This candidate gene study sought to determine if genetic polymorphisms within a variety of genes coding for cytokines and other immunologic factors confer risk for HAD. While previous studies have examined the relationship between genotype and risk for HAD, this study also considered the contribution of exogenous factors (current drug dependence and major depression), endogenous factors (age and ethnicity), and virologic factors (length of infection) to this risk. Further, the cohort used for this study was thoroughly characterized with regards to neurologic diagnoses and neuropsychological functioning.
Current MDD conferred an almost fivefold risk for HAD. Such a finding raises important questions about the association between the neuropsychiatric and neurocognitive manifestations of HIV. Notably, a recent study found incidence of MDD during a two-year period to be associated with symptomatic HIV disease at baseline, as well as prior psychiatric history.32
Depression has also been found to be associated with accelerated disease progression.33
Such findings suggest an association between depression and the biological processes underlying HAD. Indeed, there may be a common biological pathway leading to HAD and depression. There is now ample evidence to support the hypothesis that pro-inflammatory cytokines released during viral infection are involved in depression.34,35
It is plausible that significantly elevated cytokine levels detected in those with HAD36
also result in depression, as detected by the PRISM in the current study. Alternatively, it may be that a more indirect relationship exists between depression and HAD. For example, depression may lead to delayed initiation of HAART and also greater noncompliance.37,38
As such, HIV may be allowed more time to penetrate the CNS and proliferate, thus leading to neuropathological changes and ultimately HAD. Still another putative explanation is that those with HAD experience greater rates of reactive depression in response to deteriorating cognitive functioning and loss of independence. This question will need to be further addressed, but our results undoubtedly provide strong evidence for a link between depression and HAD.
The only other variable that offered statistically significant prediction value was the rs1130371 SNP within the CCL3 gene. In our sample, individuals with HAD had almost double the frequency of homozygosity for the nonancestral T allele (31% versus 16%). CCL3 (also commonly called MIP-1α) is a natural ligand of the CCR5 HIV co-receptor and may be partially responsible for CD8+ T-cell suppression of HIV.39,40
The rs1130371 SNP is a synonymous C→ T substitution at position +868 in exon 2 of the CCL3 gene,41
suggesting that it does not alter the function or expression of CCL3. However, it is in high linkage disequilibrium with a variety of nearby SNPs not only in the CCL3 gene, but also the CCL4 and CCL18 genes.41
There are few studies describing the role of CCL3 in HIV. Modi and colleagues41
genotyped 21 SNPs, including rs1130371, in a 47-kb interval on chromosome 17 containing the genes for CCL3, CCL4 (MIP-1β), and CCL18 (DC-CK1/PARC/AMAC-1). Seven highly correlated SNPs spanning the three genes examined were significantly associated with more rapid disease progression among Europeans Americans, but not African Americans.41
While rs1130371 was not among the seven SNPs, it was in high linkage disequilibrium (LD) with some, including those within the CCL4 and CCL18 genes. In another study, Gonzalez and colleagues examined HIV-1 transmission and disease-modifying effects of CCL3 haplotypes based on the rs1719134 and rs35511254 polymorphisms.42
Three haplotypes based on these two SNPs were identified, and they varied markedly between the African Americans, European Americans, and Hispanics. For example, the authors reported one haplotype (TT) that was common in African Americans but observed in less than 1% of European Americans. Among the European Americans in their sample, the length of time before developing AIDS was shorter in individuals lacking the common ancestral haplotype (CC). While no disease-modifying effect was observed for any of the CCL3 haplotypes in African Americans, possession of the TT haplotype among this group was associated with a significantly lower risk of HIV-1 acquisition. In short, the Gonzalez study indicates considerable variation in genotype between races, and differential disease modifying effects depending on ethnicity. Importantly, our analyses did indicate a significantly different distribution of rs1130371 genotypes between African Americans and European Americans. However, after entering an interaction term for ethnicity-by-genotype, our logistic regression was still significant, as was the rs1130371 SNP. Despite this, it is clear from the existing literature on CCL3 that further examination of this stretch of the genome coding for this chemokine is required to determine the specific polymorphism or haplotype that is driving the findings reported here.
The role of CCL3 has also been investigated in other neurodegenerative conditions. A relationship between CCL3 and Alzheimer’s disease (AD) was previously reported.43
In fact, the peripheral T-cells of individuals with AD demonstrate over expression of CCL3, which promotes T-cell transendothelial migration across the blood-brain barrier via binding to CCR5 on brain endothelial cells.44
Such phenomena may be particularly germane to unraveling the neuropathogenesis of HAD. Further, like CCL5 (RANTES) and CCL4, CCL3 is a beta-chemokine and natural ligand of CCR5. CCL5 and CCL4 are considered ARGs due to their modulatory effect on HIV/AIDS disease progress.45
Like those ARGs, CCL3 may act to slow disease progression by inhibiting HIV entry into the cells within the CNS, or cells that would ultimately enter the CNS.
Several unavoidable factors limit the interpretation and generalizability of the results from this study. First, while our sample size was sufficient for the statistical analysis conducted, it is preferable to have very large samples for genetic studies due to the small effects genotype at a single SNP may have upon measurable phenotypes. However, while small, our sample was very well characterized with regards to the phenotype of interest. Thus, the exclusion of many hundred cases not meeting our strict inclusion criteria would have reduced the overall power given perfect data, but probably strengthened the validity of our findings. Second, we decided to conduct the statistical analyses on a sample of mixed ethnicities to increase our sample size of clean data. This strategy seems justified by the homogeneity of the results we found among the ethnic subsets. Further, while we observed significantly different CCL3 allele frequencies between European Americans and African Americans, our statistical analysis indicated that ethnicity was not a factor in the statistical analysis findings. Fourth, the diagnosis of HAD is often difficult considering the high frequency of co-morbid medical and psychiatric conditions in this population. This difficulty was clearly shown in an inter-rater reliability study of HAD diagnosis in this sample,30
and supports our choice of strict inclusion criteria. It is hoped that with more recently refined diagnostic nosology for HAD,46
this problem will be overcome in future analyses. Finally, as discussed above, the high LD in the area of the genome surrounding rs1130371 will require a follow-up study that genotypes this region more densely.