Sex Differences in HIV Infection

The risk of HIV seroconversion per heterosexual act is estimated to be approximately twofold higher for the female compared to male partner [1], with multiple contributing factors. The unique characteristics of the female genital tract as compared with rectal and penile mucosal surfaces confer differences in transmission risk. Inflammation at the cervicovaginal mucosa lowers the barrier to HIV infection [2‐5], and both the vaginal microbiome itself [6] and sexually transmitted infections [7‐11] are important determinants of the levels of local inflammation. The association of depot medroxyprogesterone (DMPA) hormonal contraception with enhanced risk of infection (hazard ratio of 1.4) [12‐14] underlines the sex-specific risk associated with hormone exposure, which also impacts the vaginal microbiome. Clearly, these factors have distinct manifestations in the male and female genital tracts and these basic differences have important implications for prevention interventions discussed below.

Sex differences in both adverse effects and the efficacy of protective responses to vaccination are well described [15]. These differences are of clinical significance as seen in the higher rates of vaccine-associated severe viscerotropic yellow fever disease in women [16, 17] and the HSV glycoprotein vaccine that was protective only in women [18]. The mechanisms driving these differences are not totally clear; no specific immunologic correlate was reported for the sex differences in the HSV vaccine trial [18] although subsequent work suggested that specific epitopes may be preferentially recognized in women [19]. Systems biology analysis of gene expression profiles after yellow fever vaccine identified sex-specific programs of gene induction [20], highlighting the potential for studies of sex differences to identify correlates of successful protection. In HIV vaccine trials, there has not been clear evidence of sex differential effects. In the RV144 study, protective efficacy was estimated 25.8% in men (n = 4875) and 38.6% in women (n = 3085), with no statistical difference associated with sex [21]. In terms of immune correlates of protection, differences in humoral and cell-mediated immune responses have been seen in multiple vaccines [20]. Mechanistically, there is evidence for more potent induction of inflammatory pathways in cytotoxic T cells from women [22]; sex comparison of the magnitude and breadth of T cell responses induced by vaccines would be of interest. Likewise, there is data to suggest that somatic hypermutation is enhanced by estrogen [23] and that antibody glycosylation patterns are influenced by sex [24] suggesting that biological sex may influence both antibody affinity and non-neutralizing functions.

Moving forward, sex-specific analyses of both efficacy and immune correlates of protection should be leveraged to enhance responses. For example, sex-specific induction of type 1 interferons or the inflammasome might indicate a role for specific adjuvanting strategies in men versus women. Given the challenges of vaccine development, all avenues for optimization bear consideration.

In the absence of an effective vaccine, pharmacologic strategies have become a critical adjunct to the prevention of transmission. Notably, despite initial studies showing high levels of efficacy for PrEP in men who have sex with men [25] and in serodiscordant couples [26], studies of PrEP exclusively in women showed no efficacy, results that were attributed to very low adherence to study drug [27, 28]. Clinical pharmacology studies have highlighted differences in drug concentration at the rectal mucosal and cervicovaginal tissues [29] that may obligate different levels of adherence in women versus men to maximize effectiveness. To circumvent this, topical delivery designed for the vaginal microenvironment is another potential route to modulate risk of infection in women; the CAPRISA 004 study reported a 39% risk reduction with tenofovir gel [30], although the VOICE study, which was limited by low adherence, did not show efficacy in the vaginal gel arm [27]. The topical approach using a vaginal ring preparation of the novel antiretroviral dapivirine has recently demonstrated a significant but modest reduction in the risk of HIV acquisition (27–31%) [31, 32]. Importantly, recent work has shown that adherence is not the only challenge to the topical approach. Local metabolism of tenofovir itself by components of the vaginal microbiome is associated with reduced efficacy of protection [33]. As studies defining the effects of topical exposure at the rectal mucosa have suggested that tenofovir may increase certain inflammatory mediators [34], specific assessment of the in vivo cervicovaginal effects is also warranted. Further studies are necessary to define the optimal approach to risk reduction in both men and women; advantages of topical preparations must be considered in light of adherence challenges, and careful studies are necessary to fully define sex-specific modulators of efficacy at the sites of acquisition. Taken together, the data suggest that there are sex-specific features of risk perception and medication adherence, along with critical differences in pharmacologic properties and the microenvironment at sites of acquisition in men and women. Considering these differences will be critical in the design and implementation of chemoprophylaxis strategies.

Sex is a clear contributor to disease pathogenesis in multiple infectious diseases [35], and HIV follows this paradigm. Across most studies, women have lower HIV viral loads early during infection but despite this difference, disease progression is comparable between the sexes [36‐46]. Substantial differences in immune activation may underlie this apparent paradox; women have higher CD8⁺ T cell activation at a given level of HIV viremia, corresponding to activation seen in men at one log₁₀ higher viral load [47]. Similarly, the expression of interferon-stimulated genes was higher in women when controlling for HIV viral load [48]. Given the role of immune activation in driving HIV disease progression [49, 50] and in comorbidities that emerge during effective ART [51, 52], the sex difference in immune setpoint likely has clinical consequences.

In selected individuals, HIV disease progression is attenuated, with either spontaneous control of viral replication in the absence of drug therapy [53‐55] or sustained viral suppression after interruption of ART (post-treatment controllers; PTCs) [56]. The factors that allow natural control of HIV are not fully defined but include host genetics, highly efficient immune responses, and in select cases, viral fitness [53‐55]. Cohort studies have reported that women are more likely to be categorized as spontaneous controllers of HIV [57, 58] although the determinants of this advantage have not been elucidated. Similarly, women are overrepresented in cohorts of post-treatment control: women were 36% of PTCs, 43% of low viremia patients (viral load between 50 and 500), and only 14% of post-treatment non-controllers in one study [59]. Again, sex-specific mechanisms of protection have not been defined within this group, and it should be noted that the total numbers evaluated are very low. Thus, although limited by biases in case finding, women more frequently demonstrate phenotypes of viral control. This suggests that identifying sex determinants of immune response and viral setpoint may shed light onto features of a successful host response.

Consistent with sex differences in pharmacokinetics/pharmacodynamics, drug metabolism, body composition, and drug distribution, the rates of adverse reactions with the early generation of antiretroviral therapies showed sex variation [60, 61]. Efforts to analyze these differences are hampered by the limited enrollment of women in trials of new therapeutics [62]. In response to this challenge, the GRACE (Gender, Race And Clinical Experience) trial specifically enrolled women to determine the sex-specific efficacy of a darunavir-based ART regimen [63] and yielded critical insights into the barriers to participation by women (discussed further below) [64]. Recent subgroup analyses of therapeutic trials have largely demonstrated similar efficacy in men and women, consistent with the improved therapeutic index of modern antiretrovirals [65‐67]. However, unanticipated effects of antiretrovirals, such as the recently reported weight gain associated with integrase inhibitor regimens in a predominantly male cohort (14% women in integrase inhibitor subgroup) [68], should be carefully evaluated for sex effects. In addition, the response to treatment as measured by CD4⁺ T cell recovery has been reported to favor women, although with unclear implications for immune competence [69]. Complications of immune reconstitution such as the immune reconstitution inflammatory syndrome (IRIS) have not been reported to have a particular sex predilection. However, this is difficult to clearly establish given the heterogeneity in case definitions of IRIS, bias for women to be enrolled in resource-limited settings, and lack of disaggregation of data by sex in some larger studies.

Treatment-induced changes in biomarkers of inflammation also show discordance; in one cohort, women had higher baseline high-sensitivity C reactive protein (hsCRP) levels and less change with therapy, along with higher levels of soluble CD163, a marker of monocyte activation [70]. Other cohorts have reported similar differences in response to treatment, although inconsistent differences in baseline levels [71]. Further work will be necessary to dissect the direct contribution of HIV and ART as compared with concurrent inflammatory stimulators such as coinfections and smoking, and modulators such as sex hormones given the potential for direct effects of estrogen on some markers such as CRP [72]. Overall, women and men can both achieve viral suppression with ART but differences in residual immune activation and reconstitution may remain, with consequences for comorbid conditions.

With the advent of effective ART, morbidity and mortality among people living with HIV has shifted to non-AIDS events including cardiovascular disease, cancer, and neurocognitive dysfunction, many of which are driven by inflammatory consequences of HIV infection. Biological sex is one contributor to the multifactorial determinants of these comorbidities [52]. The excess risk of cardiovascular events in people living with HIV [73] is amplified in women [74] and linked to higher levels of circulating markers of monocyte activation [75]. Likewise, the increased risk of cerebrovascular events in HIV-infected individuals [76, 77] is exaggerated in women [78]. Of note, the epidemiology of these comorbid conditions varies significantly across different social and geographic contexts obligating thoughtful design of trials to assess for the contribution of sex [79]. The differences in risk profile between men and women highlight the potential for studies of sex differences to identify causal pathways or biomarkers of disease pathogenesis.

As noted above, there is an association with DMPA contraceptive use and enhanced rates of infection. The precise mechanisms are unclear, as the progestin-associated thinning of the vaginal mucosal observed in non-human primate models [95‐97] has not been seen in women [98‐102]. Recent data identified endogenous and exogenously administered progesterone-induced variations in the frequency of cervical HIV-susceptible target cells [103]. There are additional intersections between sex hormone levels and inflammation induced by microbiome composition and concurrent infections [104]. Given the global need for effective family planning methods and widespread use of hormonal contraception, determining the mechanisms of hormonal contribution to risk of infection and potential pathways for modification is of critical importance.

Outside of acquisition, estrogen is also a direct modifier of HIV transcription. Previous work has demonstrated that the estrogen receptor can be indirectly recruited to the HIV-1 long terminal repeat (LTR) and act to repress transcriptional activity [105]. More recently, using an unbiased small hairpin RNA screening strategy, the estrogen receptor was identified as a potent inhibitor of HIV transcription in latency models and primary cells (Karn et al., IAS, 2015; Das et al., submitted). Ex vivo studies using primary cells from both men and women confirmed that estrogen is repressive to latency reversal, and that blockade of the estrogen receptor can enhance reactivation (Karn et al., IAS 2015; Das et al., submitted).

Sex hormones have also been reported to have a variety of direct effects on immune cell function. Both estrogen and progesterone have been reported to modulate plasmacytoid dendritic cell IFNα secretion [47, 106‐109]. Cytotoxic T cells from women have higher expression of inflammatory/cytotoxic pathways after ex vivo restimulation, and multiple genes in these pathways have estrogen responsive elements in their promoters [22]. Of note, the presence of estrogenic compounds in standard cell culture media components [110, 111] and the difficulty in replicating the in vivo balance of hormones with in vitro studies obligates careful interpretation of these studies. However, hormonal pathways can be safely modulated in vivo and offer a potential adjunctive therapeutic pathway that may be of use in studies of HIV cure.

Sex-specificity of the microbiome composition in the genital tracts is one determinant of the local immune environment. Further, recent work identified novel features of this relationship, with specific microbiome components associated with alterations in wound healing [112] and direct microbial metabolism of tenofovir associated with reduced efficacy of PrEP in the female genital tract [33]. Aside from this direct role, animal studies have demonstrated that sex hormones impact microbiome composition in the gut, with implications for sex-specific susceptibility to autoimmunity [113, 114]. Studies have confirmed sex variation in gut microbial contents in humans [115‐117] and further work will be necessary to determine if these differences have consequences for inflammation in HIV disease. Interventions to reshape the microbiome (e.g., with probiotics) may offer therapeutic benefits.

The sex-specific chromosomal complement is an additional pathway to biological differences. The X chromosome carries critical immune genes including TLR7, which encodes a pathogen sensor, FOXP3, a transcription factor critical for regulatory immune responses, and 10% of all microRNAs which have pleiotropic regulatory roles [118].

As some sex differences including lower viral loads in females are present prior to the onset of puberty, non-hormonal mechanisms including genetics are likely to play a role [119]. Gene dosage effects are attenuated by X chromosome inactivation, but the enhanced risk of female predominant diseases such as systemic lupus erythematosus in phenotypic males with XXY karyotype suggests that this is incomplete [120]. Growing evidence demonstrates that up to 20% of X chromosome genes escape inactivation [121]; this has clinical implications, with recent work suggesting that these genes may determine a portion of sex-specific susceptibility to cancer [122]. The role of sex chromosome-encoded genes in differential vaccine responses, HIV pathogenesis, and cure efforts is undefined; it is notable that the HIV controllers genome-wide association study to assess for genetic determinants of spontaneous control was restricted to autosomes [123]. Studies to identify polymorphisms in sex chromosomal genes should be pursued.

Of note, research has also demonstrated sex-specific transcriptional programs related to both chromosomal determinants and ongoing hormonal programming [124]. Analysis of the methylation patterns and transcriptome of immune cell subsets identifies differences between men and women, supporting a potential role for epigenetic regulation in sex differences in immune responses [125]. Given the potential use of epigenetic modifiers in latency reversal, sex-specific patterns of epigenetic regulation should be explored.

The combined effects of sex hormones, microbiome, and chromosomal complement contribute to distinct immune profiles. Preliminary work suggests that the relationship between residual virus activity and T cell activation and exhaustion phenotypes may be different between men and women, with men showing more activation and exhaustion and more correlations with measures of viral reservoir (Scully et al., Abstract 281,CROI 2017). Previous work has also demonstrated sex differences, partially mediated by estrogen, in antibody features including subclass, levels of hypermutation, and Fc glycan modifications [23, 24]. Sex-stratified comparisons of the humoral responses to vaccines may provide insight into the critical features of a successful response.

Also notable is the role of sex hormones in lipid metabolism that is in turn linked to innate cellular activation and risk of non-AIDS morbidity and mortality in HIV infection [126, 127]. Of note, recent data suggests that there may be sex-specific responses to lipid-lowering therapy, with women showing qualitatively greater reductions in sCD163 after treatment with pitavastatin [128]. In studies of soluble markers of inflammation, sex differences in baseline levels and in the changes after ART initiation have been reported; neopterin (marker of cellular activation associated with HIV-related neurocognitive disease) was higher in women with impaired cognition, a finding not observed in men alone, and TNF-RII was similarly elevated in cognitively impaired women but not in men [129]. In a heterogenous cohort of men and women from multiple global sites, women were reported to have lower baseline levels of CRP, lipopolysaccharide, and soluble CD14 (sCD14) but less decrease in CRP and sCD14 and more increase in TNFα after ART [71]. In contrast, in a more homogenous cohort comparison, women had lower CRP than men did at baseline but again showed limited change after initiation of ART [70]. In total, the data are far from definitive and the multiple determinants of inflammatory outcomes including coinfections, microbiome differences, sex hormones, and immune setpoints will need to be carefully parsed to guide interventions. What is clear is that sex is a modifier of immune responses and may also dictate which biomarkers are predictive of risk for a particular population.