Admixture analysis among african-americans with schizophrenia and controls

Objectives: To determine if an association between schizophrenia and HLA DQB1*0602 is due to case-control differences in admixture. Background: We have reported a significant negative association of schizophrenia with HLA DQB1*0602 among female African-Americans [Nimgaonkar et al., 1992]. The cases fulfilled DSM-III-R criteria and the controls were African-American adults screened for absence of psychosis, depression, drug abuse, or autoimmune diseases. We have replicated these findings (n = 149, OR 0.56, 95% CI 0.32, 0.97). The negative association is also detectable at DQCAR and HLA DQA1, loci in tight linkage disequilibrium with HLA DQB1. Methods: As reported previously, the association with HLA DQB1*0602 is not evident in comparison with unscreened controls (cord blood samples from African-American neonates). This raised the possibility of a spurious association due to differing levels of admixture among cases and controls. Therefore, we examined D7S657 and D5S421, two short tandem repeat polymorphisms (STRPs) among cases, adults, and neonatal controls (n = 46, 43, and 58, respectively). These STRPs were selected as their allelic distribution is significantly different among Caucasians and Africans. We detected five significant differences in frequencies of individual alleles among the three groups (uncorrected for multiple comparisons). When the overall distributions of alleles were compared among these three groups, the following significant differences emerged: cases vs. adult controls (D7S657, P < 0.01); cases vs. neonates (D5S421, P < 0.02). Conclusions: The association with HLA DQB1*0602 could be "spurious." This possibility is being evaluated further using RB2300 and FYnull, two bi-allelic markers which discriminate better between Caucasian and African chromosomes.