Important Points Made by the Individual Speakers
- Lack of insurance coverage greatly affects the access to and use of preconception genetic testing.
- The increasing number of genetic tests is producing significant educational challenges for providers and patients, which is exacerbated by insufficient numbers of genetic counselors.
- Genetic variants of uncertain significance can present major challenges in the preconception and prenatal period and produce extremely difficult counseling scenarios.
- Secondary findings from a genomic test must be dealt with on a case-by-case basis.
- Genomic data can provide direct information about a patient, but it also can have relevance to family members as well.
- A framework should be developed to put value on the information being provided by genetic tests.
- New or revised tools for more effectively conveying information to patients will need to be developed in order to accommodate the significant amount of information from genomics-based testing.
- Using genomics as a tool for avoiding unnecessary procedures or treatments has significant potential to save costs for the health care system.
In the first scenario discussed at the workshop, a woman who is contemplating pregnancy seeks counseling:
In 2012, a 35-year-old Ashkenazi Jewish female smoker in good health is seen for a preconception visit. Under the current standard care model, targeted carrier status testing is offered. In terms of high effect size variations that would be detected by traditional genetic testing, she is found to be a carrier for Tay-Sachs. In addition, if testing were extended in this scenario beyond what might be considered to be current standard of care, she would be found to harbor a prothrombin gene mutation, as well as variations in CYP2C9 and VKORC, indicating that she is likely to be highly sensitive to warfarin anticoagulation. She is also homozygous for ApoE4, but does not have familial hypercholesterolemia. She can be expected to have lower risk variants and variants of unknown significance in accordance with expected population frequencies for the conditions under consideration.
Siobhan Dolan, associate professor of clinical obstetrics and gynecology and women’s health at the Albert Einstein College of Medicine, noted that 50 percent of pregnancies are unplanned, so most women do not come in for this type of clinical assessment prior to conception. She also noted that when a woman is not pregnant, many insurers will not cover genomic screening. “For many of our patients, they don’t have access if it is not covered,” she said. Even when a patient has insurance coverage, clinicians have to spend time checking to be sure which genomic tests are included in that coverage. “That is a rate-limiting component of access for many women in this country,” Dolan added.
Targeted Mutation Testing
There is an immediate decision point for the patient and provider on what screening to perform for targeted testing, said Dolan. The American College of Obstetricians and Gynecologists (ACOG) recommends screening for nine conditions that are more common among Ashkenazi Jewish populations. Certain philanthropic programs, however, are currently offering screening for 19 conditions, including diseases in which there is a higher risk in this population, such as cystic fibrosis, as well as diseases in which there is not an elevated risk but for which screening is generally recommended, such as spinal muscular atrophy. The number of tests offered has a tendency to rise over time, said Dolan, which presents educational challenges as well for both providers and patients in understanding the appropriate use of new tests and being able to interpret and act upon results.
With a personal or family history of autism spectrum disorder or intellectual disability, testing for Fragile X syndrome would be considered. Though this patient did not seem to have either of these conditions, some researchers call for offering this screening to everyone, Dolan said, but guidelines today require a history before doing so. Likewise, for spinal muscular atrophy, ACOG recommends offering screening only to women with a family history, whereas the American College of Medical Genetics and Genomics recommends it for all women. “It is difficult for clinicians to act in that setting,” said Dolan. In addition, Dolan would discuss maternal age and the risk for aneuploidy.
One in four Ashkenazi Jewish individuals will be found to be a carrier of at least one condition through genetic screening. This high incidence rate points toward the value of screening in this population. Dolan cautioned, however, that DNA screening may not be appropriate for other high-risk groups because current mutation testing may not be effective outside of Ashkenazi Jewish populations. Dolan recommended that enzyme testing and not DNA screening be offered instead for these individuals. She also suggested that Jewish groups should get enzyme testing in addition to genomic screening, which would present a challenge for whole genome testing. Many genomic tests have enzyme tests used as an adjunct to screening, and if large numbers of conditions were being tested, this adjunct testing could become onerous.
If the woman is identified as a carrier, the partner needs to be tested to have any impact on the pregnancy and the health outcome. But for various reasons, partners often cannot be tested. For example, they might be out of the country for work, overseas in the military, incarcerated, or uninsured and unable to pay for testing. “When we don’t have partners, we certainly increase the maternal anxiety in the pregnancy. We certainly spend lots of money. We don’t necessarily improve any health outcomes. It is a side effect to keep in mind,” Dolan said.
Genetic variants of unknown significance can present major challenges in the preconception and prenatal period. Women need to make very difficult reproductive decisions, which can produce extremely difficult counseling scenarios. Dolan noted, “We traumatize many, many women. These pregnancies become incredibly stressful, whether the outcome is good or bad.”
Dolan said that she would focus on informing the patient of the availability of testing options and supporting autonomy in her decision making. She also would offer partner testing for Tay-Sachs. If the partner is Ashkenazi Jewish and is negative for Tay-Sachs mutations, he has a residual risk of 1 in 560 of being a carrier, she said. He should also be offered the enzyme assay, because 11 percent of Ashkenazi Jewish carriers will be missed if enzyme testing is not done, which will bring his residual risk of
being a carrier to 1 in 1,451 if both tests are negative. The residual risk to the fetus then needs to be reported to the couple: the risk is 1 in 5,800 if the partner has full testing, and it is 1 in 2,240 if he has mutation testing alone.
If both partners are carriers, further counseling is needed. Preimplantation genetic diagnosis and in vitro fertilization is an option for some, but it may be too expensive for others. Additionally, some couples will choose to have an affected pregnancy, which is “an interesting challenge to this whole scenario,” Dolan said. Is the goal of testing to produce informed choice or to decrease the rate of affected pregnancies? This is a deep ethical question at the core of the scenario, said Dolan.
The biggest impact a clinician could have on this patient is to help her quit smoking, Dolan observed. A meta-analysis of 20 prospective studies on preterm delivery comparing any maternal smoking versus no maternal smoking found an odds ratio of 1.27, signaling an increased risk of preterm birth with smoking (Shah and Bracken, 2000). Furthermore, prematurity creates huge health care costs (Russell et al., 2007), and it is a very common outcome, representing more than 12 percent of births (Martin et al., 2012).
Targeted Results Plus Actionable Variants
The second model for the provision of information envisions whole genome sequencing with the return of data relevant only to the current clinical situation and a handful of “actionable variants.” In this case, Dolan suggested that an expanded panel of testing for carrier conditions could be offered, such as the Counsyl Universal Genetic Test, which includes more than 100 conditions (Srinivasan et al., 2010). Testing generally starts with the mother, and the father is then tested for either the full panel or conditions for which the mother was a carrier. These conditions have varying disease prevalence, and the sensitivity and specificity of the testing for each condition vary widely, so each disease has to be considered separately if a result is positive.
This particular patient is reported to harbor a prothrombin gene mutation, which increases her risk of thrombosis, the development of blood clots. The woman’s risk for venous thromboembolism per pregnancy with no history is less than 0.5 percent. If she has had a previous venothromboembolic event, however, her risk would go up to 10 percent; prothrombin gene heterozygotes account for 17 percent of all venous thromboembolisms (Lockwood et al., 2011). On the basis of this information, Dolan would also want to test her Factor V Leiden mutation status because this mutation has also been associated with increased risk of thrombosis and “60 percent of venous thrombosis cases in pregnant women” (Grody et al., 2001). Still, a major question is whether genetic information can signal risk before a sentinel event, but no data are yet available to make that determination.
“With no prior events entering pregnancy, there is really no clear guidance [on] how to treat her other than to watch,” Dolan said.
She would also ask the woman whether she wanted to know about her possible BRCA1 and 2 mutations. Most women, however, do not want to talk about breast cancer while they are thinking about pregnancy. Dolan said, “When you are excited about your new pregnancy, while it is true that your mother and your sister had breast cancer, you may not be receptive to that information at that time.”
Dolan would also suggest possibly examining CYP1A1 and GSTT1 status because one study of 700 women demonstrated that specific genotypes modified the association between maternal cigarette smoking and infant birthweight, suggesting an interaction between metabolic genes and cigarette smoking (Wang et al., 2002). Eleven women with this relatively rare genotype delivered on average 5 weeks earlier. Dolan asked, If the woman had this genotype, could this evidence provide impetus to help her quit smoking? “This is research data, but I do think it could help us target who is at risk and what we could offer them,” she said.
Many of these conditions are rare, so Dolan said that if she had a patient with a particular test result, she would seek out an expert on the natural history of the condition. But given that people all over the country need counseling, rare conditions are an additional challenge for the limited number of counselors available. A related question is how to reimburse genetic counselors and other staff for the huge amount of time that will be spent counseling patients about tests and the results of those tests.
A challenge with this expanded panel is that Counsyl is currently not accepting New York State Medicaid, so it is not affordable for some women, even though the $350 charge is a cost-effective way of testing for all these conditions. The question then becomes whether such testing will widen disparities if certain segments of the population will not have access.
As specific follow-up to the expanded genetic information, Dolan said that she would do a hematology workup and continue to emphasize smoking cessation. She would also consider anticoagulation therapy early in pregnancy, although no clear guidelines exist for the management of these patients, particularly in the absence of any prior event.
The Whole Genome Sequence—What Are We Paying For?
In the third and final model, whole genome sequencing is conducted, and data relevant to the current clinical situation as well as other potentially significant secondary findings are made available to the patient with the best current data for interpretation. Dolan said that as a provider, she would not be excited about having a lot more information. The information will include variants with lower effect sizes and of unknown significance.
Will the woman want to know that she is an ApoE4 homozygote? What if the emotions associated with learning that information affect her decisions about having children? Did her family history reveal any information about the penetrance of that variant? Did her family history suggest any other potential risks? For example, diabetes is a huge public health challenge in the Bronx, which leads counseling strategies toward exercise and diet without genetics playing a substantial role.
The economics of genetic testing can be very difficult, Dolan concluded, because “we don’t really know what we are paying for.” Informed decision making is a laudable goal, but testing and counseling are expensive and will not necessarily lead to fewer affected infants. Nevertheless, the amount of disease that can be prevented is tremendous, as is the excitement surrounding genomics.
Arthur Beaudet, Henry and Emma Meyer Professor and chair of the Department of Molecular and Human Genetics, Baylor College of Medicine, discussed the case from the perspective of a futurist, looking at the kinds of capabilities and information that might be available 20 years from now. At that point, whole genome sequencing conducted in the first trimester using noninvasive techniques is likely to be common, he said. For the woman in the scenario, such testing would identify risk to her offspring caused by inherited conditions. It also would identify genetic risks related to new mutations, such as trisomies, point mutations, and deletions or duplications.
Beaudet divided the effects of genetic mutations into two categories. In the first category are debilitating conditions where individuals cannot live fully independently. Individuals affected in this way typically cannot advocate for themselves.
The second category includes conditions with milder severity. For these disorders, preimplantation genetic diagnosis becomes more of an option, said Beaudet. Examples might include breast cancer mutations, hereditary deafness, and achondroplasia.
From an economic perspective, whole genome sequencing will be more expensive than targeted testing, at least initially, but could be cost-effective if very expensive conditions are avoided, Beaudet said. If information provided to a patient or family from a whole genome sequence is limited, that restriction will be done not by designing a different test but by limiting the information to be shared. Also, from a multigenerational perspective, whole genome sequencing is far more cost-effective. If everyone has a whole genome sequence done at birth, it can be used throughout life, and the relevant information can be applied to other family members.
Beaudet said that it will be important to identify the causative mutations for all individuals with serious Mendelian disorders. This information will allow for a better understanding of the type of variation found for each disease and the clinical utility of identifying various mutations.
De novo mutations that are not present prior to conception, such as trisomies, genetic deletions or duplications, and point mutations, can only be detected through intrapartum testing. Such testing, however, could also produce secondary findings that would pose a challenge to patients and providers. For example, Beaudet said, “we are doing now quite a large amount of prenatal diagnosis using copy number arrays where we encounter copy number variants of unclear significance.”
These secondary findings must be dealt with on a case-by-case basis, he said. He also stated that, in his opinion, more information is almost always better. He added, however, that “this is a personal opinion and not one I recommend for everybody.” But a physician giving a physical does not avoid listening to a patient’s heart because of the possibility of hearing a heart murmur. “We have the information that comes with the society and the technology that we currently live in,” he said.
The behavior of providers must be regulated through provisions such as the 2008 Genetic Information Nondiscrimination Act to prevent abuses. But most patients can be counseled through informed decision making, said Beaudet, even with findings of uncertain significance. Interpretation and annotation, though expensive today, could drop in price as informatics develop. Nevertheless, delivering information to patients will almost certainly involve considerable time and resources.
Beaudet concluded by pointing out that the pediatric community already has considerable interest in whole genome sequencing. The Baylor College of Medicine began offering whole genome sequencing in November 2011, and after several months it received between 10 and 20 samples per month. The most recent month saw 69 samples. Most were from children, but a few were from adults who were looking for an underlying genetic cause for a disease. In about 30 percent of the samples, testing is revealing a disease-causing mutation. The use of whole genome sequencing “seems to be on the rise,” he said.
The woman in the scenario may not know what to expect, said Michelle Gilats, a genetic counselor at the Chicago Center for Jewish Genetics. Though she is likely to have at least heard of Tay-Sachs disease, the patient is unlikely to know much, if anything, about the other conditions for which she is being tested. She is likely to expect that the testing will tell her whether her child will be at risk for certain conditions about which she
may have to make different reproductive decisions. But she may not know what her options are if she is found to be a carrier, and she may not be aware that options exist. Also, she may not be expecting to receive genetic information about her own health, especially about conditions for which the implications are unclear and the significance unknown. Because of this lack of knowledge, pretest counseling is imperative, said Gilats.
Preconception screening to date has been determined mainly by ethnicity. But not everyone knows his or her ethnic background, and many people have mixed ancestry. This situation creates an advantage for larger screening panels because they are more universal in scope and reduce the need to rely solely on patients’ self-assessment of their ethnicity.
The Center for Jewish Genetics uses the Counsyl panel for testing but gives people the option of doing a more limited Ashkenazi panel for 18 conditions. Most people choose the larger panel because it provides more information at the same cost. The downside is that a large panel can provide too much information. For example, it can produce results for conditions that do not have clear-cut responses, such as hereditary hearing loss, or conditions or traits that are not lethal or may not be very life altering. “Yet, because people have the information, they [can] feel they need to act on it,” Gilats noted.
Gilats said that she used to work in prenatal genetics and often encountered patients who had maternal serum screening in pregnancy without being fully informed of the implications of possible results. When she would explain that the results indicated an increased likelihood of Down syndrome, patients could become angry because they had no intention of changing their pregnancy plans and would have refused the test if they had known what it might tell them.
Most patients have a poor understanding of genetics and the residual risk, said Gilats. Even in the well-educated population with which she works, the concept of residual risk for recessive disorders after carrier screening is often misunderstood. For instance, she recently told a patient that she was a carrier for a condition and that her husband was not, yet the patient was still not sure whether she should be concerned. Another patient, a doctor, was confused after being told that she and her husband still had a risk of having a child with cystic fibrosis even though she screened negative. And because many of these conversations occur in a clinical setting after a pregnancy occurs, rather than before conception, decisions made on the basis of this information can be even more difficult.
This lack of understanding will only be exacerbated with whole genome screening, said Gilats. It will not be possible to educate people about all the different disorders and results that are possible. Rather, explanations will have to be broad, outlining categories and examples of results. Having whole genome information for the woman in this scenario, for
instance, would entail determining whether she would want to know her risk of developing certain cancers or currently untreatable diseases such as Alzheimer’s. She would also need to be informed that the results could impact family members as well. Her options would need to be discussed prior to testing and a plan put in place for how, which, and when results would be delivered.
A participant noted that genomic data inevitably raise specific issues involving families. Does a patient want other family members to know about a genetic condition? What responsibility does a physician have to relay information to other members of a family? How will payers respond to various uses of this information?
Different testing and delivery models will lead patients to different actions. Having more information causes patients to ask more questions and spend more time with their providers, discussing their options and recommendations, said Gilats. “The hope is that information can be acted upon, such as with lifestyle modifications or medical intervention. But this won’t always be the case,” she said. Most people need little evidence to be concerned about a specific mutation, whereas a great deal of information is needed to reassure them that their concern is unwarranted.
Whether a result constitutes enough information to cause patients to change behavior remains to be determined, said Gilats, and depends on the specific condition. With conditions where the effects are not known or where a prevention or treatment cannot be recommended, the action a patient would take is even less clear.
In the second and third delivery models, the patient would receive results beyond autosomal recessive disorders. Depending on her specific results, she may need to see a specialist and follow up with extra surveillance or management. What are the costs of this follow-up? Would it be covered by insurance? Even if it is, the copayments alone may be cost prohibitive for some patients, Gilats said.
Costs may also be incurred by family members, either because they also are at risk or because they become a means to further assess a variant of unknown significance. Another potential cost is unnecessary screenings for surveillance purposes, said Gilats. If the patient has a mutation that increases her susceptibility to a common disease such as cardiovascular disease, but there is no family history, are extra screenings warranted?
Interpretations may and likely will change for some variants, thereby changing a patient’s risk over time. Changes in the assessment need to be conveyed to patients, which will require reinterpreting and recontacting patients after the initial results are delivered, said Gilats.
Whole genome sequencing has the potential to deliver great benefits to patients, but the results need to have meaning, Gilats concluded, and the
patients need to want the information that the testing can provide. “With great power comes great responsibility,” she said.
Scott Grosse, research economist and associate director for health services research and evaluation in the Division of Blood Disorders, National Center on Birth Defects and Developmental Disabilities, Centers for Disease Control and Prevention, pointed out that measures of cost-effectiveness and value to patients will not necessarily coincide. They will in a case such as screening for Tay-Sachs disease, where doing so in Ashkenazi Jewish populations is cost-effective and there is high value for the patient, but they may not with thrombophilia testing, where decisions on whether to prescribe anticoagulation medications are not necessarily driven by a risk-benefit balance. Grosse noted that more predictive information regarding thrombosis can be gained by looking at blood type, with Type A or Type B individuals being at two to four times higher risk of developing blood clots, than testing for rare variants such as Factor V Leiden or prothrombin (Dentali et al., 2012; Jick et al., 1969; Medalie et al., 1971; Sode et al., 2013). Nonetheless, blood type data are not factored into decisions for managing patients. The value of information, said Grosse, depends entirely on how and if it is used.
Scott Ramsey, full member in the Cancer Prevention Program, Division of Public Health Science, Fred Hutchinson Cancer Research Center, discussed how the tests mentioned in the scenario might be considered, reimbursed, adopted, and used within the current framework by which most health plans evaluate new technologies. Health plans do review evidence for genomic tests, but the decision to review is usually based on the cost of that test rather than on its purported benefits. For example, BRCA1 and 2 testing is carefully tracked by health plans because it is very expensive. But testing for individual variants such as CYP2C9 “falls under the radar” for health insurance plans because it is relatively inexpensive. If whole genome screening were to fall to a very low price, it could be below the level where health plans have the tools to identify it, though additional and possibly sizable costs will be associated with interpreting and annotating the information because these costs are not predicted to fall at the same rate as sequencing. An increase in use would also raise questions about data storage, noted one participant, particularly about who would house this information and pay for that service.
If whole genome sequencing is recognized by health plans, a major question is how it will be reimbursed. Some tests have codes that are used to identify and pay for them, but most do not, and these are often crosswalked against other tests with existing codes to make reimbursement decisions.
If genomic testing does fall under the radar, health insurance plans will see the consequences of that testing in the form of subsequent tests or procedures. “That is going to be the hardest thing that health plans are going to have to deal with,” said Ramsey. “They aren’t going to be able to pick up the individual test happening, but they are going to see all the downstream impacts in terms of health system use.”
Finally, Ramsey observed that as the costs of sequencing tests continue to fall, multiple companies will be competing with each other to sell this service. One way they will distinguish themselves is through the number of variants that they report, which will create an incentive to create and generate increasing amounts of genomic data for each person. Paul Billings, chief medical officer for Life Technologies, added that in addition to services provided, companies will compete on experience.
Innovation in genomics has been unrelenting, said Billings, driven by unmet needs and current opportunities in the market. Much of that innovation is going on in industry, which has a responsibility to figure out ways to profit from its innovation. James Evans cautioned against allowing market drivers to determine policies for implementing genomic medicine, however.
Veenstra observed that comparative studies can have great value—for example, whole genome sequencing versus standard of care. Understanding the differential impact of using various approaches can provide fundamental information on their value.
The economists also discussed consumer preferences for more versus less information in genetic testing. Some portion of the population will want as much information as possible, while others will resist even information that has high value. To date, demand has existed for tests offering more information, but continued demand will depend on the cost of the testing, who is paying for it, and the consequences of testing.
Ramsey pointed to a coming crisis caused by conflicts between the need for genetic counseling and the resources available for those services. “As these tests provide more and more information, something is going to give. You can’t provide more and more counseling given the limited reimbursement available,” he said. The question then becomes where the people being tested will turn for additional information. Will companies provide that service? Will people use the Internet? Will genetic counselors refer patients to other sources of information? Will the information that patients receive be accurate? And will they be able to make sense of the information? “It really raises a lot of problems,” Ramsey noted. He argued that better tools are needed to convey information to patients in ways that maximize their welfare.
As part of this conversation, Grosse pointed to some of the problems with using QALYs as a measure of health states. They do not necessarily do a good job of measuring people’s willingness to make trade-offs. Also,
they do a poor job of measuring acute end points instead of chronic end points—for example, people may pay a lot to avoid a 3-day food-borne illness, but that health state has virtually no effect on QALYs. And they do not measure many of the things people care about, such as a slight depression of IQ. Grosse asked, “Does that mean that preventing mild cognitive loss has no value to society? Of course not. It is just that the QALY is not designed to capture that type of end point. That’s why we need a fuller set of tools.”
One of the unheralded potential benefits of genetics will be to indicate when something does not need to be done, Ramsey said. For example, many patients who are diagnosed with low-grade, local-stage prostate cancer are treated aggressively despite the fact that 5-year survival for that cohort is very high. “Why are we doing that? We are doing that because men are worried about it, and there is an incentive for urologists to do that. If we could come up with genomic tests that told us with a high degree of certainty that that person was not going to go on and develop advanced prostate cancer, that would save the system billions and billions of dollars,” Ramsey said.
Innovators could also help remove waste and cost from the system by identifying areas in which more precision at the same or reduced price could be found, added Billings. This could be achieved by replacing human variance with quantitative measures.
One participant observed that genomic medicine is evolving within the context of a changing health care delivery system. It would behoove the system and the underlying economics to begin to change to incorporate genomic medicine because eventually it will be part of standard medical practice, whether 10 years or 50 years from now.
Ned Calonge, president and chief executive officer of The Colorado Trust, pointed out that continued expansions of coverage to include items that are cost-effective will produce some improvements in health. Nevertheless, it will also increase the total cost of health care, resulting in “a system we can’t afford.” Thus, the decision to pay for something can have a negative impact by potentially reducing access to care for everyone, especially disadvantaged populations.
Grosse pointed to a disconnect between economics and reimbursement decisions. Many current health care practices are not cost-effective or even based on much evidence, he said. To control health care costs, it would be better “to stop doing things for which there is limited evidence of effectiveness rather than trying to prevent the adoption of new technologies where there is good evidence.”
Calonge noted that newborn screening only looks at a subset of the genetic variants that can be detected using current testing technologies. It may be cost-effective to add variants to the screening panel, but this would add costs for additional testing and interventions as well as bring up issues regarding uncertainty about the effects of the variant being detected. The ultimate problem is that “I know what I am spending, but I don’t know what I am buying,” Calonge said. Genomics will face this economic reality on a much larger scale in the years ahead.
Billings added that some states have added to the subset of detected conditions to develop the information base for possible future testing. And Ramsey pointed out that the value of this information could turn out to be “extraordinarily high” and could be provided without incurring any additional immediate costs. “We could imagine a scenario that before [a test is added], we look at the value that that information would provide, the benefits and costs, before we allow [a test to be performed and results] provided as information to the patients,” said Ramsey.
In relation to the limitations of self-assessment of ethnicity, one participant suggested that whole genome sequencing could be used to make estimations of biogeographical ancestry so that clinicians would have supplemental information. For example, they could recommend enzyme testing.
In response to a question about malpractice litigation arising out of genomic testing, Beaudet said that the possibility exists. For example, if a patient were offered a limited genetic test when a more comprehensive test would have uncovered a serious mutation, could that be considered malpractice?