Validation of genetic case-control studies in AIDS and application to the CX3CR1 polymorphism

New polymorphisms have been recently identified in CX3CR1, a coreceptor for some HIV-1 strains, one of which was associated with a strong acceleration of HIV disease progression. This effect was observed both by a case-contro l study involving 63 nonprogressors (NP) from the asymptomatic long-term (A LT) cohort and Kaplan-Meier analysis of 426 French seroconverters (SEROCO c ohort). These results prompted us to analyze these polymorphisms in 244 nonprogress ors (NPs) and 80 rapid progressors (RPs) from the largest case-control coho rt known to date. the GRIV cohort. Surprisingly, the genetic frequencies fo und were identical for both groups under all generic models (p > .8). The d iscrepancy with the previous work stemmed only from the difference between GRIV NPs versus ALT NPs. We hypothesized this might be due to the limited n umber of NPs in ALT (n = 63) and in this line we reanalyzed the data previo usly collected on GRIV for over 100 different genetic polymorphisms: we eff ectively observed that the genetic frequencies of some poly morphisms could vary by as much as 10% (absolute percentage) when computing them on the fi rst 50 NP subjects enrolled, on the first 100, or on all the NPs tested (24 0 study subjects). This observation emphasizes the need for caution in cast -control studies involving small numbers of subjects: p values should be lo w or other control groups should be used. However, the association of the CX3CR1 polymorphism with progression seems quite significant in the Kaplan-Meier analysis of the SEROCO cohort (426 in dividuals), and the difference observed with GRIV might be explained by a d elayed effect of the polymorphism on disease. Further studies on other sero converter cohorts are needed to confirm the reported association with disea se progression.