How Can We Move Clinical Beyond the Hype?

Timely Analysis of Immediate Health Policy Issues October 2011 Michael L. Millenson

“All of the fruits of the tremendous explosion in innovation that’s been occurring in biomedical research – which make the molecular metamorphosis possible – fulfill their purpose only when they are translated into interventions and solutions that are applied to patients.” —Dr. Andrew von Eschenbach, U.S. Food and Drug Administration Commissioner, 2006

Introduction or no value to be marketed to challenges beyond the direct doctor- providers and patients. The result of patient interaction. Health More than a decade after the this policy vacuum is a disturbing information technology systems successful mapping of the human paradox. Genomics as a field is typically possess neither the ability to genome, clinical genomics1 is beginning to demonstrate its present genomic information in a starting to permeate important parts profound potential to improve health structured way clinicians can use nor of patient care and propel ripples and health care. At the same time, the the large data storage capabilities throughout the entire U.S. health care evidence suggests that the current use such information demands. system. of clinical genomics tests and Comparative effectiveness research often contributes to needs to focus on identifiable Although the field has fallen far short wasteful spending and even patient subpopulations where clinical of the transformational therapeutic harm. genomics interventions can have the impact once widely predicted, most powerful impact, but the particularly in regard to common A 2008 article in Health Affairs concept of “identifiable” can raise , specific interventions have warned that the low threshold for awkward questions about race and crossed the divide between rarity and allowing “unsubstantiated [genomic] ethnicity. regular use. routinely technologies to enter into practice guides in several common [has] the potential to overwhelm the Finally, the science involved is cancers and in HIV , and use health system.”3 Since then, private- complicated and often confusing to of costly molecularly targeted anti- sector efforts to encourage use of patients, payers, providers and cancer drugs is rising sharply. Six in clinical genomics have intensified, policymakers alike, making informed 10 primary care have while efforts to ensure appropriate decisions that much more difficult. ordered a genetic test.2 use have lagged. Evidence-based guidelines for clinicians remain Controversies over appropriate , the science of scarce, despite growing efforts to genetic testing related to the risk of how genetic differences can affect address the problem, and those that have already drawn in individual drug responses, has do exist can be contradictory. A all three branches of the federal become an established part of drug federal advisory committee that government and numerous private discovery and of the guidance given pushed for greater regulation and players. As issues related to clinical to physicians on the labels of drugs more reliable data had little impact genomics become a regular part of approved by the Food and Drug and was disbanded last year. access, cost and quality discussions, Administration (FDA). Meanwhile, government funding for it is time for a policy community long fluent in the argot of DRGs and Yet while increased adoption is assessment of the evidence on genomic interventions has plunged. billing codes to acquire similar partly due to growing acceptance of proficiency in the language of DNA useful interventions, it is also due to Integrating clinical genomics into and genetic codes. This paper aims to a porous regulatory structure that routine care poses multiple system assist that process by examining what allows diagnostic tests of unproven

is hype, what is hopeful and what is The result has been a surge in and opioid agents.9 starting to make a genuine difference scientific discovery unparalleled in in clinical genomics. It also makes Western .5 Consider: , chromosomal abnormalities, proteins in the blood and similar specific recommendations about the type of information that would help • More than 3,000 genes have been indicators are called biomarkers, and linked to Mendelian (single- they are the key to clinical genomics. policymakers, payers, providers and 6 the public make better decisions ) disorders. For example, certain proteins are about the use of clinical genomics. associated with more aggressive • Genome-wide association cancers, while other biomarkers can The Beginning of the studies, searching beyond single- predict the response to a medication. gene disorders, have found Bench-to-Bedside Journey significant correlations between In 2003, the FDA began asking companies to voluntarily submit When a draft map of the human more than 1,200 common genetic some types of biomarker data related genome was announced in 2000, variations and 210 traits. These to clinical usage in order to enhance followed by a complete map in 2003, ranged from the clinically drug safety and effectiveness. In breathless predictions of imminent important (abdominal aortic cases where having biomarker breakthroughs blossomed aneurysm) to the curious (the information is critical to safe and everywhere from opaque scientific ability to smell asparagus in 7 effective use, the FDA would not journals to the set of Oprah. If the one’s urine). allow voluntary submission. The expected clinical cornucopia has not • Different genes code for different FDA also now requires drug yet materialized, other metrics show enzymes. One family of companies to ensure that a substantial results. The $3.8 billion enzymes, cytochrome P450 companion diagnostic test is the U.S. government invested in the (CYP450), has been identified as approved with the drug approval. from 1988 involved in the metabolism of 90 through 2003 helped spark $796 percent of all drugs.8 Just one As of June 30, 2011, the agency had billion in economic impact and $244 member of this family, CYP2D6, approved genomic biomarker 10 billion in total personal income. In is involved in the metabolism of information for 76 unique drugs, 2010 alone, genomics research and 100 different drugs, including including the following: industry activity directly and analgesics, anti-arrhythmics, 4 • Twenty-six different genetic indirectly supported 310,000 jobs. antidepressants, beta-blockers biomarkers affecting 18 different

Precision Labeling for “Precision Medicine” therapeutic areas ranging from anti-infectives to . Language shapes perceptions, but not always accurately. Proponents of “personalized” and “predictive” medicine tend to use those terms within a • Common medications used to framework of DNA determinism that pays little more than lip service to individual treat heart disease, depression characteristics unrelated to genetic biomarkers. Yet 40 percent of the factors and pain. contributing to premature among Americans are linked to behavior, 30 percent to genetics, 15 percent to social factors, 10 percent to medical care and 5 • A preponderance of percent to environment.1 A personal genotype remains a much less powerful drugs, with more to come: the predictor of future health than a good family history. oncology drug pipeline contains an estimated 300 Phase II or As genomic medicine enters the health care mainstream, those in the field should higher candidates with the consider a description of their role that recognizes the importance of other potential for testing against a approaches. Terms such as “genetically ” or “personalized 11 genetic medicine” would make policy discussions about “precision medicine” a genetic biomarker. great deal more precise. ______

1McGinnis JM, Williams-Russo P and Knickman JR. “The Case For More Active Policy Attention To Health Promotion.” Health Affairs, 21(2): 78–93, 2002.

Timely Analysis of Immediate Health Policy Issues 2

Genomic Science: Harder Than It First Looks Almost every cell in the human body contains two copies of the genome, with one each from the father and mother. (Egg and sperm cells contain only one genome copy, while red blood cells have none.) The genome itself is a kind of “book of life” composed of 23 chapters (chromosomes), each containing a thousand or more stories (individual genes), and each of those stories is made up of base-pair DNA “words”—a total of nearly three billion of them.1 Rather than being static, though, the genome is dynamic. Genetic code translates into a functional biological effect when the gene is “expressed” as a protein. When this happens, the underlying “genotype” (the total information in the genome) is actualized as a “phenotype” (the manifestation of an individual’s genome, including characteristics ranging from eye color to a disease state). A cause-and-effect relationship between a specific disease and a specific protein was first established by Linus Pauling and colleagues in a seminal 1949 paper on sickle cell anemia. Pauling confidently predicted “significant progress in the field of medicine as it is transformed from its present empirical form into the science of molecular medicine.”2 Six decades later, testing for sickle cell disease has improved, but there have been few advances in treating it. Despite the mapping of the human genome, there are three major reasons clinical progress has been so difficult.

• There is much about the human genome that we still don’t know. There is roughly a 99.5 percent similarity between the genomes of any two human beings.3 These remaining differences amount to more than 16 million DNA variants, and although this number is tiny in relative terms, it is these variations that likely contribute to common diseases. Tests that sequence only a part of each genome may miss the bigger picture. has great potential, but given the variation in human genomes, identifying a specific gene variant responsible for a particular condition remains extremely complex.

• What a disease looks like in the exam room can be very different when seen from a genetic viewpoint. At a genetic level, the appearance of disease commonality can vanish. For instance, hypertrophic cardiomyopathy—a thickening of the heart muscle—looks like one disease on an echocardiogram, but 20 different genes could be involved. Even untangling complex genetic interactions may not yield much care improvement: “Type 2 is the most genetically understood and dissected disease out there,” notes James Evans, editor-in-chief of Genetics in Medicine. “Yet from a clinical standpoint, the [genetic] information is of questionable utility.”4

• Environmental factors can actually change the way an individual’s genetic code is expressed. “Why Your DNA Isn’t Your Destiny,” is the way a January 18, 2010 Time magazine cover story summarized the impact of epigenetics. Environmental factors can alter the “epigenome” that sits on top of the genome and tells genes to switch on or off. Given the importance of diet and stress in epigenetics, the saying “you are what you eat” may be figuratively true at a molecular level. Put all this together and the genomic medicine challenge once thought to be like a jigsaw puzzle more closely resembles a Rubik’s cube. Eric Green and Mark Guyer of the National Human Genome Research Institute write, “Our ability to generate [genomic] information has outpaced our ability to analyze it.”5 ______

1 Ridley M. Genome: The Autobiography of a Species in 23 Chapters. New York: Perennial, 2000. 2 Pauling L, Itano H, Singer SJ, et al. “Sickle Cell Anemia, a Molecular Disease.” Science, 110: 543–548, 1949. 3 Levy S, Sutton G, Ng PC, et al. “The Diploid Genome Sequence of an Individual Human.” PLoS Biology, 5(10): e254, 2007. Note: Others have used a figure of 99.9 percent similarity. 4 Evans JP. “Health Care Professionals and Personalized Medicine.” (Paper presented at Personalized Medicine in the Clinic Conference, Phoenix, AZ, March 8, 2010.) 5 Green ED and Guyer MS. “Charting a Course for Genomic Medicine from Base Pairs to Bedside.” Nature 470: 204–213, 2011.

The availability of genetic biomarker treatment decisions, and monitor data on FDA-approved labels and Dissemination With and therapy. However, two major barriers information on diagnostic tests from Without Validation stand in the way of widespread use: organizations such as the National proof of test validity and clinical Clinical genomics tests have carved Institutes of Health (NIH) suggest an utility. Test validity—whether a test out an increasingly important role in orderly process of translating accurately measures a biomarker—is improving the quality of patient care. research into patient care. The reality an issue because current federal The tests are now used to make a is far less systematic. regulation only requires that these prognosis, confirm a diagnosis, tests measure what they say they do, adjust a medication dose, make not whether what they purport to

Timely Analysis of Immediate Health Policy Issues 3

measure is of the clinical significance Failing to Protect Patients from false reading on an ovarian that is claimed. More broadly, the Harm cancer test. science underlying many studies of genetic biomarker significance has In an April 2008 report by the • A test for Down syndrome that 12 Department of Health and Human was discovered to be flawed just also been challenged. Proven 18 clinical utility matters because Services, the Secretary’s Advisory days before going to market. physicians won’t use a test unless Committee for Genetics, Health and Society recommended enhanced At a March 2011 meeting sponsored they believe it provides actionable by the National Comprehensive information that can help patients. oversight of LDTs because of serious cost and quality concerns.15 One Cancer Network, one The FDA has generally exercised example was inaccurate test results spoke of lung cancer patients what it calls “enforcement intended to diagnose a serious routinely receiving “180 or 190” discretion” in regard to so-called genetic condition in some individuals molecular tests despite a “lack of evidence” of the tests’ clinical laboratory-developed tests (LDTs). with lower back pain. Opponents of 19 In plain English, the FDA has more regulation, led by pathologists validity. A few months later, a decided not to regulate them. and clinical labs, warned that greater page-one article in the New York Historically, LDTs have been simple, oversight could delay introduction of Times detailed how IVDMIAs from well-understood tests designed by innovative new tests, and the FDA in respected academic labs produced pathologists to enable the medical June 2009 settled for publishing inaccurate treatment recommendations for cancer staff to diagnose rare conditions. suggestions on how labs could 20 Nicknamed “home brews,” they have voluntarily improve quality.16 patients. As of mid-October 2011, a low risk of patient harm. At the regulations promised in June present, LDTs need only comply A year later, the agency altered 2010 had not yet appeared. with the good manufacturing course, citing “ Separately, Senator Orrin Hatch (R- practices set out in the Clinical concerns” in a Federal Register UT) is reportedly preparing Laboratory Improvement notice that signaled an intent to legislation on this issue. Amendments (CLIA).13 CLIA regulate. “While the absence of FDA oversight may make it easier for The Direct-to-Consumer Market requires that a lab show it is able to Draws Attention perform the test accurately, not that laboratories to develop and offer tests the test itself is valid. on a rapid timeline…diagnostics The FDA, Congress and others have critical for patient care may not be paid more attention to the much However, many of today’s diagnostic developed in a manner that provides smaller, direct-to-consumer (DTC) tests are as far from “home brews” as a reasonable assurance of safety and 17 genetic test market, where questions breakfast at a fast food franchise is effectiveness,” the agency wrote. about accuracy and usefulness have from Mom’s homemade muffins. For Indeed, a July 2010 article by the generated national publicity and example, in vitro diagnostic FDA commissioner and the NIH Congressional interest. DTC tests multivariate indexed assays director added new problems caused range from the well-established (IVDMIAs) produced by commercial by unreliable LDTs to the Secretary’s (paternity tests) to the pseudo- labs examine multiple genes and use Advisory Committee’s list. The scientific (predicting a child’s sports complex algorithms to produce a risk additional issues included: abilities) to the debatable (genome- score predicting prognosis for a wide scans offering information on disease or drug response. While test- • Women falsely informed they were negative for the BRCA everything from ancestral origins to makers can voluntarily seek FDA caffeine sensitivity to predisposition approval, perhaps hoping for a , which conveys a high risk of breast cancer. to various diseases. Supporters point marketing advantage, fewer than two to the tests’ personal utility, while dozen of more than a thousand • One woman whose ovaries were skeptics refer to recreational genomic tests on the market have 14 removed unnecessarily due to a genomics; the truth may depend on FDA approval. what test is used for what purpose.

Timely Analysis of Immediate Health Policy Issues 4

The BRCA Battle: Hope, Hype, Fear, Money, Politics and Science

The economic, clinical, legal and political questions swirling around testing for the BRCA have involved all three branches of the federal government and numerous private players. It’s a story that displays both sides of the clinical genomics coin: personalized, predictive, preventive and precise medicine that can also be political, profit- and plaintiff-driven, and perplexing to both patients and professionals. The different actors involved are set forth in italics below. Women who inherit mutations in the Breast Cancer 1 and/or 2 (BRCA1 and BRCA2) genes have about a 50 percent lifetime risk of developing breast cancer and a 40 percent lifetime risk of ovarian cancer.1 Men with one or both mutations run an increased risk for prostate and even breast cancer. Both sexes have a higher risk of pancreatic cancer. The BRCA genes were cloned at the University of Utah in the early 1990s and quickly licensed to and patented by Myriad Genetics. An estimated 20 percent of the human genome is patented, and Myriad’s patents have been challenged in a long-running lawsuit pitting the American Civil Liberties Union, the Public Patent Foundation and the Association for Molecular against the U.S. Patent and Trademark Office, the University of Utah Research Foundation and Myriad. Myriad lost in a lower court in 2010, but the loss was partially reversed in July 2011 by a federal circuit court.2 That ruling, in turn, is being appealed to the U.S. Supreme Court. Deciding Who Gets the Test The financial stakes are high. Myriad sells its BRCAnalysis test for about $3,300 per patient, and it generates about $350 million in annual revenues. The U.S. Preventive Services Task Force (USPSTF) and the Centers for Disease Control and Prevention (CDC) both recommend BRCA testing only for those whose family history indicates increased risk. The USPSTF guidelines include seven different clinical scenarios, but only one in five primary care physicians could correctly identify what to do in all of them.3 The absence of easy-to-use and reliable decision support is emblematic of the situation for less-publicized interventions. The Agency for Healthcare Research and Quality (AHRQ) commissioned research on a Web-based BRCA decision support tool for doctors and patients alike, but it had no money to fund full development. A tool that draws on the AHRQ work is being developed elsewhere drawing on federal funding for breast cancer awareness. Legal Issues In 2008 an Illinois appellate court ruled in a malpractice case4 involving BRCA testing that cost concerns are not a defense for recommending against a genetic test. That decision has led some analysts to worry that malpractice fears will prompt unneeded testing. A separate legal and ethical question is whether doctors with a BRCA-positive patient are obligated to contact non-patients who are that patient’s relatives.5 Meanwhile, the one lawsuit filed in connection with the Genetic Information Nondiscrimination Act, passed in 2008 to protect individuals from actions by health insurers or employers, involved a BRCA-positive woman who allegedly lost her position at work after undergoing bilateral prophylactic mastectomy.6 The BRCA Test in Context With all the attention paid to the BRCA mutations, it’s easy to forget that they occur in just two out of a thousand women in the general population and account for less than 5 percent of all breast cancers. To put those numbers into context, two hours per week of brisk walking can reduce the risk of breast cancer by nearly 20 percent for the 99.8 percent of women without the BRCA gene.7 Perhaps the most important lesson, then, is that clinical genomics interventions should be seen as part of a larger picture. They are a means to an end. Appropriate use adds value and improves care; inappropriate use adds cost and can threaten health. ______

1 BRCA1 and BRCA2: Cancer Risk and Genetic Testing. Bethesda, MD: National Cancer Institute, 2009 www.cancer.gov/cancertopics/factsheet/Risk/BRCA. 2 Pollack A. “Ruling Upholds Gene Patent on Cancer Test.” New York Times, July 29, 2011. 3 Bellcross CA, Kolor K, Goddarfd KA, et al. “Awareness and Utilization of BRCA1/2 Testing Among U.S. Primary Care Physicians. American Journal of Preventive Medicine, 40(1): 61–6, January 2011. 4 Downey v. Dunnington SIU. Appellate Court of Illinois, Fourth District, June 13, 2008, caselaw.findlaw.com/il-court-of-appeals/1411888.html. 5 Surbone A. “Social and Ethical Implications of BRCA Testing. Annals of Oncology, 22(suppl 1): i60–i66, 2011. 6 Lowe Z. “Woman’s Complaint Could Become Test Case for Genetic Discrimination.” Law.com, June 2, 2010, www.law.com/jsp/law/LawArticleFriendly.jsp?id=1202469829813. 7 Breast Cancer. American Cancer Society, February 9, 2011, www.cancer.org/cancer/breastcancer/detailedguide/breast-cancer-risk-factors.

Timely Analysis of Immediate Health Policy Issues 5

A study of five companies selling A Lack of Data routine practice remains a genome-wide scans found they challenging task. offered information on 213 Unfortunately, the actual usage, cost conditions, with 20 conditions (such and growth rate of clinical genomic Putting Clinical Genomics as Alzheimer’s, heart disease, type 2 tests are impossible to independently into Practice diabetes, psoriasis and restless leg ascertain. That’s because individual syndrome) having the greatest genetic tests are not billed as separate Because clinical genomics can be overlap.21 Different companies use items with the Current Procedural complex and confusing, different gene combinations and Terminology (CPT) codes used for dissemination into patient care predictive algorithms. Predictions insurance reimbursement. Instead, requires building an infrastructure made by two of the most prominent billing necessitates an idiosyncratic that allows clinicians to appropriately companies about seven different “code stacking” that obscures what order, use and be paid for tests and diseases for five different individuals was ordered. However, a sweeping therapies. When a biomarker, a agreed less than half the time, one update in CPT codes that takes effect patient population that could be study found.22 A government report in 2012 is expected to include more helped and the benefit itself are well- called DTC genetic tests “misleading than 90 gene-specific and genomic defined, adoption is easier. Today, and of little or no practical use to procedures in the Tier 1 category of genetic testing routinely helps define consumers.”23 commonly performed tests, with therapeutic choices for some cancers additional codes to be added in 2013. (lung, colorectal, breast and Supporters say the tests can motivate The update will make it much easier leukemia), heart transplantation and a consumers to change behavior, for clinicians to order and be number of other conditions. In HIV although studies of the behavioral reimbursed for common genetic tests disease, genetic testing to avoid a 24 impact are equivocal at best. and will also generate data on usage. potentially life-threatening reaction Moreover, the tests’ uncertain The most recent effort by researchers to the drug abacavir has become the accuracy also means some consumers to even roughly estimate test usage standard of care. will ask their doctor for unneeded seems to have been 15 years ago, additional testing or, conversely, when a mail survey completed by Still, as one overview put it, “The falsely feel protected. In 2010, 245 clinical labs reported 175,314 number of genomic markers in following publicity about plans to put genetic tests during 1996.26 clinical practice is very small. The a genomic test on drugstore shelves, number of markers to guide treatment 28 the FDA reasserted its enforcement There is not even an accurate count decisions is even smaller.” There is authority. So far, it has sent letters to of the number of genetic tests an almost complete lack of evidence 27 DTC companies asking them to available to be billed for. The from traditional randomized demonstrate that their claims meet Secretary’s Advisory Committee controlled trials, forcing reliance on legal standards. endorsed a mandatory test registry at other forms of evidence. Over the the same time it called for increased past five years an estimated two This regulatory boldness may have test regulation. The committee dozen clinical genomics evidence been helped by the fact that the DTC suggested that CDC, CMS or FDA reviews have been funded by market accounts for less than $20 might be an appropriate lead agency agencies such as CDC, CMS and 25 27 million in total revenues and to manage the registry. More than AHRQ (including through its involves mainly small companies. By three years later, a Genetic Testing USPSTF), yet very few have been comparison, the larger molecular Registry is set to be launched shortly; able to unequivocally determine diagnostics market produces an it will be voluntary rather than clinical utility, even for interventions estimated $6.2 billion in revenue for mandatory and will be managed by that have made their way into clinical laboratories, according to the NIH, which will phase out its practice.29 research firm G2 Intelligence, and GeneTests site. Meanwhile, the has grown rapidly from 14 percent of Secretary’s Advisory Committee was Meanwhile, the funding for new clinical lab revenues in 2006 to 21 itself phased out last year when its research dwarfs the monies available percent in 2009. charter was not renewed. to translate existing research into improved care. Consider the Registries notwithstanding, following examples: integrating clinical genomics into

Timely Analysis of Immediate Health Policy Issues 6

• Less than 3 percent of genomics insurers also review selected tests two tests, while widely used, have a articles in the peer-reviewed and drugs. less solid evidence base. Though literature between 2002 and 2006 patients expect their doctors to be included information on It is not surprising that different able to explain the rationale and translating research into groups reach different and even workings of genetic tests,37 they’re 30 contradictory conclusions. In 2007, practice. likely to be disappointed. While 98 with much fanfare, the FDA percent of physicians agree that • Just 1.8 percent of grants in approved a genetic test for sensitivity genetic testing is useful in guiding cancer genetics for 2007 by the to warfarin, a commonly used and drug therapy, just 10 percent feel National Cancer Institute went to clinically important blood thinner. adequately informed about using it.38 translational research, and a scant However, in 2009, CMS reviewed Even where evidence is available, it 0.6 percent of publications on the the science and professional group has not been utilized.39 Patients and topic dealt with translational recommendations and decided 31 providers alike guess at which issues. Medicare would not cover the test’s interventions are investigational and use. On the other hand, even before which are not. One large Blue Cross • The federal office overseeing the testing for the KRAS mutation to plan has posted its genomic medicine Evaluation of Genomic guide colorectal cancer therapy was coverage policies online, but it is an Applications in Practice and embraced by the FDA, some private exception.40 Prevention (EGAPP) Working payers decided to reimburse for it.33 Group—a national effort to apply As expensive anti-cancer evidence-based medicine to Another example involves breast that provide substantial benefit to a genomics—recently had its cancer. The Blue Cross TEC small but genetically identifiable budget cut by 90 percent. determined that Oncotype Dx, one of subpopulation become more Funding for CDC’s Office of two major genetic tests that can guide common, the clinical and cost Public Health Genomics breast cancer therapy, met its consequences of reimbursement plummeted from $12.3 million in evidence standards, but the similar decisions will rise. For instance, a fiscal year 2010 to an expected MammaPrint test did not. Yet the recent study of a drug that targets the $797,000 in fiscal year 2012. MammaPrint test was FDA- enzymes involved in blood clotting approved, while Oncotype Dx was (apixaban) was hailed as “one of the NIH recently announced plans for a never submitted for FDA approval. most significant advances in National Center for Advancing (As noted earlier, FDA approval is cardiovascular medicine in the last Translational Sciences that includes not required.) Meanwhile, CDC’s five years” for its impact on mortality genomic medicine, but its impact is EGAPP Work Group concluded there and complications.41 Yet when it unclear. In contrast, the influence of was evidence for both tests, but not comes to getting these types of the private sector continues to grow. 34 enough to make a recommendation. medications into regular use, Randomized trials are underway, but Contradictions and Confusion researchers recently concluded that test usage continues to grow in the “no medical literature exists One of the most influential private- meantime. The company making suggesting how to formally integrate sector voices is the Blue Cross and Oncotype Dx says 50,000 patients [pharmacogenomic] information into Blue Shield Association’s got the test in 2009 at a retail cost of 35 the formulary decision process” used Technology Evaluation Center about $4,000 per patient. by hospitals, medical groups and (TEC), whose recommendations go 42 A national survey of randomly health plans. to health plans serving 100 million selected primary care physicians members. From 1997 through 2010, Guidelines and Infrastructure found that 60 percent had ordered the TEC evaluated 24 genomic tests Building one of four common genetic tests, and therapies, including specific either for breast or colon cancer or One solution is to improve guideline interventions and broad topics such for Huntington’s or sickle cell development and dissemination to as genetics in prostate cancer. About disease, and 74 percent had referred a practitioners. Prominent efforts two-thirds of evaluations related to 36 32 patient for genetic testing. The include those of the Clinical oncology. Large commercial latter two tests are for well-defined Pharmacogenetics Implementation single-gene disorders. The former Consortium, affiliated with NIH; the

Timely Analysis of Immediate Health Policy Issues 7

American Society of Clinical reimbursement. For instance, could shrink those requirements Oncology; and the National biomarker information in drug considerably. Comprehensive Cancer Network. A interaction databases provides value few large health systems, such as only if the patient’s genetic Genomic Therapy Goes Direct-to- Intermountain Health Care, are trying information is known. Kaiser Consumer to build guidelines related to clinical Permanente and the University of As with diagnostic tests, treatment genomics interventions into their California–San Francisco have options based on genetic test results electronic medical records (EMRs). genotyped 100,000 Northern have a direct-to-consumer (DTC) California Kaiser enrollees under an component that skirts traditional A JAMA commentary called for the NIH grant to link that data genomics community to reach out to gatekeepers. The following are just a electronically to health surveys, other few examples: clinicians with “unambiguous biological data and disease evidence that links the use of registries.45 (Kaiser is also part of a • PatientsLikeMe, a social media genomic information to improved larger pharmacogenomics evaluation site, enables tracking of genetic patient-centered health outcomes, effort by the 15-member HMO variables, treatment choices and particularly for common 43 Research Network.) The Department patient-reported outcomes for conditions.” In the meantime, the of Veterans Affairs recently ALS. for-profit sector is building its own announced the “Million Veteran implementation infrastructure: Program,” an effort to consolidate • Genomera, a start-up, plans to use “crowd-sourced health • McKesson, the giant genetic, military exposure, health and lifestyle information in a single science” to enable patients to pharmaceutical distributor, is carry out genomic research. creating electronic guidelines, database in order to study billing codes and a database to correlations between genetic • Cancer Commons, an open- track insurer payment policies. variations and disease. science initiative, helps patients CVS-Caremark and Medco, two Separately, a national collaborative individualize treatment based on of the largest benefit led by the Coriell Institute for their tumor’s genomic subtype managers, are similarly involved Medical Research is trying to and then rapidly disseminate in making it easier for providers determine whether patient-friendly what they’ve learned to to order and be paid for genetic genetic information, including professionals and other patients. interventions. genetic counseling, will lead • N-of-One and other consultants individuals with chronic disease to • DNA Direct, a Medco 46 offer to help patients find subsidiary,44 provides a change their behavior. Scalability customized, genetic cancer- may be an issue: there are only about “turnkey” service to health plans fighting therapies. 2,000 certified genetic counselors in and hospitals. There is a well- the nation. defined menu of genetic tests, The extent to which these diverse activities will contribute to value- phone and Web-based advice for Integrating genetic data into an EMR professionals and patients, and enhancing use as opposed to just for routine therapeutic use poses increasing volume and costs remains guidance on what actions to take another challenge. Few systems have based on test results. to be seen. The key is whether the capability to process or structure clinical genomics continues to be genetic data; even at academic • Navigenics, with roots in DTC treated as a separate phenomenon or medical centers, clinicians often must testing, is targeting primary care is recognized as one more enter genetic data manually. physicians in a partnership with technology, albeit a potentially Moreover, storage requirements for a national concierge medicine extraordinary one, to be integrated patient’s entire genome will be orders company MDVIP (a subsidiary into the health care system. of magnitude larger than traditional of Procter & Gamble). biomedical information. Experts Integrating genetic data into care estimate that just one genome could

management faces practical problems consume 370 GB, although the beyond guideline development and ability to focus on desired variants

Timely Analysis of Immediate Health Policy Issues 8

Race, Ethnicity and Genetics: A Difficult Mix Race and ethnicity can complicate even ordinary clinical encounters; Uncomfortable Science adding in genetics can make those conversations even more awkward. After all, it was not that long ago that genetic pseudo- Beyond the practical implications, individuals and ethnic groups may science was widely used to label some groups as intrinsically be uncomfortable with science that singles them out. The BRCA inferior. genes are more frequently found in Ashkenazi (Eastern European- origin) Jews than in the general population, but Jewish groups remain There are two separate issues involving race and ethnicity. The first uncomfortably aware that the false idea of a Jewish “race” has been a is scientific: To what extent should clinical genomics guide treatment staple of anti-Semites. of different racial and ethnic groups? The second is economic: How should we ensure access by traditionally underserved racial and Many African Americans had similarly mixed feelings when in 2005 ethnic groups? the FDA approved BiDil, a combination of vasodilators to treat heart failure, as the first racially targeted drug. The idea of a race-based The science is complicated by code words. A recent journal article therapy was controversial,5 but in the case of BiDil the argument referred to a study showing an increased risk of a potentially fatal eventually became moot. The drug’s maker had tested it on self- adverse reaction in “Asian patients” taking carbamazepine, an identified Black people, but not on any comparison population, in 1 anticonvulsant and mood stabilizer. The plain meaning refers to order to gain a new indication that would extend the drug’s patent anyone from an Asian nation, but in the United States the word is life.6 BiDil was removed from the market in early 2008 after sales used to refer to certain Asian subgroups. But genetic homogeneity is plummeted due to lower-priced generics. linked to geographically distinct populations and an ancestrally shared gene pool. “Asian” appearance can be misleading. The The conclusion that race-based drugs are only a marketing conceit, carbamazepine study was conducted in Taiwan with Han Chinese however, has also turned out to be premature. Recall that race and patients.2 Koreans, in contrast, tend to have the Alatic genetic ethnicity can sometimes be a proxy for ancestral geography. Two makeup of the Mongolian region. recent papers in the journal Nature found a unique gene structure in African Americans. The researchers plan to pursue similar research Sometimes, research can help resolve the racial and ethnic jumble. with regard to Latinos, who have a mix of European, American The blood thinner warfarin is one of the most frequently prescribed Indian and African ancestry.7 drugs in the world, but too low or high a dose can have serious consequences. Three different groups—Whites, Blacks and East Scientific and socioeconomic concerns can overlap. For example, the Asians—have different probabilities of genetic warfarin sensitivity. Northwest-Alaska Pharmacogenomics Research Network is Without more information, these racial differences potentially conducting pharmacogenomic research in American Indian, Alaska complicate a dosing algorithm. However, an 11-nation study was Native and rural Pacific Northwest populations with the aim of able to identify a single VKORC1 polymorphism correlated to identifying genetic differences and also of ensuring that those sensitivity across all racial groups, enabling one dosing algorithm for populations have access to interventions based on those differences. the drug.3 (Age, sex and vitamin K intake can also influence warfarin Researchers have also noted ways in which discrimination and other dose-response.) environmental stresses among low-income populations can have epigenetic effects. Sometimes, the way to avoid racial or ethnic concerns may be to screen everyone. For instance, a lawsuit that followed the death of a Studies have examined the use of genetic testing by physicians serving primarily minority populations (who are significantly less Black football player with sickle cell trait prompted the National 8 Collegiate Athletic Association to recommend screening all Division likely to order a test), the use of BRCA testing by Hispanic and Black women at high risk (which has increased, but is still I athletes for this condition. However, the universal screening 9 remains controversial.4 significantly lower than among non-Jewish White women), and the accuracy of a BRCA-carrier prediction tool among minorities (which But sometimes there are simpler ways to deal with these complicated generally works well for African Americans, Asian Americans and genetic scenarios. With sickle cell screening for athletes, for Hispanics, but has room for improvement with non-Hispanics).10 example, critics say the problem is not a genetic trait, but dangerously high-stress training methods. With both warfarin and Unfortunately, unfair profiling by race remains a danger in the brave carbamazepine, extra watchfulness for an adverse reaction new world of clinical genomics, and the old categories of haves and immediately following the first dose can be just as effective as, and have-nots have not disappeared. As one study put it, “Access and much less costly and complicated than, pre-dosage genetic testing. knowledge barriers continue to limit the use of this technology to the 11 ______wealthy, the well-insured and the medically well-informed.”

1 The article referred to Stevens-Johnson syndrome among those taking carbamazepine. See Mrazek DA and Leerman C. “Facilitating Clinical Implementation of Pharmacogenomics.” Journal of the American Medical Association, 306(3): 304–305, 2011. 2 Chen P, Lin JJ, Lu CS, et al. “Carbamazepine-Induced Toxic Effects and HLA-B*1502 Screening in Taiwan.” New England Journal of Medicine, 364(12): 1126–1133, 2011. 3 Limdi NA, Wadelius M, Cavallari L, et al. “Warfarin Pharmacogenomics: A Single VKORC1 Polymorphism is Predictive of Dose Response Across 3 Racial Groups.” Blood, 115(18): 3827–3834, 2010. 4 Koopmans J, Cox LA, Benjamin H, et al. “Sickle Cell Trait Screening in Athletes: Pediatricians’ Attitudes and Concerns.” , 128(3): 477-483, 2011. 5 Carlson RJ. “The Case of BiDil: A Policy Commentary on Race and Genetics.” Health Affairs (Web Exclusive), October 2005. 6 Wolinsky H. “Genomes, Race and Health.” EMBO Reports, 12(2): 107–109, February 2011. 7 Brown E. “Scientists: New DNA Maps May Help Blacks’ Health.” Chicago Tribune. July 21, 2011, sec. 1, p. 17. 8 Shields et al. 9 Levy DE, Byfield SD, Comstock CB, et al. “Underutilization of BRCA 1/2 Testing to Guide Breast Cancer Treatment: Black and Hispanic Women Particularly at Risk.” Genetics in Medicine, 13(4): 349–355, 2011. 10 Huo D, Senie RT, Daly M, et al. “Prediction of BRCA Mutations Using the BRCAPRO Model in Clinic-Based African American, Hispanic and Other Minority Families in the United States.” Journal of Clinical Oncology, 27(8): 1184–1190, 2009. 11 Hall MJ and Olopade OI. “Disparities in Genetic Testing: Thinking Outside the BRCA Box.” Journal of Clinical Oncology, 24(14): 2197–2203, 2006.

Timely Analysis of Immediate Health Policy Issues 9

The Promise of Clinical that biomarker-guided therapy might interventions in actual practice. Genomics eliminate expensive and harmful At present, the medical literature adverse drug reactions.47 However, is largely filled with anecdotes The buzz of excitement surrounding were that proposition to be presented that provide little useful policy clinical genomics is similar in some seriously to patient safety experts, guidance. ways to the early enthusiasm for there would be hard questions about computers in medicine. Both are the proportion of adverse reactions 3. Data on regulatory actions. The technologies intuitively perceived to attributable to genetic causes in the FDA could finally regulate LDTs possess enormous power to change first place, the type of patient data as promised, report on what LDT care for the better. Both have birthed and information systems needed to tests have been approved, and snappy sound bites: “paper kills,” in implement this solution, and the initiate post-marketing the one case, or personalized, probability that less costly and surveillance on their safety. FDA predictive, and precision medicine, in complicated interventions could data on pharmacogenomic the other. And both have attracted accomplish at least as much a great information available to true believers whose enthusiasm can deal more quickly. prescribing physicians could also sometimes lead them to blur the be examined, since the agency’s distinction between what is doable Policy Changes to Improve Value pronouncements on these issues today and what might be possible can be contradictory and For policymakers, payers, providers confusing.49 tomorrow. As with health and the public, the overarching information technology, the rapid imperative is for improved 4. Data for payment. Public and dissemination of clinical genomics transparency of information. This private payers could be clearer technology is strongly backed by includes: about what evidence they businesses that stand to reap require. Medicare now pays for significant profits from increased 1. Data on evidence. A recent genomic interventions on the adoption. Although there is nothing JAMA commentary noted “the basis of ad hoc decisions. The inherently wrong with their thin line between hope and hype 48 Medicare Payment Advisory advocacy, it should be recognized as in biomarker research.” Right Commission could provide such. now, there are large gaps in Congress with objective advice knowledge, either because Those similarities should sound a about how this technology should questionable studies or a lack of be implemented. Evidence-based warning about realistic expectations. data entirely. Dissemination of As the tools of DNA discovery enter criteria will become critical with information on genomics in the proliferation of expensive mainstream medicine, clinical practice is impeded by having genomics is following the same path medications targeting small information scattered across subgroups of patients, as other innovations: sometimes multiple websites. Government- underused, because of a lack of particularly those with cancer and privately funded technology and other serious diseases. understanding or reimbursement; assessments often fail to sometimes appropriately used and coordinate methodologies and 5. Data for consumers. Congress reimbursed; sometimes overused or targets of inquiry. and federal agencies could give misused because of patient or the public information and tools provider enthusiasm, legal pressure 2. Data on implementation. NIH it needs to make good decisions or other factors; and sometimes used could demonstrate that its about a technology that often entirely appropriately to significantly forthcoming Genetic Testing raises uncomfortable ethical, improve patient care. Registry will serve constituencies clinical and economic questions. beyond clinical laboratories by For instance, prenatal genetic The challenge is to ensure that as making available comprehensive clinical genomics grows in testing is becoming easier and information to researchers and much less expensive, as is testing importance, it meets the same clinicians. The government could evidence standards as other newborns for possible also fund more reliable research predisposition to a long list of interventions. For example, clinical examining the benefits, harms genomics advocates routinely assert diseases that may or may not be and economic impact of genomic clinically relevant. Given the

Timely Analysis of Immediate Health Policy Issues 10

risks and benefits of clinical The disbanding of a citizen advisory things that are ‘obviously’ better… genomic tests set forth in this committee that consistently pushed may have unanticipated costs and paper, government and private the Department of Health and Human real medical harms that should make payers alike could focus more Services for greater oversight is us all demand a higher level of closely on consumer education disturbing. “Medicine is no longer evidence before we adopt change.”50 and protection. able to simply adopt every new change or technology that comes As clinical genomics moves from Developing the evidence base and down the pike,” says Robert Davis, a bench to bedside and becomes a educating doctors, patients and senior Kaiser Permanente researcher. significant component of the U.S. payers on how to use the evidence “There is increasing awareness that health care system, it is advice well will take time, patience and money. worth following.

The views expressed are those of the author and should not be attributed to the Robert Wood Johnson Foundation, or the Urban Institute, its trustees, or its funders. About the Author and Acknowledgements Michael L. Millenson is the president of Health Quality Advisors LLC, the Mervin Shalowitz, MD Visiting Scholar at Northwestern University’s Kellogg School of Management and a senior policy consultant to the Urban Institute. The author thanks Bob Berenson and two anonymous peer reviewers for their enormously helpful comments and suggestions and Juliana Macri for her editorial and research assistance. This research was funded by the Robert Wood Johnson Foundation. About the Urban Institute The Urban Institute is a nonprofit, nonpartisan policy research and educational organization that examines the social, economic and governance problems facing the nation.

Timely Analysis of Immediate Health Policy Issues 11

Notes 2210, 2011. And Booth RA. “Assessment of 24 Rollnick S, Butler CC, McCambridge J, et al. Thiopurine s-methyltransferase Activity in “Consultations About Changing Behaviour.” 1 This paper uses clinical genomics as an Patients Prescribed Thiopurines: A Systematic BMJ, 331(7522): 961–963, 2005. umbrella term, with the understanding that Review.” Annals of , 154(12): distinguishing among genetics, genomics, 814–823, 2011. 25 Wright C. “Estimating the Size of the DTC Genomics Market.” Genome Unzipped. , epigenetics and other possible 13 categories is more important in a scientific The Clinical Laboratory Improvement September 13, 2010, paper than in a policy overview. Amendments of 1988 were passed by Congress www.genomesunzipped.org/2010/09/estimating- in response to public concern over false- the-size-of-the-dtc-genomics-market.php. 2 negative Pap smear readings from clinical labs. Shields AE, Burke W and Levy DE. 26 “Differential Use of Available Genetic Tests Information on FDA use of those provisions can McGovern MM, Benach MO, Wallenstein S, Among Primary Care Physicians in the United be found at: et al. “Quality Assurance in Molecular Genetic States: Results of a National Survey.” Genetics fda.gov/MedicalDevices/DeviceRegulationandG Testing Laboratories.” Journal of the American in Medicine, 10(6): 404–414, June 2008. uidance/IVDRegulatoryAssistance/ucm124105.h Medical Association, 281(9): 835–40, 1999. tm. 27 3 Lakhman K. “A Payor Goes Head-to-Head Khoury MJ, Berg A, Coates R, et al. “The 14 Evidence Dilemma in Genomic Medicine.” Personal communication with Elizabeth with NIH’s Genetic-Testing Registry, While a Health Affairs, 27(6): 1600–1611, Mansfield, director of the personalized medicine Pharma Focuses on Patients.” GenomeWeb, November/December 2008. staff, U.S. Food and Drug Administration, 15 October 14, 2010, July 2011. www.genomeweb.com/blog/payor-goes-head- 4 head-nihs-genetic-testing-registry-while- Economic Impact of the Human Genome 15 Project. Columbus, OH: Battelle Technology “U.S. System of Oversight of Genetic Testing: pharma-focuses-patients. A Response to the Charge of the Secretary of Partnership Practice, May 2011, 28 www.battelle.org/publications/humangenomepr Health and Human Services.” Bethesda, MD: Tajik P and Bossuyt PM. “Genomic Markers oject.pdf. Secretary’s Advisory Committee on Genetics, to Tailor Treatments: Waiting or Initiating?” Health, and Society, April 2008, Human Genetics, 130(1): 15–18, 2011. 5 Feero WG, Guttmacher AE and Collins FS. oba.od.nih.gov/oba/SACGHS/reports/SACGHS 29 “Genomic Medicine—An Updated Primer.” _oversight_report.pdf. Gurvaneet Randhawa, personal communication with senior advisor on clinical New England Journal of Medicine, 362: 2001– 16 2011, 2010. Chen B, Gagnon M, Shahangian S, et al. genomics and personalized medicine, Agency “Good Laboratory Practices for Molecular for Healthcare Research and Quality, July 15, 6 OMIM—Online Mendelian Inheritance in Genetic Testing for Heritable Diseases and 2011.

Man. Washington: National Center for Conditions.” MMWR, 58(RR06); 1–29, 2009. 30 Biotechnology Information (NCBI), Khoury et al. 17 Food and Drug Administration. “Oversight of www.ncbi.nlm.nih.gov/omim. 31 Laboratory Developed Tests.” Federal Register, Schully SD. “Translational Research in 7 Hindorff LA, Junkins HA, Hall PN, et al. “A 75(116): June 17, 2010, Cancer Genetics: The Road Less Traveled.” Catalog of Published Genome-Wide Association edocket.access.gpo.gov/2010/2010-14654.htm. Public Health Genomics, 14(1): 1–8, 2011. Studies.” Washington: Office of Population 32 18 Piper M. “Assessing the Evidence for Genomics, 2011, www.genome.gov/gwastudies. Hamburg MA and Collins FS. “The Path to Personalized Medicine.” New England Journal Genomics: Focus on the Patient.” (Presentation 8 Maréchal JD, Kemp CA, Roberts GCK, et al. of Medicine, 363(4): 301–304, 2010. at 3rd Annual Personalized Medicine Partnerships Conference, Bethesda, MD, April “Insights into Drug Metabolism by 19 Cytochromes P450 from Modeling Studies of Mulcahy N. “Molecular Testing has a Long 11, 2011.) CYP2D6-Drug Interactions.” British Journal of Way to Go, says NCCN Panel.” Medscape 33 ibid. Pharmacology, 153: S82–S89, 2008. Medical News, March 23, 2011, www.medscape.com/viewarticle/739507. 34 Wizemann T and Berger AC. “Generating 9 Caraco Y. “Genes and the Response to Drugs.” 20 Evidence for Genomic Diagnostic Test New England Journal of Medicine 351(27): Kolata G. “How Bright Promise in Cancer Development: Workshop Summary.” 2867–2869, 2004. Testing Fell Apart.” New York Times (New York Edition), July 7, 2011, sec. A, p. 1. (Roundtable on translating genomic-based 10 research for health, Institute of Medicine, Table of Pharmacogenomic Biomarkers in 21 Drug Labels. Washington: Food and Drug Swan M. “Multigenic Condition Risk Washington DC, November 17, 2010.), Administration, 2011, Assessment in Direct-to-Consumer Genomic nap.edu/openbook.php?record_id=13133&page www.fda.gov/Drugs/ScienceResearch/Research Services.” Genetics in Medicine, 12(5): 279– =R1. 288, 2010. Areas/Pharmacogenetics/ucm083378.htm. 35 Schmidt C. “Assays that Predict Outcomes 22 11 Frueh FW, Amur S, Mummaneni P, et al. Ng PC, Murray SS, Levy S, et al. “An Make Slow Progress Towards Prime Time.” “Pharmacogenomic Biomarker Information in Agenda for Personalized Medicine.” Nature, Journal of the National Cancer Institute, Drug Labels Approved by the United States 461: 724–726, 2009. 102(10): 677–679, 2010. Food and Drug Administration: Prevalence of 36 23 Shields et al. Related Drug Use.” Pharmacotherapy, 28(8): “Direct-To-Consumer Genetic Tests: Misleading Test Results Are Further 992–998, 2008. 37 Complicated by Deceptive Marketing and Other Stanek EJ, Sanders CL, Teagarden JR, et al. “National Pharmacogenomics Physician Survey: 12 See the following two sources. Ioannidis JPA Questionable Practices.” Washington DC: US Who Are the Physicians Adopting and Panagiotou OA. “Comparison of Effect Government Accountability Office, July 22, Pharmacogenomics and How Does Knowledge Sizes Associated with Biomarkers Reported in 2010, www.gao.gov/products/GAO-10-847T. Impact Adoption?” Medco Health Solutions, Highly Cited Individual Articles and in 2009, Subsequent Meta-Analyses.” Journal of the American Medical Association, 305(21): 2200–

Timely Analysis of Immediate Health Policy Issues 12

www.medcoresearchinstitute.com/pharmacogen Health Research, Southeast Kaiser Permanente, omics/physician-survey. Feb. 23, 2010.

38 ibid.

39 Burke W, Laberge AM and Press N. “Debating Clinical Utility.” Public Health Genomics, 13(4): 215–23, 2010.

40 “Genetic and Molecular Diagnostic Testing.” Regence Blue Cross Blue Shield, May 1, 2011, blue.regence.com/trgmedpol/geneticTesting/gt2 0.html.

41 Wilson D. “Trial Shows Blockbuster Potential for Blood Clot Pill.” New York Times, August 28, 2011, nytimes.com/2011/08/28/business/trial-shows- potential-for-blood-clot-pill-eliquis.html.

42 Poppe LB and Roederer MW. “Global Formulary Review: How Do We Integrate Pharmacogenomic Information?” Annals of Pharmacotherapy, 45(4): 532–538, 2011.

43 Feery WG and Green ED. “Genomics Education for Health Care Professionals in the 21st Century.” Journal of the American Medical Association, 306(9): 989–990, 2011.

44 DNA Direct is not involved with PBM activities. Medco itself recently agreed to be acquired by ExpressScripts, but the combination of the two PBMs had not been approved as this was being written.

45 La L. “DNA Study Has Details from 100,000 Kaiser Patients.” Sacramento Bee, July 6, 2011, www.geneticsandsociety.org/article.php?id=57 85.

46 Coriell Personalized Medicine Collaborative website, www.cpmc.coriell.org.

47 Ernst & Young. “The Case for Personalized Medicine.” Personalized Medicine Coalition, May 2009, www.personalizedmedicinecoalition.org/about/a bout-personalized-medicine/the-case-for- personalized-medicine.

48 PMM Bossuyt. “The Thin Line Between Hope and Hype in Biomarker Research.” Journal of the American Medical Association, 305(21): 2229–2230, 2011.

49 A widely quoted statistic used by FDA officials and by others quoting the agency is that “about 10 percent of labels for FDA approved drugs contain pharmacogenetic information.” However, the FDA’s own website lists just 76 drugs with biomarker information, much fewer than 10 percent of all approved drugs. Having genetic information in the label could be a description of the clinical pharmacology, not of a drug characteristic actionable in clinical practice.

50 Personal communication with Robert Davis, MD, MPH, director of research, Center for

Timely Analysis of Immediate Health Policy Issues 13