Docket Management
Docket: 02D-0018 - Collection of Race and Ethnicity Data in Clinical Trials for FDA Regulated Prodcuts; Draft Guid
Comment Number: EC -2

Accepted - Volume 1

Comment Record
Commentor Dr. Tony Frudakis Date/Time 2003-02-04 10:16:39
Organization DNAPrint genomics, Inc.
Category Company

Comments for FDA General
1. General Comments It certainly is encouraging that the FDA mandates respect for quality issues related to measuring population structure and demographics, but as a geneticist, the current practice of collecting sociocultural (i.e. non-anthropological) data as described in lines 69-71 is somewhat disconcerting. For the sake of the science, and the health of all, we think it is time to incorporate molecular anthropological data metrics as well. We have three issues we would like to point out: 1. Socioeconomic or Geographical measures of “race” are not as tightly linked to drug response as Biogeographical Ancestry (BGA), which is the heritable component of “race”. 2. Because it is subjective, imprecise and sometimes inaccurate, the self-reporting of race as is currently practiced obscures how and why “race” or BGA is related to drug response. 3. The rigid binning of patients into prefabricated racial groups is a practice of generalization, which again does little for our ability to learn how BGA relates to drug response. Many US residents trace their origins to multiple populations through the process of admixture. Proportional BGA can and should be measured using molecular anthropological methods in a scientifically reproducible manner. 4. Thanks to recent advances in genome data resources and testing methods, BGA proportions can now be easily measured for less than it costs to extract DNA from a sample. Pharmacogenetics attempts to identify those specific sequences linked to variable drug response, but we already know from decades of observations that the heritable component of “race” - the biogeographical ancestry (BGA) - is an important factor in the equation (Burroughs et al., 2002, Kalow 1992). The reason for this is that genetic drift, geographical and/or reproductive isolation, and regional selective pressures have molded the allele frequencies of our ancestors for compatibility with alkaloids, tannins (self-defense chemicals) and other xenobiotics found in their indigenous diets. Most drugs are derived from these chemicals, and so it is no coincidence that the family of enzymes that that allow us to detoxify these chemicals is the same family that we use to metabolize drugs (called xenobiotic metabolism genes, of which the cytochrome P450s are the most commonly studied). However, in addition to pharmacokinetics and drug metabolism, pharmacodynamics, or how a drug mechanistically interacts with the patient is also important for understanding variable drug response. We believe pharmacodynamics to be significantly more complex to unravel, requiring whole genomes be screened not just certain gene families, and we have found that the most economical and efficacious method for screening genomes relies on the measurement of SNPs across the genome that carry information about BGA and other levels of population structure. Specifically, we use a method called Mapping by Admixture Linkage Disequilibrium (MALD), first introduced by Chakraborty and Weiss back in 1988. The method requires the ability to measure individual BGA proportions and in an effort to enable ourselves access to perform MALD, as well as to help expand the influence of this method, we developed an 80-marker test for determining BGA proportions and their confidence intervals. The basic methodology for doing this was first described by Hanis et al., 1986, but our method involves a number of important improvements. Since determining individual or group BGA proportions is also useful for reducing bias interjected during the formulation of many case-controlled study designs (Terwilliger et al., 2002), we also offer the method as a separate service to the academic genomics market (as well as to a general recreational genealogical market - see for a more detailed description of the test). The test enables us to determine the relative proportionality of BGA within individuals – for example while one person may register with 80%African, 20% Native American ancestry, another may register 90% Indo European and 10% East Asian ancestry. If the primary objective of FDA approved research is to “Collect…data using standard categories to enhance patient safety” (lines 75-78), it would seem the way that population affiliation is measured today leaves a great deal to be desired. As currently measured with biographical questionnaires, little knowledge of population structure other than the obvious is obtained, and only basic connections between population structure and drug response would be apparent to the trial sponsor or the FDA. Consistency of reporting is a significant problem the FDA would like to address (line 111), but with a subjective and imprecise method of data collection, consistency would seem difficult to achieve. Rather than reformulating how questions are asked, it seems to us that consistency is better addressed by replacing the subjective nature of the exercise with objective, reproducible science. In fact, it would seem that standardization is of paramount importance for the collection of “race” data, because its measure is perhaps as subjective as for any other attribute. Specifically, the self-reporting of race is not as trivial an exercise as the self-reporting of gender. First, many people do not know what their “race” is, or are of sufficient admixture that they have trouble classifying themselves (lines 151-155). This is not uncommon here in the US, which we all know to be a melting pot. For example, a woman of European descent raised in Puerto Rico may describe herself as Hispanic and though she socio-culturally identifies with Hispanics, her xenobiotic metabolism and drug target polymorphisms would align most closely with Europeans and/or West Africans. The current guidelines potentiate this type of problem by using “nonanthropologic designations that describe the sociocultural construct of our society” (lines 69-71). Where a person was raised and lives may indeed have an impact on how they respond to a drug, so nonannthropologic metrics should still be used, but so too may BGA have an impact and this too needs to be measured accurately. Nonetheless, we (unpublished results) and others (Risch et al., 2002; Rosenberg et al., 2002) have shown that when tested against methods of reporting that rely on genome markers, majority population affiliation is quite accurately reported on questionnaires. We have amassed BGA data on over 1,700 individuals and have yet to observe a single discordant result in terms of majority ancestry (in terms of minor ancestry, family pedigrees also show concordant results). Therefore, our results suggest that the mis-reporting of majority race is not a very significant event, and that authors such as Risch and Rosenberg are correct to attach value to self-reported racial classifications, and that determination of majority ancestry affiliation is not the main problem with current self-reporting methods. The problem is that even if a patient knows their majority affiliation, classifying a patient in a single group sacrifices more subtle information related to population structure and sub-structure. Not only is this exercise inaccurate and misleading if the patient is of admixed ancestry, and many patients are, but it is also equally problematic for those who not even know that they are of admixed ancestry. We also observed considerable admixture within a significant percentage of individuals we have tested to date, and considerable structure within most majority BGA groups as well, and it is not unreasonable to suspect that this admixture may have some bearing on drug response. For example, our unpublished results using the ancestrybydna test, and those of Rosenberg et al., 2002 show there to be significant structure within the “Caucasian” or “European” population; Russians and Scandanavians commonly exhibit minor East Asian heritage (possibly through the Lapps) and numerous other examples similar to this can be found. Is this component of ancestry linked to the response of any currently FDA approved drug? Nobody knows and the way “race” is reported today, nobody will ever know. The recommended recruitment of trial participants mandated in 1993 by the NIH Revitalization Act and described in lines 288 – 300 does not allow for such information to be learned, though using tests such as that we have developed would. This problem is relevant for the FDAMA 1997 act (lines 337-346), the Demographic rule of the 21 CFR amendments of 1998 (lines 348-353), the 1999 Population Pharmacokinetics guidance (lines 355-359), in addition to others. In lines of 169-175, a separate problem is also apparent in the non-anthropological method of determining population affiliation. Individuals of the South Asian sub-continent of India are more closely linked culturally, socially and in terms of molecular genetic distance to Europeans than East Asians. The OMB Directive 15 groups them with Asians, which has meaning in a geographical but not a biogeographical context, and since the latter is more relevant for a clinical trial or any other medical practice, it is what should be measured. We suggest that the FDA should pay more attention to molecular characterization of population structure when evaluating and assisting with the construction of clinical trials. Thanks to the human genome project, we now have the means by which to determine not just majority BGA for an individual or group, but BGA proportions if the patient or group is of admixed ancestry. In fact, DNAPrint genomics performs this type of testing for less per sample than most companies pay to extract DNA from blood, and we would be happy to send a validation pack to the FDA or any other group that desires to learn more about the test. We have validated the test on parental and admixed populations, in family pedigrees, with blind challenge, repeated testing and sampling, performance on sample mixtures, and on large numbers of patients of known admixture. To be sure, collecting a buccal swab and sending it to a testing center is less convenient and more expensive than asking for self-reported race on a questionnaire (the average test costs about $160 in bulk), but given the importance of BGA to drug response which is indirectly acknowledged in recent FDA guidelines such as this, it would seem that this inconvenience would be well worth the benefits. We now have the power to get the science right, but the FDA needs to learn about and evaluate such tests so that we can utilize this power to improve our development of safe and efficacious drugs. Burroughs V.J., Maxey R.W., Levy R.A.. (2002) Racial and ethnic differences in response to medicines: towards individualized pharmaceutical treatment. J. Natl. Med. Assoc. (2002) 94:1-26. Chakraborty, R. and Weiss, K. (1988) Admixture as a tool for finding linked genes and detecting that difference from allelic association between loci. (1988) Proc. Natl. Acad. Sci. 85:9119-9123. Risch, N., Burchard, E., Ziv, E., and Tang, H..(2002) Categorization of humans in biomedical research: genes, race and disease. Genome Biology. 3:1-12. Rosenberg, N.A., Pritchard, J.K., Weber, J.L., Cann, H.M., Kidd, K.K., Zhivotovsky, L.A., and Feldman, M.W. (2002) Genetic Structure of Human Populations. Science. 298:2381-2385.

EC -2