Many investigations into genetic determinants of HIV/AIDS have evaluated the effects of Δ32, selected SNPs, and haplotypes across CCR2
on disease progression in a variety of infected populations. Studies of these markers as determinants of acquisition have usually been conducted in pairs of mothers and infants or in exposed men of European ancestry whose male sexual contacts are largely unknown [13
]. Our relatively large prospective study of heterosexual discordant African couples has produced further evidence for involvement of variants in these genes in both control and occurrence of HIV-1 infection.
HHE was associated with slightly higher VL than was seen with other haplotypes, a finding consistent with observations in a number of other studies on different ethnic groups and various modes of transmission [11
]. Further confirmation of the effect of HHE highlights its potential impact on clinical HIV-1 disease control in diverse populations, in contrast to that of the protective Δ32 variant whose distribution is confined to individuals of European ancestry. We detected an association of homozygous HHF*2 (containing CCR2-64I) with lower VL in recent seroconverters but found less certain effects of heterozygous HHF*2. This finding is consistent with previous reports [11
]. Although an early meta-analysis persuasively documented modest protection by the 64I allele against progression of HIV-1 subtype B infection [22
], results in subsequent studies have been less consistent--showing association with slow progression either among Europeans, but not African-Americans [19
] or among African-Americans but not Europeans [23
]. For populations with subtype C infection, however, no previous study is available as a basis for comparison.
As for the influence of CCR2-CCR5 alleles or haplotypes on transmission and acquisition of infection, the highly significant deviation of the distribution of haplotypes from HWE among the index, but not the exposed partners, was strong evidence of a selective effect, and the differential deviation of the seroconverters, but not the persistently seronegatives, corroborated the difference. Neither chance nor systematic selection of couples into the study cohort by their CCR2-CCR5 profile unrelated to infection seems as plausible an explanation as the direct effect on acquisition of HIV-1 infection proposed here.
variant carried by index partners was associated with an appreciable difference in transmission--not even the diplotype HHD/HHE associated with a statistically significant higher mean VL. This relative deviation in level of viremia was apparently not equivalent to the larger deviation conferred by index partner HLA-B*57, a genetic marker associated with a significantly lower transmission rate in this population [14
]. Such differential impact of the different genetic markers may reflect a threshold effect by which a deviation of VL greater than a certain level overrides any genetic influence, but the number of subjects in our cohort was insufficient to assess that possibility.
We observed a trend toward an increased rate of acquisition among the exposed partners carrying HHF*2. In another African population (Cameroon), the frequency of CCR2-64I (HHF*2) was higher in the HIV-1 seropositives (most likely of mixed viral subtype) than in the seronegatives [30
]. However, we remain skeptical about the importance of these findings for several reasons. First, the association and its significance in Zambians diminished in the multivariable analysis. Second, previous evidence for a role of HHF*2 in occurrence of infection is sparse, and there is no other report from a prospective study. Third, considerable uncertainty remains about the functional relevance of the HHF*2 polymorphism and CCR2
itself to HIV-1 infection [31
]. Further population studies alone are unlikely to clarify more precisely the true nature of this genetic contribution.
More rapid HIV-1 acquisition among exposed seronegatives occurred in association with the HHD/HHE diplotype, and the association was stronger in exposed women than men. An association with this diplotype has not been reported before, most likely because the single SNP allele that distinguishes HHD from other haplotypes is only frequent enough in persons of African ancestry. The relatively higher frequency (7%) of HHD/HHE in our population than in Caucasians or other smaller groups of Africans may have facilitated detection of its effect. Associations with higher risk of mother-to-child transmission have been reported for HHD in Africans [32
] and with homozygous HHD in African Americans [33
]. HHE has also been reported to be detrimental for HIV infection as well as disease progression, but HHD/HHE has not been studied previously as a diplotype. Although our findings do not constitute exact replication of previous work, they appear to indicate consistent effects of the two haplotypes across populations with different viral subtypes.
The effects of HHD/HHE appeared stronger in male-to-female transmission. Differences in VL among the donor groups did not explain this difference according to direction of transmission. Nor did the difference arise from any obvious difference in age or sexual exposure of the two groups. For each subgroup stratified by gender, the number of seronegative subjects carrying these genotypes (diplotypes) was relatively small. Analysis based on larger samples will be necessary to reach a reliable conclusion about such gender-specific associations.
One feature of our study worth noting is the advantage of survival analysis of time to transmission/acquisition in detecting relationships that may be weaker in the cross-sectional or case-control approach often used to assess genetic influences on HIV-1 infection. Survival methods may be more sensitive in capturing time-dependent genetic effects on infection just as they have been in the analysis of disease progression.
We did not adjust statistically for the number of genetic polymorphisms tested. Rather we have emphasized those nominally significant associations with CCR2-CCR5
variants that have previously been implicated in HIV/AIDS and de-emphasized those whose involvement was less predictable from earlier studies. The previously documented HHE association with higher VL [11
] provided ample rationale for interpreting our results as confirmatory without treating all haplotypes as equally likely to be involved. The impact of HHD/HHE on seroconversion was predicted somewhat less directly by earlier work associating HHD with a higher frequency of neonatal infection [33
]. An even more important reason why these relationships cannot be readily dismissed as chance findings is that they were observed in the context of significant deviations from HWE of the haplotype distributions in each of the seropositive groups but not the seronegative group.
A consistent effect of the frequent HHE with higher VL in subtype C HIV-1-infected Africans as well as subtype B-infected Europeans and a stronger effect of HHD/HHE could have further ramifications. Since the response to antiretroviral treatment in Europeans may be modified by (Δ32) [34
] and perhaps by other receptor variants [37
], investigators in African settings should consider whether similar studies of CCR2-CCR5
polymorphism might provide epidemiologically or clinically useful prognostic information.