Ancillary risk information and pharmacogenetic tests: social ... - Nature

The Pharmacogenomics Journal (2008) 8, 85–89 & 2008 Nature Publishing Group All rights reserved 1470-269X/08 $30.00 www.nature.com/tpj

[Ethical Economic Legal & Social] EELS PAPER

Ancillary risk information and pharmacogenetic tests: social and policy implications NB Henrikson1, W Burke1,2 and DL Veenstra1,3 1 Institute for Public Health Genetics, School of Public Health and Community Medicine, University of Washington, Seattle, WA, USA; 2 Department of Medical History and Ethics, School of Medicine, University of Washington, Seattle, WA, USA and 3Pharmaceutical Outcomes Research and Policy Program, School of Pharmacy, University of Washington, Seattle, WA, USA

Correspondence: NB Henrikson, University of Washington, Box 357236, Seattle, WA 98195, USA. E-mail: [email protected]

Some pharmacogenetic tests may provide ancillary disease risk information. To evaluate evidence and assess the social and policy implications of ancillary disease risk information associated with candidate pharmacogenetic variants, We conducted a literature search and abstract review of disease susceptibility studies for each of 42 gene variants potentially associated with drug response. Twenty-two variants (53%) had suggested association with disease risk in at least two studies, and sixteen (38%) were for diseases other than the pharmacogenetic indication. Seven variants (16%) were associated with risk for at least two different diseases. Pharmacogenetic tests have the potential to provide ancillary disease risk information, and this potential should be considered as pharmacogenetic tests are brought into clinical use. Implications will vary with each test but tests should be evaluated individually within a framework that outlines the potential implications of ancillary information. The Pharmacogenomics Journal (2008) 8, 85–89; doi:10.1038/sj.tpj.6500457; published online 8 May 2007 Keywords: ancillary information; pharmacogenetic test; genetic susceptibility

Introduction Pharmacogenetics is the study of the contribution of genes and gene variants to inter-individual variation in drug response.1 Although the study of genetic determinants of drug response began more than 50 years ago,2 the completion of the Human Genome Project has made it possible to envision the common use of pharmacogenetic testing to individualize drug therapy. Pharmacogenetics may become one of the first widespread clinical uses of genetic information in health care. Ancillary risk information can be defined as information on an individual’s disease risk provided by a genetic test that is unintended and outside the original use of the test, and can potentially arise in research or clinical contexts. It has been suggested that ethical concerns for pharmacogenetic testing may be less significant compared to other genetic tests because of its more specific purpose of improving drug treatment.3,4 Perhaps ancillary disease risk information has not been described as a potentially important issue in pharmacogenetic testing in part because of the types of variants that have shown promise. Genes that code for drug metabolizing enzymes – currently the most common clinical application of pharmacogenetics – generally have not been implicated in disease risk, so ethical analysis of pharmacogenetics has focused on access and distribution of Received 31 October 2006; revised 18 March 2007; accepted 30 March 2007; published beneficial technologies and the social and ethical implications of race- or ethnicity-based prediction of safety or benefit.3,5 Furthermore, risk information online 8 May 2007

Ancillary information and pharmacogenetic tests NB Henrikson et al

86

provided by a pharmacogenetic test would often be for the same disease as the drug treatment, potentially mitigating the implications of ancillary information. A predominant paradigm that pharmacogenetic testing is unlikely to provide significant disease risk information has therefore been suggested.6–8 However, while the existence of ancillary risk information in the context of pharmacogenetic testing is conceptually very real, neither the potential scope of the issue nor consideration of the ethical and social implications of such information has been well studied. Therefore, our objective was to conduct a literature-based inquiry to estimate the potential scope and variation of ancillary disease risk information and to discuss the social and policy implications of this type of information. Results We reviewed the abstracts of 555 studies that met our inclusion and exclusion criteria. For consistent interpretation, we have presented the results in terms of variants and their potential associations with disease risk. Twenty-two (53%) of the 42 variants we studied were reported to have a significant association with disease risk in at least two published studies. The risk in sixteen (38%) of these variants was for diseases other than the pharmacogenetic indication (Table 1). Seven variants (17%) were associated with risk for at least two different diseases. Thirty-three variants (79%) were reported to be associated with risk of a disease in only one study. We found studies suggesting conflicting results for risk of at least one disease for 22 variants (54%). We found only reports of negative results for a disease for 28 variants (67%). We describe below several examples to highlight relevant characteristics of potential ancillary disease risk information. Two studies versus only one study suggesting association with disease risk Several variants have been well-studied; in addition to established associations, these variants often also have many reported associations with disease risk that have not been replicated. For example, Goldstein et al. report that ACE (In-del variant) may predict decreased proteinuria in response to ACE inhibitors for renal disease; according to Table 1

our review, this variant may also be associated with risk for Alzheimer disease.9 However, this variant also has 21 ‘one positive study’ results, suggesting unconfirmed but potential associated risk for 21 other diseases. Also, in the TNF gene variant G308A, the drug response is immunosuppressive therapy for aplastic anemia and sensitivity to the seizure medication carbamazepine. We found studies associating this variant with risk for rheumatoid arthritis, tuberculosis, celiac disease, and ulcerative colitis.10–13 Mental or behavioral disorders Several variants were associated in at least one study with diseases that are associated with mental or behavioral disorders. The A1/A2 variant of the DRD2 gene may predict greater response to antipsychotics, and at least two studies suggest an association with risk for alcoholism.14,15 GNB3 gene variants (C825T) may predict response to antidepressants, and is also associated with risk for obesity and type 2 diabetes. The promoter VNTR variant of the SLC6A3 gene may predict responders to the ADHD medication methylphenidate, and may also predict cigarette smoking behavior.16,17 Severity or treatability of disease The E4 variant of the APOE gene is well known for risk of Alzheimer disease; there is also potential drug response use predicting non-responders to statins.18,19.Similarly, the C667T variant of the MTHFR gene may predict toxicity from methotrexate therapy, and MTHFR gene variants may also be associated with increased risk for several types of cancer, migraine, neural tube defects, and stroke.20–23 Several of the variants may provide ancillary risk information about various types of cancer (XRCC1, GSTM1, GSTP1, MTHFR, CYP2C9), many amenable to early detection and surveillance interventions that improve health outcomes.20,24–26 Based on these findings, we determined that the policy implications of ancillary information would be highly dependent on the characteristics of the individual tests: the strength of the risk estimate, the severity and potential for stigma of the associated disease, and the treatability of the disease. Tests whose associated disease risk are of high risk, are more severe or life-threatening or stigmatizing, and

Results

At least two studies indicating association for a diseasea At least two studies indicating association with a disease other than the pharmacogenetic indicationa Only one study indicating association with disease riska At least two studies indicate association with risk for two or more diseasesa Only one study indicates risk for two or more diseasesa At least two studies indicate risk for same disease as pharmacogenetic indicationa Conflicting results for risk of at least one disease No ‘positive’ association found for at least one diseasea a

These categories do not include a consideration of the number or proportion of negative studies.

The Pharmacogenomics Journal

Number of variants (n ¼ 42)

% of 42

22 16 32 7 26 6 22 28

52.4 38.1 78.6 16.7 61.9 14.3 52.4 66.7


87

that are untreatable will be more likely to have important social, ethical, and policy implications.

Discussion We conducted a review for evidence of disease risk associated with pharmacogenetic variants. Our results confirm the estimate of Goldstein and colleagues that at least 23 of 42 pharmacogenetic variants provide ancillary disease risk information and further characterize the nature of ancillary risk information.6 Our estimates provide a starting point for describing the potential scope and nature of ancillary risk information and provide a window into the dynamic state of genetic susceptibility research and the challenges in validating genetic associations. Although detailed assessment of study quality was outside the scope of this review, there appeared to be a wide range of quality in the studies we identified, with at least some studies limited by methodological weaknesses such as small sample sizes and potential population stratification. However, it is possible that overall these estimates of relationship between pharmacogenetic variants and disease variants may underestimate the true relationship since much of pharmacogenetic research is on pharmacokinetics. As more research focuses on pharmacodynamics, the scope for commonality between disease and pharmacogenetic variation expands. Our results highlight that the strength of risk is an important consideration, and the range of risk estimates we observed indicates a need for development of consensus on level of risk considered clinically significant. For example, a risk of 1.3 might be statistically significant, but not significant enough to warrant disclosure. When risk is clinically significant, the nature of the condition and its treatability become important considerations. Arguably, clinicians have a duty to disclose risks for diseases that are both severe and treatable in order to provide opportunities for early or preventive treatment. Patients also have a right to know if a test will reveal risks for untreatable disease or potentially stigmatizing condition such as mental illness. Thus, in addition to threshold of clinical significance, there is need for consensus development on appropriate informed consent procedures and follow-up. An interesting finding of our study was the large proportion of variants (79%) for which we found only one study suggesting a particular disease association; some were reported to be associated with risk for up to 20 diseases based on only one study assessing each association. These findings may be subject to publication bias – the tendency for positive studies to be published more frequently than negative ones. It is also possible that some variants are still under study or that validation studies have not been attempted. Regardless, our finding indicates the need for periodic review to update the evidence for gene-drug and gene-disease associations and determine the proportion of associations that are subsequently confirmed. Our data suggest that the ancillary risk information associated with specific variants may change as more studies

are completed and associations disproved or verified. For many variants the clinical utility of pharmacogenetic test use remains uncertain, regardless of ancillary information. Arguably only a few genetic variants may provide clinical utility, such as TPMT genotype and adverse response to 6-MP chemotherapy, and CYP29 genotype and warfarin dosing. The validity of several well-cited examples of pharmacogenetic associations – CETP variants and response to statins, alpha-adducin and response to diuretics, and ACE variants and response to ACE inhibitors – have recently been called into question.27–32 In addition, well-known and wellstudied polymorphisms such as these may be more likely to be studied for their associated pharmacogenetic effects, which may introduce bias when conducting inquiries such as ours. Ancillary risk information represents an unintended consequence of tests performed to improve health care by predicting drug response, adverse events, and dosage requirements.1 Our data indicate that some variants very likely will provide ancillary disease risk information. The benefit of treating disease using pharmacogenetics may often outweigh any potential harm posed by ancillary risk information, such as in cases where genetic information provides highly predictive information on drug response or adverse effects. In other cases, ancillary information might itself provide benefit by identifying a risk that could be reduced by early intervention. However, in some instances the potential harm of ancillary information may outweigh the benefit of the test, and evaluation of these tests for clinical use needs to consider this potential. Ancillary risk information may warrant the most careful scrutiny when there are compelling data for both pharmacogenetic benefit and disease risk; especially when the disease risk is high and the disease is severe and untreatable or stigmatizing, such as risk for Alzheimer disease, alcoholism or mental illness. Cases where significant disease risk associations are identified after patients have received pharmacogenetic testing will add a further challenge. Unless recognized proactively and included in practice guideline development and other policy-making activities, the potential harm of ancillary risk information might not be considered. Policy makers will need to consider when and how ancillary information should be provided to patients, whether formal informed consent procedures are needed, and how risks and benefits of ancillary risk information should be weighed. Many scholars have argued against an exceptionalist approach to genetic information.33–35 While genetic information is not inherently exceptional compared to phenotypic risk information of the same magnitude, genetic tests may raise the issue of ancillary information more often because of the pleiotropic effects of many variants. This study has limitations. We did not evaluate all possible pharmacogenetic variants, and other drug response associations may have emerged since the original list was published (4). For disease associations, we used the inclusion criteria of at least two positive studies, with no explicit consideration of the number of ‘negative’ studies; this definition may be



88

lenient. For example, Hirschhorn et al. used the criteria of positive results in at least 75% of published studies to define a valid association; of 166 putative associations between disease risk and genotype, only six variants met this criteria.36 As such, the pharmacogenetic associations we studied (4) and the disease risk associations we identified likely include some ‘false positives.’ Our analysis was intended to provide an overview of ancillary disease risk information, and thus did not examine in detail the clinical nuances of any one variant or its accompanying disease risk, and having only reviewed abstracts we cannot comment on the quality of the studies. Disease-specific or variant-specific inquiries into potential ancillary risk information should be pursued in more detail in future work. Nevertheless, our data indicate that a substantial proportion of pharmacogenetic tests may provide ancillary disease risk information. Evaluating the validity of both pharmacogenetics and disease risk associations is challenging, but the potential for ancillary disease risk information should be considered. We anticipate that the implications of ancillary risk information will vary widely by each test and its abilities to detect disease risk of varying severity, stigma, and treatability. Researchers and policy makers should consider ancillary information in pharmacogenetic research and policies on the use of pharmacogenetic tests in practice. It may sometimes provide an additional benefit to a patient; at other times it may represent unwelcome information that must be considered in relation to the benefit of the pharmacogenetic test.

Materials and methods We conducted a literature search and abstract review of gene variant – disease risk association studies relevant to pharmacogenetics. We chose a list of previously published by Goldstein et al.6 of 42 genetic variants that have been associated with drug response in at least two published studies as our source list of pharmacogenetic variants (Table 2). Given the difficulty of determining the validity of genetic associations these associations cannot be considered definitively validated, but the list provides a representative sample of genetic variants being studied for pharmacogenetic implications. Using PubMed, we searched for human English-language studies published since 2000 using the search terms ‘[variant name] gene’ AND ‘risk or susceptibility.’ To define potential variant-disease associations, we used the same criteria as Goldstein et al.: the publication of at least two association studies of the genetic variant of interest and disease risk. We defined a positive finding as an association with statistical significance at the alpha 0.05 level or confidence intervals for odds or risk ratios that did not include 1.0. We collected the following information from the abstracts: the variant under study, the disease association being tested; whether the finding was a positive (indicating a variant-disease association) or negative (not indicating a variant-disease association); and whether the noted association was for the same disease as the potential


Table 2

Variant list

Gene

Variant

Drug target/pathway protein ACE* Ins-del (I-D) ADRB1 Arg389Gly (missense) ADBR2* Arg16Gly (missense) AGT* Met235Thr (missense) AGTR1 A1166T (30 UTR) ALOX5 Promoter VNTR BDKRB2 C-58T (promoter) CETP* Taq1b polymorphism B1/B2 (intronic) DRD2* 30 UTR Taq1a A1/A2 DRD3* Ser9Gly (missense) DRD4* Exon 3 VNTR 4-fold (2 to 7-fold repeat), 7-fold (48 bp repeat) GNB3* Exon 10 C825 T (syn) GRIN2B C2664T synonymous HTR2A T102C (syn) HTR2A His452Tyr (missense) LIPC C514T (promoter) MTHFR* C5667T (missense) SLC6A3* 3’ VNTR 10 repeat allele (3’UTR) SLC6A4 Promoter VNTR long/short TPH1 A218C (intronic) TYMS TSER*2/3 (promoter VNTR) Drug transporter ABCB1 Metabolism BCHE* COMT* CYP2C19 CYP2C9* CYP2D6 DPYD GSTM1* GSTM3* GSTP1* GSTT1* NAT2 TPMT UGT1A1* Other ADD1 APOE* FCGR 3A HLA-B IL10* TNF* XRCC*1

C3435T (syn)

Asp70Gly; Ala539Thr (K allele); p.Asp70Gly p,Ala539Thr Val158Met *2, *3 *2, *3 *3, *4, *5 alleles *2A (splice variant) GSTM1 null (gene deletion) GSTM3*A/intronic (GSTM3*B ins/del) Ile105Val (A1578G) GSTT1 null (gene deletion) Slow acetylator *5B, *6A, *7A, *7B, *14A, *14B (missense) *2, *3A, *3C (missense) *28 promoter VNTR

Gly460Trp (missense) E4 (missense) Phe158Val (missense) *5701 (N/A) A-1082G (as part of haplotype promoter) G308A Arg399Gln

*Indicates association for disease risk in at least two studies. Source: Goldstein et al.6

pharmacogenetic use. In compiling our data we noted, for each variant: diseases for which there was an association in one study only; conflicting results (both association and no


89

association suggested in different studies); and diseases for which no association was found. We excluded review articles and abstracts that did not clearly state the variant under study. After collecting these data we categorized the results by potential social, ethical, or policy implications to begin to understand the scope and implications of ancillary risk information as they apply to future research and policy.

15

16

17

Acknowledgments 18

This work was supported in part by the UW NHGRI and NICHD sponsored Center for Genomics and Healthcare Equality, Grant #: P50HG003374 and the University of Washington Biobehavioral Cancer Prevention and Control Training Program (NIH-R25 CA92408). We also acknowledge the contributions of the Pharmacogenetics Working Group of the UW Center for Genomics and Healthcare Equality to the design and discussion of this study.

19 20

21

Duality of Interest These authors state no conflict of interest. References 1 Weinshilboum R. Inheritance and drug response. N Engl J Med 2003; 348: 529–537. 2 Motulsky AG. Drug reactions enzymes, and biochemical genetics. J Am Med Assoc 1957; 165: 835–837. 3 Breckenridge A et al. Pharmacogenetics: ethical problems and solutions. Nat Rev Genet 2004; 5: 676–680. 4 Buchanan A et al. Pharmacogenetics: ethical issues and policy options. Kennedy Inst Ethics J 2002; 12: 1–15. 5 Sankar P, Kahn J. BiDil: Race Medicine Or Race Marketing? Health Aff (Millwood) 2005; Web exclusive: W5–W455, to W5–W463. 6 Goldstein DB, Tate SK, Sisodiya SM. Pharmacogenetics goes genomic. Nat Rev Genet 2003; 4: 937–947. 7 Netzer C, Biller-Andorno N. Pharmacogenetic testing, informed consent and the problem of secondary information. Bioethics 2004; 18: 344–360. 8 Lipton P, Afshar H, Bobrow CBE M, Caldwell J, Lindpaintner K, Rawlins M, Rose N, Starey N, Weale A. Pharmacogenetics: ethical issues, (ed. Bioethics, N.C.o.) Nuffield Council on Bioethics, London, 2003. 9 Lehmann DJ et al. Large meta-analysis establishes the ACE insertiondeletion polymorphism as a marker of Alzheimer’s disease. Am J Epidemiol 2005; 162: 305–317. 10 Rodriguez-Carreon AA et al. Tumor necrosis factor-alpha -308 promoter polymorphism contributes independently to HLA alleles in the severity of rheumatoid arthritis in Mexicans. J Autoimmun 2005; 24: 63–68. 11 Correa PA, Gomez LM, Cadena J, Anaya JM. Autoimmunity and tuberculosis. Opposite association with TNF polymorphism. J Rheumatol 2005; 32: 219–224. 12 Woolley N, Mustalahti K, Maki M, Partanen J. Cytokine gene polymorphisms and genetic association with coeliac disease in the Finnish population. Scand J Immunol 2005; 61: 51–56. 13 Yamamoto-Furusho JK et al. Polymorphisms in the promoter region of tumor necrosis factor alpha (TNF-alpha) and the HLA-DRB1 locus in Mexican mestizo patients with ulcerative colitis. Immunol Lett 2004; 95: 31–35. 14 Foley PF et al. Association studies of neurotransmitter gene polymorphisms in alcoholic Caucasians. Ann N Y Acad Sci 2004; 1025: 39–46.

22 23

24 25 26

27 28

29

30 31

32 33

34 35 36

Young RM, Lawford BR, Nutting A, Noble EP. Advances in molecular genetics and the prevention and treatment of substance misuse: Implications of association studies of the A1 allele of the D2 dopamine receptor gene. Addict Behav 2004; 29: 1275–1294. Erblich J, Lerman C, Self DW, Diaz GA, Bovbjerg DH. Stress-induced cigarette craving: effects of the DRD2 TaqI RFLP and SLC6A3 VNTR polymorphisms. Pharmacogenomics J 2004; 4: 102–109. Erblich J, Lerman C, Self DW, Diaz GA, Bovbjerg DH. Effects of dopamine D2 receptor (DRD2) and transporter (SLC6A3) polymorphisms on smoking cue-induced cigarette craving among AfricanAmerican smokers. Mol Psychiatry 2005; 10: 407–414. Pritchard A et al. Association study and meta-analysis of low-density lipoprotein receptor related protein in Alzheimer’s disease. Neurosci Lett 2005; 382: 221–226. Ordovas JM, Mooser V. The APOE locus and the pharmacogenetics of lipid response. Curr Opin Lipidol 2002; 13: 113–117. Bailey LB. Folate, Methyl-Related Nutrients, Alcohol, and the MTHFR 677C-T Polymorphism Affect Cancer Risk: Intake Recommendations. J Nutr 2003; 133: 3748S–33753. Lea RA et al. Genetic variants of angiotensin converting enzyme and methylenetetrahydrofolate reductase may act in combination to increase migraine susceptibility. Brain Res Mol Brain Res 2005; 136: 112–117. Relton CL et al. Gene-gene interaction in folate-related genes and risk of neural tube defects in a UK population. J Med Genet 2004; 41: 256–260. Kelly PJ et al. Homocysteine, MTHFR 677C-T polymorphism, and risk of ischemic stroke: results of a meta-analysis. Neurology 2002; 59: 529–536. Hong YC et al. Polymorphisms of XRCC1 gene, alcohol consumption and colorectal cancer. Int J Cancer 2005; 116: 428–432. La Torre G, Boccia S, Ricciardi G. Glutathione S-transferase M1 status and gastric cancer risk: a meta-analysis. Cancer Lett 2005; 217: 53–60. Wenzlaff AS, Cote ML, Bock CH, Land SJ, Schwartz AG. GSTM1, GSTT1 and GSTP1 polymorphisms, environmental tobacco smoke exposure and risk of lung cancer among never smokers: a population-based study. Carcinogenesis 2005; 26: 395–401. Mellen PB, Herrington DM. Pharmacogenomics of blood pressure response to antihypertensive treatment. J Hypertens 2005; 23: 1311–1325. Boekholdt SM et al. Cholesteryl ester transfer protein TaqIB variant, high-density lipoprotein cholesterol levels, cardiovascular risk, and efficacy of pravastatin treatment: individual patient meta-analysis of 13 677 subjects. Circulation 2005; 111: 278–287. Davis BR et al. Antihypertensive therapy, the alpha-adducin polymorphism, and cardiovascular disease in high-risk hypertensive persons: the Genetics of Hypertension-Associated Treatment Study. Pharmacogenomics J 2006. Tantisira KG, Weiss ST. The pharmacogenetics of asthma: an update. Curr Opin Mol Ther 2005; 7: 209–217. Feldman DS, Carnes CA, Abraham WT, Bristow MR. Mechanisms of disease: beta-adrenergic receptors – alterations in signal transduction and pharmacogenomics in heart failure. Nat Clin Pract Cardiovasc Med 2005; 2: 475–483. Binder EB, Holsboer F. Pharmacogenomics and antidepressant drugs. Ann Med 2006; 38: 82–94. Green MJ, Botkin JR. ‘Genetic exceptionalism’ in medicine: clarifying the differences between genetic and nongenetic tests. Ann Intern Med 2003; 138: 571–575. Gostin LO, Hodge JG. Genetic privacy and the law: an end to genetics exceptionalism. Jurimetrics 1999; 40: 21–58. Faden RR et al. Public stem cell banks: considerations of justice in stem cell research and therapy. Hastings Cent Rep 2003; 33: 13–27. Hirschhorn JN, Lohmueller K, Byrne E, Hirschhorn K. A comprehensive review of genetic association studies. Genet Med 2002; 4: 45–61.
