INVITED COMMENTARY Genetic : The Potential Benefits and Challenges of Using Genetic Information to Improve Human Health

Amanda A. Seyerle, Christy L. Avery

Genetic epidemiology has the potential to significantly affect of genes associated with complex chronic diseases. For human health. This commentary examines major develop- example, genes shown via linkage to be associated with rare ments in the field’s history, promising avenues of research, familial forms of breast cancer in specific families, including and possible challenges faced by genetic epidemiologists. variants in the GPT and ACP genes, showed no association with breast cancer in the general population [10]. Limitations of linkage analysis led researchers to inves- nderstanding the distribution and determinants of tigate other approaches to the identification of genes asso- Uhuman disease is increasingly a priority in an age of ciated with complex diseases, including candidate gene burgeoning health care costs and rising disease burdens. At studies. Candidate gene studies, which use a priori hypoth- the crossroads of and epidemiology is the field of eses to evaluate the evidence for association between the genetic epidemiology, which examines the role of inherited outcome of interest and variants in or near selected genes, factors in disease etiology. Current public health benefits became popular because they provided greater power for of research are numerous, including an improved detecting associations for ; they also could understanding of disease mechanisms [1], targeted cancer be performed in population-based cohort and case-control treatments [2], and better dosage regimens for pharmaceu- studies [11]. Despite leveraging biological knowledge, few ticals [3]. However, to fully appreciate the current and future candidate gene studies successfully identified associations. contributions of genetic epidemiology to improving human For example, a 2009 review of studies of candidate genes for health, it is necessary to understand the major milestones in obesity found that, of 21 genes examined, only 9 had been genetic epidemiology and the current challenges research- shown to have any association with obesity [12]. In addition, ers face in the ongoing process of deciphering the human very few positive results were replicated in subsequent stud- . ies [13, 14]. Genetic epidemiology is a relatively new field. It was Occurring in parallel with the rise in popularity of candi- first described by Neel and Schull in 1954 [4, 5], when date gene studies was the sequencing of the human genome molecular genetics was still in its infancy. Although direct by the (HGP); the first draft of the measurement of genotypes was not possible then, genetic human genome was published in 2001 [15]. The HGP, which epidemiologists investigated the inheritance of disease by sought to catalog human by identifying examining whether patterns of inheritance (eg, dominant, all human genes and sequencing the 3 billion bases in the recessive, X-linked) were consistent with phenotype pat- human genome [16], has had a lasting influence on the prac- terns observed in large families. For example, early segre- tice of genetic epidemiology. A significant advance made gation analyses of breast and ovarian cancer suggested a possible by the HGP was the ability to conduct large-scale strong genetic etiology with an autosomal dominant mode human genome studies, including genome-wide association of inheritance [6, 7]. studies (GWAS), which allow researchers to test associa- Linkage analysis, the framework for which was first tions between traits of interest and changes in a single base published in 1980 [8], was a natural extension of segrega- pair. Such changes, which are referred to as single-nucle- tion analysis. Made possible by technological advances otide polymorphisms (SNPs), are spaced throughout the that allowed the direct measurement of genotypes, linkage Electronically published November 19, 2013. analysis used populations of related individuals to assess Address correspondence to Ms. Amanda A. Seyerle, Department of the genetic basis of disease; this technique successfully Epidemiology, University of North Carolina at Chapel Hill, Franklin identified the genes responsible for numerous monogenic Street Plaza, 137 E Franklin St, Ste 306, Chapel Hill, NC 27514 (aseyerle disorders, including Tay-Sachs disease, Huntington disease, @email.unc.edu). N C Med J. 2013;74(6):505-508. ©2013 by the North Carolina Institute and cystic fibrosis [9]. On a population level, however, link- of Medicine and The Duke Endowment. All rights reserved. age analysis failed to make inroads into the identification 0229-2559/2013/74611

NCMJ vol. 74, no. 6 505 ncmedicaljournal.com human genome [17]. The availability of large-scale GWAS vent blood clots and embolisms. Today, genes encoding the greatly increased the coverage of candidate gene studies, VKORC1 and CYP2C9 proteins are routinely evaluated in which typically evaluated a small number of SNPs in a hand- clinical settings when warfarin dosage regimens are being ful of pre-specified genes. Using GWAS, genetic epidemiol- assigned [3, 36, 37]. Similarly, the discovery of variants in ogy has made significant progress in unraveling the genomic the CYP2C19 gene that reduce a patient’s ability to metabo- etiology of complex diseases. As of August 2013, GWAS had lize clopidogrel prompted an FDA announcement warning found more than 11,000 SNPs to be associated with one or that this drug, an anticlotting medication used to prevent more of a wide variety of diseases or risk factors [18]. stent thrombosis, may not be effective in patients with spe- Notably, GWAS often identify pathways that are largely cific genetic variants [38]. In response to that announce- ignored by candidate gene studies. One such example is ment, companies have developed new drug therapies that the association between obesity and the FTO gene, which work effectively in all patients, regardless of their CYP2C19 encodes a messenger ribonucleic acid (mRNA) demethyl- genotype [38]. However, pharmacogenomic research needs ase [19] and was identified through a fused-toe phenotype to be expanded to examine diverse traits and to include in mice [20]. Although not a compelling candidate for an racial and ethnic minorities. The inclusion of minorities obesity gene, FTO was one of the first obesity loci identified is especially important because most pharmacogenomic by GWAS, and this finding has been successfully replicated research has examined populations of European descent, across studies and populations [20-25]. Novel findings such even though many genetic variants—including many of the as the FTO gene have also prompted new avenues of inquiry clinically actionable pharmacogenetic SNPs—vary substan- in obesity research, including closer examination of the rela- tially in frequency in ethnically diverse populations [39]. tion between the control of energy expenditure and obesity When the human genome was first sequenced, many [12]. This highlights the ability of GWAS to illuminate novel believed it would positively impact both medicine and public biological pathways underlying disease etiology. health. Warfarin and clopidogrel highlight the translational Despite notable successes, genetic epidemiology has yet potential of genetic epidemiology research for improved to fully elucidate the genetic basis of disease. With obesity, patient care, but most of the findings in genetic epidemiology for example, studies have consistently suggested that a sub- to date have not had a noticeable impact on public health. stantial fraction (16%–85%) of the variation in body mass One reason for the perceived lack of impact is that many of index (BMI) is genetic in origin [26-31]. Although GWAS the current benefits of genomic research are indirect. In 2011 have identified 36 genetic regions associated with BMI, the National Human Genome Research Institute published a these regions only explain a tiny proportion (less than 2%) perspective on genetic medicine, observing that “the most of the estimated of BMI [32, 33]. One possible effective way to improve human health is to understand source of the missing heritability is gene-environment inter- normal biology (in this case, genome biology) as a basis action [34]. In epidemiology, it is commonly understood that for understanding disease biology, which then becomes the diseases are caused by both genetics and the environment; basis for improving health” [1]. Therefore it remains difficult this is true even for monogenic disorders such as phenyl- to fully ascertain the future promise of genetic epidemiology ketonuria (PKU). For example, a child who has the geno- for the advancement of public health. type for PKU will not develop the disease unless he or she Equally important when assessing future contributions is exposed to phenylalanine. In a population-wide context, of genetic epidemiology is the realization that disease eti- studies of gene-environment interactions involve determin- ology is complex, and that genetic risk does not equate to ing whether an environmental factor governs the variation genetic determinism. The complex relationship between in the magnitude of association between a genetic variant genetics and disease poses an ethical dilemma for health and a complex trait. Because environmental exposures are care practitioners regarding what to tell patients about the often more amenable to intervention than are genetic fac- results of genetic tests. Genetic tests may yield incidental tors, gene-environment studies may offer the best avenue findings in addition to the sought-after results, because it is by which genomic research can contribute to improving often more economical to sequence the entire genome than public health; of course, these studies also help identify the to genotype only specific regions. For example, a test for the missing heritability for complex diseases [17]. Huntington could also yield results for Tay-Sachs One promising area in which to study gene-environment disease, breast cancer, or hereditary hemochromatosis interactions is , which involves examin- [40]. There is debate as to whether patients should be told ing the genomic underpinnings of drug response in order to about these incidental findings, which may be of potential better understand adverse drug reactions and to tailor indi- medical value. Recently, the American College of Medical vidualized treatments [35]. As of 2011, there were 70 drugs Genetics and Genomics (ACMG) published a set of recom- for which the US Food and Drug Administration (FDA) had mendations that emphasize the need to alert patients to approved new labeling that includes information on genetic such incidental findings; these guidelines even outline how variants that affect the metabolism of the drug [17]. The specific incidental findings should be handled [41]. However, classic example is warfarin, an anticoagulant used to pre- the ACMG article also points out that for many genetic risk

506 NCMJ vol. 74, no. 6 ncmedicaljournal.com variants there is insufficient data to make a clear recommen- ated with warfarin dose in African-American individuals: a genome- dation [41]. wide association study. Lancet. 2013;382(9894):790-796. 4. DeWan AT. Five classic articles in genetic epidemiology. Yale J Biol Incidental findings are also problematic in research set- Med. 2010;83(2):87-90. tings, and many large studies have performed extensive 5. Neel J, Schull W. Human . Chicago, IL: University of Chicago genotyping of study participants. For example, in 2012, Press; 1954. 6. Go RCP, King M-C, Bailey-Wilson J, Elston RC, Lynch HT. Genetic the Electronic Medical Records and Genomics (eMERGE) epidemiology of breast cancer and associated cancers in high-risk Network—a collection of archives and biorepositories, some families. I. Segregation analysis. J Natl Cancer Inst. 1983;71(3):455- of which perform GWAS—identified a subset of genetic 461. 7. Williams WR, Anderson DE. Genetic epidemiology of breast can- abnormalities about which data are sufficient to warrant cer: segregation analysis of 200 Danish pedigrees. Genet Epidemiol. informing patients of their test results. However, the deci- 1984;1(1):7-20. sion of whether to inform patients of their results was left to 8. Botstein D, White RL, Skolnick M, Davis RW. Construction of a ge- netic linkage map in man using restriction fragment length polymor- the individual repositories [42]. phisms. Am J Hum Genet. 1980;32(3):314-331. Unfortunately, physicians and patients are not well- 9. Grace VM. Human genome epidemiology: reviewing the stakes. Int J informed about how to interpret incidental findings [43]. Health Serv. 2008;38(1):143-159. 10. King M-C, Go RCP, Lynch HT, et al. Genetic epidemiology of breast Reporting these results will only be valuable when both the cancer and associated cancers in high-risk families. II. Linkage anal- physician and the patient understand what the genetic infor- ysis. J Natl Cancer Inst. 1983;71(3):463-467. mation means and how it can be incorporated into clinical 11. Tabor HK, Risch NJ, Myers RM. Candidate-gene approaches for studying complex genetic traits: practical considerations. Nat Rev care [1]. Genet. 2002;3(5):391-397. Although genetic epidemiology is potentially beneficial, 12. Loos RJ. Recent progress in the genetics of common obesity. Br J Clin it is important to understand that the role of genetic epi- Pharmacol. 2009;68(6):811-829. 13. NCI-NHGRI Working Group on Replication in Association Studies; demiology in medical care is largely undetermined, with Chanock SJ, Manolio T, Boehnke M, et al. Replicating genotype-phe- a few notable exceptions. Even with advances such as the notype associations. Nature. 2007;447(7145):655-660. HGP, many important questions remain unanswered, includ- 14. Siontis KC, Patsopoulos NA, Ioannidis JP. Replication of past candi- date loci for common diseases and phenotypes in 100 genome-wide ing how the genome differs in diverse populations, how the association studies. Eur J Hum Genet. 2010;18(7):832-837. environment affects the genome, and what role is played by 15. Venter JC, Adams MD, Myers EW, et al. The sequence of the human (the study of heritable mechanisms that alter genome. Science. 2001;291(5507):1304-1351. 16. Collins FS, McKusick VA. Implications of the Human Genome Proj- gene expression without altering the underlying genetic ect for medical science. JAMA. 2001;285(5):540-544. sequence). Given these questions, it is not surprising that 17. Khoury MJ, Gwinn M, Clyne M, Yu W. Genetic epidemiology with a it could easily be 2020 or beyond before genetics has any capital E, ten years after. Genet Epidemiol. 2011;35(8):845-852. 18. Hindorff LA, MacArthur J, Morales J, et al. A Catalog of Published significant impact on public health [1]. Even as genetic find- Genome-Wide Association Studies. National Human Genome Re- ings become more translational, the general public and even search Institute Web site. http://www.genome.gov/gwastudies/. many health care professionals do not yet have sufficient Accessed September 18, 2013. 19. Jia G, Fu Y, Zhao X, et al. N6-methyladenosine in nuclear RNA is knowledge to use genetic information effectively. Efforts to a major substrate of the obesity-associated FTO. Nat Chem Biol. educate both patients and practitioners are therefore inte- 2011;7(12):885-887. gral to realizing the full potential of this quickly advancing 20. Fawcett KA, Barroso I. The genetics of obesity: FTO leads the way. Trends Genet. 2010;26(6):266-274. field, and this education should parallel current advances in 21. Dušátková L, Zamrazilová H, Sedláčková B, et al. Association of genetic epidemiology. obesity susceptibility gene variants with metabolic syndrome and related traits in 1,443 Czech adolescents. Folia Biol (Praha). Amanda A. Seyerle, MSPH trainee, cardiovascular disease epidemi- 2013;59(3):123-133. ology, Department of Epidemiology, Gillings School of Global Public 22. Frayling TM, Timpson NJ, Weedon MN, et al. A common variant in Health, University of North Carolina at Chapel Hill, Chapel Hill, North the FTO gene is associated with body mass index and predisposes to Carolina. childhood and adult obesity. Science. 2007;316(5826):889-894. Christy L. Avery, PhD assistant professor, Department of Epidemiology, 23. Ibba A, Pilia S, Zavattari P, et al. The role of FTO genotype on eat- Gillings School of Global Public Health, University of North Carolina at ing behavior in obese Sardinian children and adolescents. J Pediatr Chapel Hill, Chapel Hill, North Carolina. Endocrinol Metab. 2013;26(5-6):539-544. 24. Scuteri A, Sanna S, Chen WM, et al. Genome-wide association scan Acknowledgments shows genetic variants in the FTO gene are associated with obesity- Financial support. A.A.S. is supported with funding provided by related traits. PLoS Genet. 2007;3(7):e115. NHLBI training grant T32HL7055. C.L.A. is supported with funding from 25. Xi B, Cheng H, Shen Y, et al. Study of 11 BMI-associated loci identi- NHLBI grant R00HL098458. fied in GWAS for associations with central obesity in the Chinese Potential conflicts of interest. A.A.S. and C.L.A. have no relevant children. PLoS One. 2013;8(2):e56472. conflicts of interest. 26. Adeyemo A, Luke A, Cooper R, et al. A genome-wide scan for body mass index among Nigerian families. Obes Res. 2003;11(2):266- References 273. 1. Green ED, Guyer MS; National Human Genome Research Institute. 27. Allison DB, Kaprio J, Korkeila M, Koskenvuo M, Neale MC, Hayaka- Charting a course for genomic medicine from base pairs to bedside. wa K. The heritability of body mass index among an international Nature. 2011;470(7333):204-213. sample of monozygotic twins reared apart. Int J Obes Relat Metab 2. Xing M, Haugen BR, Schlumberger M. Progress in molecular- Disord. 1996;20(6):501-506. based management of differentiated thyroid cancer. Lancet. 28. McQueen MB, Bertram L, Rimm EB, Blacker D, Santangelo SL. A QTL 2013;381(9871):1058-1069. genome scan of the metabolic syndrome and its component traits. 3. Perera MA, Cavallari LH, Limdi NA, et al. Genetic variants associ- BMC Genet. 2003;4 suppl 1:S96.

NCMJ vol. 74, no. 6 507 ncmedicaljournal.com 29. Pietiläinen KH, Kaprio J, Rissanen A, et al. Distribution and heri- characterization of 32 CYP2C9 allelic variants [published online tability of BMI in Finnish adolescents aged 16y and 17y: a study of ahead of print, June 11, 2013]. Pharmacogenomics J. doi:10.1038/ 4884 twins and 2509 singletons. Int J Obes Relat Metab Disord. tpj.2013.22 1999;23(2):107-115. 38. Holmes DR Jr. Pharmacogenomic testing and antithrombotic 30. Platte P, Papanicolaou GJ, Johnston J, et al. A study of linkage and therapy: ready for prime time? Rambam Maimonides Med J. association of body mass index in the Old Order Amish. Am J Med 2013;4(1):e0005. Genet C Semin Med Genet. 2003;121C(1):71-80. 39. Ramos E, Doumatey A, Elkahloun AG, et al. Pharmacogenomics, 31. Yang W, Kelly T, He J. Genetic epidemiology of obesity. Epidemiol ancestry and clinical decision making for global populations [pub- Rev. 2007;29:49-61. lished online ahead of print, July 9, 2013]. Pharmacogenomics J. 32. Xia Q, Grant SF. The genetics of human obesity. Ann N Y Acad Sci. doi:10.1038/tpj.2013.24. 2013;1281:178-190. 40. Green RC, Berg JS, Berry GT, et al. Exploring concordance and dis- 33. Monda KL, Chen GK, Taylor KC, et al. A meta-analysis identifies new cordance for return of incidental findings from clinical sequencing. loci associated with body mass index in individuals of African ances- Genet Med. 2012;14(4):405-410. try. Nat Genet. 2013;45(6):690-696. 41. Green RC, Berg JS, Grody WW, et al. ACMG recommendations for 34. Manolio TA, Collins FS, Cox NJ, et al. Finding the missing heritability reporting of incidental findings in clinical exome and genome se- of complex diseases. Nature. 2009;461(7265):747-753. quencing. Genet Med. 2013;15(7):565-574. 35. Bochud M. Genetics for clinicians: from candidate genes to whole 42. Fullerton SM, Wolf WA, Brothers KB, et al. Return of individual re- genome scans (technological advances). Best Pract Res Clin Endo- search results from genome-wide association studies: experience of crinol Metab. 2012;26(2):119-132. the Electronic Medical Records and Genomics (eMERGE) Network. 36. Daneshjou R, Tatonetti NP, Karczewski KJ, et al. Pathway analysis Genet Med. 2012;14:424-431. of genome-wide data improves warfarin dose prediction. BMC Ge- 43. Attia J, Ioannidis JT, Thakkinstian A, et al. How to use an article nomics. 2013;14 suppl 3:S11. about : C: what are the results and will they help 37. Niinuma Y, Saito T, Takahashi M, Tsukada C, Ito M, et al. Functional me in caring for my patients? JAMA. 2009;301(3):304-308.

508 NCMJ vol. 74, no. 6 ncmedicaljournal.com