Pediatric NIH Workshop: Session I: Nonhuman Primate Models in HIV Prevention Research A Historical and Comparative Perspective

Abstract

During the past 4 decades, a growing number of nonhuman primate (NHP) models of HIV infection have been developed and gradually refined. Progress has been aided by technological advances that resulted in better resources, including test compounds, reagents, and assays to monitor viral and host markers of infection and pathogenesis, and integration with New Approach Methodologies (NAMs). Although no NHP model is perfect, carefully designed experiments with selection of input variables can address many questions related to basic pathogenesis and can provide proof-of-concept of prophylactic and therapeutic intervention strategies. Because of significant differences between pediatric and adolescent/adult HIV infection, NHP models that cross the life span – from infancy to adulthood - have been developed. In this review, we will highlight some of the early challenges, the milestones and key findings of the infant and adult macaque models, with examples of discoveries that have already led to clinical applications, and how the recent development of highly effective long-acting drug formulations and delivery methods provide much promise for the future.

Introduction

Although human immunodeficiency virus (HIV) was identified as the cause of acquired immunodeficiency syndrome (AIDS) more than 4 decades ago, the pandemic continues to affect many communities across the globe, especially marginalized people and those living in resource-poor areas. It is estimated that in 2023, almost 40 million people were living with HIV, of which an estimated 1.4 million children (≤ 14 years of age), and 1.3 million new infections occurred (UNAIDS 2024).

There has been major scientific progress in identifying effective strategies to reduce infection rates and effectively treat people living with HIV (PLHIV), and the tools exist to reach the ambitious goal of making the next generation free of AIDS. However, reaching that goal relies on our ability to scale up implementation of proven strategies by mobilizing resources and breaking down barriers of human rights violations, poverty, stigma and discrimination. Despite major progress over the past 2 decades, unfortunately, and very worrisomely, this progress is at high risk of being lost due to political cut-off of funding to agencies and vital programs that provide these services to those most in need, and due to severe budget cuts in biomedical research that leave researchers paralyzed. At this pivotal time in history, it is useful to reflect back on the journey of scientific discoveries and understand how supporting a strong foundation of preclinical basic and translational research is important for continued progress towards clinical applications.

While in vitro systems (including organ-on-a-chip) and computer models, including the use of artificial intelligence, collectively called New Approach Methodologies (NAMs) provide increasingly useful resources and tools as a first screening step, consistent with the 3 Rs of animal research (Replacement, Reduction, Refinement), they do not recapitulate the full spectrum of HIV pathogenesis and cannot predict 100% accurately the effects of antiviral strategies, and the many interactions among them. This is where animal models come in, as the next step. Advantages of animal models of infectious diseases include our ability to control many input variables (such as host factors, virus inoculum, timing of antiviral interventions), and to do procedures that are difficult to do in humans (e.g., access to tissue samples via biopsy or euthanasia; administration of specific cell-depleting antibodies). Accordingly, using a limited number of animals, and in a defined time window, experiments can be performed to target specific research questions and explore novel hypotheses that are logistically difficult or unethical to test in humans.

The ideal animal model is one where an animal species that is close to humans, exposed to the same virus that infects humans, develops infection and a disease that highly resembles that in humans, and allows use of the same surrogate markers to track disease progression and the effects of interventions. Currently such animal model of HIV does not exist. While various small animal models (such as humanized mouse models; reviewed in^{1, 2}) can be useful in the early stages of preclinical research, they have limitations.

Nonhuman primates (NHP’s) such as macaques, are phylogenetically much closer to humans, and their physiology, including immunology, pre- and postnatal development, resembles that of humans. A major discovery was reported in 1985, when it was demonstrated that rhesus macaques can be infected with a similar lentivirus, simian immunodeficiency virus (SIV), and develop a disease that very much resembles AIDS³.

Since the first discovery of SIV, NHP models of HIV infection have faced many obstacles, including funding, scarcity of technical resources, and skepticism of the ability to extrapolate their results into HIV infection of humans. However, the ingenuity and perseverance of researchers has led to gradual and major progress, including recognition of their high value and relevance. A number of major HIV prevention and treatment strategies that are now used globally are founded on discoveries first made in NHP models. This review, without aiming to be comprehensive, will highlight the historical development of some of the main NHP models.

The road of development of NHP models of HIV: overcoming hurdles

The increased value of NHP models of HIV infection is a result of the development and optimization of different models with select host and viral factors, and the availability of better resources (including reagents, assays and many technological advances)⁴. Many of these advances occurred at different but overlapping times, leading to stepwise progress. Without being comprehensive, some milestones in SIV/SHIV research are presented in Figure 1 with special emphasis on prevention strategies for which proof-of-concept was first demonstrated in NHP models prior to demonstration in clinical trials.

Figure 1. Timeline of milestones in NHP models for prevention of infection.

The increased value of nonhuman primate models of HIV infection stems from the development and optimization of diverse models incorporating specific host and viral factors, along with the availability of improved resources—including reagents, assays, and numerous technological advances. Figure Abbreviations: CAB: cabotegravir. RVP: rilpivirine; LEN: lenacapavir; PrEP: pre-exposure prophylaxis; ARV: antiretroviral.

Upon discovery of SIV, the earliest studies used juvenile or adult macaques, and used the intravenous route of virus inoculation, usually with a large inoculum dose to assure infection. As HIV can transmit across the full life-span (“from womb to tomb”), SIV infection models have been developed to mimic these different stages of life, and to also better mimic the natural routes of transmission, which usually involve mucosal exposure (Figure 2). To develop models of pediatric HIV infection, early experiments studied natural transmission of SIV by inoculating pregnant or lactating female macaques with SIV, but the results (including frequency and timing of transmission) were inconsistent, which made these models less practical (reviewed in⁵). While currently no good NHP model exists to study in utero transmission, to mimic the oral exposure that occurs intra-partum but especially also post-partum through prolonged breastfeeding, newborn or infant macaque models with direct oral virus inoculation were developed in the mid-1990’s^{6, 7}. Similarly, to mimic the routes of sexual transmission, mucosal inoculation models, using the intravaginal or intrarectal routes, have been developed with juvenile and adult macaques^{8, 9}.

Figure. 2. Development of macaque models to mimic human development and natural routes of transmission.

The most common routes of HIV transmission in humans can be modeled in the macaque models. Image generated with Biorender and Iconfinder. Persons by Universal Icons from Noun Project (CC BY 3.0)

Rhesus monkeys become sexually mature at approximately 2.5 – 3.5 years of age, reach adult stature by 8 years of age, and females undergo menopause by about 25 years as summarized in Figure 2. The rate of aging ratio of monkey to human years can be summarized as 1:4 from birth to sexual maturity, 1:3 during young adulthood, and 1:2 for females before menopause¹⁰. These factors may be relevant to vaginal transmission models, as there are significant changes that occur in the female reproductive tract with age and sexual maturuty and menopause that may affect HIV susceptibility¹¹.

Initial mucosal inoculation models used relatively high doses of virus, so that a single or double inoculation assured nearly 100% infection rates of control animals. While such models are highly relevant to address particular research questions, such as the study of early events following exposure, including infection of cells at the mucosal site with subsequent systemic dissemination, these high-dose inoculation models may underestimate the efficacy of prophylactic strategies. Accordingly, repeated low-dose inoculation models which more closely mimic natural transmission, in particular the transmission of a limited number of variants across the mucosal barrier, have been developed for the various mucosal transmission models^12–14.

Researchers discovered that different isolates and molecular clones of SIV could have different degrees of virulence, as defined by virus replication and disease progression, covering the whole spectrum from high virulence (such as SIVmac251) to low or no virulence (such as SIVmac1A11)¹⁵. Despite the high value of SIV isolates and clones to study pathogenesis, preclinical testing of antiviral strategies in the SIV-macaque model is most relevant if the viral target and its susceptibility to inhibition naturally resemble, or can be engineered to resemble, that of HIV-1. SIV, which is mostly closely related to HIV-2, has some significant sequence differences with HIV-1, which affect its susceptibility to certain antiretroviral drugs or immune-based strategies, such as those that target the viral envelope. To overcome this, chimeric Simian-Human Immunodeficiency Viruses, called “SHIV’s”, have been engineered, in which one or more genes of the SIV backbone have been exchanged by their HIV counterparts. The most commonly used SHIV’s are (i) the env-SHIV, that have HIV-envelope, often selected and engineered further to replicate better, or to reflect transmitter/founder viruses, to increase their relevance for vaccine immunogenicity and efficacy studies (reviewed in¹⁶), and (ii) RT-SHIV, which due to having the HIV-1 reverse transcriptase (RT), is also susceptible to nonnucleoside RT inhibitors (NNRTI)¹⁷. For some SIV and env-SHIV strains, stocks with genetically engineered bar-coded variants have been produced, and are useful to trace the kinetics of transmission, distribution and persistence of individual variants (reviewed in¹⁶).

In the early years of the NHP models, monitoring infection status of animals after inoculation was limited to detecting virus through virus isolation, proviral DNA polymerase chain reaction (PCR), or monitoring evidence of immune responses such as seroconversion, which basically provided “yes/no” answers regarding infection status. The lack of sensitive quantitative assays to monitor virus replication meant that one could not detect if an intervention, even if it did not protect against infection, resulted in partial reduction of virus replication. Starting in the early 1990’s, a few macaque studies used quantitative viral outgrowth assays, by doing limiting dilution cultures of plasma or lymphoid cells with susceptible cell lines, as a first attempt to quantitate viral replication¹⁸. However, these assays have the limitations of being not only very labor-intensive, but also not very sensitive, especially to monitor virus replication during the asymptomatic stage of infection, when infectious virus levels in plasma of infected juvenile or adult macaques are usually below the limit of detection. A breakthrough to circumvent this problem was the development of the newborn macaque SIVmac251 infection model. Infant macaques, inoculated early after birth with SIVmac251, developed persistently high infectious plasma viremia, had rapid immunosuppression (as indicated by poor antiviral immune responses), and a rapid disease course, as animals had poor weight gain and most animals developed clinical immunodeficiency that met the criteria for euthanasia within 3 months¹⁹. This more consistent fulminant disease course and persistent viremia allowed detection of antiviral drug efficacy using relatively small animal group sizes. Indeed, in this SIVmac251 newborn macaque model, early initiation of zidovudine therapy within one week of infection delayed and reduced the initial peak of virus replication, and reduced the immunosuppression, which allowed the animals to mount antiviral immune responses which likely contribute to reduced virus replication. Although virus replication was not fully suppressed, and zidovudine monotherapy selected for the emergence of viral mutants with reduced susceptibility, the final outcome was a significantly delayed disease course^{19, 20}. In hindsight, although zidovudine had its limitations of having relatively weak antiviral potency, and a rather complicated dosing regimen, these early experiments were important by demonstrating (i) that, at a time of controversial debates about early versus deferred initiation of zidovudine therapy for PLHIV²¹, early antiviral therapy reduces virus replication and resulted in clinical benefits, and (ii) the value of NHP models and the need to further support their development, especially at a time when some critics raised doubts about their relevance. As explained later in this review, the availability of NHP models in the mid-1990s played an essential role in the preclinical development of much more potent drugs (including tenofovir), which have become the cornerstones for prevention and treatment regimens.

A major milestone was the development of assays that could quantitate viral RNA in plasma, with increasing sensitivity, such as quantitative reverse transcription polymerase chain reaction (RT-qPCR)^{22, 23}. These assays were important as they allowed better use of NHP models for several reasons: (i) virus replication could also be monitored during the asymptomatic stages of infection, including in adult animals; (ii) because plasma viral RNA levels correlated with the rate of disease progression, they became a very valuable surrogate marker of disease progression, and thus reduced the need to monitor animals for prolonged periods of time until their clinical onset of disease; (iii), because viral RNA levels, together with CD4+ T cells, were also found to be strong predictors of disease progression in PLHIV (reviewed in²⁴), this allowed better direct comparison of results obtained in NHP models and humans.

Important considerations for HIV prevention studies

HIV prevention strategies that are effective across the lifespan require the consideration of age-specific factors in their implementation (Figure 3). On the virus side, the route of acquisition, the frequency of exposure, and the number of different viruses per exposure is distinct for different age groups. Infants now primarily acquire HIV by breastfeeding and while exposure occurs multiple times per day, exposure is generally limited to the dominant maternal virus variant. This is in sharp contrast to adolescents and adults who acquire HIV primarily via sexual transmission and, to a lower percentage, by intravenous drug use²⁵. Therefore, frequency and exposure to different virus strains depend on the frequency of unprotected sexual encounters and the number of different partners, or, in the case of IV drug users, on the number of exchanges of dirty needles with multiple HIV-infected individuals. Host factors that can impact the success of prevention strategies include the developmental status of the immune and central nervous systems, age-dependent hormonal changes, sex, host microbiota, and host genetics. Additional factors include geographic region and socioeconomic parameters that impact living conditions.

Figure. 3. Considerations for HIV prevention studies addressable in NHP models.

Primate models can address important questions across the lifespan including immunologic and neurologic changes during pregnancy, neonatal periods, and in aging.

Differences in the functional maturity of the immune and Central Nervous System (CNS) in infants, and to a certain degree even in adolescents (e.g., frontal lobe development), may not only translate into the need of different dosing regimens and the inclusion of immunomodulatory components, but also require additional safety assessment to ensure no negative consequences of antiretroviral drugs or other prevention means on normal immune, organ, and CNS development. The sheer number of exposures via breastfeeding starting immediately after birth also implies that an exclusively vaccine-based approach to prevent HIV infection in infants is of low feasibility. Instead, including other prevention tools, such as passive immunization via long-acting broadly neutralizing antibody cocktails, would be highly valuable. In older adults (>50 years of age), a population that has experienced a rise in HIV infections in the last decade²⁶, prevention strategies may also require the addition of immunomodulatory agents because senescence and inflammaging²⁷ may limit immune responses. Furthermore, due to the increasing potential of comorbidities with age, various prescription medications may interfere with antiretroviral drug therapies. The following section will focus specifically on the need of pediatric and adolescent NHP models, their unique features and their translational value.

The need for pediatric and adolescent NHP models to test HIV prevention therapies

Drug and vaccine implementation are strictly regulated. The time from early drug/vaccine development to optimization, assessment of safety and toxicity in relevant animal models, to safety and efficacy testing in human clinical trials, and approval for human use often spans several years to decades. Moreover, clinical trials are performed in a stepwise manner, starting with healthy adults and only if proven safe and effective, drug therapies or vaccines will move for testing into other populations, such as immunocompromised individuals, pregnant women or children. Thus, implementation of prevention strategies for the pediatric population is generally delayed by several years after successful introduction into adults. Furthermore, while clinical trials in adults include a dosing component, drug and/or vaccine doses for the pediatric population are often not based on actual data, but are somewhat randomly selected, most often rationalizing that smaller size and lower weight should translate into a lower dose. Bridge studies in relevant pediatric and adolescent NHP models that are performed in parallel with late-stage human clinical trials could provide critical data on optimal dose, dosing interval, safety and efficacy of prevention strategies to reduce the major time gap in the implementation of effective HIV prevention strategies. The Pediatric Research Equity Act, that was written into US law in 2003, provides the legal framework to advance studies of new drugs, formulations, and vaccines to the pediatric population²⁸. The FDA Modernization Act 2.0, signed in 2022, paved the way to permit data from certain in-silico model systems to accelerate the approval process²⁹. This was solidified by the Modernization act 3.0 which replaced “animal” tests, data, studies, models, and research with “nonclinical” tests, data, studies, models, explicityly allowing non-animal/human-biology-based approaches such as cell-based assays, microphysiological systems, bioprinted models, and computer models³⁰. Importantly, this did not prohibit animal studies, and animal tests remain one permissible type of nonclinical test.

Nonhuman primate models to test antiretroviral drugs for prevention: a successful journey from infancy towards maturity

Although the FDA defines the pediatric populations as ranging in age from birth to 16 years, it clearly defines the specific age ranges for neonates (birth to 1 month), infants (1 month to 2 years), children (2 to 11 years) and adolescents (12 to 16 years). The age groups within the pediatric population are necessary as age-dependent differences in weight, body fluid percentage (in particular water), blood volume per kg, liver and kidney function, the gastric pH, and CNS barrier function can all alter drug absorption, distribution, half-life, and clearance. The main drug-metabolizing enzyme systems also differ in their activity^{31, 32}. Conversely, the drugs can alter the metabolomic function and potentially negatively affect normal developmental changes in immune, bone, and CNS development.

In the case of HIV infection, the NHP model has proven instrumental in the implementation of novel drug regimens in infants and adults. While initial studies testing antiretroviral drugs failed to show efficacy in NHP³³, a milestone in HIV prevention was a publication in 1992 of a study that demonstrated that a 6 week thrice daily oral dosing regimen of the antiviral drug zidovudine (AZT), which was at that time the only approved HIV drug, was effective in protecting infant macaques against infection following intravenous inoculation with a low dose of the highly virulent SIVmac251³⁴. The promising results of this study guided clinical trials which subsequently demonstrated that zidovudine administration to pregnant women and their newborns drastically reduced the rate of vertical transmission of HIV³⁵.

The next major milestone in antiretroviral drug prevention occurred in 1995, when studies with the at-that-time novel RT inhibitor tenofovir (originally called 9-[2-(phosphonomethoxy)propyl] adenine; PMPA) were initiated in macaques. A first study demonstrated that a once-daily tenofovir injection regimen for 4 weeks was 100% effective in protecting juvenile long-tailed macaques again infection, even when treatment was started 24 hours after intravenous SIVmne inoculation³⁶. Tenofovir also proved to be very highly effective in protecting newborn macaques against oral SIVmac251 infection, even with short regimens of 1 to 2 doses of tenofovir around the time of virus inoculation^37–39. This high prophylactic efficacy of tenofovir in the infant macaque model, including with short regimens, was afterwards also confirmed in juvenile and adult macaques, including protection against mucosal (vaginal or intrarectal) transmission; it was also found to be highly effective as chronic therapy to treat established infection (reviewed in^{4, 40}).

The high efficacy of tenofovir in the various infant and adult macaque models was highly predictive of the efficacy in humans in clinical trials. Accordingly, tenofovir, in its oral prodrug formulations, became the cornerstone not only of many HIV combination treatment regimens, but also of the first Pre-Exposure Prophylaxis (PrEP) regimen (Truvada; FDA approval in 2012) as a daily pill to prevent HIV infection for people with high-risk behavior. The observations that PrEP efficacy correlated highly with adherence sparked high interest to develop long-acting drug formulations and delivery devices, to overcome hurdles of daily drug adherence. Again, prophylactic success observed in NHP models predicted the outcome in humans. Several of these approaches were found to be highly effective in adults and are described later in this article. Further studies with these long-acting antiretrovirals are warranted to study their potential to prevent pediatric HIV infection.

Passive immunization studies in pediatric NHP Models – Proof-of concept for broadly neutralizing antibody therapies to prevent HIV infection

The potential of antibodies to protect against HIV acquisition was established long before broadly neutralizing antibodies (bnAbs) had been isolated and developed as tools in HIV prevention. In 1998, the administration of plasma from SIV-infected rhesus macaques was demonstrated to protect neonatal rhesus macaques against high-dose oral SIV challenge⁴¹. Subsequent studies by a different group confirmed this finding⁴². They further established that antibody administration within a limited window of time after SIV or SHIV exposure could ameliorate the normally fast progressing disease course in infant rhesus macaques^{43, 44}. Twenty years later, the concept had evolved to the administration of specific monoclonal broadly neutralizing antibodies that could prevent oral SHIV acquisition in neonatal macaques if given prior to exposure or reduce viremia if given within 72 hours of SHIV exposure^{45, 46}. The success of the Antibody-Mediated Prevention (AMP) trial in humans⁴⁷ was validated in infant NHP models further demonstrating efficacy of bnAbs⁴⁸. The development of bnAbs with an increased half-life of up to 6 to 9 months positions bnAb cocktails at the forefront of effective prevention tools against breastmilk transmission of HIV. As the breastfeeding period is of relatively limited duration, and the risk of HIV infection in childhood postweaning is minimal, the risk for the emergence of viruses with resistance mutations in treated infants is neglectable.

Active HIV vaccine strategies and their implementation in infancy

Lessons from past vaccine trials, increased understanding of HIV pathogenesis, advancements in structural biology, and the development of novel vaccine delivery platforms and adjuvants, among many other scientific and technological advances have culminated in the now widely accepted idea that a successful HIV vaccine needs to induce bnAbs to protect against HIV acquisition. Yet, the induction of B cell lineages and their maturation towards production of bnAbs still faces several obstacles. BnAb maturation is a lengthy process that depends on the tightly regulated progression of a germline B cell receptor (BCR)-expressing B cell to a B cell that acquired numerous, but specific mutations that are essential for a specific bnAb family. In fact, in adults living with HIV, only a small subset will develop such bnAbs and only after several years. The factors driving the maturation of bnAbs towards targeting the CD4 binding site (CD4bs), the V3 glycan site, the V2 glycan site or V1/V2 loop, the membrane-proximal external region (MPER) of the gp41 subunit, or the gp120-gp41 interface of the HIV envelope are not defined. Essential mutations are distinct for these different bnAb classes. As a consequence, a single immunogen will not be sufficient to induce bnAb maturation; instead, a multi-dose regimen consisting of different HIV Envs containing these specific mutations in the correct order and in a step-wise fashion will be required. Historical and epidemiological data indicate that multi-dose vaccine regimens are more difficult to implement in adults. Furthermore, one of the major risk groups for HIV acquisition are young adults. In adolescents between the ages of 15–19 years of age, HIV acquisition has remained relatively stable over the years, despite progress in successful introduction of pre- and post-exposure prophylaxis in other age groups. Thus, a successful HIV vaccine should induce immune responses that can protect against HIV infection prior to sexual maturity. Considering the need for a multi-dose vaccine to induce bnAbs, implementation of a HIV vaccine regimen in infancy constitutes a logical inference. Not only does the infrastructure for a multi-dose vaccine regimen exists due to the Extended Program of Immunization, but there is also evidence that infants living with HIV develop bnAbs in a shorter time period and these bnAbs require fewer mutations^{49, 50}.

Studies in infant rhesus macaques have clearly demonstrated that the magnitude of envelope-binding antibodies in plasma or serum is comparable to those achieved in adults⁵¹. Recently, it was reported that 3 of 5 infant rhesus macaques who were immunized with adjuvanted BG505 GT1.1 SOSIP, an Env trimer targeting a germline BCR, develop precursors to CD4bs bnAbs within 18 months and 1 additional animal developed that signature after a heterologous boost by two years⁵². A monoclonal antibody isolated from one of these animals exhibited broad neutralization and structurally resembled human VRC-01-like CD4bs bnAbs⁵². Vaccine studies using the same germ line-targeting (gt) strategy in infant and juvenile rhesus macaques provided further evidence that the infant immune system is better equipped to promote the maturation of bnAb precursors compared to older animals (unpublished data, Permar and De Paris). These findings support the idea that the development of a HIV vaccine is feasible and could be implemented in infancy and childhood to provide lifelong protection against HIV acquisition.

The NHP model provides the opportunity to test this hypothesis because of the accelerated aging process in NHP compared to humans (Figure 2). Thus, we can immunize infant rhesus macaques and implement challenge by the vaginal or intrarectal route to test vaccine efficacy in 3- to 4-year old macaques, reflecting sexual acquisition of HIV in human adolescents. This relatively faster age progression requires immunization intervals adapted to this timeline. To ensure translatability of findings in the infant NHP model to humans, studies are in process to adjust the infant vaccine schedule with adjuvanted BG505 GT1.1. SOSIP to the human pediatric vaccine schedule. One limitation of the NHP model is the small group sizes due to the tremendous costs associated with infant rearing and maintaining animals for several years. Ongoing studies also explore the interactions between standard pediatric vaccines and HIV vaccination in the NHP Model. To even more closely recapitulate human conditions, and because studies have demonstrated that nursery-reared versus dam-reared infant macaques have differences in the development of gut microbiome and immune system⁵³, in recent studies, newborn and infant rhesus macaques are, instead of being reared in a nursery, dam-reared during the early immunization period and then housed in social groups after weaning. Without a doubt, the model is not a substitute for human trials, but it allows to perform critical preclinical proof-of-concept studies. Among those are ongoing studies testing various HIV vaccine immunogens and doses and formulations of various age-appropriate adjuvants that could provide key information for subsequent human trials⁵⁴.

Non-Vaccine HIV prevention studies in adult nonhuman primates:

Prevention of HIV infection of infants could start with protection against maternal infection. Infants differ in both the way they are infected and in their response to HIV infection. For example, HIV infected infants have higher viral loads and more rapid progression to disease^{55 56}. Further, pediatric HIV infections occur through oral ingestion of infected maternal fluids during birth or from breastfeeding soon thereafter⁵⁷. Due to the risk of breastfeeding, formula feeding is usually recommended to HIV-infected mothers in developed countries. However, postnatal transmission of HIV is more complex in developing countries, where formula feeding may not be safe or affordable⁵⁸. Preventing HIV infection of mothers, especially around the time of pregnancy would be the best option for reducing pediatric HIV infections.

The above statement should be considered with caution as increasing evidence suggests that hormonal changes associated with pregnancy may even increase a woman’s susceptibility to vaginal HIV infection^{59 60}. For the last several years, we have been examining ways to prevent vaginal HIV transmission using macaque models. However, assessing the safety and efficacy of a drug for a disease such as HIV, which has a relatively low vaginal transmission rate, requires large numbers of subjects. Nonetheless, using macaque models we have shown that hormones play a major role in susceptibility to vaginal transmission. Macaque research showed that administration of progestins resulted in thinning of the vaginal mucosal epithelium and markedly increased vaginal SIV and SHIV transmission rates^{61, 62}. This was later confirmed in humans when it was shown that injectable progestin-based contraceptives increased HIV transmission risks⁶³. The mechanisms of this are debatable, but multiple studies suggest it is the loss of estrogen dominance in the vaginal vault resulting in thinning of the protective squamous layers of epithelium, bringing the underlying T cells closer to the surface where they are more susceptible to HIV exposure from the lumen. In published studies, we have also confirmed that thinning of the vaginal epithelium occurs in women during the luteal phase of the menstrual cycle while in the follicular (pre-ovulatory) phase of the cycle, the epithelium is thickened in response to estrogenic effects^{64 65 66} However, during the luteal phase (progesterone dominance) there is thinning of the layers of protective squamous epithelium, resulting in increased exposure of lumenal antigens (or virus) to the underlying target T cells⁶⁷ Studies have confirmed activated, tissue resident CD4+ T cells are indeed the first cells infected in the vagina^{68, 69}. Thus, protection against infection of these CD4+ T cells could be an effective strategy for prevention of vaginal transmission. Alternatively, modifying the inflammatory environment of the vagina also appears to influence vaginal HIV transmission. Topical IFN-β (a type I IFN) vaginal treatments showed protection against multiple SHIV vaginal challenges⁷⁰. This fits with the hypothesis that specific types of local inflammation increase susceptibility to vaginal HIV transmission. Thinning of the vaginal epithelium may simply bring viral particles in closer proximity to primed CD4+ T cells in the mucosa Notably, the thickness and differentiation of the macaque vaginal epithelium vary markedly with reproductive and hormonal status; relatively thin, poorly keratinized epithelium occurs naturally in juvenile and reproductively senescent macaques, while late pregnancy is associated with pronounced mucification epithelial thinning, and complete keratinized epithelium⁷¹. Although it is impossible to collect vaginal biopsies from pregnant women, this could explain the higher HIV infection rates in pregnant women, especially in late pregnancy, and the early postpartum period when there is progesterone dominance⁵⁹. This is another reason NHP research is invaluable, as these studies simply cannot be performed or even confirmed in pregnant women for ethical reasons. Therefore, an NHP vaginal transmission model was developed using the human contraceptive Depoprovera^®, which renders an adult female macaque highly susceptible to vaginal SIV or HIV transmission^{72, 73}. Administering 30 mg Depoprovera to adult female macaques a few weeks prior to a single vaginal viral challenge results in a 90% transmission rate, enabling the use of small numbers of animals to test safety and efficacy of HIV prevention products⁷⁴. Although subtle effects of a vaccine or prevention may be missed using this “high dose” model, this is a stringent test of safety and efficacy against vaginal HIV infection.

The hormone model of vaginal transmission has been widely used to test the safety and efficacy of several topical HIV prevention products. These studies have also revealed much about the early events involved in vaginal HIV transmission. For example, vaginal administration of antibodies or compounds that prevent CD4 binding of HIV (antibodies to HIV env binding sites) results in protection from vaginal transmission⁷³. Similarly, compounds that block CCR5 binding also completely prevent vaginal transmission^74–77. In fact, some topical gels containing fusion inhibitors were safe and 100% effective when used within 30 mins of exposure^{76, 78}. However, this level of efficacy did not persist for hours, making these “coitally dependent” methods less viable candidates for prevention. Moreover, human clinical trials of such coitally dependent gels resulted in failure, largely due to lack of compliance⁷⁹. Thus, longer-acting prevention methods are now being pursued, including vaginal rings, subcutaneous implants, and long-acting injectables or orally administered antiviral compounds for prevention.

Long-acting delivery devices:

Vaginal rings:

Vaginal rings are currently approved for contraception in the US, and are a popular choice based on convenience and efficacy. Thus, a vaginal ring that could release controlled and sustained levels of antiviral compounds in the vagina could be a safe and effective method of HIV prevention. Macaques have been invaluable for testing vaginal rings. Using rings specifically designed for smaller macaque vaginal vaults, studies with CCR5 inhibitors (Maraviroc) and another CCR5 fusion inhibitor (CMPD167) showed high levels of sustained release in vaginal pharmacokinetic studies⁸⁰. An NNRTI (MC1220) in vaginal rings also showed highly significant levels of protection from SHIV challenges compared to placebo ring controls⁸¹. Vaginal rings containing tenofovir also provided complete protection against SHIV challenges in macaques⁸² Other macaque vaginal ring studies documented excellent levels of release of dapivirine and darunivir⁸³. These studies led to several human trials, and now dapivirine rings are available for use in some countries, albeit not yet approved in the USA. There are however concerns with vaginal rings as some studies have reported increased transmission rates compared to controls, so adequate testing in these NHP models is necessary prior to advancement to human trials. If improperly designed, vaginal rings loaded with certain microbicides may induce inflammation, which increases vaginal HIV transmission rates. For example, detergents such as nonoxynol-9 prevented SIV transmission in single dosing models but failed to predict increased transmission in humans with repeated use, which induced inflammation⁸⁴. This fits with the hypothesis that inflammation increases susceptibility to vaginal HIV infection. On a more positive note, modifying the inflammatory environment of the vagina also appears to influence vaginal HIV transmission. As mentioned earlier, topical IFN-β (a type I IFN) vaginal treatments protected against multiple SHIV vaginal challenges⁷⁰. Therefore, reducing certain types of local inflammation may be a different approach to HIV prevention.

Subcutaneous implants:

Another promising technology for HIV prevention or treatment of infants includes subcutaneous implants that deliver sustained effective levels of antiviral drugs without the need for daily dosing, which can be especially difficult in the pediatric patient. Other advantages include stable levels of drug delivery without peaks and troughs. It is hoped implants may be identified that release effective levels of antiretroviral drugs for up to a year or longer⁸⁵. We have previously tested implants in macaques for sustained delivery of tenofovir derivatives⁸⁶ and cabotegravir (CAB)⁸⁷ and are currently testing combination implant formulations of other ARTs with collaborators at Research Triangle International (and see session IV in this special issue⁸⁸).

Testing in animals is critical, as we have found some antiretrovirals induce marked inflammation and discomfort. For example, tenofovir alafenamide implants induced chronic inflammation when administered by slow-release implants, despite being a very safe and effective drug when delivered subcutaneously⁸⁶. In contrast, CAB implants were very safe and induced no detectable inflammation even after months of implantation⁸⁷. Further, cabotegravir implants protected against rectal challenges in macaques⁸⁹ Thus, some drugs are well tolerated when formulated in implants, whereas others, which are safe orally and parenterally, cannot be delivered in such long-acting devices. This makes such testing in animals, in which recruitment and retention of inflammatory cells can be different than in vitro models, essential for testing HIV prevention methods.

Oral and injectable prevention methods:

As highlighted earlier, the promising results obtained in NHP models with injectable and oral regimens of tenofovir (including its disoproxil fumarate and alafenamide prodrug formulations) with or without emtricitabine, predicted the high efficacy of these drugs as oral PrEP in humans⁹⁰. This sparked interest in testing other compounds and developing long-acting formulations. There are currently several antiretrovirals (ARVs) being used as injectable PrEP in a limited number of countries. This includes the integrase strand transfer inhibitor CAB, the non-nucleoside reverse transcriptase inhibitor rilpivirine (RPV), and the capsid inhibitor lenacapavir (LEN)^{91 92}. Notably, all of these were extensively tested in macaques prior to advancement and found to be very effective^92–96. Very recently, a twice-yearly subcutaneous injection regimen of LEN, found to be highly effective in protecting adults against HIV infection, was approved in the USA and EU^{97, 98}.

Discussion/Conclusions:

The development of antiviral strategies can be accelerated by efficient and predictive in vitro and in silico approaches, and other New Approach Methodologies (NAMs), complemented with animal models, capable of screening and selecting the best products at each decision point. An opportunity to discard ineffective products early in the development process (and thus save much time and resources), however, is lost whenever an available appropriate animal model is bypassed. In addition to directly testing efficacy, animal models can be very useful to define surrogate markers of efficacy, which can help understand the mechanisms of action, guide clinical trials and accelerate future product development.

Although there are limitations to animal testing, even when using nonhuman primates, the risks outweigh the benefits, and more often it is our failure to understand the disease processes that lead to misconceptions about the animal model. NHP models are not perfect when trying to predict the complexities of viral infections, but we cannot know what will happen in a living animal system until we test it in an appropriately designed model. For example, the selection of challenge virus, NHP species, route of challenge, and dose may affect the results of studies in different ways, and experiments showing efficacy in macaques may not predict efficacy in humans in unexpected ways. For example, early prevention studies in macaques using CXCR4 strains of SHIV for challenge did not show efficacy in humans who we later learned all become infected with CCR5 using strains of HIV⁸⁴. Moreover, topical vaginal microbicides that showed efficacy in macaques failed in humans due to lack of compliance in the latter, which could not be predicted by NHP studies. Nonetheless, nonhuman primates are clearly the most directly translatable model for testing the safety and efficacy of HIV prevention methods. Notably, unlike rodent or other animal models, most reagents used in NHP research are the same as those used in human research due to the similarities of the immune systems. Moreover, we now know that HIV evolved from simian immunodeficiency viruses⁹⁹, and thus further rigorously designed studies of SIV/SHIV infections and prevention in NHP models will continue to shed light on the pathogenesis and treatment of HIV infections of humans, especially for pediatric HIV prevention, which is challenging to study in human infants.

In summary, over the past decades, despite a slow start, stepwise progress has turned the NHP models into valuable and relevant resources that were essential in providing proof-of-concept of novel intervention strategies that guided clinical trials. The development of long-acting antiretroviral PrEP regimens with very high efficacy, such as twice-yearly lenacapavir, provides high hope that we can further reduce the number of new HIV infections. Nonetheless, despite this optimism, an HIV-prevention strategy that relied exclusively on lenacapavir would have limitations, including development of drug resistance¹⁰⁰ and the development of HIV vaccine remains an essential component of a global, multipronged strategy needed for ending the HIV epidemic. (reviewed in¹⁰¹). NHP models, complemented and integrated with NAM’s, are expected to continue playing a vital role in these efforts for HIV vaccines and other preventative and therapeutic strategies.

Data sharing:

Data sharing is not applicable to this review article as no new datasets were generated in this review of previously published data.