Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 33
4 Vision of the Future Wylie Burke, M.D., Ph.D. University of Washington Extraordinary promises about the future of genomics in health include individualized health care based on testing for inherited risk, improved clinical management based on molecular characterization of disease, and new therapeutics. Although genomic technology is still nascent, there are already compelling examples of technologies that fulfill these promises. It is becoming standard practice to test for mutations in the BRCA or MLH1 and MSH2 genes to identify genetic susceptibility to breast cancer and colon cancer or to test for genetically based hypersensitivity to certain drugs (e.g., HLA-B*5701 and abacovir). Gene expression profiling of tumor tissue has given providers new information to manage disease and make treat- ment decisions. New therapeutics (e.g., fomiversen, imatinib, trastuzumab) have also been developed based on a genetic or molecular understanding of disease. Despite these important advances in genomic technology that have affected health care and health outcomes, many uncertainties have yet to be fully addressed. How does one match the potential promise of genom- ics with particular health conditions? What strategy will yield the greatest benefit for a given condition? What is the scope of harm from the applica- tion of genomic technologies? How many good ideas will fail during the development process? What will be the effect on the costs of health care, and how will genomic technologies strain the system? Strains to the system are already apparent. There is a tremendous need for rapid assessment of emerging technology, but it is difficult to gather adequate study popula- tions, secure funding, and determine the most appropriate way in which to assess comparative effectiveness. There is lack of trustworthy processes 

OCR for page 33
 INNOVATIONS IN SERVICE DELIVERY IN THE AGE OF GENOMICS that use rigorous and transparent methodology to create guidelines and produce the educational tools to ensure their implementation. Access to genomic technologies is unequally distributed by factors such as geography, socioeconomic status, race, and education level. Genomic screening for colorectal cancer is one possible application of technology that could serve as an example of the potential promises and pitfalls of genomics. Colorectal cancer is the fourth most common cancer in the United States (NLM, 2009), and family history has long been known as a risk factor. Having a first-degree relative with a history of colon cancer almost doubles an individual’s risk, and it shifts the risk to an earlier age. For these reasons, colorectal cancer screening is recommended at age 40 for people with a family history, instead of the generally recommended age of 50. The assessment of family history is not as clear-cut as it may seem, however. There is a continuum of family history. A history of colorectal cancer in the family could be anything from a grandfather diagnosed at age 80, to a grandmother diagnosed at age 65 and a mother with a polyp at age 52, to colorectal cancer at a young age in every generation. Some family histories indicate an increased risk of colorectal cancer, and some do not. About 7 to 10 percent of the population has a family history indicating a moderately increased risk of colorectal cancer, while fewer than 1 percent have “high-risk” family histories (such as those with early-onset cancer occurring sequentially in multiple generations) indicating the presence of rare genetic conditions such as Lynch syndrome and familial adenomatous polyposis (FAP). Assessing family history information can, therefore, be time-consuming. Compounding the difficulty is that family histories are often imperfect because people may not know or may fail to recall the medical problems of relatives. Another potential problem is the case of the “vanishing family history” (Figure 4-1). Thanks to improved screening and treatment, instead of a diagnosis of colorectal cancer at the age of 60, a family member may have a polyp in his or her 50s, get screened and treated early, and avert cancer. Other family members may never know that the individual had an adenomatous polyp (a non-cancerous growth that can progress to cancer); the risk information previously provided by the family history of colorectal cancer is, for positive reasons, no longer available. There are several ways to address the problems involved in relying on the collection of family history to estimate an individual’s risk of colorectal cancer. Information technology could be used to obtain self-administered data, process them, interpret them, and deliver them to the patient and the provider. Electronic medical records from multiple family members could be integrated to provide a more accurate and thorough picture of family history. Alternatively, one might ultimately be able to do away with family history collection and rely solely on genetic testing.

OCR for page 33
 VISION OF THE FUTURE Adenomatous Polyp, 50s Colorectal Cancer, 60 FIGuRE 4-1 Vanishing family history of colorectal cancer. NOTE: Top boxes represent father; circles and squares beneath represent female and male children, respectively. Slash through box indicates that father is deceased. SOURCE: Burke, 2008. R01463 Figure 4-1 As genome-wide association studies identify all of the genetic contribu- tors to colon cancer risk, including common gene variants with modest effects on risk, it will likely be possible to create a genomic test to identify the 7 to 10 percent of people with moderately increased risk, rather than relying on traditional assessment of family history. A test of this kind could have several benefits: It wouldn’t depend on imperfect recall or knowledge of family history, it might result in reduced misclassification, the blood test could be less costly than spending the time to collect family history, and screening of patients could potentially be tailored better than with family history if gene variants in the test panel were associated with polyp dwell time (the time a non-cancerous growth takes to evolve into cancer) or age of onset. There are also pitfalls to using genetic testing. If a provider is perform- ing a genomic test to identify people at moderately increased risk, it is likely that he or she would also want to test for highly penetrant single-gene dis- eases that are also associated with colorectal cancer, such as FAP or Lynch syndrome, and this could raise problems. FAP has a frequency of about 1 in 32,000 in the population and results in a nearly 100 percent lifetime risk of developing colorectal cancer. If it is assumed that the test for FAP performs very well, with a sensitivity, specificity, and negative predictive value of 99.9 percent, .001 percent of those tested will get a false positive result simply because the low prevalence of the condition affects the posi- tive predictive value of the test. In other words, 32 individuals out of every 32,000 tested would be told that they have an FAP mutation when they do not, resulting in further expensive, time-consuming, and uncomfortable testing and medical procedures. If a genomic screen for colorectal cancer

OCR for page 33
 INNOVATIONS IN SERVICE DELIVERY IN THE AGE OF GENOMICS of this type were vastly expanded and offered as a standard of care to all approximately 300 million people in United States, there would be 300,000 false positives for FAP and 9,365 people correctly diagnosed with FAP. This may be an acceptable ratio, but there might be better ways to identify high- risk individuals, such as careful family-based detection after a diagnosis is made. This example indicates that genomic profiling for risk of colorectal cancer is a complex issue, and the benefits and drawbacks must be weighed and considered. In the process, emerging, potentially disruptive technolo- gies would also have to be considered: effective virtual colonoscopy or therapy to inhibit polyp formation could change the risk–benefit analysis for genomic risk profiling. In addition to the advances that create the possibility of genomic profiling for disease risk, there are rapidly expanding opportunities for dis- ease characterization and new therapeutics. Important developments have occurred in gene expression profiling of tumor tissue, and there is strong reason to believe that the development of proteomics and metabolomics will expand the opportunities to use molecular tools for disease character- ization (Gowda et al., 2008; Latterich et al., 2008). Recent genome-wide association studies have exponentially increased the ability to identify genes associated with disease and biological pathways involved in disease processes. This new information is likely to lead to insights into potential therapeutics and drug targets that are necessary for drug development. As these rapid advancements in scientific knowledge arise, it is crucial to ask the question, how much health improvement and at what cost? There is an urgent need for rapid assessment of these emerging technolo- gies in order to obtain comparative effectiveness data as well as to study questions about the cost, access, and harm that are likely to increase with research success. One particular harm that is likely to accompany advances in genomic technologies is the “cascade effect.” The cascade effect, as described by Deyo (2002), occurs when an unnecessary test is performed or a false positive result is obtained from a test that was not clinically necessary. Further testing and treatment follow, and avoidable adverse effects—or even morbidity or mortality resulting from intervention—occur. Currently, the best example of the cascade effect is in the field of radiological imaging. As imaging becomes more sensitive, health care providers are discovering unanticipated incidental findings of uncertain clinical significance. The medical follow-up from these incidental findings may cause greater prob- lems than the problem originally intended to be addressed with imaging. Deyo lists common “triggers” for the cascade effect, several of which are particularly applicable to genomics, including

OCR for page 33
 VISION OF THE FUTURE • shotgun testing, such as large-scale genome profiling or sequencing; • underestimation of the likelihood of false positives; • screening inappropriately simply because tests are available; • errors in interpretation of highly complex information; • patient demand—that is, as potential benefits of genomic tests are publicized, patient demand will be stimulated; and • low tolerance of ambiguity by patients and providers, which leads to further testing. Substantial harms may potentially result from an expansion of genetic risk assessment, whether in the form of testing for inherited risk or charac- terizing a disease process. For example, there is the potential for unwanted information, that is, information that will not help solve the problem at hand and may trigger the cascade effect. In addition to false positive results, there is the possibility of finding variants of unknown clinical significance. In BRCA testing, for example, approximately 6 percent of women of Euro- pean descent with a mutation will have a variant of unknown significance, and a greater percentage of positive results in minority patients will be of unknown clinical significance. Patients are undergoing these tests to answer important questions, and findings of unknown significance do nothing but confuse the situation. Some of the incidental or clinically insignificant find- ings may not only lead to further testing and treatment but actually lead to downstream adverse effects. For example, while testing as part of a smoking cessation program, a physician may find that a patient has genetic variants associated with an increased likelihood of substance abuse. In the future, that physician may be reluctant to provide the patient with adequate pain treatment when needed. The advances in genomic technology, along with the potential harms and benefits, have generated an intensified discussion about the scope and coordination of oversight. The Secretary’s Advisory Committee on Genet- ics, Health, and Society (2008) recently released a report on oversight of genetic testing. The direct-to-consumer movement has resulted in new calls for more regulation. There are interesting developments and continued discussion about accreditation and licensure. There is an increased need for high-quality health technology assessment. All of these trends are likely to continue and illustrate the need for careful assessment of emerging genomic tests and technologies. Eisenberg (1999) wrote, “Technology is rarely inherently good or bad, always or never useful. The challenge is to evaluate when . . . it is effective, for whom it will enhance outcomes, and how it should be implemented or interpreted.” Lessons learned from the health technology assessment process described by Eisenberg can be applied to emerging genomic tech- nologies. For example, innovation and flexibility are needed when con-

OCR for page 33
 INNOVATIONS IN SERVICE DELIVERY IN THE AGE OF GENOMICS ducting genomic technology assessment. Furthermore, assessment must not be limited to randomized controlled clinical trials but rather must be an ongoing process that uses a variety of study methods. Assessment must also examine various outcome measures such as quality of life. The process must incorporate information from the real world of health care and avoid redundancy by encouraging sharing of knowledge. In addition, it is criti- cal that those who are assessing genomic technologies engage with both providers and the public. The methods used in developing guidelines for decision making about when and how genomics enters the health care system must be independent, transparent, and trustworthy. The resulting guidelines and associated edu- cational material need to be made available in direct and simple language that is readily accessible by the public and health care providers. DISCuSSION Wylie Burke, M.D., Ph.D., Moderator One participant asked that Burke envision the year 2020 and describe her idea of the role of a genetics professional. Burke first noted that many of the ideas discussed earlier in the workshop were quite relevant to this question, for instance, delivering testing via the web or directly to con- sumers. She added that, ideally, this type of delivery system would be linked to high-quality, assessment-driven guidelines. It is also crucial that there be appropriate follow-up using a system that is convenient for patients and providers. Burke referred to a previous statement made by Levin about the amount of time it can take a patient simply to get to a clinic and agreed that receiving genetic results via the web would sometimes be more effi- cient. She stated that the pharmacogenetics ordering system at Cincinnati Children’s Hospital was a very interesting model, and as more and more tests are developed, this kind of assistance will become evermore important for providers. Finally, Burke stated that reimbursement schemes will have to be redesigned to permit genetic counselors to be reimbursed for giving advice to primary care providers. With the small number of geneticists and the increasing amount of genomic technology, there needs to be a systems- level change to allow reimbursement for this type of expert consultation. Another participant cautioned that one of the main risks of new inno- vations is premature belief in their effectiveness. The health care community must take a very critical look at what the data tell us before making recom- mendations about the use of new technologies. Frederick Chen brought up

OCR for page 33
 VISION OF THE FUTURE the example of the full-body CT (computed tomography) scan, noting that it was a technology that was marketable because it made practical sense and had intuitive value to the average consumer. Because clinical utility did not exist, however, neither the medical community nor the insuring community supported the use of this technology. The result was that full-body scans were not nearly as successful as business models suggested they would be. Full-body CT scans are just one example in a long line of failed assump- tions in medicine. The medical community, noted Chen, has two options for assessment: Either wait until these new technologies are in practice or collect evidence prior to implementing the new technology. One participant asked Burke whether genetics is different from other types of highly cognitive areas of practice. Burke responded that at present two things are particularly important about genetics and genomics. One is the dramatic pace of discovery. The other has to do with risk assessment. Although most genetic risks are similar to other kinds of risks (e.g., cho- lesterol level, blood pressure), there are two ways in which genomic risk assessment may pose particular problems: the breadth of the proposed risk assessment process and the fact that some data suggest that people respond differently to risk assessments delivered by a DNA test versus a family history. A participant added to this discussion by noting that genetics is unique from other forms of medicine because there is an opportunity for interventionary genetic therapies and genetic manipulation, which are new and potentially dangerous uses of scientific knowledge. One participant stated that the American Health Information Commu- nity is attempting to gather and make evidence available by centralizing and standardizing decision support systems. He also mentioned the efforts of Evaluation of Genomic Applications in Practice and Prevention (EGAPP), which looks at and attempts to answer questions such as: What is the level of certainty? How large are the impacts? How do they compare to various alternatives? The participant emphasized that information needs to be pre- sented in a systematic manner in order to help make decisions. Shields commented that reimbursement policy can be used as a very effective lever to incentivize the provision of genetic services. For example, many years ago in Massachusetts it was extremely difficult to provide health care for the homeless. Once Medicaid allowed for reimbursement of health care provided in a homeless shelter, that situation changed. Data that are being collected on genetic services are generally used for reimburse- ment rather than to track the clinical effect of genetic medicine on health outcomes. The coding used for reimbursement simply notes that a genetic test was conducted, not the specific type. Shields suggested that advocates, private payers, and the Centers for Medicare and Medicaid Services (CMS) work together to add routine collection of clinical information that could be used for long-term understanding of clinical efficacy.

OCR for page 33
0 INNOVATIONS IN SERVICE DELIVERY IN THE AGE OF GENOMICS Naomi Aronson, a representative of Blue Cross/Blue Shield and a member of the Roundtable, responded that payers are also concerned about coding and are open to the idea of building other coding systems with more room for clinical information. Representatives of testing laboratories noted that attempts had been made in the past to use modifier codes, but disagreement between the laboratories and the payers resulted in discon- tinuing those efforts. One participant stated that while it is possible to put enough SNPs together to obtain a risk ratio for several diseases (e.g., prostate cancer, type 2 diabetes, breast cancer), that ratio is not enough to establish clinical utility. Unless a result changes the course of prevention or treatment, genetic testing does not provide any new information for either the provider or the patient. Another participant said that the standard provider reaction to a genetic test is to give the patient the same preventive message given to all patients, even when the patient’s genes signal that there is something dif- ferent about his or her biology and that standard preventive measures may not have the same effect for that patient. Another participant added that taking action on very low relative risks could be quite dangerous and that only higher relative risks of 9 or 10 should be used. It is also necessary to ensure that the data are correct. Finally, another participant said that there is money in the system for finding gene variants but little money or other resources for the much more expensive task of determining whether or not these variants have any clinical utility. In a final comment, one participant noted that despite all of the research and the committees that are developing new data, medical practice ulti- mately comes down to the physician, who must sort through a variety of information on many factors, filter out that which is not useful, and make decisions for each patient. He said that physicians are generally hesitant about new innovations.