Human Genetics

, Volume 110, Issue 6, pp 553–560

CYP3A4-V and prostate cancer in African Americans: causal or confounding association because of population stratification?

  • Rick A. Kittles
  • Weidong Chen
  • Ramesh K. Panguluri
  • Chiledum Ahaghotu
  • Aaron Jackson
  • Clement A. Adebamowo
  • Robin Griffin
  • Tyisha Williams
  • Flora Ukoli
  • Lucile Adams-Campbell
  • John Kwagyan
  • William Isaacs
  • Vincent Freeman
  • Georgia M. Dunston
Original Investigation

DOI: 10.1007/s00439-002-0731-5

Cite this article as:
Kittles, R.A., Chen, W., Panguluri, R.K. et al. Hum Genet (2002) 110: 553. doi:10.1007/s00439-002-0731-5

Abstract.

CYP3A4-V, an A to G promoter variant associated with prostate cancer in African Americans, exhibits large differences in allele frequency between populations. Given that the African American population is genetically heterogeneous because of its African ancestry and subsequent admixture with European Americans, case-control studies with African Americans are highly susceptible to spurious associations. To test for association with prostate cancer, we genotyped CYP3A4-V in 1376 (2 N) chromosomes from prostate cancer patients and age- and ethnicity-matched controls representing African Americans, Nigerians, and European Americans. To detect population stratification among the African American samples, 10 unlinked genetic markers were genotyped. To correct for the stratification, the uncorrected association statistic was divided by the average of association statistics across the 10 unlinked markers. Sharp differences in CYP3A4-V frequencies were observed between Nigerian and European American controls (0.87 and 0.10, respectively; P<0.0001). African Americans were intermediate (0.66). An association uncorrected for stratification was observed between CYP3A4-V and prostate cancer in African Americans (P=0.007). A nominal association was also observed among European Americans (P=0.02) but not Nigerians. In addition, the unlinked genetic marker test provided strong evidence of population stratification among African Americans. Because of the high level of stratification, the corrected P-value was not significant (P=0.25). Follow-up studies on a larger dataset will be needed to confirm whether the association is indeed spurious; however, these results reveal the potential for confounding of association studies by using African Americans and the need for study designs that take into account substructure caused by differences in ancestral proportions between cases and controls.

Copyright information

© Springer-Verlag 2002

Authors and Affiliations

  • Rick A. Kittles
    • 1
  • Weidong Chen
    • 1
  • Ramesh K. Panguluri
    • 1
  • Chiledum Ahaghotu
    • 2
  • Aaron Jackson
    • 2
  • Clement A. Adebamowo
    • 4
  • Robin Griffin
    • 1
  • Tyisha Williams
    • 1
  • Flora Ukoli
    • 3
  • Lucile Adams-Campbell
    • 3
  • John Kwagyan
    • 1
  • William Isaacs
    • 5
  • Vincent Freeman
    • 6
  • Georgia M. Dunston
    • 1
  1. 1.National Human Genome Center, Howard University, 2041 Georgia Avenue, Washington, DC 20060USA
  2. 2.Divison of Urology, Howard University Hospital, Washington, DC 20060USA
  3. 3.Howard University Cancer Center, Washington, DC 20060USA
  4. 4.University College Hospital, IbadanNigeria
  5. 5.Johns Hopkins Oncology Center, Johns Hopkins University, Baltimore, MD 21231USA
  6. 6.Department of Preventive Medicine and Epidemiology, Strich School of Medicine, Loyola University Medical Center, Maywood, IL 60153USA