The specificity and diversity of human major histocompatibility complex (MHC) (i.e., HLA) products closely reflect the molecular interplay in infection and autoimmunity accumulated during human evolution and migration (10
). There is a growing consensus that specific HLA alleles and heterozygosity at HLA class I loci can collectively exert a strong impact on disease progression in human immunodeficiency virus type 1 (HIV-1) infection (6
); much of the effect appears to correlate with cytotoxic T-lymphocyte (CTL) responses directly restricted by class I allelic products in both humans (3
) and chimpanzees (2
). HLA markers associated with delayed HIV-1 disease progression may present epitopes found more frequently in various HIV-1 proteins (45
), and the two most prominent alleles (B*27 and B*57) have been found experimentally to induce immunodominant CTL responses to conserved HIV-1 epitopes (15
). The same favorable HLA alleles were overrepresented in canarypox-HIV vaccine recipients with repeatedly detected CTL responses to certain viral proteins (32
). These findings imply that HLA typing data may be useful in predicting the relative population coverage (58
) of epitope-rich HIV vaccine constructs (9
On the other hand, the impact of HLA alleles on HIV-1 disease progression can differ in magnitude and consistency by cohort, ethnicity, and stage of chronic infection. Most of the work to date has been performed in populations infected with clade B viruses; information on differences by viral clade is sparse. Our analyses of HLA and HIV-1 viral load (VL) in 259 Zambians revealed a pattern of HLA associations with unusually high or low VL that is distinct from that reported in other populations. The findings now imply HIV-1 clade-specific differential antigen presentation.
(Data derived from this study were presented in part [abstr. 327-w] at the 9th Conference on Retroviruses and Opportunistic Infections in Seattle, Wash. [24 to 28 February 2002] and at the 27th Annual Meeting [abstr. 217] of the American Society for Histocompatibility and Immunogenetics in San Francisco, Calif. [13 to 17 October 2001]).
HLA class I polymorphism has been shown to be a major determinant of early HIV-1 RNA level (27
) as well as of later disease progression (6
) in seropositive Caucasian populations where clade B virus dominates. In our analysis, the main influential HLA class I alleles and haplotypes showed more differences than resemblances between clade C virus-infected Zambians and clade B virus-infected Caucasians. These differences suggest that evolution of viral clades represents a response to the distinctive HLA class I population profiles and that this hypothesis provides an explanation for population-specific HLA-mediated effects. One particular difference between clade C- and clade B-infected groups is especially noteworthy. Despite the strong association of B*35 alleles (or a subset of those) and B*53 alleles in Caucasians and African-Americans with both rapid disease progression (6
) along with elevated early HIV-1 RNA level (R. A. Kaslow et al., unpublished data), those same alleles lacked any appreciable effect in Zambians. Although no Zambians carried the reportedly more unfavorable B*35 subtypes, subtle allelic heterogeneity in nucleotide sequence encoding residues surrounding the peptide binding groove could not entirely explain this discrepancy, because the only Zambian B*53 allele is B*5301, the one reported to account for rapid disease progression in clade B-infected persons and to prefer epitopes with the xPxxxxxxX motif (13
). Conversely, several class I allele (B*13 and B*39) or haplotype (A*30-Cw*03 and A*23-B*14) associations observed in Zambians have not shown consistent effects in Caucasians, perhaps because their low frequency would have made detection difficult or again because relationships are clade specific. B*14, B*18, B*35, and B*53 alleles from another native African population have been reported to present a variety of HIV-1 epitopes for CTL responses (33
); however, apparently only a subset of CTL epitopes were differentially recognized in highly exposed seronegative individuals showing some resistance to HIV-1 seroconversion. Thus, the timing of specific CTL responses, along with the presence of different competing CTLs induced by relatively favorable and unfavorable HLA class I alleles coexisting in the same individual, can add to the complexity of genetic effects.
Focusing attention on newly identified markers with the most statistically significant relationships can raise concern about potentially spurious associations generated by multiple comparisons. That concern is ever present in studies of numerous highly polymorphic loci, but particularly so in an assessment of previously unreported relationships in relatively small numbers of patients, as was the case in this clade C-infected Zambian population. General caution against overinterpreting the statistical significance of any original finding, pursuit of further comparable population studies in clade-C-infected cohorts, and demonstration of the functional importance of the putative markers are more appropriate ways to refine and clarify the findings described here than arbitrary correction of P values.
Besides the newly recognized cohort-specific differences noted above, certain consistencies between Zambians and other cohorts corroborated HLA allelic effects previously shown to transcend the boundaries of race and clade. For example, the presence of HLA-B*57 has already emerged as uniformly favorable in persons of both Caucasian and African ancestry and infected with both clade A and B viruses (8
). Our work now extends those findings to native Africans bearing HIV-1C infection. Recognition of HLA-B*57 (specifically and exclusively, B*57031) as the single most favorable marker in Zambians, despite LD patterns that differ between this and all other cohorts, highlights B*57 itself rather than its haplotypic lineages as broadly capable of suppressing HIV-1 RNA to a relatively low level during the early stages of infection (44
). B*5701- and B*5703-mediated immunodominant responses to conserved HIV-1 Gag epitopes have been documented in African-American adolescents (P. Goepfert and R. A. Kaslow et al., unpublished data). Cross-reactive HIV-1 p24 epitopes restricted by B*5701 and B*5703 have been simultaneously detected in seropositive European and Kenyan subjects defined as slow progressors (15
). These new findings lend further weight to the importance of B*57 alleles beyond race- and clade-specific immune responses.
Although less impressive, the association of two A*23-related haplotypes (A*23-B*14 and A*23-Cw*07) in Zambians with high VL parallels the association of A*23 with accelerated course of disease progression in Caucasians with clade B infection (30
) and that of A23 with increased transmission of HIV-1A in Kenyans (39
). Two other comparisons are also worth mentioning: the association of Cw*16 on the B*45-Cw*16 haplotype with high VL here was consistent with its overrepresentation in rapid progressors in the GRIV cohort (21
), whereas higher VL in Zambians carrying B*14-Cw*08 and A*23-B*14 contrasted with the associations of B*14 and Cw*08 with a favorable prognosis of disease progression reported elsewhere (21
). Ongoing efforts to evaluate CTL profiles in Zambians with both favorable and unfavorable HLA alleles and haplotypes should provide important clues to mechanisms of immunologic control and escape during HIV-1C infection.
The negligible risk of increased VL from homozygosity at individual class I loci with incomplete four-digit allele resolution was consistent with the minor effects seen in clade B-infected Caucasians with single locus homozygosity at similarly incomplete resolution (Kaslow et al., unpublished). The unequivocally strong influence of class I homozygosity reported in general for the progression of disease in both Africans (63
) and Caucasians (6
) may be due to an increasing importance of diverse alternatives for antigen presentation, as evolving viruses attempt to escape from initially effective HLA-mediated CTL control. Longitudinal comparison of rates of viral mutation and divergence (17
) in homozygotes and heterozygotes could help verify or refute that explanation.
Several implications can be drawn from the diminishing number of relationships seen of both nongenetic host factors and HLA markers to variation in HIV-1 viremia as more parsimonious models were applied. First, nongenetic factors such as age, gender, and duration of infection may modulate genetic effects, but variable impact of those other factors in different settings may make it difficult to assess the multiple host influences on HIV-1 viremia. Second, because HLA effects may covary strongly as a result of LD or reciprocity (increased frequency of one allele compensated by a decrease of one or several others), neither univariate nor multivariable analysis can completely distinguish the colinear effects of closely related markers without comprehensive stratifications and subset comparisons. Finally, a clear host genetic contribution to variability in HIV-1 viremia may not always translate directly to an appreciable difference in later disease progression and vice versa (65
). Complex genetic effects probably depend on differential regulation of events throughout the pathological process, not simply on relationships detectable only at early or late stages of HIV-1 infection.
HLA polymorphisms probably influence plasma HIV-1 RNA concentration relatively early in the infection process (28
). Evaluation of host genetic influence on HIV-1 RNA levels should prove increasingly useful, because early in the course of infection, those levels have profound implications not only for the rate of subsequent progression of both untreated (4
) and treated (4
) disease but also for the transmissibility of HIV-1 from infected to uninfected individuals (12
). If the three consistently favorable HLA markers (Table 7
) prove more broadly protective in Zambians and other southern Africans, individuals carrying them should experience a relatively benign course of disease progression without treatment. Over successive generations of HIV-1C-infected native Africans with corresponding HLA profiles, favorable antigen-presenting molecules should accumulate as unfavorable ones diminish concomitantly. Such evolution may gradually alter the genetic composition and responsiveness of populations in greatest need of effective HIV-AIDS vaccines. Consensus through continuous monitoring of disease progression in diverse ethnic groups carefully characterized for their HLA profiles should improve the design of epitope-rich vaccine constructs (9
) and help predict their relative population coverage (58