The impact of host genetics on HIV infection and disease progression in the era of highly active antiretroviral therapy

Introduction Suspicion that host genetic factors determine variability in the course of natural HIV-1 infection deepened throughout the 1980s and early 1990s [1,2], but only relatively recently have we recognized the impact of human genetic variation on HIV/AIDS as a complex phenomenon [3–11]. As the control and prevention of HIV-1 infection with highly active antiretroviral therapy (HAART) became more widely available, the growing opportunity for pharmacogenetic studies [12–14] was also accompanied by uncertainty about the residual host genetic influences on HIV/AIDS. The critical assessment of existing immunogenetic and pharmacogenetic findings in HIV/AIDS is not an easy task because other factors such as age [15–18], sex [19–21], ethnicity [22], and viral characteristics [23–28] can also play important roles. Evaluation is further complicated by the variety of terminology (Table 1) and laboratory techniques used by individual investigators. Nevertheless, consensus data from some of the more widely recognized and well-designed studies have provided useful benchmarks in further experimental and translational research; moreover, both the concepts of genetics and the functional consequences of human polymorphisms reflected in studies of HIV/AIDS are being explored in other disease systems [29–34]. These rapidly expanding enterprises should eventually promote a more comprehensive view of the evolving host–pathogen relationships and their relevance to public health.Table 1: Glossary: genetic and related terminologya as used in the text.The impact of host genetics on HIV/AIDS before highly active antiretroviral therapy As captured in various reviews and commentaries [2,5–7,9–11,35–42], major accomplishments in immunogenetics of HIV-1 infection have largely concentrated on the human leukocyte antigen (HLA) complex (genes for class I, class II, and accessory molecules) and the chemokine receptor (CR) and chemokine ligand (CL) systems participating in both innate and adaptive immunity (Fig. 1) [43,44]. Two other systems are also gaining prominence: those involved in natural killer cell activating and inhibiting receptor/ligand function and in the cytokine networks [45–48]. The molecular and cellular interactions during HIV-1 infection universally relate to products of these established as well as other candidate systems. More recent studies have extended these themes [18,43,49–60] and corroborated the multigenic modulation of HIV-1 infection and pathogenesis [7,61,62].Fig. 1.: Consensus findings derived from immunogenetic and pharmacogenetic studies of HIV/AIDS. Host genetic effects can be detected throughout the course of natural (a) and treated (b) HIV-1 infection but are often confounded by other factors (c). Markers consistently showing favorable (F) or unfavorable (U) associations concentrate at HLA class I loci on chromosome 6 and CCR2-CCR5 on chromosome 3. Further evidence suggests that subsets of HLA-B*35 and specific genotypic combinations involving CCR2-CCR5 HHG*2 (both underlined) can account for much if not all of the known effects [43,44]. Several markers with equivocal (?) associations may be more important in certain ethnic groups than in others because of their different frequencies or the contribution from other genes to their distinctive local and extended haplotypes. HLA, Human leukocyte antigen; RANTES, regulated upon activation, normal T cell expressed/secreted; HH, human haplogroup.HLA and related genes Although numerous relationships have been reported in single studies of HIV/AIDS [1,2,42], the effects observed most consistently at the population level can be attributed to relatively few HLA class I markers (Fig. 1). HLA-B22 (serological specificity involving HLA–B*55 and −B*56 in Caucasians) [63,64], HLA-B*27, B*35 (or a subset of this group), B*53, B*57, and class I homozygosity [63–68] seem to modify the course of disease progression. Class I allele sharing between an infected donor and a susceptible recipient can accelerate HIV-1 transmission [64,69,70] and viral immune escape [71]. HIV-1 subtype B-infected European Caucasians initially showed associations of the relatively common HLA-B*08 or its haplotype A*01-B*08 (A1–B8 by serology) with rapid CD4 T-cell loss and disease progression [1,2], but those relationships have not been reproduced in later, larger studies [63,65]. Recent work has explored the hypothesis that clusters of HLA class I molecules with peptide-binding pocket homology should confer a similar functional advantage or disadvantage in controlling HIV-1 infection. For example, HLA allelic products corresponding to B*08, B*35, B*53, B*55, and B*56 associated with unfavorable outcomes can all be grouped into the B7 supertype based on their shared peptide binding motifs [72,73]. However, other HLA-B alleles (e.g. B*07) within the B7 supertype have not shown clear relationships. A further suggestion that HLA alleles classified into nine HLA-A and -B supertypes may account more fully for the effects attributed to individual component alleles, including HLA-B*57 and B*58 in the B58 supertype [74,75], will require evaluation in other populations. By contrast, new data on high-risk seronegative Caucasian men who have sex with men and seroconverters [59,76] support another proposition [44] that a subset of B*35 alleles with more specific antigen-presenting characteristics (i.e. preferred binding of epitopes with a non-tyrosine residue at the P9 anchoring position) are more unfavorable for both disease progression and an excess risk of infection. Additional studies have not supported the corollary that B*53 and other alleles closely related to B*35 can operate in the same way [18,70,77]. Apart from a few B*35 alleles [59], no other class I markers with clear pathogenetic effects in infected and untreated individuals have shown comparable effects in those amply exposed but uninfected. The most striking example of disparity in effects between infection and disease is B*57. Recent work has highlighted its dominant control of viremia exerted through cytotoxic T-lymphocyte (CTL) responses mediated by this allele early in the disease process [78,79], while also confirming its ineffectiveness at preventing infection [18,54]. Appreciation that the HLA class I system critically modulates the effectiveness of the CTL response both to natural virus, as summarized above, and to candidate constructs for T-cell-mediated vaccines [80] has been enhanced by other molecular and immunological investigations with broad public health implications. In populations of diverse ethnic ancestry, efforts to map conserved and evolving HIV-1 epitopes commonly recognized by prevalent HLA class I alleles have revealed that certain subregions of HIV-1 proteins are particularly promising targets for incorporation in a vaccine [81,82] because of rare escape mutation and promiscuous binding to multiple class I alleles [83,84]. One new by-product of the exhaustive work on epitope mapping is a model documenting viral evolution under HLA-mediated pressure [85]: evidence that regions of HIV-1 sequences in viral isolates from an individual more readily mutate at the epitope sites specifically targeted by class I alleles to which the virus has been exposed [85,86]. HLA-B and -C molecules not only mediate antigen-presenting function but also interact with the highly variable inhibiting and activating products of genes in the killer immunoglobulin-like receptor (KIR) system. In particular, a Bw4 motif (aa79–84 in the α3 domain) commonly displayed by certain HLA-B alleles is known to stimulate natural killer function. The carriage of alleles with the Bw4 motif appeared to account partly for a favorable course of infection [48]. A biologically persuasive explanation for such an advantage has been the interaction between certain Bw4 motif-bearing alleles and one activating natural killer receptor encoded by the KIR3DS1 gene [51]. The near absence of that KIR locus in African populations probably accounts for their failure to show similar Bw4 protection [87]. Studies of new populations will undoubtedly help define the relative importance of KIR and other genes that govern natural killer activity. Chemokine receptor and ligand genes A convincing indication of involvement of chemokine receptors in HIV-1 infection was the discovery that homozygosity for CCR5-Δ32, a null allele resulting from a 32 base pair deletion in the open reading frame of CCR5, conferred nearly complete resistance to infection in heavily exposed seronegative Caucasians [88–91]. Additional polymorphisms in CCR2 and the promoter region of CCR5 also appear to modulate HIV-1 RNA levels early in infection as well as the ultimate course of disease [43,92–95]. A new analysis of time-specific associations of both CCR5-Δ32 and CCR2-64I in cohorts with well-documented effects has suggested that the former protects more uniformly throughout the course of infection, whereas the latter may exert much of its protective effect during early infection [96]. For both the acquisition and progression of infection, comprehensive work on haplotypic relationships of polymorphisms in the CCR2-CCR5 region suggests that HHG*2 (the haplotype carrying CCR5-Δ32) and HHF*2 (the haplotype carrying the CCR2 allele for 64I) are not the only contributory factors [43,92, 97,98]. The HHE/HHE genotype (Fig. 1) appeared to account for the unfavorable promoter effect described earlier [99], and other genotypes containing HHE (e.g. HHC/HHE, the most frequent in north American Caucasians) may also show outcome or race-specific associations with the rapid progression of infection [43,92]. The correlation of CCR5 promoter variants with receptor expression differences is not understood well enough to explain the epidemiological observations [98]. Some encouragement has come from an in-vitro system in which the ease of the CCR5-dependent infection of Langerhans cells differed in a pattern often consistent with the differential population effects of the CCR5 promoter and coding sequence polymorphism [100]. Further advances in elucidating the functional significance of these genetic variants will inevitably accompany the ongoing development of inhibitors of fusion between HIV-1 and its co-receptors [101]. As for the chemokine ligands, elevated levels of several β chemokines (e.g. regulated upon activation, normal T cell expressed/secreted; RANTES) [102] are also known to protect against infection [103,104]. Genotyping data for CCL5 (RANTES) and CCL3 (macrophage-inflammatory protein 1α) have suggested possible roles for different single nucleotide polymorphisms (SNP), either as simple allelic forms or as specific haplotypes, but these relationships are far less clear than those observed with receptor gene variants [49,105–107]. Cytokine genes Preliminary analyses of cytokine gene variants, mostly SNP, have revealed modest and mostly unconfirmed relationships with HIV/AIDS. In north American Caucasians and African-Americans, the IL10 promoter −592A (5'A) allele has been associated with adverse outcomes after infection [47]. The IL4 promoter variant −589T, initially found in conjunction with elevated serum IgE levels and increased conversion to syncytium-inducing viral forms in seropositive Japanese individuals [46], was subsequently associated with more favorable responses (lower viral load and slower rate of CD4 T-cell depletion) in HIV-infected European Caucasians [52]. More recent studies in Caucasians have not yet resolved these inconsistencies [58,60]. Pharmacogenetics: advances and limitations In the earliest studies, the protective effect of CCR5-Δ32 (HHG*2) in the context of potent antiretroviral therapy (ART) implied that CCR5 genotyping has the potential to inform clinical decision-making, particularly for patients with atypical responses to therapeutic agents [108–111]. However, success at teasing apart immunogenetic and treatment effects has not been uniform [55,112], and is unlikely to come easily as the wide spectrum of available therapeutic agents, the timing of intervention, and adherence to effective regimens can diminish the discriminatory power in studies of typical design. The likelihood of uncovering pharmacogenetic determinants of atypical or adverse effects of ART has been exemplified by the strong and consistent association (odds ratio > 10) of HLA-B*57 and its extended haplotype with potentially life-threatening hypersensitivity to abacavir [13,14]. This striking relationship also reflects the importance of population-specific effects: retrospectively analysed data from treated Australians suggested that excluding B*57-positive HIV patients from treatment with abacavir would have reduced the overall frequency of hypersensitivity by threefold [13], but in north Americans, B*57 (or its haplotype) was a much less specific predictor of the abacavir reaction [14]. Moreover, individuals of African ancestry rarely carry the implicated haplotype [Cw*06-B*5701-TNF (−238A)-DRB1*0701-DQB1*0303] (J. Tang and R.A. Kaslow, unpublished); instead they typically carry the B*5703 allele in haplotypes such as Cw*07-B*5703 and Cw*18-B*5703-DRB1*1303-DQB1*0301 [18,113, 114]. In the MDR1 gene encoding the P-glycoprotein, an ABC transporter commonly associated with multidrug resistance, relationships between isolated SNP variants and drug uptake/metabolism may prove useful in guiding clinical trial design and even individual therapeutic decisions [115]. Attempts have usually been made to correlate MDR1 genotypes defined by a single synonymous SNP (3435C > T) with serum P-glycoprotein protein levels [116–119], the uptake of efavirenz, nelfinavir, and other drugs [12,120–123], and degrees of immune reconstitution after ART [12]. However, in both immunogenetic and pharmacogenetic studies, findings based on a single SNP are often difficult to interpret. For example, genotypes defined by the MDR1 3435C > T SNP are highly differentially distributed among various ethnic groups [124,125], and relationships seen in Caucasians are not immediately applicable to other ethnic groups. In addition, an increasing number of MDR1 SNPs and their haplotypes [126,127] can be involved in regulating the function of MDR1 [119,128–130]. MDR1 expression is also regulated by epigenetic factors (e.g. CpG methylation within MDR1 promoter sequences) [131–135] as well as cytokines [136], whose production is, in turn, governed by their own SNPs and haplotypes [137,138]. Finally, as with the multiplicity of immunogenetic effects, polymorphisms in genes encoding other drug transporters such as cytochrome P450, CYP3A4, CYP3A5, CYP2D6, CYP2C19, and multidrug resistance-related proteins (MRP1 and MRP2) [139,140] will also require formal evaluation, either alone or in combination with MDR1 SNPs and haplotypes [12,122]. Host genetics in the era of highly active antiretroviral therapy: opportunities, challenges, and implications ART and HAART have been successful in reducing the rates of vertical HIV-1 transmission as well as HIV-1-related morbidity and mortality [141–144]. In particular, the mother–child transmission of HIV-1 can be reduced by more than 50% by zidovudine or nevirapine alone [145,146], whereas complete viral suppression and rapid immune recovery are typically achievable after HAART [147,148]. However, HAART has not been readily accessible in regions with the most severe social and public health problems arising from HIV/AIDS [149–151]. For the foreseeable future, the extension of immunogenetic studies to non-European populations [18,94,152,153], distinct in ancestry or HIV-1 subtype exposure, is unlikely to be complicated by HAART. Studies of patients with drug-resistant HIV-1, atypical responses to HAART, or immune restoration diseases after HAART may prove equally informative [53,154–156]. Genome-wide studies Expensive and life-long HAART is clearly not the best solution to the global control of HIV/AIDS, so continuing immunological investigation coupled with a genome-wide search for immunogenetic and pharmacogenetic determinants should complement the search for alternative interventions, including vaccination. The number of informative SNPs required for a genome-wide survey ranges from 180 000 to 600 000 [157–159] depending on the ethnic background [160]. Research on that scale has been done for several genetically related diseases [161–167]. New information on patterns of linkage disequilibrium (the non-random co-existence of multiple variants on the same chromosome) in representative populations [168] should further reduce the degree of uncertainty about haplotypic or interactive effects. Meanwhile, the spectrum of HIV-related outcomes implies that productive genome-wide screening may need to target multiple conditions or milestones in the course of HIV infection. Refinements and confirmation The appropriate replication and refinement of preliminary immunogenetic and pharmacogenetic markers is essential for several reasons. First, because individual studies are rarely uniform in design, sample size (power), outcome measure, genotyping strategy, and statistical approach, inaccuracies of one sort or another are highly likely. Second, the viruses circulating in infected individuals and populations are evolving sufficiently rapidly [85,169–171] that an analysis of genetic effects could often be obscured by the expanded viral usage of alternative co-receptors [172–175], by immune escape mutants [71,176], and by the spread of drug-resistant strains. Third, univariate analyses of individual genetic variations at any level (allele, haplotype, lineage, supertype, allele and haplotype pairing, or zygosity) can often be enhanced by multivariable models capable of defining the relative strength and independence of associations [18,43,51,177], as exemplified in a recent study that confirmed the strong association of B22 with the rapid onset of AIDS in three cohorts of HIV-1-infected Caucasian men [178]. Fourth, as documented in the analyses of CCR5 [179], RANTES (CCL5) [47], SDF1 [180], IL4 [60], and IL10 [181], the diversity and complexity of haplotypes cannot be fully appreciated without systematic studies in different populations. Genetic profile As immunogenetic and pharmacogenetic markers are identified and confirmed, efforts to evaluate their joint impact (independent, additive, synergistic, or antagonistic) are increasingly critical. In an analysis of five HLA class I variants on chromosome 6 and two CCR2-CCR5 variants on chromosome 3, more than 50% of African-Americans had at least one of the seven and 5% carried at least one HLA plus one CCR marker. Corresponding frequencies in Caucasians would be higher as a result of the increased presence of HLA-B*27 [114] and CCR5-Δ32 (HHG*2) [88,182]. Compared with individual markers, genetic scores based on simple algorithms showed a stronger correlation with levels of disease control in HIV-1-infected Africans as well as Caucasians [54,66,183]. Alternative scores based on weighted relative hazards also worked well [184]. A demonstration that a profile of host genetic variation carries could genotyping data to the clinical but the of host genetic information in other confirmed (e.g. viral load and CD4 T cell to be In two of have important and provided clear evidence that host genetic variation modulates the acquisition and course of HIV-1 infection as well as variable responses to The functional for the established relationships involving a or so genetic markers (Fig. 1) at HLA, chemokine and other loci are being molecular is reducing to and in the whereas a growing of the specificity and complexity of genetic effects at the population level may into study and data In interactions between and are to immunogenetic and pharmacogenetic information on HIV/AIDS. The would to for the to this would also to and two for and the critical reading of an earlier for this and the own work has come from several including and from the of