to 21 percent oxygen. It never sees it. It is not a relevant situation. In contrast, the difference between exposing cells to 5 percent and 1 percent, a hypoxic insult, is enormous. So we know we can change cell function by changing oxygen tension, but remarkably, the cells have been very, very resistant to 21 percent, even though we continue to run everything at 5 percent. Dr. Levitt: Thanks very much, Mark. We have one more presenta- tion before the break. Theodore Slotkin is going to talk about animal models, to take us from cells to animals. Dr. Slotkin is professor of pharmacology and cancer biology at Duke. HOW MAY ANIMAL MODELS BE USED TO EXAMINE POTENTIAL ENVIRONMENTAL-BASED MECHANISMS?12 Dr. Theodore Slotkin Dr. Slotkin: When we are dealing with autism, humans are the animal models of choice. If we could draw a connection between specific developmental neurotoxicants and ASD or autism per se, what we would like to be able to do is to use animals to define prototypes, that is, types of compounds that define entire families that attack the developing brain in the appropriate manner, that would then enable us to guide human investigations as to what we might look for in human populations with ASD as potential causative links. Then in the reverse direction, we could take things that people had noticed from clinical or epidemiological studies and then do the animal studies that could potentially prove cause and effect. The thimerosal story would be traditionally one of those where you go from suspected human exposures back into the animal. But the real impediment is that there actually is no animal model that gives us a complete picture of ASD. Therefore, what we have to concentrate on at this stage is to identify the mechanisms that contribute to adverse neurodevelopmental outcomes, which may include ASD, but are not restricted just to ASD, to point out the types of chemicals we might want to be concerned about and thereby guide future clinical 12 Throughout Dr. Slotkinâs presentation, he may refer to slides that can be found online at http://www.iom.edu/?id=42466.
108 AUTISM AND THE ENVIRONMENT investigations. We are all awash in a sea of tens of thousands of neuroactive chemicals. That raises some very serious questions that I think animal models help us address. That is, first of all, why is it that we adults are awash in them and our brains arenât permanently affected, whereas the developing organism is? What is it about development that renders neuroactive agents capable of producing permanent alterations? Why is there a critical period for it? Why is it that exposure before or after a certain stage doesnât do what exposure during a critical period does? A key question was just addressed in the previous talk: Why do apparently unrelated agents produce similar outcomes? I am going to show you that that is not just restricted to the issue of oxidative status inside the cells, but there are very cogent reasons for a critical exposure period, potentially even more important than what that compound is. Then finally, I am going to show you an example of this trade-off of how animal studies can help guide human studies for studies of autism using an example of work we were doing that we originally thought was totally on a different topic, using the drug terbutaline and its action on beta-adrenergic receptors. We will take those in sequence. Why is it that development is special and the developing brain especially vulnerable? There is a good reason for that. One of the purposes of the developing brain is to assemble itself and learn. For example, when you are young you can learn multiple languages very easily, and after that it becomes much more difficult. Cells are the same way. They have a specific period in which they use their inputs in order to learn. In other words, the brain has a certain degree of hard-wired development, but superimposed on that is input from the environment, which influences the subtle but important connections that define us as individuals and that ultimately impact on all our important behaviors. One of the ways in which environmental input is transduced into a change in the development of neural cells is the release and actions of the same small molecules that communicate ordinarily across the synapse, that is, neurotransmitters. But during development, neurotransmitters arenât just neurotransmitters. They are trophins that influence the fate of their target cells. They do so through the very same receptors that our adult brains use for ordinary synaptic communication, and the same signaling cascades. The difference is that during development, the actions of these molecules change the repertoire of genes that are read
PROCEEDINGS 109 out of the nucleus in the target cell, and depending on the stage of development of the cell, those genes might be involved in the control of cell division or differentiation or growth or apoptosis, since programmed cell death is a part of the modeling of the brain, or what we will be concerned about hereâlearning. There is a specific period in which input to a developing neuron teaches the neuron how to respond to specific inputs, both during development and for the future, life-long function of that particular circuit. The critical thing about this is that the same neurotransmitter and the same receptors and the same signaling cascade can do all of these things in a given cell, depending on the stage of development of the cell. And each cell doesnât just have one neurotransmitter, it has multiple ones. So what that says is that if cells learn during a critical period, learn in the same way we learn language but at the cellular level, then when you provide an input during that period, you change the fate of the cell, and that changes what it is going to do permanently. Whereas, for our mature brains, input after the critical period just produces short-term responses, and if we continue to try to elicit the response, we become desensitized to that particular inputâit gets ignored in the same way that a continual sound eventually disappears from active perception. It is subject to short- term, reversible compensatory adjustments. Let me show you that as an illustration of a presynaptic nerve terminal and its postsynaptic target cell in the mature brain. If you reduce the amount of neurotransmitter in the synaptic cleft, this cell will respond by increasing the number of receptors and augmenting the ability of those receptors to produce a response. So you have small input, the cell is sensitized so as to maintain the response; the reduced input is offset by enhanced responsiveness. If you overstimulate this cell and make a lot of neurotransmitter appear in the synapse, then the cell will desensitize. It will lose its receptors from the surface and it will uncouple them from the ability to elicit a response inside the cell. So this is a negative feedback loop. A small amount of input results in a boost in the ability to elicit a response so that you can make that synapse work. Stimulate the synapse too much, and the postsynaptic site shuts off and terminates the signal. In contrast, during development, because these synapses are learning, that relationship is reversed. A lot of stimulation during development promotes the development of the appropriate receptor for that neuro-
110 AUTISM AND THE ENVIRONMENT transmitter, and therefore augments the response. This is a positive feedback, so that is a good thing for learning. A little bit of stimulation teaches. The good side is that this is what enables you to learn, or as I will now illustrate, even survive getting born. For this illustration, I am going to concentrate on peripheral responses, because I am going to come back to this at the end for messages for the future. Stress hormones, especially adrenaline, rise precipitously at birth, and if this were the adult you would expect to see desensitization of the responses for that transmitter. Instead, in the developing organism, that response increases during the period in which the stimulus is rising. Now, if I use terbutaline, a stimulatory drug that acts like adrenaline on its target receptorsâthe beta-adrenergic receptorsâin the adult that would desensitize the response, but in the newborn it instead sensitizes the response. It teaches the cell: Beta-receptor stimulation in the immature organism leads to increased expression of the receptors and increased coupling of receptors to responses. Now, what if we block the stimulus? If, for example, we destroy the nerves that are supplying that response, then the cells will never learn how to respond to this particular input. In this illustration, you can see the normal response in rats at ages ranging from adolescence to adulthood: the early denial of input results in permanent loss of responsiveness. In other words, there is a critical period in which you must teach those cells what their fate is going to beâthis is learning at the cellular level. The same principles operate in the developing brain. They are universal for all neural circuits, not just peripheral systems. This inputâ output relationship originated very early in evolution and in fact, they can be seen in lower organisms such as sea urchins and bacteria, where they use the same molecules and receptors as a way of controlling their own growth and development. Thus, if neural input programs response development, that means there is always a critical period in which any disruption of input, positive or negative, will permanently alter function. If you send the wrong signal, it is going to produce the wrong outcome. It is going to teach the cell to respond incorrectly to the wrong kind of input. It will make that change permanently. What kinds of things will do that? Any drug or chemical that is neuroactive works ultimately by reinforcing or blocking the actions of a neurotransmitter. This will mean that exposure to drugs of abuse
PROCEEDINGS 111 or therapeutic agents that are neuroactive, environmental contaminants, whether they are organometals, which influence things like signaling and oxidative stress, insecticides, all of these can lead to misprogramming of responsiveness. Let me give you an example of this with terbutaline and its potential connection to ASD. Terbutaline is a drug that is commonly misused in preterm labor. It is a beta-adrenergic stimulant, so it works the way adrenaline does. It inhibits uterine contractions, but it is effective only for about 48 to 72 hours in doing that. Then as you would expect, in the adult it desensitizes the response so that preterm delivery proceeds anyway. Accordingly, it is not supposed to be used for maintenance tocolysis, but nevertheless it still is frequently used for that purpose. We were doing studies on this agent in the 1980s and 1990s, primarily to study receptor regulation. As predicted, terbutaline produces permanent changes in responsiveness because it acts just like overstimu- lating the nerves supplying those receptors. Coincidentally with that, there were some clinical and epidemiologi- cal findings indicating that maintenance tocolysis with terbutaline was resulting in adverse neuropsychiatric outcomes. In the last decade, we performed definitive studies to show that terbutaline is indeed a developmental neurotoxicant that affects specific targets in the brain, notably cerebellar Purkinje cells, areas of the hippocampus and cerebral cortex. For our purposes, though, the important finding is that terbutaline treatment in rodents shares morpho- logical and behavioral characteristics that are found in autism. It does involve things like oxidative stress and neuroinflammation. There is a critical period corresponding to the second trimester of brain develop- ment as it would be in the human fetus. There are also peripheral changes in cardiovascular reactivity and other peripheral functions that are similar to those that are reported to act in autism. Consequent to our studies in animals, which were not originally conceived with a relationship to autism, a study done by Dr. Andrew Zimmermanâs group and others pursued the connection between the use of terbutaline in preterm labor and the incidence of autism. They found that if you gave terbutaline for 2 weeks or longer, there was an increased risk of autism spectrum disorders. If you superimposed that on a receptor polymorphism for the beta-2-adrenergic receptor that prevents the receptor from desensitizing, the risk becomes much higher. They concluded that prenatal overstimulation of beta-adrenergic
112 AUTISM AND THE ENVIRONMENT receptors by terbutaline by itself or in combination with these genetic polymorphisms were responsible for autism in this particular cohort. What I want to point out is that that is an example of how animal models can lead to unsuspected mechanisms that can be responsible for neurodevelopmental disorders, an approach that is not strictly limited to autism. How does this point out a path for future research efforts? First, rather than relying on accidental connections, the way our two laborato- ries did for the case of terbutaline, there is a lot of information available on developmental neurotoxicants from the animal literature and databases that could point to retrospective examinations relating these exposures to ASD. On a population level, even GIS information might prove useful for examining how having been brought up in an area where, for example, organophosphate pesticides were used heavily, given that these agents are suspected as contributors to the increased incidence of ASD. Terbutaline and organophosphates are potential examples of a rational approach to using existing findings to search out new connections between exposure to chemicals that may be contributing to the rise in ASD. As we just heard, agents that produce oxidative stress are also good candidates for this database or literature approach to trying to draw mechanistic connections from existing information. Our study showed that the results from animals can be used to trigger the comparable studies of human populations for exposures and outcomes. I think there is an added value from examining an outcome where you can readily document the exposure. Terbutaline is different from most environmental exposures in that regard because the exposure is documented in an individualâs medical records. But there are certainly many other compounds or classes of compounds that can provide similar types of leads for future investigation. The basic problem is that we have potentially as many as 50,000 chemicals we have to screen for developmental neurotoxicity, so we are not going to be able to proceed one at a time, or at least not in animal models. We are going to need high-throughput models. As just one approach, cell culture models for neurodevelopment are plentiful and easy to use. We can then use all of the armamentarium of siRNA or humanized cells to explore the role of specific genes or genetic polymor- phisms that are found in autistic human populations, and insert them into cells and see what they do to vulnerability to developmental neurotoxi-
PROCEEDINGS 113 cants. There are efforts going on at NIEHS as well as in our laboratories and elsewhere to use lower organisms as ways of screening for these. As I pointed out, the peripheral surrogates that are predicted by common outcomes for autonomic input from the central nervous system to the periphery might then be an additional role for future use of studies. To sum up, the big impediment here is that there are too many candidate molecules to study, too little time, and too little money, and consequently designing high-throughput screens for developmental neurotoxicity is a mandatory first step to drawing connections between environmental chemical exposures and ASD as well as other neurodevel- opmental disorders. Dr. Noble: I would just like to reinforce that. Our work shows that it is trivial to screen chemicals. With these high-sensitivity systems that we have, we have multiple parameters now that let us analyze things as fast as we can grow dishes of cells, frankly. It is very, very simple, so long as you use the right cells. It is very critical. If you use established cell lines it is pointless because they have mechanisms that protect them against these kinds of insults. You have to use early progenitor cells to get high sensitivity to outcomes. A lot of the biomarkers that are coming out from this look pretty intriguing, particularly when I look at the studies from Dr. James. The changes in metabolic components of oxidative state that are being reported in autism are very much like things that we are seeing. We also know how these proteomic markers that we are finding very reliably, some of which are completely independent of transcriptional changes. So when we see changes in receptor levels, there are no transcriptional changes at all. So I think we are building up between the transcriptional work, the protein work, the metabolic work, we are building up the tool kit. So we need to do this. Dr. Schwartz: In the human study that you showed us, you showed us a relative risk of 2.4 and 4.4, but what percentage of the cases was it associated with? Dr. Slotkin: That is not my study. Iâm just citing the literature. Iâm not familiar with the details of it. Dr. Schwartz: And in your own studies, did you see any strain differences in responsiveness? Dr. Slotkin: We were using outbred rats, so strain differences is a
114 AUTISM AND THE ENVIRONMENT moot point. Participant: It might be very naive, but are there any natural animal models of autism? And if not, why not? And if there are, wouldnât you expect epigenetic environmental factors to impact on that? Dr. Levitt: Why donât we save that for the general discussion? Our next speaker is Ira Lipkin. He is the director of the Greene Infectious Disease Laboratory and other major titles at Columbia University. He is going to be talking to us about infection and immunity in autism. AUTISM, INFECTION, AND IMMUNITY: WHAT ARE THE POTENTIAL CAUSATIVE ENVIRONMENTAL FACTORS AND HOW CAN THEY BE IDENTIFIED AND PRIORITIZED?13 Dr. W. Ian Lipkin Dr. Lipkin: My task is to describe technologies that might be applied to answer questions relating to infection and immunity in neurodevelopmental disorders. We like to think of these technologies as âpeace dividendsâ because they were developed with support from biodefense funding that came online after 9/11. To set the stage for consideration of how diagnostic and surveillance technologies can be implemented, I must first discuss the mechanisms by which infectious agents can cause disease. This introduction is critical because if infectious agents play a role in pathogenesis of autism spectrum disorders, common conceptions of mechanism may not apply. We typically think of infections as associated with acute illness at the sites where the agents replicate. Poliovirus, for example, is an enterovirus that causes gastrointestinal dysfunction in many people, meningitis in a smaller subset, and paralysis due to death of motor neurons in brain and spinal cord in a still smaller subset. Vibrio cholera infects the intestine to cause diarrhea known as âcholera.â However, less obvious relationships may also be important. Clostridium botulinum grows in the skin or the gastrointestinal tract, and elaborates a toxin that travels systemically to cause paralysis through a remote effect at the neuromuscular junction. There are also instances, such as infection with hepatitis B or hepatitis C, 13 Throughout Dr. Lipkinâs presentation, he may refer to slides that can be found online at http://www.iom.edu/?id=42467.
PROCEEDINGS 115 where the agent itself has only a modest direct effect; however, immune responses to infection and the associated inflammation results either in death of cells and organ failure, or neoplasia. Some microbes can cause immunosuppression, resulting in disease due to infection with other, frequently opportunistic organisms. In our era we think of HIV as the prototype for this scenario; however, virus-induced immunosuppression was first described by von Pirquet in the 1800s in the context of measles. Persistent viral infections can have subtle effects on differentiated cell functions such as the capacity to make an enzyme or a neurotransmitter. Although this mechanism has not yet been shown in humans, its plausibility has been demonstrated in animal models of persistent viral infection resulting in dementia, type 1 diabetes mellitus, and hypothyroidism. Molecular mimicry, a mechanism by which an antibody or T-cell mediated immune response against a microbe results in damage due to cross reactivity to a normal host component, was described by Sue Swedo this morning. We must also consider effects of infection that are locked into specific windows of vulnerability. This is particularly pertinent in autism where such windows have been defined vis-Ã -vis exposure to thalidomide. Ezra Susser highlighted related examples wherein risk of schizophrenia was increased by prenatal stressors, including infection. The hope, of course, is that discovery of agents that cause disease by any of these mechanisms will facilitate the development of preventive and/or therapeutic strategies that promote public health. A major recent advance along these lines was the development and approval of a vaccine for papillomaviruses that is anticipated to have a profound effect on the incidence of cervical cancer. Whether we will be able to achieve similar success in autism remains to be seen; however, I think that is highly unlikely. My view is that no single agent or group of agents will be implicated. Instead, to the extent that environmental factors are important in pathogenesis, we will discover that toxins, infection, and stressors of various types, can activate similar pathways to cause similar effects. Finally, although I have been asked to focus on the role of patho- gens, I want to emphasize that our working model is one with three dimensions where genetic susceptibility, environmental triggers, and temporal context act in concert to cause disease. Thus, a comprehensive investigation must address the intersection of all three components.
116 AUTISM AND THE ENVIRONMENT This slide from the New Yorker, published during the West Nile virus outbreak of 1999, shows medical staff examining a pharaoh in a hospital bed in classic garb. The point is that differential diagnosis of infectious disease is rarely this straightforward. Hence, we need laboratory diagnostics. The next several slides describe systems that together define a suite of diagnostic tools for use in differential diagnosis of infectious diseases. The first system is MassTag PCR, a multiplex PCR platform that allows us to rapidly query 20 to 30 different infectious agents simultane- ously. Throughput is rapid and assays are inexpensive. The second is the GreeneChip, an array that allows us to consider thousands of agents, but at lower throughput and higher cost. The third is metagenomic sequenc- ing. More time-consuming and expensive yet, however, extraordinarily powerful and uniquely suited to cataloging microflora and discovering new pathogens. MassTag PCR panels have been established for detecting the vast majority of infectious agents causing respiratory disease, hemorrhagic fever, meningitis, encephalitis, and diarrhea. In 2005, application of this method allowed us to discover a new rhinovirus. GreeneChips are glass slides, similar to the types of slides that many of you may use for transcript profiling, that are decorated with thousands of probes. Nucleic acids in samples are amplified by PCR, and applied to slides. Binding of amplification products and of a fluorescent label allows detection of microbes. Various formats have been developed for different applications: all viruses, respiratory pathogens, all known vertebrate pathogens. Analysis has been automated such that images from slides can be submitted from remote locations and analyzed. In this instance, we identified influenza A virus in a nasopharyngeal swab; more detailed analysis allowed us to determine that it is H5N1. Implementation of diagnostic technologies allows one to discover not only new pathogens associated with disease, but also known pathogens in asymptomatic individuals. In the context of a study of environmental triggers of type 1 diabetes mellitus in a Norwegian cohort, we recently examined stool samples from children using MassTag PCR and GreeneChips. To our surprise we found evidence of frequent infection with enterovirus subtype 71, a virus typically associated with paralytic illness in Asia. Genome sequencing and phylogenetic analysis revealed mutations in the Norwegian viruses that we predict impair replication and ability to cause disease.
PROCEEDINGS 117 Our most expensive and labor-intensive platform is metagenomic sequencing. Although applications of this platform in whole-genome sequencing are well described, its use in pathogen discovery is relatively recent. In essence, thousands of short sequences are aligned and assembled into continuous strings. Host sequences are subtracted, yielding a series of candidates that represent potential pathogens. Candidate sequences are analyzed for similarity at the nucleotide level and at the protein level for viruses, bacteria, parasites, or fungi. Metagenomic sequencing can be employed in real time. Here is an example wherein we investigated a cluster of deaths in organ recipients linked to a single donor. No agent was identified through classical methods such as culture, serology, or PCR. Thus, tissue samples were referred to us for study. After failing with MassTag PCR and GreeneChips, we moved to metagenomic sequencing. Analysis of 140,000 sequences led to detection of a novel arenavirus. This virus was subsequently implicated through specific PCR and serological investigations. And nowâÃ la Monty Pythonâfor something completely different. Several weeks ago, at another IOM conference, I met Diana Cox-Foster, an entomologist and microbiologist studying Colony Collapse Disorder (CCD). This is an extraordinary phenomenon wherein honey bees inexplicably leave their hives and donât return. Given the importance of pollination for agriculture, we began using metagenomic sequencing to examine the microflora of bees with and without CCD. This work yielded intriguing leads that may provide insights into CCD; however, the point I want to make is that metagenomic sequencing allows one to simultaneously define complex microflora in a sample: bacteria, viruses, parasites, and fungi. We have been asked to comment on what resources might be helpful in addressing the role of the environment in autism pathogenesis. In this spirit I would like to note the Autism Birth Cohort (ABC), a prospective birth cohort based in Norway of 100,000 children and their parents, that collects biological samples and clinical data beginning at the 17th week of gestation. ABC collections, joined with the diagnostic platforms for microbiology, toxicology, and genetics, will enable new strategies for examining geneâenvironmentâtiming interactions in health and disease. Those of you interested in learning more about the ABC may wish to visit its website at www.abc.columbia.edu.
118 AUTISM AND THE ENVIRONMENT In summary, microbial pathogenesis is complex. Susceptibility is a function of genes, age, and other factors. Mechanisms can be direct or indirect. Expression of disease may be delayed. The microbiome is largely uncharted; however, new tools for microbial surveillance will change the landscape for understanding chronic as well as acute diseases. I will close with this wonderful quote from Einstein. In the period that he was active as a professor, a student remarked, âThe questions on this yearâs exam are the same as last year.â âTrue,â Einstein said, âbut this year all the answers are different.â With new models and new strategies for addressing them experimentally with clinical materials, we may shed new light on the role of environmental factors in the patho- genesis of autism and related disorders. Dr. Levitt: We have time for some questions. Dr. Leshner: You went through a slide very suddenly and then got off it, and I canât remember my question, but I can remember what was on the slide. Dr. Lipkin: Neurodevelopmental disorder? Is that the slide you wanted? Dr. Leshner: No, one more slide. Dr. Lipkin: The next slide is this. This is the slide you want. Dr. Leshner: This is an incredibly gigantic project which you skipped over rather rapidly. Dr. Lipkin: That is because I was told I didnât have time. Dr. Leshner: I know. What is this? Who is doing it? Who is paying for it? 100,000 children? Dr. Lipkin: The National Institute of Neurological Disorders and Stroke (NINDS) is supporting it. It is called the Autism Birth Cohort. It is nested within something called the mother and child cohort. It is in Norway. There are several people who are working with this cohort at present: Ezra Susser, Deborah Hertz, Mady Hornig, Alan Wilcox, and then we have counterparts in Norway as well. This was conceived to do the same sorts of things that all the other birth cohorts are conceived to do, but because we were later to get started, we were able to focus more on proteomics and transcriptomics and viromics and any other -omic you want to think of, so that we could try to address these kinds of questions. Our vantage point was that it was going to be geneâenvironmentâ timing interactions. That is the principle that guides this study. Dr. Leshner: My question is, when do you start taking samples?
PROCEEDINGS 119 Dr. Lipkin: We started taking samples already. Dr. Leshner: No, I meant in development. Dr. Lipkin: The first visit, which often occurs as late as 17 to 18 weeksâ gestation. The mother has consented and blood is collected. Then there is blood collected from there on. So you can do serology and look for changes in titer and so forth. Dr. Levitt: I wanted to ask one technical question. The different platforms that you used, some are qualitative and some are quantitative. Dr. Lipkin: Those are all qualitative. They are purely surveillance tools. You have to follow them, using real time or an equival. Dr. Levitt: So presenceâabsence. Dr. Lipkin: Correct. Participant: I had a question about the developmental disorders and children with autism. I think there is a deeper question here. Children with developmental disorders and also autism had more infections from multiple pathogens, but what about the effect of the pesticides or other neurotoxicants on responses? Dr. Lipkin: I have no clue. Dr. Levitt: We are going to talk about that toward the end during the discussion time. I think it is going to come up. Dr. Schwartz: Ian, I had just one clarifying question for you. Just to be totally clear here, your microbiome is part of this 100,000 cohort study? Dr. Lipkin: Yes. Dr. Schwartz: And how many patients are you doing it on, what time points? Dr. Lipkin: At present, all we have right now is the tool kit that was built for detecting West Nile virus and avian flu. But the remainder, which is the collection of these samples and so forth, that is supported, and those materials have been collected. In fact, the first samples have only recently come out of Norway to Columbia for analysis, and those are cord bloods. So we havenât done anything with those samples as yet except some of the stool samples, which I showed you. We just found a wide variety of viruses. Dr. Levitt: Thanks very much, Ian, thatâs great. Our next speaker is Jill James, who is professor of pediatrics at the University of Arkansas School of Medical Sciences.
120 AUTISM AND THE ENVIRONMENT ENVIRONMENTAL FACTORS AND OXIDATIVE STRESS: HOW MAY OXIDATIVE STRESS IMPACT THE BIOLOGY OF AUTISM? WHAT FACTORS MAY BE CAUSING THIS OUTCOME?14 Dr. S. Jill James Dr. James: For the next 15 minutes, I would like to explore with you pro-oxidant environmental exposures and the possible implications of redox imbalance in autism. Let me begin by explaining that redox imbalance is actually a relative term. In fact, it is a continuum from subtle shifts in redox homeostasis up to more severe imbalance that is associated with oxidative stress and pathology. Most reactive oxygen and nitrogen species are generated endoge- nously from normal oxidative metabolism. They are also generated exogenously, and most relevant to our discussion today are pro-oxidant environmental exposures. Multiple or chronic exposures to pro-oxidant environmental toxicants can sustain redox imbalance and lead to oxidative damage and promote complex disease. Counteracting these sources of oxidative stress is a wide variety of antioxidant defense mechanisms. Chief among these is glutathione. The ratio of the reduced active form of GSH to its inactive oxidized form, GSSG, is considered to be the best indicator of intracellular redox status. A decrease in GSH and/or an increase is GSSG will negatively affect intracellular redox homeostasis. Most research in oxidative stress is focused on the damaging effects, but I would like to make a point that is less commonly appreciated, and that is that the small subtle shifts in redox balance are in fact beneficial and represent essential signal mechanisms for normal cell function, as Dr. Nobel has elegantly demonstrated. Redox signaling is important for cell cycle status and for the activity of a multitude of redox-sensitive enzymes. The redox status of the cysteine in the active site will activate or inactivate these enzymes. In addition, gene expression, transcription factor binding, and chromatin remodeling are affected by small changes in redox status as well as the activation of the innate immune system and the inflammatory response. 14 Throughout Dr. Jamesâs presentation, she may refer to slides that can be found online at http://www.iom.edu/?id=42484.
PROCEEDINGS 121 On the other hand, chronic or severe shifts in the redox ratio can be irreversible and promote a self-perpetuating cycle of oxidative stress and damage. These include glutamate toxicity, inhibition of redox-sensitive enzymes, protein misfolding, mitochondrial dysfunction, and cell death. These more severe pathologic changes can lead to accelerated aging and contribute to the pathogenesis of complex diseases, particularly neurologic disease. I think an important point for us to consider in the discussion today, and Dr. Nobel has referred to this as well, is that we are exposed to a wide variety of structurally different chemicals in the environment which include multiple metals, solvents, pesticides, industrial chemicals. Importantly, although they are all very different structurally and chemically, they all share a common mechanism of action, which is to induce oxidative stress and deplete glutathione. They are all pro-oxidant. That suggests that there may be a common molecular mechanism of toxicity underlying these very different and diverse chemicals present in our environment. A common mechanism of action would certainly simplify the search for an association between autism and the environment. Most of these chemicals, their safe levels and toxic levels, are analyzed individually, but we all know we are exposed to complex mixtures. Multiple simultaneous pro-oxidant exposures are additive and can even be synergistic in toxicity. This implies a very important point in toxicology: multiple subtoxic exposures can become toxic when they are combined. Glutathione is not only the major intracellular antioxidant; it has important detoxification functions as well. Heavy metals have a high affinity for the sulfhydryl group on glutathione and bind spontaneously. There is a wide family of glutathione S-transferases that will enzymati- cally create water-soluble glutathione conjugates that are then metabo- lized and excreted in the bile and the urine. So glutathione is not only the bodyâs major natural chelator, it is a major mechanism for the elimina- tion of many of these environmental toxicants. Resistance or vulnerability to pro-oxidant environmental exposures depends largely on intracellular glutathione levels. Depleted glutathione reserves will increase sensitivity to pro-oxidant exposures. With very robust glutathione reserves and a high GSH/GSSG ratio, toxic insults will be buffered and will never reach a toxic threshold. On the other hand, with fragile or depleted glutathione reserves, the same toxic insult
122 AUTISM AND THE ENVIRONMENT can precipitate toxicity and pathology. We have recently shown that autistic children have a lower glutathione/redox ratio. We and several other investigators have started to identify metabolic biomarkers that suggest that many of these children may be under chronic oxidative stress. This is a flow diagram of the metabolic pathway that we have found to be abnormal in many autistic children. Many children have low levels of methionine and its product, S-adenosylmethionine, which are essential precursors for cellular methylation reactions. These metabolites also lead to the synthesis of cysteine, the rate-limiting amino acid for glutathione synthesis. We find that the active reduced glutathione levels are decreased and the oxidized form, GSSG, is increased in many autistic children. The only reason that GSSG would be increased in the plasma is that it is being exported from cells under chronic oxidative stress as an attempt to normalize the intracellular redox environment inside the cell. These metabolic pathways are important not only for redox homeostasis and methylation, but are also essential for error-free DNA synthesis and cell proliferation. So perturbation of these metabolic pathways would negatively affect normal immune function, methylation, redox homeostasis, and clearly affect normal development as well. We questioned then whether there could be a genetic basis for the metabolic imbalanceâincreased vulnerability to oxidative stress in autistic children. We are using a targeted approach to autism genetics because we have a phenotype. We are using this metabolic phenotype as a guide to the selection of candidate genes. We plan to evaluate more than 30 genes that affect this pathway to see if some of these genetic variants are increased in autistic individuals compared to controls. So far we have found an increase in the genetic variant for the reduced folate carrier (RFC1) that regulates folate transport into the cell and would be expected to negatively affect this pathway. Because these genes may affect the synthesis of glutathione, they may be particularly important candidates for geneâenvironment interactions. We think that the evaluation of multiple polymorphisms that affect a common pathway may provide clues and a plausible explanation for the redox-vulnerable phenotype that we see in autistic children. A purely genetic approach to autism is a daunting challenge. As you know, it has been estimated up to 100 different genes may be required for the phenotype. Beyond that, there may be different combinations of genes in different autistic individuals. Then if there is a genetic
PROCEEDINGS 123 susceptibility that requires an environmental trigger, these same genetic risk factors may be present also in unaffected controls, and that is going to really confound the search for autism-relevant genes. And of course, genetics does not encompass the timing or the severity of the environ- mental exposures or the heterogeneity that is autism. Many of us have started looking beyond the brain. Martha Herbert has eloquently suggested this topic. We are looking more at what we call the gutâbrainâimmune axis. We now know that these three systems do not function in isolation. They are all mutually interdependent; they all talk to each other. We know that all three systems are highly sensitive to oxidative stress, particularly during critical developmental windows. All three systems are developmentally immature at birth. They require environmental cues to develop normally. So the developmental trajectories of all three systems depend on appropriate signals, and an inappropriate signal could derail normal development. This implies that you could have a toxic insult to any one of the three systems; it is going to affect the developmental trajectory and the function of the other two. I am trying to think more broadly here. There are many neurotoxins that are known to equally impact the gut and the immune system. The two that come to mind are mercury and lead. They negatively impact all three systems. A healthy brain needs to develop in the context of a healthy immune system and a healthy gut. This brings us to new questions. Do we need a broader paradigm for pathogenesis, a more systemic approach beyond the brain, and could there be a component of metabolic encephalopathy that could be treatable? We think that the oxidative stress hypothesis that we are pursuing at least encompasses the possibility of the gutâbrainâimmune interaction and geneâenvironment interactions as well. How do we get from epidemiology to mechanism, which is where we all want to be? We know that the genetic background clearly affects the vulnerability and resistance to environmental insults. We know the environment will alter gene expression. But we have multiple additive and variable genes as well as multiple and additive and variable environmental factors that make this link to behavior quite tenuous. They are both necessary, but neither is independ- ently sufficient. So what if we interject a metabolic endophenotype? This can lead us closer to relevant genes. It might lead us closer to and give us clues to relevant environmental factors, and could lead us possibly to a mecha-
124 AUTISM AND THE ENVIRONMENT nism. In our case it would be redox imbalance or methylation dysregula- tion that might be more closely related to behavioral abnormalities. Equally important, I think it gives us treatment targets and raises the question, if we can normalize the metabolic imbalance that we see, can we affect behavior? So for our future research agenda, I would like to suggest that a targeted metabolic signature in fact is an integrated reflection of genes, environment, and nutrition on a relevant pathway of interest. Single metabolites lack the context that we need and global metabolomic approaches are not yet mature. We think that this targeted approach can provide metabolic context as well as insights possibly into molecular mechanisms of the disease, candidate genes, and also treatment targets. For our recommendations, we would suggest focusing on candidate metabolic pathways to provide clues for environmentally relevant candidate genes, not only in the childrenâand I think so many of the answers are in the children as wellâbut we can look also into animal models and apply this to cell models to look for relevant signaling pathways or metabolic pathways, including redox, detox, immunologic, mitochondrial to name a few. We can also use metabolic biomarkers as targets for treatment strategies and treatment efficacy. I also think we need to invest in the children who already have autism. The treatment options for these children are very limited. We need to invest in placebo-controlled, double-blind studies to try to advance standard of care for these children. Within these trials, I think it is very important not just to look to see if we have a mean difference between groups, but look at which children responded and then characterize those children very carefully genetically and metabolically. Also, equally important are the nonresponders or the negative respond- ers, so that we can begin to individualize treatment for these children because the parents will tell you, each child is an individual case. I think it is also important to do comparative studies to look for differences in CSF plasma and urine to try to differentiate central from peripheral differences. Quantitative difference in metabolic patterns may be able to distinguish subpopulations within the autism spectrum. We also need predictive biomarkers. Examples would be to evaluate high-risk children that present with developmental delay because that is usually how the parents bring the children inâwith speech or develop- mental delay. This is a high-risk population as well as siblings and discordant twins. If we can come up with predictive biomarkers that
PROCEEDINGS 125 would go along with the behaviorâagain, they are not going to be specific, but if you have behavior plus a biochemical biomarker, that would be a huge step forward, and would suggest possible targeted intervention strategies. For infrastructure, we would need many different high-tech analytical instruments and a repository or bank of biologic samples. I would like to point out, this repository or bank of samples is a trickier proposition than it might seem. For our samples, we get them fasting, the same time of day, and they have to be analyzed within a certain amount of time, because a lot of these metabolites are unstable. So the sample preparation, the sample storage, has to be very carefully monitored if we are going to have a sample bank or repository of biological samples from these children. Thank you very much. Dr. Levitt: Do we have a few questions of clarification before we open up for the general discussion? Dr. Noble: Jill, with the comments we have heard from parents today about children who have responded to chelation therapy, have you had any kids that you have actually been able to analyze before and after such therapies to see if these metallic profiles change? Dr. James: We would love to do that. I am not a physician. There are trials now going on with chelation. I think that would be a natural thing to do, to look to see if they were under more oxidative stress, and when you pull out these different heavy metals, does that improve? That would be a fascinating question, but we are not doing it. Dr. Newschaffer: I just want to ask real quickly, the model you put up where you showed that the gutâbrainâimmune systems are all immature at birth and are all developmentally susceptible, are they the only three, or were there others? Dr. James: The reason we picked those is because many autistic children have gut issues and immune issues. Dr. Newschaffer: No, I understand, but are there others that meet those criteria? Iâm just curious. Other systems, endocrine, cardiac? Are there other systems that meet those criteria, too? Dr. James: Oh, of course, yes. But we picked those specific to autism. Dr. Goldstein: Iâm just curious, are there data on children right now blinded, where you donât know whether they have autism or not, and you see different patterns? If you took 100 samples of children with autism
126 AUTISM AND THE ENVIRONMENT blind and 100 children who were developmentally typical, could you pick them out blinded? Dr. James: Actually, that is how I got into this field in the begin- ning. We were doing a study of children with Down syndrome, and we needed a control. Control children are very difficult to get. So I had this great idea that they could bring in their siblings, because for Down siblings are a fine control. One mother had twins, and one had Down and one had autism. When I looked at that n of 1, I couldnât tell which was which. I couldnât differentiate the control. It was very different. So our very first study was simply a follow-up on that n of 1. We now have done over 150 kids, and there is variation. We now do a metabolic screen before we are doing some intervention studies. I would say about 25 percent look fine. But what I look for is that methylation ratio and the glutathione ratio, and cysteine is low in I would say 70 percent of the kids in a larger sample. Dr. Alexander: Do you have any data showing the stability of this phenotype over time within the same individuals? Second, do you have any more evidence for the specificity vis-Ã -vis other nonautistic developmental disorders? Dr. James: Actually we donât know that, but I think when we do it at the same time of day, fasting, over a large number of kids, we see a lot of the same pattern. In our new study that we are doing, I am going to do two baselines, so we get an idea of the variation, because I donât know that. DISCUSSION Dr. Levitt: We have a lot of time for discussion. There are a number of issues that came up. This is a session that focused on environmental factors that impact fundamental biological processes, and many of us get caught up in trying to understand fundamental biological processes. In the context of trying to understand this, we often donât reflect upon the impact that it has, not just on the individual biological system, whether it is a metabolic pathway or a cell, but on the individual child and the family as well. We talked about this this morning a little bit in the context that there is a lot to be learned and gained from listening to families and parents and listening to clinicians, at least from a basic scientist perspective, in terms of clues.
PROCEEDINGS 127 I think some of that relates to what we were going to be discussing and talking about. One is trying to understand timing of these things that we have discussed. Why is it that there is such neurobiological specific- ity to this in the kinds of things that we have talked about today, which were general changes in metabolism, for example? Why is timing so important? Is there credibility for this concept of heavy genetic load, and what does that mean in the context of timing and specificity, and how does that relate then to the heterogeneity that we talked about quite a bit this morning, and that we heard about from Sue Swedo? So these are some difficult questions that we are going to need to grapple with. One of the things that I wanted to pose initially is, sometimes we measure things because they are easy to measure, and we donât measure things that are more difficult to measure. So we make associations and then we assume those are part and parcel of the etiology of whatever we are trying to understand. So I pose this question to the group, where there was a decided focus on metabolic pathways. There are lots of disorders that involve disruption of metabolic pathways. So while those disruptions and those measures may be valid, where is the selectivity and specificity in the application of this? I want to throw that out initially. Dr. Noble: I can have a go at synthesizing some of what I heard and some of what we are doing. I think those questions are actually the same question. When we talk about why is timing important, what most people say is, because something happens at a particular time. That is not a very satisfactory answer, because you want to understand the mechanistic underpinning of why timing is important. What it appears to us, in the context of what we have been discussing today, is that for progenitor cells, and we are beginning to get data also for differentiated neurons, at times in their development when they are making decisions, am I going to differentiate or not, or am I supposed to have a lot of dendritic growth or not, they become extremely sensitive to these changes in redox state. Their biology is very, very different than other times. They are buffered against redox changes. So it looks to us like that may have a mechanistic contribution. Dr. Herbert: You laid out a whole bunch of questions. What I am saying is not going to follow what Mark Noble said, but I just want to respond to one of your many points. In my own brain imaging work, I did a series of studies comparing
128 AUTISM AND THE ENVIRONMENT brain volumes in autism and developmental language disorder, otherwise known as specific language impairment. It was almost impossible for me to tell the difference between those two disorders according to brain volume, except that the autistic brain volume deviations from controls were somewhat more. But it was a question of degree and not a question of kind in just about every single measure that I did. So this raises some interesting questions that the design of my data gathering didnât allow me to address. There is literature in both autism and in language disorders about the role of immune abnormalities in those disorders. We had no immune data. Immune is related to a number of the other pathophysiological processes that have been discussed such as oxidative stress. We had no data like that. I personally am not clear that we can answer the kinds of questions posed by the kind of brain data I gathered without having data at other levels of pathophysiology; what is it that makes autism, autism and specific language impairment not so much as autism. Because there are children who when they lose their diagnosis have specific language impairment, and because I am given anecdotal evidence from some of the baby sibs data that a surprising number of the infants who were at risk for autism who did not develop autism developed language impairment, it makes me feel like there are some underlying relationships going on here whose mechanisms, if we were to explore themâand I believe to explore them we would need to be doing some of the work that we have talked about todayâwe need to do this sort of work. I agree also with Jill James about targeted metabolic studies, and Dr. James is much more of a biochemist than I am, and I am not an immunologist, either, but certainly focusing on environmentally sensitive pathways and so forth. If we had a way of giving brain imagers like me and other people who are not metabolically oriented researchers a way of collaborating without a huge activation energy with others who could provide a means to gather these other kind of data, we would have the interdisciplinary capability to answer why it is that something similar at one level is different at other levels. I think that we will need to do that. Dr. Schwartz: A general comment that I have that I want to pose to the folks who gave these presentations in this last session, which I thought were really fantastic: It seems to me like we got a little bit off course, in the sense that part of the intent of this session was to figure out what we know about the biology of autism and what that tells us about
PROCEEDINGS 129 etiology. So I want to pose that question to everyone who thought about it and who presented, as a way of trying to get at the question that we were trying to answer during this session. Dr. Levitt: So more specifically, is it what do we know about the neurobiology of autism? Dr. Schwartz: Precisely. Dr. Levitt: The neurobiology of autism. I want the responders to answer carefully. Dr. Schwartz: Right, as opposed to the behavioral biology of autism. I am talking about the neurobiology of autism. Dr. Noble: I can synthesize one part of this. What Martha did not talk about is her extremely interesting studies on myelination abnormali- ties in autism. Those myelination abnormalities and also what is reported in the literature, abnormalities in the latencies in the auditory brain stem responses, which are also indicative of myelination deficiencies, are precisely as we would predict from our developmental studies and from Jill Jamesâs work. They are absolutely precisely as we would predict. I can talk about that with you in detail, but I have been really struck by how close this concordance is. One of the implications of that in terms of the auditory brain stem response is that what happens is, you get a spreading of the latencies between peaks. The whole nervous system as we all know is based upon having highly synchronized information transfer. If you have a dismyeli- nation disorder in a nerve trunk, so that you get a spread of information transfer, you have signals being delivered at different times. What is described in the literature for iron deficiency in autism and in a variety of things that we have been studying is that you have the spread of interpeak distances that would be predicted from a lack of enough myelin. Moreover, we are now seeing as we analyze data in more detail that you have a spreading of each individual peak, which is again exactly what we predict. Some fibers are myelinated normally, others are not. What happens in the auditory system as it has been explained to me as a consequence of this is that phoneme parsing can become very difficult, which of course ties right into language acquisition. Dr. Schwartz: So what you are telling me is that neurobiology connects to the physiology, not to the etiology. Dr. Leshner: Can I just ask, are we moving out of this session a little bit?
130 AUTISM AND THE ENVIRONMENT Dr. Levitt: No. I think the question was, if we are trying to understand how environmental factors impact fundamental biological processes, it is in the context of what are the fundamental neurobiologi- cal or other organ system processes that are disrupted. There is evidence from imaging. I would characterize it differently. I donât think we really know whether it is dismyelination. I think we know there are some long tracks that are smaller and some long tracks that are larger, and we donât know the reason why they are smaller or larger. So that is number one. Dr. Herbert: I can clarify a little bit. But you finish your summary, and then I would like to say a couple of things. Dr. Levitt: Do you want to clarify the white-matter part? Dr. Herbert: Oh, yes, I could do the white-matter part. What my own particular work identified was an enlargement in the white-matter compartment in the outer white matter that we call the radiant white matter, that was most pronounced in the areas that myelinated latest. But using T1 weighted imaging, we had no way of deciding whether this is myelin or anything else that existed in that compartment. So studies are under way, including my own, of more multimodal imaging to characterize this. But I want to say that there is some interesting unpublished data, and I have some slides of it that I got from Carlos Pardo. He was inspired by my imaging localization to go back and stain brains in the distribution of gray matter, outer white matter, and deep white matter for the activated neuroglial, astroglial, and microglial cells, and found that there was a greater amount of astroglial activation in the area where I found increased white-matter volume, the outer area, not the inner area, with microglial activation in the cortex, which is not the same thing as saying reduced myelination. This is just a few brains. So there are a whole bunch of links here which havenât been replicated a lot, but it does raise the question there. You can have from that a lot of exciter toxicity or altered modulation of transmission in synaptic activity, which is another mechanism that would still get you to the same place in terms of disruption of signaling from the very earliest parts of signaling. So I would agree with your conclusions, but I would argue that there are a number of other ways of getting to that. Dr. Slotkin: We also have been collaborating with Carlos Pardo with our terbutaline model, and got exactly the same results. Dr. Herbert: In the mice?
PROCEEDINGS 131 Dr. Slotkin: In the rats, in the same pattern that you guys saw with the autism brains as well, identical. Dr. Herbert: The astrocytes in the outer white matter? Dr. Slotkin: Microglial activation in the white matter, cerebellum, and cerebral cortex. Dr. Levitt: What is the period of sensitivity? Dr. Slotkin: In the rat postnatal, two through five. Dr. Levitt: And outside of that they were resistant? Dr. Slotkin: They were resistant outside of that. It is a window that corresponds to the second trimester of human brain development. Dr. Levitt: Women donât get terbutaline duringâ Dr. Herbert: But they get other exposures. Dr. Slotkin: ACOG withdrew recognition of maintenance tocolysis with terbutaline in 1995, but it continues to be used. I will quote from a couple of people I spoke to: I use it and it works. Never mind that the placebo control trials show that it doesnât. But Carlos did exactly the same studies on our rats that you guys did, and got the same results. Dr. Herbert: But there is also probably a final common pathway in there, just to point out. Dr. Slotkin: Absolutely, overactivation of beta receptors during that period also causes oxidative stress. So I think what we are seeing, and we have all been alluding to this, what we are seeing is that a whole series of apparently unrelated insults that really differ in the way we think about mechanism in the classic sense nevertheless converge on a common set of final pathways that then produce the consequence. Then it becomes an issue of, can you feed into these common final pathways from fill in the blank, and is the timing right to cause damage to the parts of the brain that are most likely involved. Dr. Levitt: The other thing that I wanted to say about the neurobiol- ogy is that we actually can say what it isnât more than what it is. So in all the studies that looked at changes in cell numbers, for example, or neuronal numbers or changes in cyto architecture, the organization of different parts of the brain, there are no findings that I know of that show even moderate differences. There are reports of changes in individual neuronal structures here or there, but they are counterbalanced by reports in which there are no differences. But when you look at really careful stereological studies, this is a disorder not unlike other neuropsychiatric disorders, in which there
132 AUTISM AND THE ENVIRONMENT are not profound structural changes in the brains. I think it is fair to say that. Dr. Insel: There is only one stereological study. Dr. Levitt: Jeffrey Huchsler has published and David Amoral has published, so there are two. Dr. Insel: That is the sum total of the literature. Dr. Levitt: Listen, the cupboard is relatively bare in terms of this. The other thing to keep in mind that I always emphasize when I talk to students about this is that the neurochemical findings in the brain are all based on measurements that are done in, for the most part, adult brain. For any of us who do basic science, we know that the danger is there of trying to extrapolate a steady-state measure, irrespective of issues about samples or things like that, steady-state measures in the adult and trying to understand what happened developmentally. The third thing I want to state is that from a neurobiological perspective, studies that have nothing to do with autism talk about development and developmental trajectory, and have quite important findings to tell us now as developmental biologists that we need to pay attention not to what we are measuring at any individual point in time, but it is the trajectory of development, the trajectory of change that really matters. I think Jake Eadesâs study from NIMH points to this, in terms of measuring gray-matter volume doesnât correlate with IQ. What correlates with IQ is the change in gray-matter size over time in any particular individual. Dr. Goldstein: I wonder if you would agree that some of these mechanisms that are being defined are not going to be specific to autism. We are looking at these interesting interactions that several of you have talked about, but I think if you were studying a different neurobehavioral outcome, a different neurodevelopmental disorder, including Carlos Pardoâs work, when you start looking at the other nonautism disease controls, they have very similar changes. So I donât know that these mechanisms are going to get at the specificity of what we call autism. Maybe they are very relevant to aberrant brain development. Iâm not sure we yet know why these things could result in the clinical picture of autism. So as you are studying these mechanisms, and maybe it is a whole nosology issue here of what is autism and what distinguishes it. So we shouldnât lump different autisms together; there are other neurodevelop-
PROCEEDINGS 133 mental disorders of mental retardation and motor disorders that are going to have very similar pathologies and very similar aberrations in redox and cell proliferation and differentiation, just to confuse this. Dr. Slotkin: I donât think that is confusing at all. I think it is extremely relevant. Letâs posit the possibility that you could have two brains with identical morphological changes, one of which comes from a child with autism and the other of which comes from a child with a different neurobehavioral disability. I would actually find that entirely plausible because the brain doesnât just sit there passively and take a developmental hit. There are things that happened afterward like plasticity and adjustment that are influenced by environment, that enrichment. We already know, for example, that the incidence of learning disabilities and lowered IQ in the offspring of women who smoke can be completely obviated by an enriched rearing environment. So why should we adopt the idea that we are going to see a morphological phenotype that says autism or neurochemical type that says autism, when the odds are that it is far too complex to be defined within those rigorous bounds? Dr. Goldstein: I think that is what is happening. Dr. Akil: I also wanted to say that sometimes, a matter of degree might wind up eventually resulting in a different qualitative difference. For example, you can have somebody who has a small degree of language problem or hearing problem, parsing problem, but it can be overcome by something, parental training and so on. You can go just one step beyond, so it is still the same problem, just a matter of degree. But it can be so isolating that it can have social implications, emotional implications, family interaction implications, and you could wind up with a somewhat different syndrome. I can imagine that that could happen, where one kind of symptomatology would wind up just as a matter of degree facilitating other types of symptomatology. So I would be careful about not putting too much stock into this variation at that level, and still thinking about dimensionality. So the language analogy that you are seeing in your imaging stuff I think is very telling, very exciting, and would be very interesting to see in siblings and so on. Dr. Insel: I think there needs to be a little urgency added into the discussion. I am really concerned that we havenât thought about how to focus this area of science on where the needs would be greatest. This same forum maybe 6 weeks ago, something like that, with
134 AUTISM AND THE ENVIRONMENT Dennis Choi chairing the meeting on biomarkers, that was our first such meeting. When biomarkers came up today, some of us are trying to be thinking about that. We were thinking about biomarkers for clinical neuroscience very broadly, and there were lots of discussions about even the difficulty of getting a good biomarker for Huntingtonâs disease, where we have a gene that is actually diagnostic, trying to find physio- logical transcriptional changes before the onset of symptoms. Walter Korschetz described efforts to do that in a simple Mendelian disorder. What strikes me here is that the really urgent need for a biomarker that could be as Jill James was saying predictive, something that you could use at 1 week or 1 month or 6 months, well before you have to begin thinking about what the early detection behavioral paradigm might be. If there is an opportunity for that, I would think that would really drive a lot of the research, and it would be one of the most important things that could come out of trying to identify the pathophysiology. Dr. James: One caveat to our results is that we are looking at this metabolic profile in children who already have autism, so there is no way to know whether it is a cause or a consequence. We are funded to look very earlyâsee if this metabolic profile is there at 12, 18 months, before diagnosis. This is looking at the developmental delay clinic population, and then following the diagnosis to see whether the ones that had the abnormal profile go on to diagnosis more often than the ones that end up with just developmental delay. Another important issue is talking about lack of specificity. Do we really care if oxidative stress is an important modulator, whether it be ADHD or what, if we can correct it early, it may impact much wider a population of neurologic disorders rather than just autism. Dr. Insel: I think we really do care, though. That is one of the issues we talked about. The FDA was at that discussion to talk about how do we qualify a biomarker, something they care a lot about. It was all about specificity and sensitivity. I donât think we are hearing either of those in this discussion. Iâm not sure what it would take to get them. Dr. Akil: But it doesnât have to be a disease, right? It could be a dimension. You could have a biomarker for retarded language develop- ment that would work across five disorders. FDA would accept that, just like cognitive problems in schizophrenia. Dr. Insel: They would accept it for what purpose? For an indication for treatment?
PROCEEDINGS 135 Dr. Akil: Yes, and a biomarker. Dr. Lipkin: Biomarkers might differ depending on when you do the collection. Given what we know about the bloodâbrain barrier early in development, you might be able to find things in cord blood that you wouldnât find later. Iâm not saying that is the case because we havenât looked. But that is something that we need to bear in mind. So when you talk about biomarkers, it is important to find when we look and how we look. Dr. Herbert: I think there are a time and place for sensitivity and specificity and a time and place when they are not indicated. I think the FDA is rightfully concerned about biomarkers that are sensitive and specific at times when they are being used to make decisions about interventions that could be deeply harmful to the individual, such as chemotherapy or surgery. I think that there are other times when the pathophysiologyâin this case, it looks like there is a fair amount of nonspecificity to at least some components of it. My own personal hypothesis is that the specificity of the behavioral phenotype may come at the level of network interactions in brain connectivity and not at the level of other things, so that it is a computational outcome more than something that we would measure with the level of biomarkers that we may be talking about here. On top of that, it may be that the markers that characterize what is specific about autism may not be the same as the markers that character- ize where we can treat. I think it is really important to keep that question alive, because otherwise we may be so insistent upon the traditional sensitivity and specificity criteria that we will march way down that path and miss things that are low-hanging fruit, but that are more generic. Dr. Insel: Dr. Herbert, you should maybe expand on how that might work, because I think that is a very important point for this discussion, just to think through some specifics. What would be some targets, for instance, that could be ripe for interventions? Dr. Herbert: Letâs say you have a child who is 3 months old and starts having ear infections every month. Frequent ear infections are commonly seen in children who later become autistic. Frequent infections set up an inflammatory process and a depletion of redox capacity, which could lead to a lot of different problems. In my own clinical practice, I take this history on everybody who walks in the door, and I find this historyâand this again is not an epidemiological study, it is just my clinical experience with children with
136 AUTISM AND THE ENVIRONMENT autism, with children with ADHD, a lot of that, with a lot of neuropsy- chiatric conditions, nonspecific. Is it possible that if you supplemented these children with antioxi- dants from an early period of time, you would reduce the severity or even prevent whatever catastrophic flipover is involved, if there is such in a metabolic transformation where there is some failure to be able to do whatever it was you could do before. Some people think of autism regression as a kind of energetic failure. There are various metabolic theories. We donât know enough to say what it is, but if you could address the metabolic depletion that could potentially have been occurring progressively before it went over a threshold, it wouldnât really matter if it was specific to the disorder. The specificity may come from something that is almost incidental to what it is that you can treat. So that is one example. I think Jill James and some others, Mark Noble, you may be able to comment on that. Dr. Levitt: Before you get to that, I was going to ask Art Beaudet, have studies been done in the syndrome disorders, independent of trying to link it to an association with the co-occurrence of autism diagnosis in those syndromes, in terms of metabolic studies and other things? Dr. Beaudet: I donât know. As I listen to this discussion, I think thatâ Dr. Levitt: Youâre not going to answer the question? Dr. Beaudet: No. I think that there is the syndromic group. I believe they are largely genetic and they greatly complicate trying to study this other group, which is potentially milder, not dysmorphic, and where there is a lot of uncertainty about the etiology. The dysmorphic group are a lot like mental retardation, and most of them are mentally retarded. So I think if you are trying to ask some of these other questions where intervention is going to be dramatic, it is not going to be dramatic in the individuals who have an underlying genetic abnormality in their neurological function. So I think those could be weeded out, and then I think these other approaches would best be focused on more normal looking children. I think also, we are to the point where some substantial fraction of the patients can have a pretty convincing underlying diagnosis. If somebody is going to be imaging or this or that or look at oxidation, I think one would want to know, is this patient somebody we know the underlying etiology or not, or is it in the unknown group.
PROCEEDINGS 137 Dr. Levitt: What is the metabolic state with the individual with Rett syndrome? Dr. Beaudet: My impression would be, looking across all these disorders in mental retardation, that it is extremely heterogeneous. Dr. Levitt: Within Rett syndrome? Dr. Beaudet: No, within disorder. But the problem is, probably no one diagnosis accounts for anymore than 1 or 2 percent of the popula- tion. It is like saying a phenylalanine diet works in PKU (phenylketon- uria), so why donât we try it in Down syndrome. We understand the pathophysiology in PKU, and you can have some rational input, but I am very skeptical for the group that have definitive heterogeneous defects that there is going to be any intervention that is going to be dramatic other than just supportive intervention, learning processes. Dr. Swedo: One of the ways we might go after this is, and Sophia Colamarino, you might want to comment on this from your workshop, but just a very simple experiment of taking Fragile X and MECP-2 and other individuals who have autism and who do not, and begin to evaluate the similarities and differences, so your question could be answered about whether there are specific metabolic defects associated with autism by looking at individuals in which we know what the genetic defect is, and then look at the additional factor whether or not they have autism, because not all of them do. Dr. Beaudet: I think it would be very interesting to particularly compare the children who are at the extremes of some of these diagnoses. But you have to be careful, because in Fragile X you are talking about a male or a female, how large is their expansion. They donât have a pure single genotype. Dr. Pessah: I think we also need to extend this to those genes or genetic markers that have strong evidence for linkage to autism, the Met gene, Cav1.2, and there are several others that were mentioned today, and construct the animal models and see if they have an inherent oxidative stress when it comes on board, does this influence their metabolic status, and then go back into the kids you have identified with the genetic problems who have the strongest linkage, and who are not as profoundly affected as Rett syndrome, and decide whether or not they are under oxidative stress. I am very concerned that if you go into every autistic individual and treat with an antioxidant, what about those kidsâand I know this study
138 AUTISM AND THE ENVIRONMENT hasnât been replicatedâthat do better with fever? Talk about oxidative stress as being a cure, or at least a mitigating response. Dr. Levitt: I wanted to pose another question. Robert Strausberg wasnât able to make it, and he was going to talk about environmental factors and mutations and draw upon his work in cancer, where it is clear that there are certain genes that are more susceptible to the environmental factors than others that cause de novo mutations. I wanted to put you on the spot to deal with this issue a little bit more. I think in the cancer literature, it is very rich in trying to under- stand the role of environmental factors in perturbing fundamental biological processes. Mostly in cancer it is pretty straightforward, because it is dealing with this issue that Mark Noble talked about, about proliferation of differentiation. But what do we know about that in terms of some of either the candidate regions or copy number variation that has been reported recently, and how environment may play a role in that? Dr. Beaudet: The thing we know the most definitively to be associated with new mutations is paternal age. That is really dramatic, convincing, and relatively well understood in terms of its molecular basis. I thought that Dr. Susser gave an interesting example of how, if you were fully deficient and you incorporated U instead of T into your DNA, that this could bring about a risk of mutation rate. In the case of these copy number variants, there are differences among individuals in the population as to what they have for exact genomic structure. Some of these are prone to having a de novo event, more prone than others. There is some of that kind of data evolving, so that there will be slightly unusual rearrangements in a parent that arenât deleterious themselves, maybe an inversion, but then predisposes to de novo events in the offspring. Dr. Levitt: So are there going to be p53 or p53-like genes that we are going to identify here that relate to autism as opposed to cancer? How many different mutations have we identified with p53? Hundreds, maybe more, thousands. Dr. Beaudet: I think there are going to be lots of genes, and they are going to have lots of different mutations in this heterogeneous group. If there is some other residual group we donât understand, then there, I think a lot of these other things are going toâI think one has a better chance of success at testing environmental factors if you weed out the people with frank genetic abnormalities before you start looking at it, or
PROCEEDINGS 139 at least look at them separately. There, I think they have a genetic abnormality that is moderately overwhelming relative to the environmental effect. There will be some other effects, but I think if everybody would effectively weed out two waysâwith lab data it is somewhat possible, although the lab tools are still pretty primitive, and on the basis of dysmorphic features. Personally, if I wanted to look for an environmental effect that would affect the possible impact, I would want to deal with male patients who looked perfectly normal and go in that population. I think we would have a much better chance of finding something that is making a difference in that group. Dr. Noble: I think there are some conceptual issues there that are quite strikingly important. What we need to ask is important, but the fact is, because the FDA wants it doesnât mean it is the right question, with all due respect to the FDA. For example, to take a cancer example, BRCA-1 is a great predictor of whether or not you are going to get breast cancer, but only for those patients who have a mutation in BRCA-1. It identifies a small percentage of the individuals who are going to get breast cancer. It is useful to have as a diagnostic and may put you on different paths. You can get to the same endpoint in different ways. Letâs take C-Met as an example we discussed. In an individual who has compromised C-Met function because of a mutation in it, from the little understanding we have now, from what Pat has published, and we have published some other stuff in the literature, would we be able to correct that with some kind of an antioxidant therapy? I donât think so. If the compromised C-Met function is because of what Jill James is saying, then we may have a real shot at having a correction. So I like this idea of trying to understand how to screen out the genetic populations so that we can focus attention on the other aspect. But I think we have to understand, particularly from the stuff that Jill James has presented, that a lot of the genetic mutations may actually be giving us these metabolic disorders that even though they are genetic are still going to be broadly treatable by metabolic modifications. So maybe the FDA is going to ask us for something that for here isnât exactly the right question. Dr. Susser: Can I say one thing about environmental? Just to extend what Art Beaudet just said, when you separate the group, and I do think there will be a reasonably sized group that will have genetic mutations,
140 AUTISM AND THE ENVIRONMENT they will have heterogeneous genetic mutations that will have large effects, even if they are not completely deterministic of the disease. But that is a particularly interesting group in which to look for environmental antecedents of those mutations. That was implied in what I was saying. Studies of environmental factors are particularly important in that very group. Dr. Spence: One of the things that we talked about in our session and you brought up was the idea that we have to understand whether those mutations are in fact functional. So the determination of these variants, because there is a lot of evidence, if you look hard enough you can find lots and lots of different polymorphisms. The question is what is the relevance? Just a case in point, Fragile X permutation status. There was a great paper that Randi Hagermanâs group did that said this is associated with autism, even in the permutation. She can show, because she looked at the RNA, that those kids with autism with a permutation have differential RNAs. But on the other hand, it turns out Fragile X permutation status is actually very, very common. If autism is 1 among 50 and Fragile X permutation status is pretty common, it could have just been the two are unrelated. So I think we have to be careful that one of your gaps was assigning functional function to these polymorphisms, and I think we canât do that. Dr. Noble: That is exactly right. We know so little about this area. From a historical scientific viewpoint it is fascinating, because it looks like early entry points to a number of fields, except there are all these other data that one can talk about, that you think you should be able to understand. It is a bit overwhelming. Something I am concerned about is that I am really listening to the urgency from the patient representatives who we have here. As we have experienced in all scientific fields, we are always trying to balance our step-by-step progression in science with that urgency. I do have a concern that we have to be running both tracks at the same time. It is too early to say that this is right. Dr. Leshner: Iâve been waiting for somebody to say that, that is, to take either exclusively a general biomarker or a specific biomarker approach seems very dangerous to me, particularly given that we do eventually have to get to the specificity of the specific disorder that we are talking about.
PROCEEDINGS 141 Dr. Insel: I think there is a crosscurrent here that still needs to be clarified. If I am hearing this right, we are getting two messages today. One of them is the general issues that have to do with metabolic stress through development, which is almost certainly not specific to autism, but may be a robust finding that could be a signal for an intervention of some sort, but it doesnât tell us all that much perhaps about the specific pathophysiology of this illness. The other current that we heard more this morning was that autism itself is so general and it is so many things that even there we need to drill down and get much more specific, much more selective. We talked about doing n of 1 studies using individuals as their own control, defining kids perhaps by what they responded to and calling that a subgroup. I do think we have to get clear about where the most traction will be going forward, because those are two very different approaches, and we donât have enough money to do everything. My own bias is that if you look at other areas of medicine, generally you see progress best. Asthma is such a great example. When you can find a way to define a subgroup even beyond what you see clinically, and come up with groups that can be defined by some pathophysiological variable that now allows you to go after these other factors that help to grow that out, I donât think we are going to get there if we start taking very general kinds of pathophysiological markers that donât in any way take us to this endpoint. So just an opinion, but based mostly on what I see in the rest of medicine. Dr. Beaudet: Iâd just like to comment. Iâm sure I come at this from a very genetic perspective, but to me the most urgent question, which it is obvious has been urgent for more than a decade, is whether the incidence is really changing or not. This is not rocket science. I think it is the CDCâs area. If the incidence is really changing, there are environmental factors, there are things to be prevented, there are interventions to be done. If the incidence is not really changing, that is a very different situation. I think also, my impression is if the incidence is changing, the percentage of males relative to females should be rising with that, because that is the group that is going to be more likely to be involved in environmental interactions. But I think the most urgent question is to know if the incidence is changing or not.
Session IV New Approaches and Discussion with Workshop Attendees Participant: My name is Kelli Ann Davis. First of all, I want to show you my son, Miles. He is now almost 15 years old. When he was a baby, I think autism rates were 1 in 5,000, and they are now 1 in 150. This is my son at two and a half months old. If you look closely, he is trying to mimic speech. Two and a half months old, he was completely fine. Here he is at 1, and he is completely fine. Here he is at about 3 years old, and if you look in his eyes, you can see he is not really there anymore. Here he is holding his baby sister when he is 6 years old, and you look at him and look how sad he is. My son is now almost 15 years old. There are a lot of smart people in this room. Iâm just a mom, but I am asking for your help to find out the truth about what happened to my son. I believe it had to do with vaccines. I believe mercury had something to do with what happened to my son. I am here for the truth. That is what I have always wanted to pursue, is the truth. I am just encouraging you all to remember when you are talking, it is about our kids. I have got to tell you that the first time I heard Marthaâs talk a couple of years ago at the symposium, I had to go upstairs to the hotel room. I couldnât even hear her talk because I thought about what has happened to my son and what was going on in his brain. You are all scientists and you are looking at it from the scientific perspective, and we need that. But there are thousands of parents out there who are heartbroken, and when they hear the descriptions of the brain and the white matter, it is almost too much to take. So I guess I am just pleading with you all, first of all I want to thank you all for being here, but to please keep that in mind, and remember the kids who arenât babies anymore. We need the help as quickly as possible. I just appreciate everybody being here. Thank you so much. Dr. Leshner: Thank you for that. It underscores the urgency I think we all feel. So thank you. Mr. Blaxill: It also suggests a tiering around what kind of bio- markers we really care about. They are diagnostic biomarkers that are 143
144 AUTISM AND THE ENVIRONMENT final common pathways, they are prognostic biomarkers, they are treatment response biomarkers. It is easy to have prospective studies and think about the kids that arenât diagnosed yet or are not affected. The constituency out there needs attention to treatment response. It may not be now, and it may not be specific. It may overlap with all sorts of other things, GI diseases and the language impairment, but there is a prioritization implied in the biomarkers. Ms. Redwood: I just wanted to make a comment, too. I guess this is moving into the general session, since the parents are talking. What I have heard today is that we are looking at the potential for there being an environmental toxicant that may have caused our childrenâs disability. One of the questions I have for the panel is, why arenât we testing the children? We went into Brick Township and we tested the water and we tested the dirt. We tested everything we could think of, but nobody ever tested the children. I hear over and over again that mercury is one of those metals that might be causing this. I know for my son, he had over five times EPAâs action level of mercury in his body. I am just wondering why we are not testing the children. If there are multiple toxicants, letâs look at the kids and see what they have. If it is mercury, lead, PCBs, to me that is the study that we are ignoring right now. So I would like to ask the panel if anybody is looking at that, if anybody is doing urinary porphyrin levels in these children, what are the plans to test the kids? Dr. Swedo: I have mentioned that the NIMH M.I.N.D. phenome project is the pilot, and that is absolutely child focused, family second, home third. My colleagues can talk about their own studies, but I think your point is very well taken that if you are going to find it, you need to look where they are affected. That is one of the tensions between the need as Art Beaudet talked about, to find out what the change in incidence is. Those are large-scale expensive studies that have to be done. On the other hand, if this was leukemia or another medical illness where we could look at a cell system and know exactly what was wrong, we wouldnât spend a lot of time looking at the unaffected to figure out what was happening with those subtypes. So I think if I were speaking to the urgency, and I am trying to, my plea would be that we do this kind of meeting where everybody is using
PROCEEDINGS 145 the same platform where possible, and drawing on the strengths of each of the different kinds of advances. Dr. Schwartz: I agree with what Lyn Redwood just said. I would go further by saying that what we should do, we should try to figure out what studies are underway in populations of kids with autism that we can build in the state-of-the-art, but admittedly somewhat limited envi- ronmental measures that could be done on the biospecimens that are available within those studies. There is no reason we shouldnât do that. If that takes expanding the studies, we just need to look at what it will cost and try to pull those funds together to use in the best way possible to make that information available. The question I was trying to get at this morning is, what are the cohorts that are ongoing that we could leverage to append these additional studies to that would get at the answers? I think they are not perfect. As Dr. Insel pointed out, we donât have 39,000 assays, we have 20 assays right now, but maybe 2 or 3 years from now we will have a thousand assays. While we have 200 assays, we may as well make use of them. Dr. Falk: This has been a very interesting progression from Tom Inselâs comment all the way through to here. It seems to me that a large part of today, the discussion has been around mechanisms and pathways which could be impacted very significantly by environmental agents, but not nearly as much discussion about specific environmental agents. And of course, the pathways themselves may not be fully specific. In truth, there have been various times where environmental etiologic factors have been identified even before pathways are understood, and only afterward does one go back and understand the pathway. I guess one conclusion that I do draw from this is that perhaps there ought to be certainly more attention to specific environmental factors, both experimentally as well as in terms ofâin the epidemiology we will discuss tomorrow morning, we will have the opportunity to see just what those opportunities are. But it strikes me that that is an important area that, as I am seeing this all put together, is not fully addressed, perhaps. Participant: I have a question for Dr. Noble. You talked about thimerosal toxicity at the very end of your talk. Are all of the oligoden- drocyte precursors selectively vulnerable, more so than other types of brain cells? Dr. Noble: This gets to the issue of when in the life period of a cell
146 AUTISM AND THE ENVIRONMENT or lineage you see vulnerability. If you wanted to design a system to enable you to study problems like this, you would come up with something like what we now know about the oligodendrocyte lineage, because it has so many advantages to doing this kind of work. One of them is that myelination occurs at different time periods in different parts of the nervous system. The organism creates cells that intrinsically have different timings. One of the ways that we learn a lot about these problems is to try and understand what controls those timings, and it has turned out to be this intrinsic redox biology. So when we look at other cell types like embryonic cortical neurons, we find that similar principles apply, but we have to study them at the right stage in order to do this. I think in terms of this issue of specific toxicants, although I think this is an oversimplification, I have to say that what we keep seeing is that all the cell cares about is, is it oxidized? It doesnât care who is doing the oxidation. Obviously at other levels there are chemical specificities, but this is what we are seeing. Can I ask a question? I am trying so hard to understand this area. I think I see an experiment, but I want to ask whether it is a good one, that ties together some of the things that we have heard. Following on from Dr. Schwartzâs and Ms. Redwoodâs important comments about looking at the kids, and what we have heard from the parents about those who have used chelation therapy, is it a good question to ask, if you have a child with autism, and you now screen these other parameters, mercury load, lead load, PCB load, get an environmental toxicant profile on them. Now you do the chelation therapy. Is it the case that the kids in whom that works, are those kids in whom we have higher levels of heavy metals? And is it the case that the parameters that Jill Jamesâs studies normalize, or that auditory brain stem response normalizes? Is that a type of focused question that one can ask to get some traction? Ms. Bono: That is basically the recovered kids study, which you are talking about. There are DAN doctors throughout America who have kids whom they started working with, 3, 4, 5, 6 years of age, specifically the younger ones are the ones that have the best recovery rate. Some have Jill Jamesâs profiles, these kids have shown methylation problems, oxidative stress. They start chelating, but they also do other things, giving them glutathione, cysteine, all of the things to help with that.
PROCEEDINGS 147 So there are those entry-level treatment biomarkers that the doctors have when they walk in the office. Then they have it tracking as they go along. Dr. Noble: Can that data be made available to us? Ms. Bono: The DAN doctors have said that they would be very willing to have data mining go into their offices and pull that type of information. Ms. Redwood: There is one of the clinicians I saw here a few minutes ago, Nancy OâHara, who has a very large practice, who might want to share her data. Participant: I am Nancy OâHara. I have been working with children with autism for 25 years, first as a teacher, the last 9 years solely with children with autism. First I want to thank the researchers, because they have given us the information that we need to see why our kids are biochemically sick, and they are, but also to see how we may begin to treat them. They are treatable. We do have that data. We have the urines, we have the stools, we have the leads pretreatment and posttreatment, in recovered kids and in kids who are not recovered. Believe me, it is not 100 percent, but the data are there. We need help mining that data and taking that data from a large group of clinicians now who have it, but we just donât have the resources to then pool the data together and use them. But we have them. There was a lot of talk this morning about inflammation and also this afternoon. They mentioned tonsils on one of the slides. David Gozal from Kentucky has very interesting literature on the very damaging effect of repetitive hypoxia. He has looked at it in cell culture systems and also in clinical systems, and found that children with learning disorders and also sleep disorders often had tonsil problems, and when they were removed they actually improved considerably. So his repetitive hypoxia paradigm would also fit in with some of the redox type of studies people have been talking about, and I think should be thrown into the big picture. Dr. Swedo: Before we go down that field though, as a pediatrician I just have to remind folks that tonsils in a 7- to 8-year-old are very different than those in an infant and neonate. Dr. Leshner: Somebody here was going to respond to the last question. Dr. Akil: It was a comment about physicians and other people in the
148 AUTISM AND THE ENVIRONMENT community who have information. Tom Insel and I were talking at lunch about how one might engage physicians in the community, whether it is part of a CME (Continuing Medical Education) or volunteer or whatever, whether we need a medical informatics national project that sits back and thinks about what kind of information is needed to do this in a systematic way, meaning something that one can participate in where the kinds of information that are needed, the kind of diagnostic criteria that are required, the kinds of treatment, the kinds of levels, what assays are approved or not approved, whether we can put something like that together either in a trial in a few centers to begin with before expanding it, and bring information into it and see if there is any way to begin to rely on people who are doing the footwork but feel isolated, and have scientists who are good at data mining or analyzing. But if you patch it like a patchwork, very pell-mell, it would not be useful. We need to come back to you and say here is the information we need, and then get it from the people in the trenches. Participant: I agree with you, but you also have to address the urgency which a lot of these parents feel. If you start prospectively and ignore all the data that are already there. Dr. Akil: You eat it. You eat what you have. You eat the data that you have so far. Participant: But you have to use that, and they may not be as clean as the data you want to use prospectively. But I think you have to use some of the data you have now to be able to start. Dr. Swedo: I think that is a fabulous idea. We thought it was such a great idea that we started 3 years ago to develop a national database for autism research, which allows clinicians in the field to become research- ers by providing them with the clinical tools they need to do systematic assessment of their patients. Our group has been very impressed with the DAN practitioners and are grateful to them for what we are learning from them, are hoping to partner with them even more in the future. But in addition there is another network, the Autism Treatment Network (ATN), which started out on the West Coast and Boston. They now have a dozen sites. They are hoping to get 20 different sites, both academic and clinic community based. They are gathering data from their patients, and if Paul Law is still here, he can speak to the Autism Speaks registry, IAN, Interactive Autism Network, that allows parents to input their data directly about their kids and get instant feedback.
PROCEEDINGS 149 I think we are poised from an informatic standpoint to meet this need of urgency and get the data very quickly, start looking for similarities and differences across this group, and then do more in-depth systematic study as the patterns emerge. Ms. Bono: I agree with you, all three of those things are very good. With the DAN doctors, with the huge practices, they need to have more of a systematic approach, where three or four people are on the payroll, and they come in and they mine that data based on whatever criteria they have, and then they move to the next one. These doctors just donât have the time to go back and try and put it together and give it to you, but it is there. Dr. Insel: I think one of the great things about the National Database for Autism Research is, it does give you the standardized assessments. All the tools are there in place, and anyone can use them anywhere. It is totally public access, or will be in September when it goes fully live. The relevance to this meeting specifically goes back to David Schwartzâs and Lyn Redwoodâs point, though. What we donât have are the large repositories of biological samples on all of the thousands and thousands of kids who have been treated. They may exist someplace. We have a relatively small brain bank, we have small banks of other kinds of samples, but clearly there is a need to do here what was done for childhood leukemias 25 years ago. You find a way to organize, consolidate, and then go from nâs of 10 and 20 and 30 to 2,000, 3,000 or 20,000 or 30,000. In a complicated area like this, you are going to need those kinds of large nâs to be able to find the subgroups that really will give you ultimately that rigor. We will have the clinical piece. One of the things that would be great for this group to weigh in on is what would be the biological samples, when should they be collected, and what would you want to do with it. Dr. Leshner: I think that one we should hold for tomorrow. Dr. Schwartz: And maybe what could we do with them in the absence of an absolutely ideal study. We have probably at NIEHS two or three epidemiology studies that we are funding in autism. Iâm sure you probably have a half dozen or a dozen or in aggregate. There are a number of epidemiological studies in autism that have been done. They are not using the same tools necessarily, they may not have the same diagnostic criteria, but there are areas of overlap that we would agree are critical elements across all of those studies that could
150 AUTISM AND THE ENVIRONMENT serve to bring those studies together in a biobank that we could then mine for environmental data and genetic data and other data that could push the field forward before the ideal population has been acquired. Dr. Falk: I am very supportive of these ideas that are coming forward, particularly if the chelation data, for example, are one of the strongest indicators for there being environmental agents. They should be looked at in detail for any group like that that is thinking about environ- mental agents. But if I may go back to something Mark said before. I want to make sure I understand this correctly. I understood what you were saying to be that there are so many environmental factors which could conceivably affect redox status and pathways, it is almost immaterial to look for the environmental agents? I donât know whether you quite said that, but maybe that is what I was thinking. You were implying I think that the specific environmental agents might be so numerous that. . . . Dr. Noble: No, it is the second one. I think that if one wants to test the hypothesis that mercury is the primary causative agent in autism, that that is the wrong hypothesis. If one wants to test the hypothesis that mercury is one of many environmental factors that may contribute to this outcome, that looks like the right hypothesis. So if we look at what we can look at now, there is a limited number of agents where the sensitivity of analyses are sufficient to enable us to do reasonable studies, the organic materials, PCBs, a few other effects. That data may turn out to be extraordinarily compelling, particularly because of what we are hearing about the heavy metals and the chelation therapy. At least heavy metals are something that can be analyzed pretty well. What I am specifically concerned with is thatâwith all respect and admiration and concern for the parentsâ groups and everyone who has been trying to pursue the idea of a specific environment toxicant or a specific vaccination, just from a biological point of view, it doesnât sound like a great hypothesis. It sounds like these may be pieces, that they happened at a particular time, but they are not going to apply to all the kids. From the cellâs point of view, I donât think it matters. Am I oxidized because of an inflammation? Am I oxidized because I got mercury? I donât care, Iâm oxidized, I am in electron deficit. The data that I am hearing just keeps agreeing with that. Even this idea of astrogliosis and the white-matter tracks, when we take these oligodendrocyte progenitors
PROCEEDINGS 151 and expose them to oxidative stress, they turn into astrocytes. So even that is a really intriguing outcome. We have to look at specific astrocyte populations there. So that is what I mean, that these all could be players. It doesnât sound likely at this stage that any one of them is a player of central importance. Participant: I am Lee Grossman. I am president and CEO of the Autism Society of America, and more importantly, the dad of a young man, a 19-year-old with autism. First an observation. I want to thank all of you for this wonderful assembly. Some of the best minds in the scientific community are working on this. The autism community, the parents, are very grateful for your efforts and everything that you are doing. I think there is one oversight here that I do want to point out. For future planning purposes, I think it is essential for a person on the spectrum to be included on the panel as well as the planning. Hopefully that will be corrected as we move forward. The comments I am making now have been supported by some of the other people here. I have been wanting to make them since Sarah Spence presented her observations this morning, when she presented the information in terms of the variations in autism, which could be in terms of millions, perhaps billions, of variations out there, when you include all the extraneous information that is out there for environmental interven- tions as well as the genetic components of this. Then Tom Insel presented his two models exploring this, and bringing in the phenotype data as well as looking at it, which would be certainly meeting the scientific rigor that I think all of you want to meet. I wanted to proposeâand I think some of the people here, Nancy and others, have started to discuss thisâa third model that I think would fit well into that, and that is a treatment model. What you are talking about here is wonderful and it is what needs to be done obviously, but we are looking at another generation of children as you do this. I donât think the community can wait any longer. There is enough anecdotal and proven information out there in terms of treatment that should be explored and followed and implemented. I think if we put into the two models that Tom recommended this morning, and also incorporate a treatment-guided model into that, the three can work in collaboration with each other. We can develop treatment protocols that can, I believe, meet the scientific rigor that you
152 AUTISM AND THE ENVIRONMENT are looking for, where these kids can be evaluated, we can see what is working, what is not working, and then move forward in that regard. In the meantime, kids are getting helped, they are getting improved, and we are learning. I think we are going to accelerate the pace of our knowledge and our learning, and certainly help the folks that are out there today. Thank you. Mr. Blaxill: I just want to amplify that, and just underscore some of the scientific deficiencies of approaching that. A lot of what we like to call evidence-based medicine is designed to ration the access to market of small molecules that pharmaceutical companies sell. That is a very useful rationing procedure. It is good for safety management and that sort of thing. I think the types of therapies that we are talking about here with environmental illness are more regimens and ways of life, diets, things that are less invasive and less dangerous than some of these things are, potentially dangerous. So there are special methodological problems, and I think it just argues for some degree of risk taking, comfort with complexity or messiness. I hate to argue for relaxing standards, because I donât think any of us want less rigorous work, but we also need to be roughly right rather than precisely accurate 20 years from now. I just think it is important to come to grips with the special types of therapy and regimen interventions that we are talking about, so that we donât throw out the baby with the bath water. I can imagine all sorts of negative studies coming out that miss the main point. Sue Swedo, I have talked with you about this. I was saying to Pat Levitt, some of us, because we donât know what else to do, we canât wait for the clinical science to take 20 years to solve all these problems. We have to act today, we have to act on some model. We donât know whether we are right, but my daughter is 11; I canât wait that much longer. I was saying to Dr. Levitt, it is like advertising. Iâm sure that 50 percent of the therapies that we are trying are absolutely worthless. I just donât know which 50 percent, and sorting that out is a challenge. I just call that out as a methodological challenge, a scientific challenge, because there is a real risk that you get a collision between the request for rigor and the movement and all of this enthusiasm about helping kids. Those ought to come together and be mutually supportive. It is the potential for them to get antagonistic, and that is something to
PROCEEDINGS 153 keep in mind. Dr. Leshner: I think you articulate well the obligation that the scientific community frankly has to help you meet that need. I like to think, I hope to think, that that is what we are doing here. I hope that this will in fact significantly move forward the research agenda in some way. But I think the point is extremely well put, and it underscores the obligation that I think everybody feels. But it is good to articulate. Thank you. Dr. Levitt: It is also underscored byâwhen you look around the table at the scientists who are here, how many scientists here who are actively doing research started doing research or were trained to do research in autism? Raise your hands. Two. That is a reflection that there has been a sea change culturally in the way science is getting done in a lot of disciplines, that this is a poster child for the willingness of scientists to not do the kinds of standard things they do, which is to keep looking for more and more rigor and being unwilling to take some stands and work together, and come from different fields. So I think that is happening pretty rapidly, and it needs to happen more obviously. But I think that is a reflection of what you just said. Dr. Noble: I think we are trying to find out how to meet you, if not halfway, some way in the middle. There are 14 million kids in the United States with some kind of neurological disorder, and parents are trying everything. From the point of view of someone in stem cell medicine, I am desperately interested in keeping this stuff regulated, because there are so many cowboy clinics out there. But most of you are not doing stem cell transplantation, you are doing things that have less of a risk. You are going to follow multiple areas of research. But if I look at this as a scientist and I can look at a minimum dataset that says, here is a kid, here is a metabolic profile, here is a heavy metal profile, here is a toxicant profile, here is what their behavior is like, and you do whatever you do, and we get those measurements at the end and can say, the kid is behaviorally changed, are there any of these measurements that have changed? We may learn quite a bit from that. It would be nicer if there were standard protocols that people were using. Dr. Herbert: I really agree with what Mark just said. I think that it is unrealistic at this stage to try and discipline people into specific standardized protocols, since we donât know how to characterize the heterogeneity, and because people are going to do what they are going to
154 AUTISM AND THE ENVIRONMENT do, anyway. I would propose from a sociological point of view that there could be usefully some support for the self-organization of the parents and some of the treating physicians, like the DAN docs, like Nancy and others, to have more support to build a platform of communicating what classes of data are available, a status report of what is being collected. It is not enough for you to stand up and say for 30 seconds that you have these kind of data. I think it would be really nice if we could have support for a description in more detail of what is going on, focusing in classes of data to facilitate the interface with NDAR and other kinds of data collection mechanisms. Iâm not clear that it would work coming from NDAR to the parent and treating community. I think there needs to be some support for the treating community, which is exhausted and overworked beyond all description with this incredible burden of cases. It is not just people coming in and mining data. There needs to be preparation for that, so that there is some kind of a systematic approach. So I think to make this happen, in order to meet in the middle, there can be a transitional infrastructure of setting up what it would mean to do that. Otherwise the activation energy to make it start happening isnât going to happen. Dr. Beaudet: It seems to me that it would be interesting to know if the two camps could come together around a truly blind chelation trial, in which certain patients got infusions of placebo, and this went on for a year. Some parents would have to take the risk that their child might or might not be on placebo chelation for a year. But I think this would take considerable investment of both sides to agree to something like that. I would be pretty impressed if such a study could show something is going on. Dr. Swedo: I just wanted to say that such a trial has been developed in collaboration with the DAN practitioners. We are using their protocol and breaking those elements down. The gluten-free, casein-free diet is already under investigation at one of the START centers as well. So I think the individual components, the hard question and the thing we really have to grapple with is this issue ofâit appears that one of the successes of the DAN approach is that it is very broad and deep. There are a lot of things going on with those kids all at the same time, so trying to figure out which components of it are useful is something that is going to take some more work.
PROCEEDINGS 155 Dr. Oberdorfer: Just to follow up on Martha Herbertâs point, when you have observational studies that you are going to be undertaking that you havenât planned yet, if you do what you suggest, you are going to see what sort of samples that you can take from a number of different studies in a much more global sense. That way you would have some commonality in a tool kit. Right now, my impression is that they are going in cross purposes. They are not collected in the same ways, they are not in the same times, but they are moving in that direction. It seems to me that even if the studies go in different directions, you will have samples that you can compare homogeneously. I think that is very important, these kinds of toolboxes. We can do that now. Dr. Lipkin: Alan, this is not a comment. Maybe if Sue Swedo could summarize for us what is going on, we might save some questions. We are continually going back to asking what is being done at NIMH. If you could just summarize what is being done in terms of treatment, then maybe some of these questions would already be addressed. Dr. Swedo: All right. The new intramural research program is about a year old. We started with two major types of studies. One is an in-depth phenotyping effort, making use of the anecdotal literature and the clinical experience from clinicians across the country, but also the CHARGE, CADDRE, and other data that had been collected. It is everything from family history of medical illness and environmental exposures to neuroimaging, genetics, and other evaluations. Within that, we have a regressions substudy that looks at children specifically with regression for additional factors, such as microbial triggers or inflammatory responses. We also have a treatment compo- nent. Intramural does best novel treatments. Tom Insel called us a SWAT team. We go where we see a lead. For example, we are using meno- cycline for its effects on NF kappa B to try and see if that would decrease neuro inflammation. We have a trial in antiglutamate agents, seeing what effect that would have not only in repetitive behaviors, but overall autistic behaviors. The chelation study is currently on hold because of some recent reports of a rat study that reported cognitive deficits in DMSA- alone treated animals. We are going back to the IRB (Institutional Review Board) on May 1 to look at that question. That is what we are doing in-house. My colleagues from extramural can talk about the new A centers. But I think that many of the things that
156 AUTISM AND THE ENVIRONMENT are happening, some of them are underway. Probably the most important is this issue of common measures and trying to get as much richness of clinical data as we can from every subject that is studied with NIH funding. That includes, as I mentioned, standardized diagnostic assessments, behavioral assessments, neuroimaging if it is being done on a common platform, as well as obtaining genetic material and biological samples. Dr. Lipkin: A constellation of toolboxes. Dr. Swedo: Right. I think we have already heard about some of the ongoing efforts in which supplements are being done to get the same kind of biological data. Now one of the questions is how do we organize it and go after the hypotheses. Participant: My name is Becky Peters. I have worked in the autism community for the last 5 years. I missed the very beginning, but I donât think until Sue Swedo just mentioned it that I have heard anything about the possibility of food allergies. I read a lot about and heard lectures on things like the gluten- and casein-free diet, the specific carbohydrate diet, and how for some children with autism, it has not only improved some, but even caused recovery in some. So I was just wondering if anyone in the research community is looking into the possibility that food allergies or certain foods that maybe children are genetically predisposed to be more sensitive to could be environmental factors in causing autism. Dr. Leshner: Does somebody have a very brief answer? Participant: A brief answer is the recent data on microflora associated with obesity, for instance. It is something that occurs in response to a specific diet. We are finding that there is increasing research that tells us that you can change metabolism and adipose cytokines and adipose tissue can change in response to the diet in conjunction with the microflora. We have the opportunity now with the tools that exist to begin to explore those types of issues. It may not be the standard allergies. It could also be other models that we need to also think about. Participant: Claudia Miller from U.T. (University of Texas) Health Sciences Center in San Antonio. We have worked extensively with adults with food intolerances and environmental intolerances. Until you eliminate all of the things that bother that particular person, you donât see the problems reversed.
PROCEEDINGS 157 Now we have started doing the same things in autism. The caution is that if you start doing a few things and just gluten and just casein, you may get some reversal but you may not get all of it. You may get other intolerances that develop, which is why you have to have a very comorbidity protocol. Participant: My question is about it being part of the etiology. I know that diet is out there and it is helping people, but I donât feel like anybody here has addressed the possibility of the food allergies causing that problem. I was just wondering if anyone is considering that, or if that is not on the table for research. Participant: I am Dr. Richard Deth from Northeastern University in Boston. There is something missing here from this discussion that unifies many of the observations and the questions that have come up during this afternoonâs discussion. That is reflecting some of the work that we did and we continue to do on the dopamine D4 receptor. The D4 dopamine receptor is involved in methylating membranes of neurons. It uses methyl groups from the folate pathway through the enzyme methionine synthase. We discovered that a certain number of years ago. The D4 dopamine receptor is linked to ADHD, and is now recog- nized as the most important genetic risk factor for ADHD. The D4 dopamine receptor is linked to lead toxicity and the role of lead in contributing to ADHD. The D4 receptor is linked to IQ. It is a risk factor for IQ reductions. So it has all the characteristics of a candidate receptor, dopamine included. This is the only receptor that utilizes sulfur pathways. It uses the enzyme methionine synthase that is turned off by oxidative stress. When oxidative stress occurs, be it mercury or be it pollutants or be it pesticides, that enzyme turns off to make more glutathione and robs that system of its methylation ability. The role of the D4 receptor is to synchronize that gamma synchrony that is gamma-frequency synchronized, synchronization of the brain during attention, a system that is deficient during autism as well as ADHD. As we have pursued this line of investigation, we have recently found that there is alternative splicing of the gene from the mRNA from methionine synthase in the human cortex. We have found that this alternative splicing is related to aging; it is complete in 80-year-olds, and it is incomplete in 20-year-olds.
158 AUTISM AND THE ENVIRONMENT As a result of recognizing the central role of this process, I think you can find mechanisms to explore. The enzyme methionine synthase utilizes methyl B12, which is a treatment that approximately 30 percent of people respond to. So I would say that there are all the elements here to start building from, even though it might not be a complete story. I just recommend that area of science to the panel members, because it can unify many of the things that they are concerned about. Dr. Leshner: Have you published it? Participant: I published several papers. The first paper about the D4 receptor was published in Molecular Psychiatry. The second paper was about methionine synthase being inhibited by ethanol, mercury, lead, thimerosal, because glutathione levels are low and methyl B12 synthesis is impaired. There is a lot to know about this. You just have to look into it. Dr. Leshner: Thank you. We will put that onto the list. Very helpful. Participant: I am from CDC. I want to make three points and try to make them quickly. The first is, I hope that everyone will be here tomorrow morning, because there will be several epidemiologic studies that will be presented, and they can be built upon in terms of specific environmental questions that are not being addressed. The CHARGE study will be presented and the CDC CADDRE study will be presented. It sounds like a lot of the questions that people are asking about studies and cohorts might be answered tomorrow morning. The second point is, Dr. Schwartz was asking about large cohorts that are available for study. I wanted to mention one in China. These are children of mothers that received folic acid. These children are about 12 years old now. There are about 200,000 children that are going to be characterized for autism. So I just wanted to go on the record to say that is a cohort that could be studied. The third point is to talk about the National Childrenâs Study a little bit more. Dr. Landrigan did mention that study, but there are a lot of environmental agents that are going to be studied as part of that study in terms of levels in the children, and autism is an outcome. There will be a research protocol that will be available for public comment. If that study is not addressing some of the questions that people have, if we donât have the right chemicals identified, if we donât have the right confounders, mediators, and modifiers described in the
PROCEEDINGS 159 study, then I would encourage people who are here today to comment upon that. That study was designed to answer a lot of these questions about specific environmental exposures. So I just wanted to remind people that there are some studies in addition to the ones that Sue Swedo had mentioned within NIH. There are some studies underway that might be able to shed some light on some of these questions. Dr. Leshner: Iâll just reiterate your point about tomorrow morning. A lot of very important talks are going to touch on an array of these issues that we have been talking about already. But we allow free talking. Thank you. Participant: My name is Harold Grahams. I am a private practi- tioner in Pennsylvania. To address Dr. Inselâs issue about the urgency of a biomarker, there is a tool that has come across my radar a few years ago that has been used in chemistry labs and hospitals and university settings all across the world that has been underutilized in autism. It is a high-performance liquid chromatography. There is a gentleman, Wayne Madsen, from PSA Labs who did work with lead studies 20, 30 years ago. Wayne Madsen did some unpublished studies with a controversial group up in the Philadelphia area probably about 10 years ago. I think it is a tool that might answer a lot of questions that all of us as practitioners and anybody could use as a reliable biomarker. What Wayne Madsen found with a group of kids is that we could give him bloodwork, and he could tell usâif we give him the blood tubes, he could run it through his chromatography, look at all these metabolites. What he could then spit back was that this was a cerebral palsy kid, this was a Down syndrome kid, or this was an autistic kid, just from those metabolites. The nice thing about itâyes, pretty impressive, right? But it was unpublished. Participant: How do we see it? How do we see the data? Dr. Leshner: People have to get the data into the system. Participant: I understand that. Dr. Leshner: I questioned unknowingly before about publication. Particularly in this field I think we have to be extremely careful that we not lead families astray, lead the scientific enterprise astray. So if you could tell Sue Swedo or somebody, people who are actively involved about this, maybe they can get access to the data.
160 AUTISM AND THE ENVIRONMENT I am a journal publisher, so I am obsessed with peer review and publication and making sure that whatever it is that we communicate is going to be as scientifically rigorous and credible as possible, lest we lead people astray. Participant: Oh, I understand that. I hate to stand up here and say there is this fabulous tool that has not been worked. But just to throw it out, that it is a tool that has been preliminarily looked at with maybe 100 or so kids. I think it also allows usâthe tool also has the fingerprint for the individual child. So what Martha Herbert was talking about, as having a way to track our treatments, I think we should as physicians be allowed the freedom to do whatever we do, because each doctor may be fixing a certain subset of children, but if we donât know, that is the frustrating world of the clinician. Dr. Leshner: I agree with that. I would just urge you to somehow get the individual attached to these networks that are developing, because we donât want to lose the opportunity if there is something particularly in this. So perhaps you could refer the physician to the networks that are developing. Participant: That is my frustration as a clinician. How do we know whether what we are doing is valuable and have the time to collect the data that the scientific communityâwhen I went to UCâDavis (Univer- sity of CaliforniaâDavis) too many years ago, I know the rigors that science wants, and we just donât have it available, but we are doing something. So while we are doing something, we might as well be collecting a yardstick so we can measure what anybody is doing, and we donât have that kind of biomarker. But I think this is a potential tool. Participant: One model that might apply here, and maybe we can talk about it more tomorrow, is the cystic fibrosis (CF) model, in which they started with a few very focused research centers, encouraged them to begin the training. That very rapidly got out to regional centers, and the regional centers began to work with the private physicians, and they markedly improved survival rates for individuals with CF. So I would be very thrilled to try to help spearhead that effort. Obviously it is going to be a major undertaking, but I think working together we can probably get that done. Participant: The bloods are already being collected so you donât need a whole lot of blood. Participant: Right. We would need to make sure that the diagnoses were accurate and blinding was done. So I think the testing of that
PROCEEDINGS 161 particular hypothesis, absolutely, that can happen very quickly. The larger question is how physicians can be providing feedback and families can be providing information. It is something I think we need to organize both sides in at the same time. Participant: I am a grandparent of two autistic grandsons. I thank you for inviting the public. I hope to be one of the taxpayers and voters who gets you the funding to go on with your research. I must say, when I saw the advertisement for this on the Internet, I was distressed to see that mercury was not going to be discussed in the context of vaccines. I thought there was going to be a white elephant roaming around in the middle there, and everybody would avert their gaze. But I see that there is frank discussion on that, and I am very encouraged. I think the pursuit of science, obviously you have to go where the truth leads you. Also, as a taxpayer funding this research, I think it is very important for you to understand that the people who are going to be out there getting political want some basic questions answered. They want you to look at the mercury hypothesis and tell them why it is not mercury, why it is not repeated environmental insults, and the number of shots they get. It is mercury and it is an immune assault to get these inoculations, and there are so many of them. So they need an explanation for why that is not the cause. I think we canât move beyond and do good research until we answer that question, put that one to bed. So letâs not ignore it. Letâs address it head on, and tell parents why they neednât feel guilty. There are so many emotional issues involved here. Another thing is, people donât really trust their government all that much anymore, CDC, FDA, and beyond. They need to trust their government more in their research by knowing that certain areas are not off limits. If we do not allow an explanation for mercury and vaccines, it will be like doing lung cancer research and saying, but letâs not say ciga- rettes. Nobody would believe it, and they would know their money was wasted. So that needs to be on the table. I have another tack altogether. Psychology doesnât seem to be represented here. I know that is perhaps not a good fit with environ- mental issues, but it is something that should participate in any funding for autism. In our situation we have gotten a lot more bang for the buck with the
162 AUTISM AND THE ENVIRONMENT biomedical. We have pretty much given up a lot of the behavioral therapies. They have been good, but they havenât been as good as the other. So we need guidance. Then one more thing. I have a natural experiment to suggest. Rho- negative mothers are subjected to a standard of care that calls for a RhoGAM injection at 5 months of gestation, so we give them a little extra environmental assault and a little extra mercury there. I understand that Rho-negative mothers have a higher percentage of autistic children than others, so what is the explanation for that? Is it Rho-negative mothers? I donât know. These are some questions we would like answered. Thank you very much. Participant: I would like to comment on the RhoGAM issue. There is a small study which is not very well done, and it shows this kind of finding. Since then there has been a more systematic study which I hope is in press. I canât give more details, but it is from somewhere very respected in the United States, which has done a large sample in a study of that kind, and it showed there was no association between RhoGAM and being the mother of an autistic child. It is in press. Participant: Just going back to something both Mark Blaxill and Mark Noble said before, we are looking at first of all a very complex set of circumstances here. I donât think there is going to be one thing that is found. I think it is going to be multifactorial, and I think this group is saying that. But I think as we address that from a clinical point of view, we have to look at not just one treatment modality, but also what the child is experiencing overall. When we look at what Jill James was showing us between the gut and the immune system and the brain, looking at all of those factors before we say, how does chelation affect a child? I think as we set up the studies that Sue Swedo is doing, for instance, if we set it up in such a way that we are just looking at the effects of DMSA on a child without looking at whether that child has had and still has gut or immunologic abnormalities, we would have very different outcomes than if we look at a child that is otherwise healthy and then looking at how chelation does it. So it is a very messy set of data and we have to look at all those parameters going in and coming out, or else we are going to have data that show us nothing. Participant: My name is Heather Elias. There is a subgroup of
PROCEEDINGS 163 children with autism, and a critical factor in treating them is regulating their hormone levels, particularly their testosterone levels. We know that boys are more likely to have autism than girls, but the girls tend to be more severely affected by the autism. That also makes me think that testosterone plays a big effect on how these childrenâs treatment should go. We know that Risparitol, Lepran, Spironolactone, all pharmaceutical drugs, are effective treatments for certain behaviors associated with hormone levels. I am curious if any of the scientists here are doing any kind of research on the hormones and how it affects the behaviors of people with autism and regulating those, how it helps them, if there is any kind of study going on about that. Participant: The study that we will be describing tomorrow, one of the domains of research that we are investigating, is related to hormone abnormalities. Dr. Leshner: Good, that will make you come back tomorrow. Participant: It will be including also immune dysfunction, which will address a lot of the other issues that have been raised today. Participant: There is someone in the United Kingdom whose name is Simon Cohen, who has been doing studies in relation to autistic symptoms or traits in the children who are born from these pregnancies. He has shown some relationship. It is very preliminary. It is not looking at autism as a disorder, as an outcome. Participant: I just also wanted to mention, I have a daughter who has autism, and we have followed the DAN protocol, and I am so grateful for these doctors doing all of this research. But we now are looking at the data. Just keep in mind that the parents have basically spent thousands of dollars to get these tests done, and they have really sacrificed a lot to get that information. So just remember the children when you are looking at the data. Thank you. Dr. Leshner: That is what this is about. You can be the last word. Participant: My name is Scott Bono. I live in Durham, North Carolina. I thank the panel for convening and taking up this topic. It is deeply personal. Last month I filed to retain guardianship of my 18-year- old son. I have two other children in college, and I never expected that when my son was born I would have to do this at age 18. It is very personal. I know what happened to my son was inexplicable, but you all are looking into it, and I appreciate that.
164 AUTISM AND THE ENVIRONMENT The most relevant question that has been asked today was asked by Dr. Choi. That is, for 15 years nobody has been looking at the urine and the blood of the children that we are talking about right now. Most parents, when they go to a pediatrician and they are told that their child is autistic, they are dismissed. That childâs illness is dismissed on the basis of behavior. I am so grateful that each of you is looking into some of the systems that have gone wrong in my son and so many other children. Treatment should always be what is in your mind as you proceed here, because I want my son back. Everybody wants their child back. I really want to thank all of you for coming. And Jackson thanks you. Dr. Leshner: Thank you, sir, and thank you for reminding us. We are going to stop for the day. This has been from my perspective a wonderful day. I want to thank the speakers, I want to thank the audience. This was for me a veryâI guess the right word is dramatic, but it was a wonderful example of how the patient and family community and the scientific community can work together. You have to come back tomorrow. I am really tough, so if you donât come back tomorrow I am going to chase you. But I didnât want to leave without making the comment that I am not sure that I have seen as good an interaction between the scientific community and the patient and family community, and I really appreci- ate it greatly. I am very grateful to the family members for coming and sharing your experience and your insight with us.