© 2001 by Oxford University Press
Journal of the National Cancer Institute, Vol. 93, No. 2, 156-157,
January 17, 2001
© 2001 Oxford University Press
CORRESPONDENCE |
Re: Population Stratification in Epidemiologic Studies of Common Genetic Variants and Cancer: Quantification of Bias
Correspondence to: Robert C. Millikan, D.V.M., Ph.D., CB #7400, Department of Epidemiology, University of North Carolina, Chapel Hill, NC (e-mail: bob_millikan{at}unc.edu).
Epidemiologists have long suspected that, in populations with racial or ethnic subgroups, spurious associations may arise between genetic markers and disease (1). Incomplete mixing of subgroups, known as population stratification or admixture, can lead to bias if one more or subgroups carries both a higher prevalence of an allele and a higher risk of disease (2). The magnitude and direction of the resulting bias are not well understood. Recently, a variety of alternative study designs have been proposed to avoid the problem of population stratification, including case–parent and case–sibling methods (2).
In a recent issue of the Journal, Wacholder et al. (3) provided evidence that population stratification leads to minimal bias in epidemiologic studies of non-Hispanic U. S. Caucasians of European origin. The authors calculated a confounding risk ratio (CRR), the ratio of the crude and race-adjusted relative risks for the effect of genotype on disease. With the use of empiric data as well as data simulations, the authors found that the range of CRRs for European-Americans was 0.78–1.22. The majority of CRR values centered around the null value of 1.00, indicating no confounding by race.
I performed similar calculations using previously published data from the Carolina Breast Cancer Study, a population-based, case–control study of inasive breast cancer in North Carolina. The study was approved by the Institutional Review Board at the University of North Carolina School of Medicine, Chapel Hill, and written informed consent was obtained from all study participants. The study population is 41% African-American, 58% white, and 1% other groups. Data for glutathione S-transferase M1, T1, and P1 (4); N-acetyltransferase 1 and 2 (5); catechol-O-methyltransferase (6); and P57/KIP2 (7) genotypes were used. The previously reported odds ratios (ORs) for "at-risk" genotypes ranged from 0.7 to 1.4, and there was little evidence for interaction between genotype and environmental exposures. For each genetic locus, logistic regression (implemented in SAS; SAS Institute, Cary, NC) was used to calculate a confounding OR (COR), the OR adjusted for age divided by the OR adjusted for age and race (3). Self-reported race was coded as a dichotomous variable, African-American versus non-African-American. Non-African-Americans included whites, as well as seven Native Americans, three Asian-Americans, and three women who listed their race as "multiracial." CORs were also calculated after stratifying on smoking and a variety of other environmental factors.
The range of CORs for genotype was 0.94–1.12 (Table 1
). CORs deviated farther from the null (1.0) as the difference in genotype frequencies between African-Americans and whites increased, but most values were close to 1.0. After stratifying on environmental factors (data not shown), the range of CORs was 0.84–1.34, but most values remained close to the null. These results are compatible with the predictions of Wacholder et al. (3) and extend their findings to African-Americans and to whites. The results suggest that, in studies of breast cancer, failure to adjust for race does not lead to appreciable bias when estimating ORs for the genetic markers shown here. However, this finding may not apply to other cancers or other genetic markers.
|
REFERENCES
1
Beaty T, Khoury MJ. The interface of genetics and epidemiology. Epidemiol Rev 2000;22:120–5.
2
Weinberg C, Umbach D. Choosing a retrospective design to assess joint genetic and environmental contributions to risk. Am J Epidemiol 2000;152:197–203.
3
Wacholder S, Rothman N, Caporaso N. Population stratification in epidemiologic studies of common genetic variants and cancer: quantification of bias. J Natl Cancer Inst 2000;92:1151–8.
4
Millikan R, Pittman G, Tse CK, Savitz D, Newman B, Bell D. Glutathione S-transferases M1, T1, and P1 and breast cancer. Cancer Epidemiol Biomarkers Prev 2000;9:567–73.
5
Millikan R, Pittman GS, Newman B, Tse CK, Selmin O, Rockhill B, et al. Cigarette smoking, N-acetyltransferases 1 and 2, and breast cancer risk. Cancer Epidemiol Biomarkers Prev 1998;7:371–8.
6
Millikan R, Pittman GS, Tse CK, Duell E, Newman B, Savitz D, et al. Catechol-O-methyltransferase and breast cancer risk. Carcinogenesis 1998;19:1943–7.
7 Li Y, Millikan RC, Newman B, Conway K, Tse CK, Liu ET. P57 (KIP2) polymorphisms and breast cancer risk. Hum Genet 1999;104:83–8.[CrossRef][Web of Science][Medline]cancerlit;99168755
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  J. Little, J. P.T. Higgins, J. P.A. Ioannidis, D. Moher, F. Gagnon, E. von Elm, M. J. Khoury, B. Cohen, G. Davey-Smith, J. Grimshaw, et al. STrengthening the REporting of Genetic Association Studies (STREGA): An Extension of the STROBE Statement Ann Intern Med, February 3, 2009; 150(3): 206 - 215. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L.-Y. Wang and W.-C. Lee Population Stratification Bias in the Case-Only Study for Gene-Environment Interactions Am. J. Epidemiol., July 15, 2008; 168(2): 197 - 201. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. M Goldstein, M.-G. Dondon, and N. Andrieu Unconditional analyses can increase efficiencyin assessing gene-environment interaction of the case-combined-control design Int. J. Epidemiol., August 1, 2006; 35(4): 1067 - 1073. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. R. Rebbeck, C. H. Halbert, and P. Sankar Genetics, Epidemiology, and Cancer Disparities: Is it Black and White? J. Clin. Oncol., May 10, 2006; 24(14): 2164 - 2169. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. C.-c. Hsu, M. S. Bray, W.H.L. Kao, J. S. Pankow, E. Boerwinkle, and J. Coresh Genetic Variation of the Renin-Angiotensin System and Chronic Kidney Disease Progression in Black Individuals in the Atherosclerosis Risk in Communities Study J. Am. Soc. Nephrol., February 1, 2006; 17(2): 504 - 512. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. R. Rebbeck and P. Sankar Ethnicity, Ancestry, and Race in Molecular Epidemiologic Research Cancer Epidemiol. Biomarkers Prev., November 1, 2005; 14(11): 2467 - 2471. [Full Text] [PDF]
|
|
|
|
 |
|
 R. Aplenc, J. Thompson, P. Han, M. La, H. Zhao, B. Lange, and T. Rebbeck Methylenetetrahydrofolate Reductase Polymorphisms and Therapy Response in Pediatric Acute Lymphoblastic Leukemia Cancer Res., March 15, 2005; 65(6): 2482 - 2487. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Khlat, M.-H. Cazes, E. Genin, and M. Guiguet Robustness of Case-Control Studies of Genetic Factors to Population Stratification: Magnitude of Bias and Type I Error Cancer Epidemiol. Biomarkers Prev., October 1, 2004; 13(10): 1660 - 1664. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||