LEDA at Harvard Law
The Emerging Field of Race-Based Genetic Research: Can We Trust It?
Alpana Gupta, Class of 2006
May 5, 2006
This paper is submitted in satisfaction of both the course requirement and the third year written work requirement.
In June of 2005, the Food and Drug Administration approved a heart disease drug named BiDil exclusively for African Americans, provoking a fiery debate among medical researchers, sociologists, and legal scholars. On the one hand, some researchers have expressed concern that race-based drugs such as BiDil promote the notion that race carries a biological component. They emphasize that the use of such drugs will lead to harms against minorities, such as increased discrimination and the worsening of racial health disparities. Proponents of race-based drugs, on the other hand, argue that race is merely a temporary proxy for underlying genetic patterns, to be replaced by more precise indicators of drug efficacy in the future as technology advances. This paper explores the relationship between race and genetics, the significance this relationship may have in the field of medicine, and the long-term consequences race-based research may hold for minorities. This paper begins by describing the context surrounding the development of BiDil to illuminate why companies are beginning to pursue race-based research during this particular decade. The second part of the paper discusses the concept of race. It begins by outlining the way the law has historically understood the concept of race and concludes with a lengthy overview of genetic studies of racial disparities in terms of both disease susceptibility and drug response. Part III of the paper explores the dangers opponents of race-based research claim such research will have, while the final part of the paper proposes ways FDA may be able to curb such harms.
In June of 2005, the Food and Drug Administration (FDA) made a decision to approve a heart disease drug known as BiDil, provoking a fiery debate among medical researchers, sociologists, and legal scholars. This drug, manufactured by NitroMed, contains the potential to improve the lives of hundreds of thousands of patients suffering from heart disease. In fact, the drug proved so efficacious that the clinical trials for it were concluded prematurely.
What could be so controversial about a drug that has demonstrated such strong efficacy? The fact is that BiDil has been approved exclusively for African Americans, marking the first occasion FDA has ever approved a drug for a single racial group. This decision has aroused the concern of many researchers because of the implications it carries for race as a biological category. These opponents of race-based drugs emphasize various political and financial incentives that may encourage researchers and drug sponsors to pursue race-based research and overemphasize the role race plays in medicine to the detriment of minorities. On the other hand, NitroMed and other proponents of race-based research argue that race provides a merely temporary proxy for genetics that will be replaced by more precise genetic indicators of disease susceptibility as the state of technology advances. According to these researchers, ignoring race may cause needless harm to members of minorities and exacerbate racial health disparities.
How do race and genetics relate to one another? What promise, if any, does race hold for patient treatment and drug development? What harms might race-based research cause minorities? Are the benefits of race-based research worth these harms, and how can researchers minimize these harms? This paper seeks to answer all of these questions. The paper begins by describing the context surrounding the development of BiDil to illuminate why companies are beginning to pursue race-based research during this particular decade. The second part of the paper discusses the concept of race. It begins by outlining the way the law has historically understood the concept of race and concludes with a lengthy overview of genetic studies of racial disparities in terms of both disease susceptibility and drug response. The third part of the paper explores the dangers opponents of race-based research claim that such research will have, to be followed by a discussion of ways FDA may be able to curb such harms.
I. The Context Of Race-Based Drug Research: Why It Is Likely To Be A Formidable Force In The Future
Several different events have come together over the last decade to help explain the recent interest many large drug companies have taken in ethnopharmacology, which seeks to explore how different ethnic groups respond to various drugs. These events include scientific, regulatory, commercial, and legal developments.
A. The Completion of the Human Genome Project
The first of these developments which has helped position ethnopharmacology as a hot new field of study is the strides in genomics researchers have made in the last decade. Most importantly, in 2003, the Human Genome Project completed the mapping of the human genome. The announcement raised great expectations that genetics research would lead to improvements in the treatment of cancer, diabetes, and other major diseases. As some researchers have argued, the genomics revolution has actually inspired a “religious faith” in the promises of genetic research, transforming the gene into a “powerful cultural icon.” Books like Edward O. Wilson’s Consilience exemplify such a faith. In Consilience , Wilson promotes a “unity of knowledge” that goes beyond disciplinarity. While arguing that genes are the fundamental basis for human behavior, he notes that culture is able to influence the survival of certain genes. Importantly, however, he treats culture in a mechanistic manner, transforming it into a series of mental constructs which is relatively static and rigid. In other words, as one researcher puts it, within Wilson’s framework “culture is subsumed within a genetic epistemology.” In a powerful illustration of the power of genetics in our culture, the epistemological framework of geneticism thus begins to pervade and guide our understanding of other important fields.
B. Growing Governmental Concern Regarding Racial Disparities In Health
A second factor that has played an important role in the rise of ethnopharmacology is regulatory. Importantly, the federal government has increasingly taken an interest in reducing health disparities among racialized groups. In 1997, Congress passed the Food and Drug Administration Modernization Act, which required the Secretary of the Department of Health and Human Services, in coordination with the National Institutes of Health (NIH), to provide guidance for the “inclusion of women and minorities in clinical trials.” That same year, President Clinton made a highly publicized apology for the role the federal government played in the Tuskeegee Syphilis Study, in which many African Americans were notoriously exploited for decades in the name of science. More recently, in 2000, NIH published a plan to implement Healthy People 2010, a federal initiative that specially addresses disparities in minority health. The statistics the report cited were alarming in their breadth: the mortality rate for African Americans with heart disease or cancer is 30% to 40% higher than it is for Whites; African Americans, Native Americans, and Alaskan Natives have an infant mortality rate almost double that for Whites; and Hispanics living in the U.S. are almost twice as likely to die of diabetes as are non-Hispanic Whites.
Consequently, FDA may be more likely to approve drugs that have been shown to be effective in certain racial minorities. In the case of BiDil, FDA granted approval for the drug exclusively for use by African Americans, despite criticisms that the drug may increase discrimination against African Americans and that the drug had not been shown to be ineffective in other racial populations. Similarly, NIH may be more inclined to provide grants to groups which are conducting pharmacogenetic research that may help improve the health of certain racialized groups. Moreover, by requiring the sponsors of drug trials to report their results according to race, NIH may encourage researchers to find race-based differences and accordingly develop race-based treatments.
In addition to FDA and NIH, the United States Patent and Trademark Office (PTO) may provide strong incentives for drug companies to develop ethnic drugs. Notably, Dr. Jay Cohn, one of the leading cardiologist investigators of the drug trial for BiDil, acquired a patent for the drug for use in all populations in 1989. This patent was set to expire in 2007. In 1997, however, FDA denied approval for BiDil, citing that the biostatistical validity of the trial results was too uncertain. After conducting more clinical trials, the makers of BiDil recharacterized the drug as an ethnic drug and reapplied for a patent. The PTO issued a new patent for the drug, set to expire thirteen years later than the original patent, in 2020. As some scholars argue, by granting a patent for an ethnic drug, PTO has created an incentive for drug companies to conduct research (or “mine” data from previous trials) that will demonstrate the efficacy of a preexisting drug in specific racial populations to obtain new race-specific patents and thus extend their monopoly over the drug’s manufacture and sale.
C. Rising Costs Of Drug Development
The rising costs of drug testing are providing drug companies with a strong incentive to adopt a more targeted model for drug research and development. According to industry leaders, the average cost of bringing a drug to market in 2003 was 900,000 dollars, and the cost in 2010 may reach as much as two billion dollars. Through the use of pharmacogenomics, or the study of the intersection of genes and drugs, companies are better able to screen new products for efficacy, toxicity, and side effects. Consequently, companies are more likely to obtain drug approval as they enter the last stage of clinical trials, the most costly phase in the clinical trial process. Additionally, companies may be able to reduce the duration and size of clinical trials, as their drugs demonstrate efficacy more quickly. In the case of BiDil, for instance, upon narrowing its trial population exclusively to African Americans, NitroMed was able to decrease the risk of death from heart disease among African Americans by 43%. As Sidney Taurel, the chief executive of drug company giant Eli Lilly, states, “The challenge for us as an industry, as a company, is to move more from a blockbuster model to a targeted model. We need a better value proposition than today.”
On the other hand, some commentators have contemplated that such targeted medicine may actually narrow the market for a specific drug. For instance, a company may receive approval for a drug recommended for use in Asians alone. While doctors are permitted to prescribe the drug to non-Asians, insurers may resist covering the drug for non-Asians. However, where the population for whom the drug is prescribed is relatively large and no competitor specifically targets that population for a given condition, one may expect that companies will continue to possess a strong incentive to pursue the development of an ethnic drug.
D. Large Potential Markets And Influential Political Support
In addition to lowering the costs of drug development and better ensuring drug approval, the development of race-based drugs offers companies the potential for making tremendous profits. By producing a drug that has demonstrated efficacy in a specific racial population, pharmaceuticals can secure a fantastic percentage of the market and overshadow competitors which only offer the standard therapy. In the case of BiDil, NitroMed estimates that about 750,000 African Americans have been diagnosed with heart failure. In fact, shortly after FDA approved BiDil in 2005, Pacific Growth Equities, which has an investment banking relationship with NitroMed, raised its annual sales estimate for the drug to $450-$500 million, up from $300 million previously.
Further attesting to the perceived profit potential of the sale of race-based drugs, private financiers have indicated they may be interested in sponsoring such efforts. In 2001, for instance, NitroMed announced that it had raised $31.4 million from venture capital firms to support its clinical trials for BiDil. Notably, it was able to raise this sum in the wake of the “dot com” collapse of the stock market.
As an additional commercial incentive to pursue ethnopharmacology, a pharmaceutical which manufactures an ethnic drug may be able to secure the support of influential political groups. BiDil provides a compelling illustration of this scenario as well. Both the Association of Black Cardiologists (ABC) and the Congressional Black Caucus provided strong endorsements for the drug. Representatives from both of these groups appeared before FDA to encourage the agency to approve the drug. As of 2005, pharmaceutical DeCode Genetics was engaged in talks with ABC regarding the desirability of testing a new heart attack drug in an African American population.
II. Does Race Have a Coherent Meaning?
A survey of the treatment of race within the law reveals that the legal concept of race is mutable, ambiguous, and ultimately, political. The racial categories upon which the U.S. Census Bureau (Bureau) has historically relied to organize its data provide one of the most compelling illustrations of this point. Since the first census in 1790, the Bureau has constantly altered its racial classifications, at various times relying upon national origin, tribal affiliation, and physical features. For instance, in 1870, the list of categories offered by the census included only “white,” “colored,” “Chinese,” and “Indian.” The list in the 2000 census, however, offered a much greater variety of racial categories, including “White,” “Black, African Am., or Negro,” “Asian Indian, “Chinese,” “Filipino,” “Japanese,” “Korean,” “Vietnamese,” “Native Hawaiian,” and “Guamanian or Chamorro,” among others. In fact, in the twentieth century alone, the Bureau used at least twenty-six different systems to classify racial populations in the U.S.
Notably, the 2000 census asked respondents for their ethnicity, offering a choice between “Hispanic or Latino” and “Not Hispanic or Latino.” The category of Asian American consists of not fewer than twenty-five different populations. Why the Census chose to recognize the category of Hispanic as an ethnicity but Asian American as a race defies logical explanation.
The way in which the Bureau has classified African Americans over time is a particularly telling illustration of the political nature of the concept of race in the law. During parts of the nineteenth century, the Bureau classified African Americans by the percentage of African “blood” they possessed. It defined a “mulatto” as an individual with one black and one white parent while using the terms “quadroon” and “octoroon” to define individuals who had one-fourth and one-eighth African ancestry, respectively. In the 1920’s, however, in accordance with the “one-drop rule,” the Bureau began to more broadly define individuals with even one African ancestor as “black.”
While the Bureau’s historical treatment of race demonstrates the inconsistency of the law’s conception of race over time, a look at contemporary law reveals that even at a single point in time, the law fails to provide a consistent conception of race. 18 U.S.C. § 1093, which deals with crimes against ethnic, national, and racial groups, constructs a definition of race that encompasses both ancestral and physical characteristics: “The term ‘racial group’ means a set of individuals whose identity as such is distinctive in terms of physical characteristics or biological descent.” The statute here does not specify how broadly or narrowly it defines racial groups.
However, federal case law interpreting 42 U.S.C. § 1981, a law that prohibits racial discrimination in the making of contracts, provides a different conception of race. In Saint Francis College v. Al-Khazraji , a U.S. citizen brought a suit under § 1981 against Saint Francis College for denying him tenure based on his Arab ancestry. The Supreme Court, relying upon the legislative conception of race at the time the law was enacted, defined race in terms of ancestry:
In the middle years of the 19th century, dictionaries commonly referred to race as a “continued series of descendants from a parent who is called the stock,” “[t]he lineage of a family,” or “descendants of a common ancestor.” It was not until the 20th century that dictionaries began referring to the Caucasian, Mongolian and Negro races, or to race as involving divisions of mankind based upon different physical characteristics. Even so, modern dictionaries still include among the definitions of race “a family, tribe, people or nation belonging to the same stock.”
Thus adopting a notion of race as ancestry, the Court went on to recognize many different races, including Scandanavians, Chinese, Spanish, Anglo-Saxons, Blacks, Mongolians, and Gypsies. Notably, however, the Court hedged its definition of race later in the opinion:
The Court of Appeals was  quite right in holding that § 1981 “at a minimum,” reaches discrimination against an individual “because he or she is genetically part of an ethnically and physiognomically distinctive sub-grouping of homo sapiens.” It is clear from our holding, however, that a distinctive physiognomy is not essential to qualify for § 1981 protection.
The Court here holds that ancestry is the proper criterion for determining race but, by using the words “not essential,” suggests that physical features may be relevant to the definition of race as well. Like the Al-Khazraji Court, the U.S. Census Bureau defines race according to ancestry. Importantly, however, the Bureau recognizes far fewer ancestries—only five separate racial groups, plus one ethnic group—than the Al-Khazraji Court.
In addition to these various definitions of race, it is quite possible that courts applying the Civil Rights Act of 1964 will employ yet another conception of race. If courts use the meaning of race as it was defined at the time of enactment, the definition will likely be different than that used by the Al-Khazraji Court, the contemporary Bureau, and § 1093.
Notably, despite the political nature of the conception of race that the law of the present and past has applied, FDA recommends that drug companies submitting new drug applications (NDAs) report racial data using the classification system provided by the Office of Management and Budget (OMB), the same system the Bureau has adopted.
B. Exploring Genetic Explanations of Racial Difference
1. Researches Agree That Racial Groups Are Generally More Similar Genetically Than They are Different
The successful completion of the mapping of the human genome in 2003 has opened a new chapter in the discussion on the biological significance of race. As the Human Genome Project has helped show, human beings possess roughly between 30,000 and 35,000 genes. Human beings share approximately 98.56% of these genes with chimpanzees and share 99.9% of these genes with all human beings. One tenth of one percent of the human genome thus accounts for all human variations, known as alleles. Scientists estimate that ninety to ninety-five percent of these variations occur at equal rates in all racial populations. The remaining five to ten percent of variations (among the one tenth of one percent of genes that actually vary) appear to be distributed along geographical or continental lines.
While researchers vigorously debate the medical value of these observed genetic variations among racial groups, most agree that, with only a few exceptions, the variation within a race for a given trait is much greater than the variation for that trait across races. They further agree that no one gene is exclusively associated with any particular race.
2. Evolutionary Explanations Of Genetic Difference Among Racial Groups
Scientists theorize that the migration of early populations out of Africa may explain why we observe some human variation along geographical lines. After migrating out of Africa, groups of early peoples began separating from one another and settling in different continents. The first break occurred when the populations of present-day continents split from the original population in Africa. The second break occurred when the population of Oceania broke off from the newly-separated population. Subsequently, the present-day European population branched off, after which point the Asian and Native American populations similarly diverged and settled in their present-day continents.
Once separated from each other, these groups likely experienced two evolutionary phenomena which help explain the genetic distributions observed among the groups today. The first of these is genetic drift, or the “chance fluctuation of gene frequencies over several generations.” A second phenomenon is mutation, which occurs when the DNA of a certain allele is altered, and the alteration results in a deletion or addition of one or more amino acids.
Accordingly, genetic studies show that among the various populations, the populations of Oceania demonstrate the strongest genetic dissimilarity to Africa. Furthermore, studies demonstrate that people sharing ancestors from the same geographical region tend to show more genetic resemblance to one another than people who do not share such ancestors.
Importantly, however, many caveats limit the significance of such a conclusion. First, the differences which have evolved among geographical populations may not prove to be statistically significant. Even if they are statistically significant, the point at which statistical significance is achieved, importantly, is a line arbitrarily chosen by the scientific community. If statistical significance is not arbitrary, the conclusion does not necessarily follow that society should classify itself according to race; rather, it could choose to define populations using a level of generality much more specific than race. A second caveat is that even given the rough genetic identity among geographical populations, racial categories may not properly identify these populations. Some racial categories, for instance, are not limited by one geographical location: the ancestors of individuals who are designated as Hispanics/Latinos may originate from Africa, the Americas, or Europe. Racial exogamy, or cross-racial mating, further weakens the congruity of self-identified race and genetic make-up. Some studies, for instance, report that seventeen percent of the ancestors of individuals classified as African American are originally from Europe, although racial exogamy appears to occur at lower rates among other races.
3. Recent Genetic Studies On Racial Difference: Does Race Contain Medical Significance?
i. Acquisition of Disease
a. Single-Gene Diseases
Most researchers will likely agree that race plays an important role in predicting the acquisition of at least a limited number of diseases caused by a single gene, or single-gene diseases. One of the most publicized examples of such diseases is sickle cell anemia. While approximately 9% of African Americans carry the sickle cell allele, no more than .7% of White Americans are carriers (carrier status, notably, does not guarantee disease acquisition). Consequently, African Americans are far more likely to develop sickle cell anemia than whites and conversely, are much less likely to develop the disease against which the allele protects, the most severe form of malaria. Other examples of single gene disorders include Tay-Sachs disease and certain types of breast cancer, associated with Ashkenazi Jews; cystic fibrosis, associated with Caucasians; and thalessemia, which is associated with people of Mediterranean descent. The disproportionate impact these diseases have on certain racial populations is striking, and the implication that race bears at least some genetic component is hard to avoid.
Several considerations, however, greatly limit the scope of such an implication and its potential for application to the field of medicine. First, the frequency of such single-gene diseases is quite low. As one researcher notes, only 5-10% of breast cancer can be attributed to inherited autosomal mutations, such as the BRCA mutation associated with Ashkenzai Jews. Meanwhile, BRCA mutations account for only 7 percent of breast cancer among Ashkenazi Jews. In fact, Ashkenazi Jewish women are affected by breast cancer at a slightly lower rate than the rest of women, the only difference being that a gene alteration with a higher than average risk of breast cancer has been found in Ashkenazi Jews at a slightly higher rate than in the larger population, whereas similar mutations among the general population have not been discovered.
Nevertheless, medical publications discussing the genetic cause of breast cancer among Ashkenazi Jews abound. The abundance of publications on this subject may be explained by the fact that the discovery of the BRCA mutation held promise for genetic research and thus attracted a great deal of academic interest and importantly, funding opportunities. In other words, one should be wary that professional concerns may play a role in the reporting of such genetic findings, causing researchers to overemphasize the frequency of single-gene disorders or to conduct so many studies upon the disorders that the public mistakenly begins to believe they are more frequent than is the case.
Second, knowledge of a person’s race will usually be much less helpful for purposes of diagnosis than more precise information about the population subgroup from which that person has descended. In the case of Tay-Sachs disease, for instance, people of Ashkenazi Jewish descent, rather than “Whites,” share a risk for the disease. Similarly, Americans of Northern European descent are more likely than other Americans to suffer from cystic fibrosis, while sickle cell anemia is found in people with Mediterranean or South Indian ancestry, in addition to African Americans.
Finally, while some single-gene diseases may correlate well with certain racial groups, the occurrence of admixture problematizes the determination of whether an individual carries a specific disease-causing allele. Admixture occurs where individuals have ancestors from a number of different regions. As mentioned earlier, for instance, studies estimate that seventeen percent of the genome of African Americans is of European ancestry. Admixture also occurs among Native Americans. In the Gila Indian community of Arizona, for example, Pima Indians were shown to have a mean amount of admixture of European genes of 9.6%.
Notwithstanding these caveats, however, racial differences in genetic variation appear to be significant where a disease-causing allele for a single-gene disorder occurs at a relatively low rate in a particular race because this opens the possibility that the allele will not be found in any other population. Armed with this information, medical practitioners may be able to consider or at least eliminate a certain disease when presented with a particular set of symptoms.
b. Multifactorial Diseases
More controversial than the medical significance of the frequency of single disease-causing genes among racial groups is the significance of the incidence of many more subtle gene variants that combine with other factors to influence susceptibility to certain diseases. These diseases are much more common than single-gene diseases, including cancer, diabetes, and heart disease, and are likely to involve many genes, each of which has many variants. Where many genes influence disease susceptibility, one racial group may tend to have a variant of one allele while another group may tend to have a variant of another allele, diminishing the role the knowledge of such variation can play for the purposes of treatment.
Genetic Studies Of Cardiovascular Disease In South Asian Indians
Some of the most controversial findings in the debate over the susceptibility to multifactorial diseases among ethnic groups relate to cardiovascular disease. Recently, researchers have focused on the prevalence of heart disease in South Asian Indians. "Until 50 years ago it was hardly ever heard that Indians had a high heart attack risk," says cardiologist Prakash Deedwania of the University of California, San Francisco, Fresno, School of Medicine. Today, Dr. Prakash notes, many Indians are experiencing heart attacks as early as their mid-30s, and the risk is “enormously high” among Indians across the world.
Data is emerging which indicates that the frequency of heart disease among Indians may have a genetic source. Michael Miller, the director of the Center for Preventive Cardiology at the University of Maryland Medical Center, reports that Indians living in the U.S. demonstrate a high occurrence of an alteration in the apolipoprotein C3 gene, which regulates triglyceride metabolism. This alteration is rarely found in Whites. In this 2001 study, the researchers took blood samples from 99 Indians attending a festival in Northern Virginia. Meanwhile, researchers in India have found that in a genetic analysis of 139 healthy males, almost one-third of the subjects carried a related variation in the apolipoprotein gene. Moreover, they found that the variation was twice as frequent in those subjects with elevated triglycerides, a risk factor for coronary artery disease.
Genetic Studies Of Cardiovascular Disease In African Americans
Several publications in the last few decades have emphasized the differences in the rate and causes of hypertension and heart disease between African Americans and European Americans. A major study conducted in 1999 by Daniel Dries, Peter Carson, and colleagues reported that “[t]he population based mortality rate from congestive heart failure is 1.8 times as high for black men as for white men and 2.4 times as high for black women as for white women.” The magnitude of this discrepancy has led some researchers to speculate that cardiovascular disease may, in fact, be a different condition in African Americans and Whites. The Dries study, for instance, asserted that “there may be differences in the natural history of . . . left ventricular dysfunction between black and white patients.” Similarly, Clyde Yancy, a leading cardiologist of the African American Heart Failure Trial (one of the clinical trials conducted for BiDil), has argued that heart failure appears to be a “different disease” in African Americans.
Researchers have offered several different genetic theories that may explain Dries’ finding that African Americans suffer from cardiovascular problems at a disproportionate rate relative to Whites. One theory which has gained considerable attention involves the gene for nitric oxide synthase, an enzyme that produces nitric oxide (NO). A 2001 study has reported that heart failure drugs known as ACE (angiotensin converting enzyme) inhibitors work less effectively in African Americans than they do in Whites. NO, the chemical in the body that maintains the fitness of blood vessels, plays an important role in the functioning of ACE inhibitors. At the same time, researcher Dennis McNamara has found that the version of the gene for NO synthase that works best with ACE inhibitors is prevalent in 60% of Whites but only 30% of African Americans. In other words, a variation in the gene for NO synthase may help explain why African Americans disproportionately suffer from heart disease.
Some researchers, however, have questioned the finding that ACE inhibitors work less effectively in African Americans than in Whites. Daniel Dries (from the aforementioned study), for instance, has written that the ACE inhibitor “enalapril appears to be equally efficacious in black and white patients.” Another article has stated that the analysis of the data in Yancy’s study was too uncertain to provide any conclusions as to whether African Americans and Whites differed in their response to the drug. If, in fact, African Americans do not demonstrate significantly different responses to ACE inhibitors than Whites, NO synthase is likely to be a dead end as an explanation of the different rates at which African Americans and Whites suffer from heart disease.
Another group of researchers has suggested that a gene known as TGF-1 may help explain the disproportionate rate at which African Americans are affected by cardiovascular problems. TGF-1 regulates substances that act both as vasoconstrictors and as growth factors for vascular cells. Researchers led by Phyllis August at Weill Medical College of Cornell University have notably reported that TGF-1 is overexpressed in African Americans with end-stage renal disease or severe hypertension, while it is found at lower levels in Whites with the same conditions.
A third candidate that may help explain the mortality statistic in the Dries study involves the genes that regulate the response of the sympathetic nervous system to hormones like adrenaline. Stephen Liggett and colleagues at the University of Cincinnati have reported that the possession of a combination of two versions of alpha and beta adrenergic receptors increased the risk of heart failure in African Americans by a factor of ten. The high-risk version of the alpha receptor occurs almost exclusively in African Americans, at a rate of 40%. The high-risk version of the beta receptor also occurs more frequently in African Americans than Whites. The study is relevant for the use of beta-blockers, which affect the way adrenaline influences beta receptors and may be less effective in African Americans.
Recent articles, however, have raised a yield sign before the impact of such studies is exaggerated. Importantly, one researcher has discovered that the widely-cited 2:1 mortality statistic the Dries study has put forward is inaccurate in a number of ways. First, the statistic dates back to mortality rates in a 1981 study, or eighteen years before the Dries study was published in 1999. By 1999, some studies indicated that the gap between racial mortality rates had substantially narrowed. Second, the 2:1 statistic taken from the 1981 study relates to “persons aged 35 to 74 years,” a fact which significantly alters the meaning of the statistic: as the study noted, in fact, “the ratio of black-to-white rates [of mortality] was highest under age sixty-five, approaching 1 in persons seventy-five years of age and over.” In other words, the black-white ratio for mortality was 1:1 for persons beyond a certain age. Perhaps most importantly, current data from the Centers for Disease Control and Prevention report that the age-adjusted ratio of black-white mortality from heart failure is 1.1:1. Thus, while it may yet be true that African Americans suffer death from heart disease at a slightly higher rate than do Whites, and the aforementioned genetic studies may contain some truth, the primary reason for suspecting that genetics may explain racial disparities in the rate of heart disease—the abnormally high 2:1 statistic—has been dramatically deflated.
Second, critics warn that genetic, race-based studies often fail to control adequately for environmental and social factors. The Dries study is a notable example. Among environmental factors, it controlled for only two items, level of education and experience of financial distress. A great deal of medical literature, however, emphasizes that racial differences in hypertension are linked to many environmental factors such as diet, environment, exercise, and stress, many of which happen to correspond strongly with race. For instance, a 2005 study sampled Whites from eight surveys completed in the United States, Canada, and Europe and compared the results with three surveys completed by Blacks in Africa, the Caribbean, and the United States. The study surveyed a total of 85,000 subjects. Notably, when compared with the data on Whites, the data on Blacks from Brazil, Trinidad, and Cuba demonstrated a smaller racial disparity than that demonstrated by Blacks from the United States. Another study from 1991 reports that even among African Americans, significant differences in the frequency of hypertension are apparent: darker-skinned African Americans were found to suffer from hypertension at higher rates than lighter- skinned African Americans. The researchers suggested that socioeconomic influences accounted for the discrepancy: Blacks with darker skin were more likely to suffer from racism and lack of resources than their lighter-skinned counterparts.
Given the cursory attention Dries pays to the effects of environmental factors, his statement that observed racial disparities in heart failure are due to genetic differences—“differences in the natural history” of the development of the heart—seems overbroad and oversimplified. To be sure, the 2005 and 1991 studies described above do not eliminate the possibility that genetics may help account for racial disparities in hypertension and other diseases, nor do they explain the significant disparities observed in other racial groups for diseases other than hypertension. The studies do, however, lend strong support to the notion that due to problems in research design, genetics may play a less significant role in such racial disparities than some studies and statistics may lead us to believe.
ii. Response to Drugs
In recent years, the field of pharmacogenetics has gained increasing attention. Pharmacogenetics, which focuses upon the interaction of genetics and drug response, has led researchers to discover several genetic variations that increase or decrease the efficacy of drugs. For instance, researchers have found that variations in the alleles for N-acetyltransferase 2 (NAT2) affect the metabolization of a drug used in the treatment of leukemia.
As in the case of disease acquisition, researchers have discovered several instances in which race correlates with response to various drugs. Researchers who have made such discoveries do not deny that inter-group genetic variability may be low. They emphasize, however, that such variability, however slight, may importantly influence the probability that a drug will have a given response in a patient. As geneticist David Goldstein of University College in London states, "If you say on average the difference between West Africans and Europeans is slight, that does not rule out a great many variants that influence how people respond to drugs."
The most definitive evidence of racial difference in drug response may be found in the levels of drug-metabolizing enzymes researchers have observed in Whites, African Americans, and Asians. Genaissance Pharmaceuticals, for instance, has discovered a mutation in a metabolism-controlling enzyme that is found in thirty to forty percent of Asians, but less than five percent of non-Asians. This discovery may help explain why East Asians have typically required a lower than average dose of various pain, psychotropic, and heart medications.
Another study has found that variations in NAT2, which cause slow acetylation of drugs, can range from fourteen percent in East Asians to fifty-four percent in Whites. Similarly, variations of the CYP2C9 gene, which also differ in their affinity or intrinsic clearance (their metabolization) of different drugs, appear to be distributed according to race: although 98.4% of Asians and 96.3% of African Americans appear to carry the CYP2C9*1 allele, only 80.3% of Whites carry it.
Not all studies, however, have found that racial groups exhibit meaningful differences in levels of drug-metabolizing enzymes. In one study (Goldstein study), James Wilson, David Goldstein, and colleagues at University College, London, compared twenty-three markers for genes related to drug metabolism among 354 people representing eight racialized groups: White (Norwegian, Ashkenazi Jews, Armenians), Black (Bantu, Ethiopian, and Afro-Caribbean), and Asian (Chinese and New Guinean). Using what is called a model-based clustering method, the researchers found that the subjects formed four different clusters. Notably, these clusters did not correspond to racially-defined categories. It remains to be said, however, that the researchers themselves acknowledged that several individual alleles which influence drug response have been shown to differ in frequency among populations, in some cases as much as by twelvefold.
In addition to the limitations upon racial pharmacogenetic claims the Goldstein study suggests, several of the caveats which applied to the value of findings of genetic variation among racial populations regarding disease susceptibility apply to the value of similar findings regarding drug response. Admixture, for instance, will obfuscate attempts to determine whether a particular individual is carrying a certain allele. A second limitation on such findings is that in some cases, drug-related genetic variation among races may be too insignificant to influence treatment decisions. Furthermore, even if practitioners may justifiably rely upon race for certain treatment decisions, reliance upon ancestry may be a more informative choice. The reason is that variation in response among populations within a certain race may be greater than variation between races. For instance, as one study shows, the frequency of the CYP2D6*1 allele, which affects drug metabolism, varies between 22.7% in a Chinese population and 49% in a Korean population, while the variation in Whites is much narrower—33.4% in a British population and 37.1% in a Turkish population. Finally, in addition to genetic factors, drug response is influenced by several environmental factors, such as overall health, lifestyle, support system, education, and socioeconomic status. Race is likely to highly correlate with at least some of these factors, and the precise influence of each is difficult to determine. Even if one can determine that race plays an important role in the case of a certain drug, race will not be all-determinative for the purposes of treatment—inquiry about environmental factors will still be essential.
iii. Studies Of Human Genetic Diversity
A third set of studies has emerged that complicates the relationship between genes and race. As discussed before, the complete mapping of the human genome has demonstrated that at least ninety percent of human variation is found within races and roughly only ten percent between them. A geneticist named Anthony Edwards has recently argued, however, that this conclusion ignores the possibility that patterns of genetic variation are spread throughout the genome. In other words, when one compares individuals by single genes, one finds that individuals in different racial groups are more similar genetically than individuals in the same racial group. When one compares individuals by clusters of genes, however, correlations might emerge which better correspond to particular ethnic groups.
Some studies seem to support such an argument. For instance, in 2005, David Hinds and colleagues conducted a study (Hinds study) to explore the relationship between whole-genome patterns of common human DNA variation and phenotypic variation (such as height and disease susceptibility). To this purpose, they genotyped 71 Americans of African, Asian, and European descent for what are called single-nucleotide polymorphisms (SNPs), the most abundant form of DNA in the human genome. The samples were obtained from the Coriell Cell Repositories’ Human Variation Collection. While confirming prior studies which have found that most common genetic variation is shared across racial populations, Hinds found differences in the frequencies of several alleles among populations. Moreover, he found that 18% of the SNPs genotyped segregated in each of the three populations (“private SNPs”), most noticeably in the African American subset.
In a second study led by Neil Risch, also in 2005, 3,636 subjects identified themselves as White, African-American, East Asian, or Hispanic. Scientists analyzed three hundred twenty-six microsatellite markers in their DNA samples and uncovered four major genetic clusters. These clusters corresponded to the self-identified race of the subjects. In fact, only five participants demonstrated genetic clusters that failed to correspond to their self-identified race.
The finding and implications of such studies as the Hinds and Risch studies have not escaped criticism. One scholar, for instance, has emphasized that clustering studies will not yield the results Risch achieved except in strictly confined circumstances: subjects must have recent ancestors who all come from a geographically isolated location. Moreover, such clustering studies must examine microsatellites, a particular “class of non-functional DNA” that is “not typical of genes” but is selected because it is “‘maximally informative’ about group differences.” This criticism deserves particular attention because it highlights the way in which the types of questions a study asks may produce the results it finds. As one scholar emphasizes, “[T]he use of race and ethnicity in biomedical research is problematic because it is caught in a tautology, both informed by, and reproducing, ‘racialized truths.’ We assume that racial differences exist, and then proceed to find them.” This point, importantly, further begs us to ask why these questions are being raised and who is determining which questions are raised.
Even accepting the results of Risch’s study as true, New York University sociologist Troy Duster criticizes the implications it has for medicine. According to Duster, the study demonstrates a mere correlation, which does not necessarily reveal “anything about a behavior or disease manifestation.”
Duster provides an additional note of caution that applies to both the Risch and Hinds studies and similar clustering studies. As Duster admits, such clustering studies pursue goals that are well-intentioned and logical. They draw subjects from across the globe to create maps of human genetic diversity in an effort to uncover subtle genetic patterns among humans. These patterns, it is hoped, will help scientists locate genes that are relevant to disease. Duster cautions, however, that the subjects who are used to represent different regions of the world are often selected because of their accessibility and convenience. In other words, they do not always represent geographically isolated groups. Cell and tissue repositories, such as the Corriel Institute utilized by the Hinds study, collect such samples in an effort to save researchers the cost and difficulty of obtaining such samples independently. Without necessarily investigating the reliability of the collections, researchers then make claims about racial differences based on these unrepresentative samples.
The preceding section of the paper has discussed what information, if any, race may be able to provide the medical community about disease susceptibility and drug response. As discussed, researchers generally agree that most human variation occurs within racial groups, rather than between them. Recent genetic advances—such as the mapping of the human genome in 2003—have only strengthened this consensus. As the latter part of this section has shown, however, several studies have indicated that the minor variation that does occur among different racial groups—or, to be more precise, among groups from different geographical regions—bears at least some significance for both disease susceptibility and drug response. Importantly, the amount of this variation is minor, and thus the scope of its utility for medical purposes may be narrow. Moreover, even the sponsors of race-based drug trials agree that the most precise predictor of drug response is genetic profile. Given these qualifications, the significance of race-based research must be carefully weighed against the harms it may potentially cause. These harms are addressed in the following section.
III. Dangers Of Race-Based Research And The Approval Of Ethnic Drugs
A. Errors In Medical Diagnosis, Treatment, And Clinical Drug Trials
One important harm race-based research may cause in the clinical setting is errors in both diagnosis and treatment. In the case of diagnosis, for instance, if sickle cell anemia is widely associated with individuals of African ancestry, doctors may fail to diagnose the disease in other individuals in which it is known to appear, namely individuals of Greek, Italian, and Arabic descent. In a recent study, in fact, researchers examined counseling regarding testing of the BRCA1 and BRCA2 alleles, variations related to breast cancer which medical literature has associated with Ashkenazi Jewish women. According to the study, African-American women with first or second degree relatives who suffered from breast or ovarian cancer were far less likely than White women to receive counseling concerning testing for these alleles, despite the fact that their risk of carrying these alleles is no less.
In the case of treatment, race-based research may cause doctors to limit prescription of a drug to a certain ethnic group, when other populations could benefit from its use. In the case of BiDil, some doctors may prescribe the drug exclusively to African Americans, when the combination of the drug may prove to be effective in non-African Americans as well. Prescribing the drug to non-African Americans presents its own problems, however, given that the drug would not have been adequately tested in that population. In fact, in the case of BiDil, early reports indicate that this scenario is being played out. As Flora Sam, a cardiologist at Boston Medical Center stated, “I would prescribe BiDil to non-African Americans who are already on standard therapy and not doing well.”
The use of race-based drugs may lead to overprescription in yet another important way. Doctors may be too quick to prescribe the drug to African American patients without thoroughly considering environmental or other factors which may favor treatment with standard therapy. In fact, there may be good reason to believe ethnic drugs will not work effectively in at least some members of the target race due to the degree of genetic variation found within any racial group. Africa’s current population, for instance, has had the greatest amount of time to develop genetic variations and accordingly demonstrates more variation than any other racial group. Moreover, people of Latino descent are likely to exhibit considerable variation because they come from three different continents, while Asians come from a vast range of regions as well.
Even assuming that race is an important variable in disease diagnosis and treatment, patients are still likely to suffer from incorrect diagnoses and treatment because the markings of race are uncertain and variable. As mentioned earlier, a growing number of individuals in the United States has mixed origins. For instance, individuals who look White can have eighty percent West African ancestry according to their genetic profiles, while individuals who look Black can have primarily European ancestry. Consequently, some doctors may incorrectly label a light-skinned individual of West African ancestry as White and fail to provide her with the best treatment. One doctor, for instance has described a childhood friend who was not diagnosed with cystic fibrosis until she was eight years old. Over the years, doctors had described her as a Black female with a persistent cough. The fact that cystic fibrosis is much less common in Blacks than in Whites likely played a role in her misdiagnosis for so many years. By the same token, a patient may identify herself as belonging to one racial group when her genetic profile better fits that of another group. The individual in the earlier example thus might describe herself to her doctor as White when she better fits the genetic profile of a West African person for medical purposes.
Many of the same arguments apply to the utility of race-based clinical trials. Given the indefiniteness of the concept of race and the growing number of individuals with mixed backgrounds, trials are likely to misclassify or arbitrarily classify the race of participants. In the case of BiDil, for instance, Elyse Frazier was told she could participate in the trials for the drug. Frazier’s mother is half Black, half Cherokee Indian, while her father is half Black, half Blackfoot Indian. The criteria by which the trial investigators classified her as Black is unclear, and the inclusion of her and others like her may importantly distort the findings of NitroMed’s trial and overemphasize the role race has played in the results of the trial.
B. Exacerbation Of Racial Disparities In Health
As has been discussed, race-based research may lead to medical errors in diagnosis and treatment. These errors bear consequences for not only racial minorities but Whites as well. There is good reason to believe, however, that race-based research and clinical trials will cause additional harms which will uniquely affect minorities.
The federal government, as noted previously, has increasingly made efforts to illuminate and improve racial disparities in health. NIH, in fact, has conditioned research grants upon the inclusion of women and minorities in studies. Ironically, however, these well-intentioned efforts may actually worsen health disparities among racial minorities. One of the reasons this is true is because race-based drug research may fuel the mistrust some minorities have historically felt towards the medical profession. For instance, as one scholar argues, the historical exploitation of African Americans by medical researchers, as best illustrated by the Tuskegee syphilis study, has measurably stymied genetic research: as the National Health and Nutrition Examination Survey showed, blacks were less likely than whites to permit their blood or other bodily fluids to be stored for future research, notwithstanding guarantees of privacy and anonymity. Where an ethnic drug on the market is discovered to contain some grave defect, minorities may become even more reluctant to accept important educational outreach and to participate in future clinical trials. In other words, if not carefully conducted, genetic research focusing upon race may actually provide greater obstacles to continued research.
Perhaps a more obvious way in which focusing upon genetic causes of disease in minorities may heighten racial health disparities is by distracting researchers from exploring another important cause of disease: racism. Past and present racism helps to explain, at least in part, the significant disparities among ethnic groups that exist today. To name one example, racial research upon which doctors have historically relied to treat patients reports that Blacks are less likely to comply with medical treatment, less likely to be familiar with their disease, and less likely to promote their own health. Current discriminatory practices in the field of medicine may be more likely to take the form of unconscious or semi-conscious stereotyping by medical practitioners, rather than intentional racism. Other studies have suggested that racism may cause or contribute to disease by causing minorities to experience stress.
In addition to diminishing the importance of racism, focus upon the genetic basis of race may affect the study of other important psychosocial, economic, and cultural elements which affect physiological processes. Some studies, for instance, show that social isolation, coping styles, and confidence regarding one’s ability to handle challenges in life influence cardiodynamic and hemodynamic responses.
To be sure, genes likely play some role in the differences in drug response and disease susceptibility which have been observed among racial groups. Focusing on race as a genetic placeholder, however, diverts resources from exploring and addressing the environmental mechanisms described above. Such diversion is likely to be most apparent in times of financial distress. As one scholar notes, “When disease is ‘located’ within the individual, strategies to ameliorate ill health tend to be similarly focused. The social dimensions of health and disease are ignored, or at best paid lip-service only.” In fact, data from the NIH CRISP database (which provides information on all NIH grants awarded since 1975) may help provide some evidence for this point. From 1995 to 2004, use of the search term “genetics” in CRISP identified 21,956 new grants (and when additionally indexed by the term race, 181 new grants). Use of the term “racism” or “racial discrimination,” on the other hand, yielded only 44 new grants.
Importantly, as one scholar emphasizes, arguments that support the use of race as a genetic placeholder do not occur in a political vacuum. Exploring the social causes of health disparities usually implicates intervention by the government and other non-market institutions. Emphasis on the genetic causes of such disparities, however, largely implicates pharmaceuticals, the development of which is arguably best controlled by impersonal market forces. Arguments which emphasize genetic difference thus tend to have the effect of privatizing efforts to address racial health disparities; health disparities, rooted in social inequality, are transformed into health differences, best addressed by private institutions. While such an effect does not necessarily affect the merits of such genetic claims about racial differences, one should be wary that such claims may be exaggerated for the sake of non-scientific, political ends.
C. Increased Discrimination Against Minorities
Scholars worry that the harms of race-based genetic research will not confine themselves to health, but will extend to the domains of employment, insurance, and possibly even education in the form of increased racial discrimination. Current law, as the following section argues, suggests that fears about increased discrimination in employment may be slightly exaggerated. On the other hand, the state of the law suggests that concerns about discrimination in other domains may, indeed, be real and weighty.
Many scholars argue that as a consequence of race-based genetic research, some employers may avoid hiring members of racial groups associated with a particular disease for fear of excessive absenteeism, low productivity, and high insurance costs. More sophisticated employers might actually test at-risk racial groups for disease-causing genes and exclude those members who test positive for the genes. Several employers, in fact, have tested African Americans for the sickle cell trait. Most notoriously, in the 1970s until 1981, the U.S. Air Force Academy excluded all African Americans with the sickle cell trait following a series of incidents in which four African Americans carrying the relevant allele died while undergoing training at a relatively high altitude. Increasing the egregiousness of these cases is the fact that carrier status of the sickle cell allele bears no ill health consequences. This fact highlights the way in which employers who are willing to test minority applicants (rather than automatically exclude them) may be prone to misinterpret test results and consequently exclude them from employment.
Given current law, however, the likelihood that employers will be able to discriminate against potential or current employees on the basis of race where that race is highly associated with a certain disease is weak. A touchstone of this law is Title VII, which makes it unlawful for any employer—private or public—“to fail or refuse to hire or to discharge any individual, or otherwise to discriminate against any individual . . . because of such individual’s race . . .” In Arizona Governing Committee v. Norris (Norris ), the state of Arizona offered employees the opportunity to enroll in one of several deferred compensation plans offered by specified private companies. All of the companies offering a post-retirement annuity used sex-based mortality tables to calculate monthly retirement benefits. Because the tables indicated that women on average live longer than men, men who chose the plan received larger monthly payments than women who deferred the same amount of compensation and retired at the same age. The Supreme Court held that the Arizona Governing Committee had violated Title VII: the “assumption—that sex may properly be used to predict longevity—is flatly inconsistent with the basic teaching of Manhart : that Title VII requires employers to treat their employees as individuals, not ‘as simply components of a racial, religious, sexual, or national class.’” This case suggests that correlations between race and disease, even if very tight and demonstrated by actuarial tables, will not justify the differential treatment of racial minorities by employers. Admittedly, sex was the only factor used to classify individuals in Norris . Even if the holding extends only to decisions based solely on one discrimantory criteria, however, Title VII penalizes an employer for making an adverse employment decision based on race in any part, even if other, non-discriminatory factors influence the decision as well.
To be sure, Title VII does allow employers to discriminate against employees on the basis of health conditions if they can prove a “bona fide occupational qualification.” For instance, the Supreme Court recently held that the American Disabilities Act permitted Chevron to refuse to hire applicant Mario Echazabal to work in a refinery because toxins at the refinery were likely to aggravate his Hepatitis C. If an employer refused to hire any member of a certain racial group based on a close correlation between that group and a certain disease, it would be unable to do so, however, because the BFOQ defense is not available to employers in cases of racial discrimination. Morevoer, Chevron’s refusal to hire Echazabal, was based upon an individualized assessment, rather than a mere close correlation between a certain disease and a person’s race. If, alternatively, an employer targeted a specific racial group for genetic testing and then excluded certain members based on the individualized assessments, the same argument would apply: the employer is treating certain individuals differently based on race, and the BFOQ defense is not available.
In another possible scenario, rather than automatically excluding all members of a specific racial group (or singling out a specific group for testing), an employer might test all applicants for a certain disease that it knows is highly associated with a particular racial group and thereby exclude many of the members of that racial group. However, in this instance, plaintiffs would have at least two compelling legal claims. First, they could make a claim for disparate treatment (under Title VII) by proving intentional discrimination through, for instance, a showing that the employer did not test job applicants for diseases that posed similar problems in the workplace as did the disease associated with race (such as a decrease in alertness in a safety-sensitive position).
Second, a plaintiff may make a claim that the employer’s testing policy has a “disparate impact” on a certain racial group, although this claim will be more difficult to prove than a disparate treatment claim. A plaintiff is able to make a claim for disparate impact by showing that notwithstanding the employer’s intention, the percentage of the members of a certain racial group employed by the employer is less than the percentage of that population in the real labor market. Where an employer excludes all individuals who test positive for the sickle cell trait, for instance, a plaintiff could readily demonstrate this incongruence.
Problematically, however, once this showing is made, the defendant receives the opportunity to exculpate itself by showing that the challenged practice is both “job related” and a “business necessity.” At least some courts have used a fairly loose standard to define these two criteria. For instance, some courts have sustained anti-nepotism rules because they “plausibly improve the work environment,” despite the fact that the employer cannot prove “that its rule increases production.” Indeed, the Sixth Circuit has sustained employer discrimination based on cost-minimization and administrative convenience. An employer facing a claim of disparate impact based upon race-associated health conditions may thus prevail, at least in some courts, by arguing that requiring such tests is important to worker productivity, is necessary to avoid costly liability, or is necessary to a safety-sensitive position which requires alertness and energy.
Unlike the threat of discrimination by employers as a result of continued race-based research, the possibility that insurance companies and managed care institutions will increase discrimination against minorities is quite plausible. As scholars emphasize, insurers issuing individual policies might use an individual’s race to assess risk and determine premiums, rather than relying upon individualized assessments of risk. For instance, they might assume African Americans are at greater risk for high blood pressure and consequently charge them higher premiums or deny them coverage altogether.
To be sure, several recent efforts have been made to regulate the production, circulation, and abuse of an individual person’s genetic information. The federal government, for instance, has increasingly grown concerned with abuse of such information: in 2005, the U.S. Senate unanimously passed a bill that prohibits a group health plan or health insurance issuer from: (1) adjusting premiums on the basis of genetic information; or (2) requesting or requiring an individual or a family member of such individual to undergo a genetic test. A House subcommittee is currently reviewing a related bill. In the future, these efforts will prevent insurers from requiring a member of a particular ethnic group to be tested for a disease with which that group has been associated or using the member’s genetic information to charge her higher premiums.
However, such efforts focus primarily on the use of individualized genetic information. Importantly, they do not address discrimination based upon generalizations about ethnic groups, which advances in genetic research have promoted. As Jonathan Kahn states, race-based genetic research threatens to “entangle existing groups that have historically been subject to various forms of discriminatory treatment . . . with new biological categories.” A recent case, in fact, provides some support for such arguments. In Guidry v. Pellerin Life Insurance Co. , a group of African Americans sued an insurance company under 42 U.S.C. §§ 1981, which prohibits racial discrimination in contracts. The insurance company had been using actuarial tables indicating that African Americans on average die at an earlier age than Caucasians to discriminate against African Americans in both policy sales and administration. The court granted summary judgment for the insurance company, reasoning that the company’s differential treatment of African Americans and Caucasians was “based on risk, not race ” and was akin to “charging a higher premium to a smoker than a non-smoker.” Given this case, it is foreseeable that courts may permit insurers to use actuarial tables which indicate that a particular ethnic group disproportionately suffers from a certain disease to charge that group a higher premium. It remains to be said, however, that the likelihood of insurance discrimination as a result of race-based research may be mitigated slightly by the fact that many individuals receive insurance through their employers, and as Norris (in the previous section) demonstrates, these employers cannot administer (or hire other companies which administer) these policies in a discriminatory manner.
3. Affirmative Action Programs
A less obvious harm than either increased employment or insurance discrimination that may follow from race-based research is the undermining of affirmative action programs. As Jonathan Kahn notes, affirmative action initiatives use racial classification largely as a means to redress past and present social and political wrongs. By emphasizing the genetic origins of race, race-based research may question the legitimacy of such a rationale and may thus decrease support for the implementation of such programs.
D. Race-Based Research May Not Be A Mere Temporary Step On the Road To Personalized Medicine
Notably, many proponents of the use of racial distinctions in medical diagnosis and treatment acknowledge several of these dangers and openly emphasize that skin color and the other socially-defined characteristics of race do not exactly correlate with the genetic markers that influence disease susceptibility. Cohn, one of the lead researchers of BiDil, for instance, states that “skin color is only a crude indication of underlying genetics differences.” These proponents all emphasize that locating the underlying genetic variations that cause disease or differential response to drug treatment is the ideal research strategy. The use of race in the medical setting, they argue, is simply a temporary placeholder, or an interim step on the path to the development of more personalized, genetically-based medicine. Following the clinical trials for BiDil, in fact, NitroMed began examining data from ten genetic markers for heart failure, in an effort to more precisely understand the mechanisms of heart disease. Once the company identifies reliable markers, according to its chief medical officer, it plans to conduct a clinical trial, for which it will select participants on the basis of these markers, rather than race.
The likelihood that companies marketing ethnic drugs will follow NitroMed and conduct further pharmacogenetic research is probably weaker than industry officials claim, however. As M. Gregg Bloche argues, once a pharmaceutical gains regulatory approval and patent protection for a drug, it has little incentive to expend more resources locating the specific genetic markers that influence a condition. In fact, such research “risks shrinking the demand for a drug, by subtracting patients who lack the genetic markers that predict a good response.” A racial group may often be the broadest possible population group a drug can reach, as one member of the FDA advisory panel has suggested is true in the case of BiDil. Even if pharmacogenetic research is likely to expand the population base of a drug, the realities of marketing may nevertheless disincentivize such research: as sociologist Troy Duster emphasizes, “These [drug] markets are not about individual designer drugs, but about groups and population aggregates that become the target market.” Using BiDil as an example, Jonathan Kahn reiterates, “[A] drug company cannot effectively market BiDil to the biological group of individuals who have a particular genetic polymorphism that may lead to lower levels of nitric oxide. Rather, NitroMed will market BiDil to the social group known as African Americans.”
An additional disincentive for drug companies to conduct pharmacogenetic research is that such research is often costly and may take years to produce results. The recent clinical trials of Iressa, a drug designed to treat lung cancer, illustrate this point well. When Iressa proved effective in only 10% of trial subjects, data seemed to indicate that mutations in the epidermal growth factor receptor (EGFR) of these subjects explained their response. Subsequent studies have proposed several new causative markers, however, and since that time, even the utility of these markers, along with EGFR, has been questioned. The cost of studying such markers provides another obstacle to the pursuit of genetic research. Mutation analysis of EGFR, for instance, costs several hundred dollars per patient. If a mutation occurs only 20% of the time, a pharmaceutical will need to test 1,000 patients to reach statistical significance.
Even if companies feel assured that genetic research will yield successful results, yet another obstacle may prevent them from conducting such research—drug companies cannot fully embrace pharmacogenetics until doctors embrace the movement. According to Ronald Salerno of Wyeth Pharmaceuticals, doctors may resist such a movement because they fear they will be second-guessed by lawyers: "If someone gets injured from an adverse event [from taking a drug],” he asks, “will that person ask why the doctor didn't check my genotype?" Moreover, doctors will need to be trained to use and properly interpret genetic tests before drug companies can expect doctors to prescribe drugs based on a patient’s genetic profile.
On the other hand, however, some commentators have argued that market forces may actually drive pharmaceuticals to move beyond race and conduct additional research on genetic markers. As some researchers note, genetic analysis could expand the potential market for a drug by identifying individuals outside of a given race that might benefit from a particular drug. Perhaps more importantly, such research might help companies avoid liability for side-effects of a drug: “If I were a drug company executive, in addition to finding out about what works, I might be able to find out what causes problems, and save myself some liability," says Arthur Caplan, director of the Center for Bioethics at the University of Pennsylvania in Philadelphia.
In the ultimate analysis, given these incentives and the advancing state of genetic technology, drug companies are likely to look beyond race and pursue genetic research as a long-term strategy: the companies that succeed in making the drugs with the greatest efficacy and mildest side-effects are likely to drive out their competitors. This likelihood, nevertheless, does little to diminish the need to regulate carefully the ways in which companies conduct race-based research in the interim. For as long as companies conduct race-based research, the dangers of the biological reificaiton of race remain, and the wisest approach to improving minority health will require circumspection and vigilance.
IV. Proposals For Reform
While race-based research contains some promise of improving racial health disparities, it thus also carries the potential to harm members of racial minorities in far-reaching ways. Exacerbating this risk is the fact that lying in the background of these consequences is the implicit approval of racial reification by the state. As Jonathan Kahn emphasizes, federal legal and regulatory support for race-based research and drug use—via FDA, NIH, and patent law—lends the “imprimatur of the state to the use of race as a biological category.” This governmental complicity should be kept in mind as one reviews the following proposals for ways in which FDA may help curb the dangers of race-based reearch.
A. Racial Data Submission In New Drug Applications
Currently, FDA requires drug sponsors seeking an investigational new drug application (IND) or new drug application (NDA) to report the number of trial subjects tabulated by race (along with gender and age) and to provide information on drug safety and efficacy for these specific subgroups. FDA advises drug sponsors to report such racial data according to the categories recommended by the Office of Management and Budget (OMB). It reasons that the use of standardized OMB categories will improve minority health by 1) helping ensure consistency across clinical trial data and data submitted by other agencies and 2) illuminating potential differences in drug response across racial groups.
These categories, however—which FDA acknowledges are not anthropological or scientific designations —have not been designed for the purposes of medical research. As discussed earlier, they are political designations which are constantly changing and may encompass several geographical regions. By encouraging drug sponsors to use these categories, FDA discourages them from using more subtle, precise distinctions to define racial populations. Equally importantly, by requiring all drug sponsors to report racial data in the first instance, FDA provides governmental support for the notion that race plays an important role in biology. Such support will likely influence the types of studies investigators will pursue and importantly, the results they will find.
Rather than requiring drug sponsors to adopt the categories used by OMB, FDA should, as a first step, devise and recommend for use a new list of categories which is more closely tailored to geographical regions. For the important purpose of data comparison across agencies, FDA could thereafter request sponsors to group the geographically-based categories into the designated, more broadly-defined OMB categories. The advantage of using geographically-based categories is that they provide more precise indicators of drug efficacy than do racial categories, as well as deemphasize misleading connections between biology and the amorphous concept of race.
In addition to recommending the use of categories which are better designed for medical purposes, FDA should cease requiring all sponsors to provide racial (or ethnic) data on trial participants. Presumably, the present reporting requirement is intended to help illuminate and address racial differences in drug response where they exist. Race may, in fact, help accomplish this goal in some cases, but in many cases, it will not—environmental or genetic factors unrelated to race, as this paper has discussed, will be more important. Requiring drug sponsors to provide racial data as a matter of habit, regardless of its significance, overemphasizes the role race plays in drug response. Similar to the system one scholar has proposed that NIH adopt, in place of the current requirement, FDA should require sponsors to report racial data only where 1) it is statistically significant and 2) it will more likely than not lead to public health benefits, as determined by the appropriate FDA approval committee. This requirement will help illuminate racial differences in drug response that may improve minority health at the same time it prevents FDA from fueling a mindless mentality in the medical field that race and biology are integrally connected. Furthermore, it will help prevent researchers from making potentially harmful claims about racial difference which they have not thoroughly researched and analyzed.
Finally, FDA should issue guidance strongly encouraging drug sponsors to report subject data according to several environmental dimensions not necessarily related to race. As emphasized earlier, many different social, economic, and cultural factors strongly influence disease susceptibility and drug response. Notably, FDA requires pharmaceuticals to report racial data but does not necessarily require them to report this other potentially important data. Placing so much emphasis upon race risks the danger that investigators will find racial differences where they may not exist or may not be medically significant. It further implicitly provides government validation of the notion that race and biology are strongly correlated. To avoid these risks, FDA should encourage drug sponsors to report participant data in terms of factors that may include wealth, income, diet, exercise, stress level, exposure to toxins, residence, childhood residence, occupation, and regularity of medical care. Admittedly, as one scholar notes, not all studies will contain sufficient data to provide information on each of these dimensions. Rather than requiring the public to be skeptical about the research design of such studies, however, researchers should make explicit the uncertainty of variables involved in their studies.
B. Standards For Race-Based Trials
Notably, in the context of research analysis and publication, prominent medical journals, including Pediatrics and Nature Genetics , have already taken steps to curb the foreseeable dangers of race-based research. Nature Genetics , most conspicuously, has issued editorial guidelines which require the scrutiny of racial claims in article submissions:
The laudable objective to find means to improve the health conditions for . . . specific populations must not be compromised by the use of race or ethnicity as pseudo-biological variables. Nature Genetics will therefore require that authors explain why they make use of particular ethnic groups or populations, and how classification was achieved . . . We hope that this will raise awareness and inspire more rigorous design of genetic and epidemiological studies.
Similarly, FDA should exact higher standards from drug sponsors who wish to conduct drug trials on subjects of entirely one race. Importantly, given the degree of genetic similarity across populations, one can expect that many trials will fail to uncover medically significant differences in drug response among racial groups. In fact, studies such as the Goldstein study suggest that drug metabolizing enzymes do not cluster along racial lines in any meaningful way. At the same time, the use of race-based drugs may cause several harms—medical mistakes, increased discrimination against minorities, heightened mistrust of the medical community by minorities, and the diversion of resources away from environmental studies of racial health disparities. Given these dangers, FDA should require drug sponsors that wish to conduct race-based clinical trials to meet strict requirements before approving their use.
One promising test for approval that FDA may adopt resembles constitutional equal protection analysis. According to this model, a drug sponsor that wishes to conduct a race-based trial must pass a two-prong test. First, it must demonstrate that the exclusion of other races is justified by a “compelling interest.” Next, it must show that its use of race is “narrowly tailored” to meet that interest. This part of the test would require the sponsor to provide strong evidence that observed race-health correlations actually have some genetic basis. In the case of BiDil, NitroMed would likely have passed the first prong of the test—reducing the high mortality rate among African Americans suffering from heart disease serves a compelling interest. BiDil would likely have failed to pass the second prong, however. The component drugs of BiDil—hydralazine and isosorbide dinitrate—have been shown (in combination) to be effective in populations besides African Americans, belying the notion that the efficacy of BiDil is likely to have some race-based genetic component. Excluding non-African Americans from the trial was thus both overinclusive—the drug would likely not work in at least some African Americans—and underinclusive—it would likely demonstrate efficacy in many non-African American populations.
Recent case law, however, suggests that FDA will likely not be permitted to prevent drug sponsors from conducting race-based trials. In Association of American Physicians & Surgeons, Inc. v. United States FDA (AAPS), FDA issued a regulation known as the “Pediatric Rule” in response to growing data which suggested that many medicines approved for use by adults were being prescribed to children. According to this rule, FDA could 1) require drug sponsors to report data on the effects of their drug upon children (or in other words, conduct drug trials upon children) and 2) require sponsors to develop formulations of the drug suitable for children. As authority for its promulgation of the rule, FDA largely relied upon a labeling provision in the enabling statute which requires manufacturers to demonstrate that a new drug product is safe “for use under the conditions prescribed, recommended, or suggested in the proposed labeling thereof.” FDA argued that the likely off-label prescription of certain drugs to children qualified as a use “suggested” by the labeling. The court, however, rejected FDA’s argument, declaring that FDA was mistakenly conflating the term “suggested” with “foreseeable” or “likely.”
Based on this reasoning, courts are similarly likely to reject any attempt by FDA to require a drug sponsor that wishes to conduct a race-based trial to include additional racial populations. If the product is shown to be “safe” in the population in which it is tested and the product label accordingly “recommend[s]” or “suggest[s]” the drug for use in that population alone, despite the fact that the population is limited to one racial group, courts are unlikely to permit FDA to withhold approval of the drug. In other words, if the AAPS case is predictive of future decisions, then in order for FDA to adopt a restrictive program, it will first need to gain congressional approval.
At the same time, however, opponents of race-based trials may be able to make a strong case that by approving INDs or NDAs which rely upon discriminatory clinical trials, FDA violates the Equal Protection Clause of the Constitution. Equal protection law prohibits FDA from engaging in discrimination when it authorizes drug trials. By approving INDs which rely upon discriminatory clinical trials, FDA is arguably complicit in discrimination. Although FDA is not directly engaging in discrimination, courts have previously found violations where the government has been less involved with discriminatory parties.
Given this finding of discrimination, FDA may thereupon justify its conduct only by demonstrating that it has a compelling government interest and that its action is narrowly tailored to meet its objective. In many cases, FDA would likely be able to prove it has a compelling interest—the treatment of a disease that disproportionately affects a minority population. However, FDA would probably be unable to meet the “narrowly tailored” requirement in some, if not many, cases. Namely, it would likely fail to meet this requirement where an alternative means exists to predict drug response. For instance, if the racial distinction is predicated upon an environmental difference such as diet, a drug sponsor could just as easily admit subjects based upon the relevant diet as they could admit subjects based upon race. Where racial distinctions are predicated upon genetic differences, a criteria other than race is similarly available—ancestry is usually a more reliable predictor of drug response than race. Thus, conditioning trial participation upon race would appear not to pass the narrowly tailored requirement.
A second course of action FDA might recommend to drug sponsors in place of race-based drug trials is the adoption of the methodology of the Goldstein study described in the subsection on “Studies of Human Genetic Diversity.” This study was able to use a genetic cluster methodology to identify average differences in drug response among groups of people without any knowledge of race. The Goldstein approach has two strong advantages. The first advantage is medical—the cluster methodology will provide greater benefits to a larger number of people than will a race-based approach. To be sure, race-based clinical trials may provide strong benefits to some members of the minority being tested, but as was the case with BiDil, several individuals outside of that minority who may benefit from the drug will not be properly tested for drug efficacy. The Goldstein approach, in contrast, determines a more precise predictor of efficacy than race and accordingly, its results will apply to a larger part of the population. At the same time, the Goldstein methodology will reduce the number of people who may be harmed by the drug by better predicting those individuals who will experience side-effects. The second advantage is social—using genetic cluster methodology will prevent both researchers and the media from mistaking correlation with causation and further reifying racial categories.
Finally, where sponsors nevertheless choose to conduct race-based trials, FDA should continue to encourage them to conduct genetic research alongside drug testing. As scientists agree, genetic, biochemical, and physiological biomarkers, rather than race or ancestry, are the most precise predictors of disease susceptibility and drug response. In addition to avoiding the entanglement of race and biology, locating these markers can help shield manufacturers from liability where drugs may cause grave side-effects. Commendably, FDA issued a guidance document in 2005 that encourages drug sponsors to submit to it data from pharmacogenomic testing programs. Recognizing that drug companies may be reluctant to submit such data because of uncertainty regarding the ways in which FDA will use the data as it reviews NDAs, FDA outlines when companies should submit pharmacogenomic data, what format should be used, and how FDA may use this information.
FDA should also strongly encourage drug sponsors, at a minimum, to collect tissue samples from all trial subjects and should accordingly specify a single protocol for sample collection. Importantly, locating the appropriate biomarkers related to drug response is a long and difficult process, and the failure to collect such data slows progress and extends dependence upon race as a proxy for genetic makeup. A large study of the lung cancer drug Tarveca, for instance, provided promising leads in the search for a mutation responsible for lung cancer. The samples that helped produced these leads, however, were taken from less than half of the subjects participating in the trial, and many were taken from biopsies conducted before patients received the drug. Such disorganized data collection critically hinders advancement in the field of personalized medicine. In addition to urging drug sponsors to collect tissue samples from trial patients, FDA should outline standard collection procedures in order to facilitate research across different clinical trials.
There is good reason to think that the creation of more drugs similar to BiDil which are tested and marketed exclusively in a single racial population is likely to form a more important part of the developing field of pharmacogenetics. AstraZeneca, a U.K.-based company, for instance, has made plans to sponsor a six-week clinical trial which compares Crestor, a newly produced cholesterol lowering drug, with atorvastatin, an older drug, in South Asian Americans. Company spokesman Gary Bruell said that if the trial uncovers that the drug has a “profound effect” on a particular population, the company would label and market the drug accordingly. More recently, in 2005, DeCode Genetics of Iceland reported that it was considering testing a new heart drug, DG031, specifically in an African American population. The motivation behind the decision was the discovery of a gene variant that dramatically heightens the risk of heart attacks in African Americans, despite the fact that the variant is common in Caucasians. Drug company giant Pfizer has joined the wave in ethnopharmacology as well, showing interest in researching hypertension-related genes in African Americans in addition to genes that could help explain the prevalence of diabetes in both Asian Indians and Native Americans.
One can expect many of these companies, along with the media and scientific studies, to make brief caveats that race is a social construct and is a crude marker for underlying genetic mechanisms. It seems all too easy, however, to fall into a kind of objectivism about racial identity, as sociologist Michael Omi notes with regard to sociological research: “Although abstractly acknowledged to be a sociohistorical construct, race in practice is often treated as an objective fact.” As race-based drugs increasingly reach the market and their advertisements routinely cover television screens, we should be sensitive to the development of a culture in which race is effectively treated as a biological category. To be sure, as this paper has attempted to show, race-based research does contain promise for lowering health disparities among various racial groups. However, given the types of harms to which it can lead in a society with a lamentable history of racial relations, we should expect and encourage the government to closely monitor race-based research and require scientists to conduct such studies with purpose and scientific rigor.
 Stephanie Saul, F.D.A. Approves a Heart Drug for African-Americans, N.Y. Times, June 24, 2005, at C2.
 See id.
 Sharona Hoffman, “Racially Tailored” Medicine Unraveled, 55 Am. U. L. Rev. 395, 407 (2005).
 Sandra Soo-Jin Lee et al., The Meanings of “Race” in the New Genomics: Implications for Health Disparities Research, 1 Yale J. Health Pol’y, L. & Ethics 33, 36 (2001).
 Id. at 51.
 See id. at 52.
 Soo-Jin Lee, supra note 5, at 52.
 Jonathan Kahn, How a Drug Becomes “Ethnic”: Law, Commerce, and the Production of Racial Categories in Medicine, 4 Yale J. Health Pol’y, L. & Ethics 1, 17 (2004).
 Id. at 4-5.
 Soo-Jin Lee, supra note 5, at 41.
 Hoffman, supra note 4, at 408-09.
 Id. at 409.
 Kahn, supra note 12, at 13.
 Id. at 32.
 Id. at 31.
 Id. at 31-32.
 Hoffman, supra note 4, at 409.
 Alex Berenson, Blockbuster Drugs Are So Last Century, N.Y. Times, July 3, 2005, at 3.
 Robert F. Service, Going From Genome to Pill, 308 Science 1858, 1859 (2005).
 See id.
 Service, supra note 26, at 1858.
 Berenson, supra note 25.
 Hoffman, supra note 4, at 408.
 Kahn, supra note 12, at 7.
 Stephanie Saul, Maker of Heart Drug Intended for Blacks Bases Price on Patients’ Wealth, N.Y. Times, July 8, 2005, at C3.
 Kahn, supra note 12, at 25.
 Hoffman, supra note 4, at 408.
 Donna M. Christensen, Open Public Hearing Printed Statements [for approval of BiDil] before Cardiovascular and Renal Drugs Advisory Committee (June 15, 2005), available at http://www.fda.gov/ohrms/dockets/ac/05/slides/2005-4145OPH-Statements.htm.
 Nicholas Wade, Genetic Find Stirs Debate On Race-Based Medicine, N.Y. Times, November 11, 2005, at A16.
 Soo-Jin Lee, supra note 5, at 42.
 Hoffman, supra note 4, at 418.
 Soo-Jin Lee, supra note 5, at 43.
 Id. at 44.
 Id. at 42.
 Id. at 42-43.
 Id. at 43.
 Erik Lillquist & Charles A. Sullivan, The Law and Genetics of Racial Profiling in Medicine, 39 Harv. C.R.-C.L. L. Rev. 391, 408 (2004).
 Id. at 403.
 Id. at 404.
 Id. at 405.
 Id. at 404.
 Id. at 404-05.
 Id. at 408.
 Id. at 407.
 Guidance Document: Race and Ethnicity Data in Clinical Trials, Food Drug Cosm. L. Rep. (CCH) ¶ 98,932 (Sept. 2005).
 Hoffman, supra note 4, at 407.
 Id. at 412.
 Lillquist and Sullivan, supra note 52, at 409.
 Id. at 409.
 Id. at 418-19.
 Id. at 414.
 Id. at 419.
 Id. at 419-20.
 Id. at 421.
 Id. at 410.
 Id. at 429-30 n.230.
 See id. at 429-30.
 Jacqueline Stevens, Racial Meanings and Scientific Methods: Changing Policies for NIH-Sponsored Publications Reporting Human Variation, 28 Journal of Health Politics, Policy & Law 1033, 1042 (2003); Lillquist & Sullivan, supra note 50, at 434.
 Sally Satel, I Am a Racially Profiling Doctor, N.Y. Times, May 5, 2002, at 6.
 Constance Holden, Race and Medicine, 302 Science 594 (2003).
 Rene Bowser, Race as a Proxy for Drug Response: The Dangers and Challenges of Ethnic Drugs, 53 DePaul L. Rev. 1111, 11115 (2004).
 Stevens, supra note 87, at 1042.
 Id. at 1043.
 Id. at 1042.
 Id. at 1042-43.
 Lillquist & Sullivan, supra note 50, at 434.
 Id. at 434-35.
 Bowser, supra note 90, at 1115.
 Lillquist & Sullivan, supra note 50, at 434.
 Id. at 421.
 Id. at 434.
 Id. at n.250.
 Id. at 434.
 Holden, supra note 89, at 594.
 Bowser, supra note 90, at 1115-16.
 Lillquist & Sullivan, supra note 50, at 434.
 Holden, supra note 89, at 596.
 Troy Duster, Race and Reification in Science, 307 Science 1050 (2005).
 Daniel L. Dries et al., Racial Differences in the Outcome of Left Ventricular Dysfunction, 340 New Eng. J. Med. 609 (1999).
 Kahn, supra note 12, at 11.
 Dries et al., supra note 118, 616.
 Kahn, supra note 12, at 9.
 Derek V. Exner et al., Lesser Response to Angiotensin-Converting-Enzyme Inhibitor Therapy in Black as Compared with White Patients with Left Ventricular Dysfunction, 344 New Eng. J. Med. 1351 (2001).
 Holden, supra note 89, at 595.
 Kahn, supra note 12, at 24.
 Holden, supra note 89, at 595.
 Kahn, supra note 12, at 20.
 Id. at 21.
 Id. at 18-19.
 Id. at 19-20.
 Duster, supra note 117, at 1051.
 Michael J. Klag et al., The Association of Skin Color With Blood Pressure in U.S. Blacks With Low Socioeconomic Status, 265 J. Am. Med. Ass’n 599, 599 (1991).
 Lillquist & Sullivan, supra note 50, at 435.
 Holden, supra note 89, at 594.
 Id. at 594.
 Lillquist and Sullivan, supra note 52, at 435.
 Id. at 435-36.
 James F. Wilson et al., Population Genetic Structure of Variable Drug Response, 29 Nature Genetics 265, 265-66 (2001).
 Id. at 266.
 Id. at 266.
 Id. at 265.
 Lillquist & Sullivan, supra note 50, at 436.
 Bowser, supra note 90, at 1116.
 See supra Part II.B.1.
 James Kingsland, Color-Coded Cures, 186 New Scientist 42, 47 (2006).
 David A. Hinds et al., Whole-Genome Patterns of Common DNA Variation in Three Human Population, 307 Science 1072 (2005).
 Id. at 1073.
 Hua Tang et al., Genetic Structure, Self-Identified Race/Ethnicity, and Confounding Case-Control Association Studies, 76 Am. J. Hum. Genetics 268 (2005).
 Id. at 271.
 Hoffman, supra note 4, at 413.
 Id. at 414.
 Soo-Jin Lee, supra note 5, at 55.
 Kingsland, supra note 170, at 47.
 Duster, supra note 117, at 1050.
 See id.
 See supra Part III.D.
 Hoffman, supra note 4, at 420-21.
 Id. at 421.
 Emily Singer, Race-Based Heart Drug Might Stall Search for Better Markers, 11 Nature Medicine 812 (2005).
 Rene Bowser has called this phenomenon the “bedside bias.” Rene Bowser, Racial Bias in Medical Treatment, 105 Dick. L. Rev. 365, 378 (2001).
 M. Gregg Bloche, Race-Based Therapeutics, 351 New Eng. J. Med. 2035, 2037 (2004).
 Hoffman, supra note 4, at 421
 Lillquist & Sullivan, supra note 50, at 401.
 Bowser, supra note 90, at 1114.
 See infra Part I.B 21 C.F.R. § 312.33 (2006); 21 C.F.R. § 314.50
 NIH Revitalization Act of 1993, Pub. L. No. 103-43, 107 Stat. 122 (1993).
 Soo-Jin Lee, supra note 5, at 64.
 Id. at 64.
 Lillquist & Sullivan, supra note 50, at 400-401.
 Id. at 401.
 Bowser, supra note 199, at 373.
 Lillquist & Sullivan, supra note 50, at 401.
 Id. at 440.
 Bloche, supra note 200, at 2037.
 Id. at 2037.
 Kahn, supra note 12, at 28.
 Soo-Jin Lee, supra note 5, at 64.
 Nancy Krieger, Stormy Weather: Race, Gene Expression, and the Science of Health Disparities, 95 Am. J. Pub. Health 2155, 2159 (2005).
 Jonathan Kahn, Misreading Race and Genomics after BiDil, 37 Nature Genetics 655 (2005).
 Kahn, supra note 12, at 36-40.
 Hoffman, supra note 4, at 425.
 Id. at 424.
 Kahn, supra note 12, at 39.
 Hoffman, supra note 4, at 424.
 42 U.S.C. § 2000e-2 (2006).
 463 U.S. 1073, 1075-76 (1983).
 Id. at 1077.
 Id. at 1083.
 Id. at 1077.
 42 U.S.C. 2000e-2m (2006) (“an unlawful employment practice is established when the complaining party demonstrates that race, color, religion, sex, or national origin was a motivating factor for any employment practice, even though other factors also motivated the practice”); See Desert Palace, Inc. v. Costa , 539 U.S. 90 (2003) (holding that a mixed-motive claim for discrimination may be established through either direct or circumstantial evidence).
 42 U.S.C. 2000e-2 (2006).
 Chevron U.S.A., Inc. v. Echazabal , 536 U.S. 73 (2002).
 42 U.S.C. § 2000e-2 (2006) (permitting bona fide occupational qualification defense for national origin, sex, and religion).
 See Chevron , supra note 243, at 76.
 Id. (outlining “burden of proof in disparate impact cases”).
 See Dothard v. Rawlinson , 433 U.S. 321, 329-330 (1977); EEOC guidelines, cited in a footnote of Connecticut v. Teal , 457 U.S. 440, 444 n.4 (1982), provide that a “selection rate . . . which is less than four fifths .. . . of the rate for the group with the highest rate will generally be regarded . . . as evidence of adverse impact.”
 42 U.S.C. § 2000e-2k (2006); Griggs v. Duke Power Co., 401 U.S. 424, 431 (1971) (“The touchstone is business necessity. If an employment practice which operates to exclude Negroes cannot be shown to be related to job performance, the practice is prohibited.”)
 Yuhas v. Libbey-Owens-Ford Co., 562 F.2d 496, 499-500 (7th Cir. 1977).
 EEOC v. J.C. Penney Co. Inc. , 843 F.2d 249, 253 (6th Cir. 1988).
 Soo-Jin Lee, supra note 5, at 60.
 Hoffman, supra note 4, at 425-26.
 Kahn, supra note 12, at 40.
 Genetic Information Anti-Discrimination Act of 2005, S. 306, 109th Cong. § 2753 (2005).
 Genetic Information Anti-Discrimination Act of 2005, H.R. 1227, 109th Cong. (2005).
 Kahn, supra note 12, at 41.
 364 F.Supp.2d 592 (D. La. 2005).
 Arizona Governing Committee v. Norris , 463 U.S. 1073 (1983) (holding that employer who permitted employees to choose from annuity plans provided by private companies which discriminated on basis of sex violated Title VII).
 Kahn, supra note 12, at 33.
 Bloche, supra note 200, at 2036.
 Kahn, supra note 12, at 26 (quoting Dr. Cohn).
 Bloche, supra note 200, at 2036.
 Malorye A. Branca, BiDil Raises Questions About Race as Marker, 4 Nature Reviews Drug Discovery 615 (2005).
 Singer, supra note 197, at 812.
 Bloche, supra note 200, at 2036.
 Id. at 2036-37.
 Singer, supra note 197, at 812.
 Bowser, supra note 90, at 1125-26.
 Kahn, supra note 12, at 28.
 See Branca, supra note 265, at 615-16.
 Id. at 616.
 Service, supra note 26, at 1860.
 Meredith Wadman, Drug Targeting: Is Race Enough?, 435 Nature 1008, 1009 (2005).
 Kahn, supra note 12, at 33.
 21 C.F.R. § 312.33(a)(2)
 21 C.F.R. 314.40(d)(5)
 Guidance Document: Race and Ethnicity Data in Clinical Trials, Food Drug Cosm. L. Rep. (CCH) ¶ 98,932 (Sept. 2005).
 Soo-Jin Lee, supra note 5, at 42-43.
 Id. at 66.
 Stevens, supra note 87, at 1075.
 See infra Part III.B.
 Stevens, supra note 87, at 1075.
 See id. at 1074.
 See id.
 See infra Part III.B.
 21 C.F.R. 312.33; 21 C.F.R. 314.50
 See Soo-Jin Lee, supra note 5, at 55 (“[T]he use of race race and ethnicity in biomedical research is problematic because it is caught in a tautology, both informed by, and reproducing because it is caught in a tautology, both informed by, and reproducing, ‘racialized truths.’ We assume that racial differences exist, and then proceed to find them.”)
 Stevens, supra note 87, at 1074.
 Id. at 1076.
 Soo-Jin Lee, supra note 5, at 65.
 Editorial, Census, Race and Science, 24 Nature Genetics 97 (2000).
 See infra Part II.B.1.
 Wilson, supra note 158, at 265-266.
 See infra Part III.
 Kahn, supra note 12, at 44.
 Id. at 12-13.
 226 F.Supp.2d 204 (D.C. 2002).
 Id. at 207-08.
 Id. at 215.
 Id. at 215-16.
 Lillquist & Sullivan, supra note 50, at 464 n.399 (citing two cases, Norwood v. Harrison, 413 U.S. 455 (1973), and Graham v. Evangeline Parish Sch. Bd., 484 F.2d 649 (5th Cir. 1973), that have found government financial support for racially segregated private schools to be unconstitutional).
 Id. at 463.
 See id.
 Id. at 463-64.
 Even if courts find that FDA’s involvement in discriminatory trials does not violate the Equal Protection Clause, opponents of race-based trials may be able to file suit against the drug sponsors for violating 42 U.S.C. § 1981, which bars racial discrimination in contracts—including contracts made between private parties. Courts are likely to view the agreement made between a drug sponsor and a trial subject as a contract for the purposes of § 1981. The contract in a race-based clinical trial is conditioned upon race and thus likely violates the mandate of § 1981. Importantly, § 1981 does not provide an affirmative defense to such violations. Lillquist & Sullivan, supra note 50, at 463.
 See supra Part II.B.3.ii.
 Wilson, supra note 158, at 268.
 See supra Part III.D.
 Guidance Document: Pharmacogenomic Data Submissions, Food Drug Cosm. L. Rep. (CCH) ¶ 99,033 (Nov. 1, 2003).
 Branca, supra note 265, at 615-616.
 Id. at 616.
 Holden, supra note 89, at 596 (2003).
 Wade, supra note 46.
 Holden, supra note 89, at 596.
 Kahn, supra note 12, at 35.