Non-Human Primate Models of Hormonal Contraception and HIV

Authors


  • The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the Centers for Disease Control and Prevention (CDC).

  • A part of this work was previously presented at the Symposium on Hormone Regulation of the Mucosal Environment in the Reproductive Tract and the Prevention of HIV infection, May 2013, Boston, USA, a satellite symposium of the International Society for Immunology of Reproduction/American Society for Reproductive Immunology meeting, June 2013.

Abstract

Problem

Recent concerns that hormonal contraception (HC) may increase risk of HIV acquisition has led to keen interest in using non-human primates (NHP) to understand the underlying mechanism and the magnitude of the risk. This is, in part, because some experiments which would be difficult or logistically impossible in women are more easily conducted in NHP.

Method of study

NHP models of HIV can inform HIV acquisition and pathogenesis research and identify and evaluate biomedical preventions and treatments for HIV/AIDS. Widely used species include rhesus, pigtail, and cynomolgous macaques.

Results

This paper reviews past, current and proposed NHP research around the intersection of HIV and HC.

Conclusion

NHP research may lead to the identification of hormonally regulated biomarkers that correlate with HIV-acquisition risk, to a ranking of existing or next-generation HC along an HIV-acquisition risk profile, and inform research around new biomedical preventions for HIV.

Introduction

This article reviews macaque models used for studies of HIV acquisition in relation to the effect of reproductive hormones, whether endogenous or exogenous (reviewed in ref. [7]).[1-6] Exogenous reproductive hormones are used in some macaque models to improve the reproducibility of vaginal infection with either simian immunodeficiency viruses (SIVs), or recombinant SIVs expressing HIV genes (SHIVs) and have been also used to explore the increased risk of S(H)IV acquisition that may be associated with hormonal contraception. There is renewed interest in how macaque studies can improve our understanding of the extent and mechanisms by which reproductive hormones, particularly contraceptive hormones such as Depo-Provera® (a widely used depot form of medroxyprogesterone acetate, DMPA), enhance HIV acquisition. This interest has been driven by recent observations from several clinical trials where secondary analyses suggested that Depo-Provera® might enhance risk of HIV acquisition approximately twofold and from older human studies that provided conflicting evidence both for and against a Depo-Provera-associated risk (reviewed in ref. [8]). To definitively answer this important public health question, it would be necessary to conduct a very large, randomized controlled clinical trial with many women enrolled into multiple study arms including Depo-Provera, possibly non-Depo-Provera hormonal contraceptive, and non-hormonal contraceptive arms. The trial would likely take many years, being very expensive and difficult to conduct.[9] During discussions of the pros and cons of this type of trial, scientists debated whether macaque models of hormonal contraception and HIV could provide insight. Questions arose as to how well rhesus macaques reflect human vaginal physiology, and whether pigtail macaques may be a more suitable non-human primate model. While macaque models of vaginal HIV acquisition may not perfectly replicate all aspects of how women acquire HIV, such models have proved invaluable in translational research that led to new biomedical preventions for HIV. As an example, pre-exposure prophylaxis (PrEP) with Truvada, a combination of the antiretroviral drugs tenofovir and emtricitabine, is now FDA approved for women (and men) to prevent acquisition of HIV.[10, 11] Macaque data showing that PrEP could prevent SHIV infection were critical in guiding clinical PrEP trials[12-16] (reviewed in ref. [17]) demonstrating that oral PrEP, when used correctly, could reduce heterosexual HIV acquisition in humans by >60% (reviewed in ref. [18]). Macaque and human studies are concordantly identifying protective levels of PrEP drugs needed at mucosal sites (reviewed in ref. [21]).[12, 19, 20] Previous macaque studies have shown that Depo-Provera at high doses enhances SIV acquisition in rhesus macaques[2] by approximately eightfold, a finding consistent with some of the human data, although greater in magnitude. These preclinical studies indicate that macaque models of HIV and hormonal contraception are of public health relevance. Macaque models should primarily be used for targeted studies to answer questions or obtain data that may be difficult or impossible to obtain in the human setting. Well-designed macaque models could answer some important questions around the intersection of hormonal contraception and HIV. Can hormonal contraceptives be ranked in terms of relative risk for HIV acquisition? Which biomarkers correlate with acquisition risk and are directly modulated by reproductive hormones? To understand the importance of macaques in answering these questions, this review will discuss macaque studies that have been informative for understanding hormonally regulated events around HIV acquisition.

Types of Non-Human Primate Models for Studying How Hormonal Contraception may Influence HIV Acquisition

Two main macaque species have been used for studies of reproductive hormones and HIV, the pigtail macaque (PTM, Macaca nemestrina) and Asian rhesus macaques (RM, Macaca mulatta) of Chinese or Indian origin. The cynomolgous macaques (CM, Macaca fascicularis) is another macaque species used for studies of biomedical preventions for HIV, sometimes in the presence of DMPA.[22-25] CM will not be extensively discussed in this review because, to our knowledge, they have not been used for studies of how reproductive hormones influence acquisition of S(H)IV. Their smaller body weight and vaginal cavity also make them somewhat less suitable for intensive reproductive studies.[26] The baboon, a larger non-human primate (NHP) with a uterus more similar in anatomy to humans, is also not discussed, as while useful for female reproductive studies,[27] this species is not widely used to model HIV susceptibility. Table 1 compares several reproductive factors between RM and PTM.[6, 25, 28-57] First, PTM cycle throughout the year. RM cycle mainly in the fall and winter, although in captivity, they may cycle throughout the year[30, 49, 50] (M. Lewis, personal communication). The menstrual cycle is generally slightly longer in PTM than in RM and humans.[6, 30, 49] Vaginal microflora in RM and PTM contain some of the same bacteria seen in humans, but tend to be more similar to those seen in women with bacterial vaginosis, in that there is a reduction in or absence of Lactobacillus species.[43, 51-53, 58] The vaginal pH of RM and PTM also tends to be more basic (pH 6–8) than the human vaginal pH, which is usually around pH 4,[33, 43, 44, 52, 53, 59] and these differences should be taken into consideration in model selection.

Table 1. Comparison of Female Pigtail and Rhesus Macaques for Variables Relevant to Vaginal SIV/SHIV Susceptibility
Feature/VariableMacaca nemestrina (pigtail macaque)Macaca mulatta (rhesus macaque)References
  1. BV, bacterial vaginosis; CT, Chlamydia trachomatis; HSV, Herpes simplex virus; RLD, repeat low-dose; SIV, simian immunodeficiency virus; SHIV, simian–human immunodeficiency virus; STI, sexually transmitted infection; TV, Trichomonas vaginalis.

  2. a

    Lewis M (personal communication).

Cycling throughout the yearYesSeasonal breeders but if experimentally housed may cycle throughout yeara [30, 49, 50]
Length of menstrual cycle (median, days)3228 [6, 30, 49]
Comparison of vaginal microflora to humanBV-likeBV-like [43, 51-53]
Vaginal pH6–87 [33, 44, 52, 53]
Comparison of histology of vaginal tissues to humansSimilar to humansMore keratinized [54, 56]
Thickness variation in vaginal epithelium during menstrual cycle++ [33, 45, 56, 63]
Thickness variation in vaginal epithelium post-hormonal treatment++ [2, 38, 47, 48, 54]
Data on DMPA metabolism++ [38, 47, 48]
Menstrual cycle synchronization++a [40]
Immune, proteomic, genomic reagents available++ http://dpcpsi.nih.gov/orip/cm/primate_resources_researchers
Characterized viral stocks (SIV, SHIV)++ www.aidsreagent.org
RLD virus infection model+++ [37, 41, 54, 55]
Vaginal infection with human STI+ (CT, TV)+ (HSV-2) [32, 35, 36, 42]

In humans and macaques, HIV/S(H)IV may enter through reproductive tissues in the vagina, cervix, and possibly endometrium or Fallopian tubes.[60-62] Many studies have focused on the histology of the vaginal wall in PTM and RM.[2, 33, 38, 45, 47, 48, 56, 63] RM vaginal tissues are reportedly more keratinized than those of PTM,[38, 45, 64] and the degree of keratinization varies with phase of menstrual cycle.[45, 56, 63] A common feature of humans, RM, and PTM is that all exhibit thinning of vaginal epithelium during the luteal phase, although the extent may vary among species (Fig. 1).[33, 56, 65, 66]

Figure 1.

Menstrual cycle variation in pigtail macaque vaginal tissue. Vaginal biopsies from pigtail macaques in follicular (left) and luteal (right) phases of the menstrual cycle, stained with hematoxylin and eosin and shown at 10× magnification. 4.2-mm2 punch biopsies were obtained as previously described,[90] and phase of cycle was determined by monitoring progesterone levels as described.[6] The vaginal epithelium (from basal lamina to surface) is thickest during the follicular phase. In the high-progesterone luteal phase, the keratinized (sparsely nucleated) and non-keratinized (nucleated) layers of the epithelium are reduced in thickness, and the keratinized layer may be absent altogether. This may be due to dissociation, as we have occasionally observed partial dissociation of this layer, as seen here (*) and in other biopsies from the luteal phase or other high-progesterone states.

Exogenous reproductive hormones also influence vaginal thickness. In women, Depo-Provera induces vaginal thinning similar to the luteal phase.[65] In RM, progesterone implants, Depo-Provera, and Norplant-II thin the vaginal epithelium,[2, 38, 54] while estrogens have an opposite, or keratinizing effect.[3, 4] DMPA also induces vaginal thinning in PTM, and ongoing studies suggest this is dose dependent[48] (E. Kersh, unpublished observations).

There are significant challenges in determining the human-equivalent dose for DMPA or other reproductive hormones for NHP because the pharmacokinetics and pharmacodynamics of a given hormone may vary between human and macaque. For example, the half-life of MPA is shorter in RM than in humans.[38, 65, 67] In humans, the gold standard for HC efficacy is prevention of pregnancy. Traditional HC prevent ovulation with suppression of progesterone and estradiol as biomarkers of such effect. In non-human primates, progesterone and estradiol are relatively easily measured and suppression of their levels by progestins has been accomplished in many studies modeling HIV in RM.[2, 38, 68] There is debate as to whether the 30 mg dose frequently used in RM studies is too high and induces excessive vaginal thinning. Some researchers find a 5 mg dose suffices to suppress endogenous progesterone and estradiol (Lewis M, personal communication). In PTM, we have shown that a desogestrel and ethinyl estradiol combined oral contraceptive (administered orally) suppress progesterone and estradiol.[40] MPA, given by injection, also suppresses progesterone and estradiol in PTM[47, 48] (and E. Kersh, unpublished observations). Finding the human-equivalent dose of exogenous hormones requires extensive pharmacokinetic and pharmacodynamic studies over several months, using a combination of measures including levels of progesterone and estradiol and vaginal thinning. These may not be the only relevant biomarkers, particularly for studies of S(HIV) susceptibility (see below) and studies of mucous, vaginal and cervical fluid components, cervical structure (e.g., ectopy), endometrial properties as well as immune and innate factors such as dendritic cells, lymphocytes, cytokines, and TLR should be included. Table 1 indicates other features of RM and PTM macaque models that may be relevant in studies of HC and HIV susceptibility. Menstrual cycle synchronization of a group of macaques, often used to facilitate breeding in agricultural animals, can be induced using short-term hormonal contraception, as we have shown in PTM[40] and can be carried out in RM (M. Lewis, personal communication). Animals then continue on synchronized cycles for months in the absence of any exogenous reproductive hormone. This may facilitate logistical aspects of S(H)IV challenge studies, as was carried out in microbicide studies in cynomolgous macaques.[25]

Reagents and tools to evaluate how reproductive hormones influence NHP host factors that may be relevant for HIV acquisition are available for immune, genetic, genomic, and proteomic approaches (information can be found at NIH primate resources Web site, http://dpcpsi.nih.gov/orip/cm/primate_resources_researchers). Systems biology strategies may identify hormonally regulated factors that influence early events of HIV infection at mucosal sites. Using such an approach, we have found that certain innate proteins associated with acute-phase immune responses may be upregulated in PTM during the follicular phase, while others, which could play an HIV-enhancing role, are upregulated in the luteal phase.[69]

Well-characterized virus stocks of SIV and SHIV are available for studies of HIV susceptibility in RM and PTM, many of which are available through US NIH AIDS Reagent Program (www.aidsreagent.org). The choice of virus is critical for the study in question, as many of the viruses vary in their co-receptor tropism, replication kinetics, and ability to be neutralized by antibodies or to escape T-cell immunity. Literature reviews of differences in properties of different SIV strains and SHIVs[70-72]can guide the selection of virus stock best suited for the species or purpose, as well as need for analysis of major histocompatibility complex (MHC) TRIM5 or other host genes. Certain virus–host combinations may require exclusion or randomization of macaques with genotypes that influence virus replication, and much data in this regard are available in RM. For example in RM, Mamu*A01 is associated with better control of SIVmac251 and TRIM5 CypA restricts replication of SIVsmE660 (reviewed in ref. [71, 74, 75]).[73] In the last decade, studies of vaginal exposures to S(HIV) in macaques have expanded from single high-dose exposures to include repeated exposures to small, human-equivalent doses of S(H)IV, providing a more realistic model and adding statistical power with small group sizes per study arm because each exposure is considered independent. This repeated low-dose (RLD) approach is widely used in both RM and PTM.[12, 14, 16, 41, 76-78] Other factors that may be important to consider in NHP reproductive studies include age, parity, whether the macaque has had caesarian section, zoonotic infections, cost, and availability. In the USA, PTM are generally more expensive and less readily available than RM and captive-bred PTM are preferred because their numbers are declining in the wild. Although there is no evidence that ketamine, a frequently used anesthesia, alters reproductive hormones, too frequent use of anesthesia can reduce overall macaque health and, for repeat procedures, some researchers are using non-anesthesia-based restraint devices.[79] PTM, compared to RM, have larger sex skin swelling, which can be used to determine cycle phase visually.[30]

To summarize, while RM are readily available and better characterized genetically, some of the advantages of PTM are that they consistently cycle throughout the year and are documented to be reproducibly infected vaginally with S(H)IV in the absence of exogenous hormone treatment.

Reproductive Hormones and HIV Acquisition in NHP Models

Table 2 summarizes literature on the impact effect of endogenous or exogenous reproductive hormones on acquisition of HIV in RM and PTM using SIV or SHIVs. Studies of natural variation in endogenous hormonal levels and risk of SIVmac251 acquisition in RM throughout the menstrual cycle showed that there appeared to be a higher risk of acquisition during the luteal phase.[5] This observation was more definitively confirmed in a PTM model of repeated low-dose exposures of SHIVSF162p3 throughout the cycle, where SHIV infection was more frequently detected in the follicular phase.[6] Due to the lapse in time between infection and detection of viral RNA, these data indicate that susceptibility to infection peaked during the late luteal phase.[6] Subsequent analyses of additional PTM receiving the same low-dose SHIVSF162p3 exposures and analyzing the cycle in four phases confirmed that susceptibility peaks during the late luteal phase and extends through the menstrual phase (Fig. 2).[1] This difference in susceptibility throughout the menstrual cycle in women had been hypothesized based on known variability in vaginal thickness and other immune factors (reviewed in ref. [80]). However, as proof of such variable susceptibility is virtually impossible to obtain in humans, these studies show the utility of macaque models in expanding our understanding of reproductive issues around HIV risk in women. Moreover, the data indicate that studies of the magnitude of risk associated with hormonal contraception in women or macaques need to account for the known baseline variability in susceptibility.

Table 2. Studies of Reproductive Hormones and Vaginal SIV/SHIV Acquisition in Pigtail and Rhesus Macaques
Hormonal source and typeMacaque SpeciesVirus (Dose)Study DesignObservationReference
  1. SIV, simian immunodeficiency virus; SHIV, simian–human immunodeficiency virus.

EndogenousMenstrual Cycle PhaseRhesusSIVmac251 (640 TCID50)Virus exposure in follicular and luteal phases3 of 6 (50%) infections in luteal phase; 1 of 7 (14%) infections in follicular phases (trend) [5]
PigtailSHIVSF162p3 (50 TCID50)Analysis of infection time points during repeated virus exposures throughout the menstrual cycleHigh susceptibility in late luteal phase; significant difference in detected infections between phases (P < 0.0001) [1, 6]
ExogenousProgesterone ImplantsRhesusSIVmac251 (640 TCID50)Virus exposures in macaques with progesterone implants or in others during the follicular phase7.7-fold difference between infections in implanted macaques (14 of 18; 78%) and those in follicular phase (1 of 10; 10%). [2]
Estrogen/Progesterone ImplantsRhesusSIVmac251 (640 TCID50)Virus exposures in ovariectomized macaques with either progesterone or estrogen implants6 of 6 (100%) infections in progesterone-implanted macaques; 0 of 6 (0%) in estrogen-implanted macaques [3]
Topical estradiolRhesusSIVmac251 (640 TCID50)Virus exposures in macaques receiving estradiol (Ovestin®) or placebo creamninefold difference between placebo-treated (6 of 8; 100%) and Ovestin-treated (1 of 12; 8.3%) macaques [4]
DMPA (Depo-Provera®)Rhesus & PigtailVariousReview of studies utilizing IM DMPA (30 mg) to facilitate vaginal infectionWidely used approach; impact on infection rates not explicitly tested [55]
Figure 2.

High susceptibility to simian–human immunodeficiency virus (SHIV) infection in the late luteal and menstrual phases. Schematic diagram of the menstrual cycle showing hormone levels (dashed and solid lines) in relation to ovulation and the periods of high susceptibility to SHIVSF162p3 infection (black triangles) based on analysis of dates of first RNA detection and phase of menstrual cycle in 43 naturally cycling pigtail macaques.[1]

The bottom part of Table 2 summarizes data on studies of several types of hormonal contraceptives studied in RM or PTM in relation to S(H)IV acquisition. A consistent finding is that progestin-containing products enhanced risk of SIV acquisition[2] (reviewed in ref. [7]) while systemic or topical estrogen products reduced risk.[3, 4] Questions raised in these studies include whether human-equivalent doses of these hormones were given and whether the magnitude of the increased risk is reflective of what would be observed in women in a clinical trial. Marx et al.[2] chose a progesterone dose (200 mg s.c.) that mimicked luteal phase and suppressed progesterone and ovulation and found a greater than sevenfold increased risk with SIVmac251 in rhesus macaques. Would such a risk have been found in a repeated low-dose exposure model with naturally cycling pigtailed macaques, and would this model provide more information about biomarkers and the magnitude of the risk? Until further information is obtained, the value of previous data in RM compared to data in PTM in informing human studies about choice of hormonal contraceptive is unknown.

The ideal hormonal contraceptive for women would be effective, have no systemic side-effects, and would not amplify HIV acquisition risk. The renewed concern that certain hormonal contraceptives may be increasing women's risk of HIV infection has spurred industry to identify a broader range of hormonal contraceptives. Unfortunately, other than vaginal epithelial thickness, there is no consensus as to which hormonally regulated biomarker would indicate ‘low risk’ for HIV infection. Even in the case of vaginal thickness, it is not known to what extent other concurrent changes in vaginal composition (cell type, receptors, cytokines) contribute to the increased risk associated with vaginal thinning (reviewed in ref. [81]). Human studies of biomarker variation throughout the menstrual cycle and with a variety of hormonal contraceptives are ongoing in several centers. These are being complemented by ex vivo studies to examine the effect of hormonal contraceptives on HIV entry or replication in tissues [Updated information on such research funded by the US NIH can be accessed through the NIH Research Portfolio Online Reporting (RePORT) tool using search terms ‘HIV and contraception’; http://report.nih.gov]. Limitations of all of these studies are that the biomarkers that correlate with risk are not known and that ex vivo and explant studies imperfectly model the early events of HIV infection. These limitations lead back to the value of macaque studies of hormonal contraception in relation to S(H)IV risk. These studies could (a) provide a ranking of hormonal contraceptive risk in relation to each other and in relation to cycle phase (see proposed relationship, Fig. 3) and (b) identify biomarkers that correlate with risk and may be true biological indicators of how a given hormonal contraceptive influences HIV risk. Such information can then be used to select contraceptives with the ideal combination of properties for women in an iterative way, where the macaque studies inform human studies and vice versa. The biomarker information could also guide safety studies in lower-order mammals or using ex vivo approaches.

Figure 3.

Reproductive hormones and states and risk of HIV infection. Schematic diagram of putative relationship between hormonal states, HIV (or SIV/SHIV) acquisition risk, progesterone, vaginal thickness, and other potential biomarkers. Hormonal states include endogenous variation in progesterone during the menstrual cycle from low (follicular phase) to somewhat higher (luteal phase) levels and states induced by exogenous hormones with medium (e.g., some oral contraceptive) or higher (e.g., depot medroxyprogesterone acetate) progestin amounts and are ranked in that order in terms of a putative correlation with increasing HIV acquisition risk. Biomarkers may correlate inversely (vaginal thickness), directly (biomarker X), or not at all (biomarker Y) with the progestin levels, and may (or may not) be in the causal pathway for how they influence HIV acquisition risk. Such a model may guide both human and non-human primate studies in terms of developing an informative matrix for evaluating existing or next-generation hormonal contraception. SIV, simian immunodeficiency viruses; SHIV, simian–human immunodeficiency virus.

Other Considerations, Challenges, and Need for Future Studies

The value of well-conducted macaque studies to inform the field of biomedical prevention is established. They have already shown that progesterone-containing hormonal contraceptives increase SIV risk. They are being used to determine whether HC negatively affects PrEP efficacy and emerging data are encouraging in this regard.[47, 82] The models could likely be used to rank-order existing or next-generation hormonal contraceptives for potential HIV risk and to identify biomarkers of risk. The models should be used for targeted research that cannot be conducted using in vitro, ex vivo, or human studies. Continued research to improve macaque models to best inform questions around HIV risk in women is ongoing. For example, most women at the highest risk for HIV infection are not only on hormonal contraception but are coinfected with one or more STI (reviewed in ref. [85]),[83, 84] and it is well established that STIs, per se, increase risk for HIV infection approximately twofold (reviewed in ref. [89]).[86-88] We have established a PTM STI model of Chlamydia trachomatis and Trichomonas vaginalis coinfection which increases SHIV infection to a similar extent.[35, 36] Crostorosa et al.[32] also describe a similar risk enhancement during herpes simplex virus type-2 (HSV-2) infection of RM. Studies of hormonal contraception and HIV risk could potentially be performed in an STI coinfection model. Challenges include study design and statistical approaches that account for within-cycle variation in susceptibility and are powered to detect two- to fivefold increased risk. Other variables that could be included in NHP modeling of hormonal effects on HIV acquisition include semen, cell-associated virus, and effects of trauma.

Acknowledgments

We thank David Garber and Mark Lewis for assistance with Table 2 and for helpful comments on the manuscript. Prachi Sharma and Ai Tsuki helped with vaginal histology. Some of the work cited in this article was funded by CDC and by an Interagency Agreement AAI 12041 between CDC and NIH/NIAID.

Ancillary