Detectability of genetic influences on AIDS progression in seroconverter and seroprevalent cohorts

A variant (64I) of the CCR2 chemokine receptor gene has show an association with delayed progression to AIDS outcomes in several cohort studies of HIV seroconverters. However, the effect is not seen in studies of HIV seroprevalents. It has been argued that this results from a intrinsic loss of information in seroprevalent studies, since times to outcomes are less accurate and may be biased due to early failures in the rapid progression group. We use data on HIV infection times from the San Francisco hepatitis B vaccine study and data on disease progression from the MACS to perform a realistic simulation of infection and disease progression. For the baseline hazard we assume a Weibull distribution of times to AIDS 1993, and exponential distributions for the steps AIDS 1993 to AIDS 1987, and AIDS 1987 to death. With appropriate parameters this model yields survival curves which closely approximate the MACS data. For each subject we record data as it would be observed with the participants enrolled before seroconversion, and as it would have been observed in the SFMHC seroprevalent cohort. We simulate a genetic effect similar to the reported effect of the variant CCR2 chemokine receptor. We show that when the protective genetic effect is constant over time, the effect actually tends to be enhanced in a seroprevalent cohort due to an increase in statistical power as the cohort is enriched in the protective rarer genotype. If the effect of the protective genotype is to reduce the hazard early in infection, it tends to be less observable in a seroprevalent cohort, and in particular an enrichment of protective genotypes among long term survivors is unlikely to be detectable in a cohort with fewer than 500 HIV-1 infected individuals.

See More