Abstract for presentation at 11th International Congress of Human Genetics

Multiple studies in several different disorders have reported significant association with the same genetic locus, but with opposite risk alleles. Do such “flip-flop” associations indicate a confirmation of association, or evidence for spurious association? We hypothesized that “flip-flop” associations may be attributable to multi-locus effects. In the current study, we tested the single-locus association under four different models of multi-locus effect. The theoretical modeling shows that the direction of allelic association may flip when the target risk allele is inversely correlated with another risk allele at another locus, or positively associated with a protective allele at another locus. The likelihood of “flip-flop” associations depends on allele frequency and inter-locus correlation. This phenomenon is only seen when these two loci act in concert to influence risk of disease. We then applied these findings to previous reports. We found the population variation in linkage disequilibrium across the catechol-O-methyltransferase (COMT) gene may lead to “flip-flop” associations between a COMT gene polymorphism and schizophrenia in different populations. Additionally, we re-examined the association between the glyceraldehyde-3-phosphate dehydrogenase (GAPD) gene and late-onset Alzheimer disease (LOAD). The variation in correlation between a GAPD gene polymorphism and the apolipoprotein E ε4 allele (a confirmed risk factor for LOAD) might result in “flip-flop” associations of this GAPD gene polymorphism with LOAD in different populations. In summary, opposite effects of the same allele may be found in populations characterized by different patterns of inter-locus correlations. A genuine risk allele may appear to be a relatively protective allele (and vice versa) when a multi-locus effect is not taken into consideration. Caution needs to be exercised when interpreting "flip-flop" associations.