- Split View
-
Views
-
Cite
Cite
Taylor W Foreman, Smriti Mehra, Andrew A Lackner, Deepak Kaushal, Translational Research in the Nonhuman Primate Model of Tuberculosis, ILAR Journal, Volume 58, Issue 2, 2017, Pages 151–159, https://doi.org/10.1093/ilar/ilx015
- Share Icon Share
Abstract
Infection with Mycobacterium tuberculosis predominantly establishes subclinical latent infection over the lifetime of an individual, with a fraction of infected individuals rapidly progressing to active disease. The immune control in latent infection can be perturbed by comorbidities such as diabetes mellitus, obesity, smoking, and coinfection with helminthes or HIV. Modeling the varying aspects of natural infection remains incomplete when using zebrafish and mice. However, the nonhuman primate model of tuberculosis offers a unique and accurate model to investigate host responses to infection, test novel therapeutics, and thoroughly assess preclinical vaccine candidates. Rhesus macaques and cynomolgus macaques manifest the full gamut of clinical and pathological findings in human Mycobacterium tuberculosis infection, including the ability to co-infect macaques with Simian Immunodeficiency Virus to model HIV co-infection. Here we discuss advanced techniques to assay various clinical outcomes of the natural progression of infection as well as therapeutics in development and novel preclinical vaccines. Finally, we survey the translational aspects of nonhuman primate research and argue the urgent need to thoroughly examine preclinical therapeutics and vaccines using this model prior to clinical implementation.
Introduction
The nonhuman primate (NHP) model of tuberculosis remains the gold standard for modeling Mycobacterium tuberculosis (Mtb) infection. Rhesus macaques, Macaca mulatta, and Cynomolgus macaques, Macaca fascicularis, have long been used in experimental tuberculosis research. Macaques were first used in the late 1960s when scientists hypothesized that macaques increased susceptibility to natural tuberculosis infection could be utilized to rapidly test therapeutics and to study natural and vaccine-induced immunity to infection (Clarke 1968; Good 1968; Schmidt 1966). The first vaccine study occurred in the early 1970s and sought to test the efficacy of the already 50-year-old Bacille Calmette-Guerin vaccine through intravenous administration followed by aerosol infection with virulent Mtb (Barclay et al. 1970, 1973). This study, among others, brought the NHP model to the forefront of tuberculosis research, and has since been ever-growing in importance for clinical application of novel therapeutics and preclinical testing of vaccines (Pena and Ho 2016). This review focuses on the strides NHP research has brought to the field of tuberculosis research, the adaptation of the model to advanced technologies, and the clinical implementation of current ongoing NHP research.
The Development of the NHP Model and its Use
Rationale for Use of NHPs
While humans have been coevolving with Mtb throughout our long history, mice are not natural hosts of Mtb. Mice have not experienced the genetic evolutionary pressure that would have selected for increased control of Mtb infection, and are thus permissively susceptible to infection characterized by severe dissemination of bacteria (Kramnik and Beamer 2016). However, due to their high genetic similarity to humans, the macaque response to Mtb accurately models all aspects of tuberculosis disease. Critical amongst these are the ability to develop long-term subclinical latent disease due to local immune control and to progress to clinically apparent active disease (Darrah et al. 2014; Dutta et al. 2010; Foreman et al. 2016; Kaushal et al. 2015; Mehra et al. 2012, 2013, 2015; Phillips et al. 2015; Slight et al. 2013). In comparison to human tuberculosis and in contrast to mouse models, macaques form highly structured granulomas with readily-apparent lymphocytic and macrophage tropic layers, central caseation, and severe neutrophilic immunopathology seen in active disease (Mehra et al. 2013). The clinical manifestations mimic those seen in human tuberculosis infection, including dyspnea, pyrexia, cachexia, positive PPD and IGRA responses, detectible bacilli in sputum, and pneumatic inflammation upon x-ray imaging (Capuano et al. 2003; Dutta et al. 2010; Gormus et al. 2004; Kaushal et al. 2015; Mehra et al. 2012). Furthermore, while the immune mediators required for the control of pulmonary bacterial infection are present in both mice and primates, subtle differences exist. These include differences in specific macrophage killing mechanisms and disparate utilization of several immunosuppressive pathways (Su et al. 2014). This has led to debate about the specific mechanisms of control of Mtb infection and their relative importance in mice vs humans (Liu et al. 2006; Nathan 2006; Thoma-Uszynski et al. 2001). The progression of infection in the wake of immune response optimal for a specific model can thus generate diverse endpoints. This discourse led to the refrain that some of the most important experiments can be performed only in mice and not humans, reflecting a key gap that precluded progress in understanding both the pathogenesis of and immunity to tuberculosis. Immunological studies have, however, highlighted the overlapping immune responses to Mtb infection both in peripheral blood and cells isolated from bronchoalveolar lavage, thus demonstrating the immune response in the macaque model parallels that in humans (Ernst et al. 2012; Mothe et al. 2015). In addition, Mtb-infected macaques demonstrate remarkable heterogeneity in both disease outcomes (Capuano et al. 2003; Mehra et al. 2015) as well as granuloma pathology (Kaushal et al. 2012). Hence, studies in macaques can fill an important gap between the mouse model of tuberculosis and humans.
Numerous reasons, however, historically restricted the use of macaques more widely to study the progression of infection and to understand correlates of disease. These included expense, requirement of specialized containment facilities and husbandry staff, and lack of availability. The AIDS epidemic resulted in the development of the macaque models of simian immunodeficiency virus (SIV) infection to primarily study HIV infection and this was conducted mostly by one of the several NIH-supported National Primate Centers. However, biosafety concerns initially kept these centers from developing tuberculosis research programs. UCLA (Walsh et al. 1996) and the University of Pittsburgh (Capuano et al. 2003) were amongst the first facilities in the modern era to study tuberculosis in macaques. While the use of primates can often be limiting due to capabilities of facilities and monetary resources, throughout this review the authors suggest that all novel therapeutics and vaccines ought to be tested in the NHP model as a checkpoint prior to clinical implementation. Though the immune response and subsequent clinical manifestation to infection mimics that of humans, there are subtle differences between rhesus and cynomolgus macaques, including strengths and weaknesses for each species that must be considered when designing NHP studies.
Different Species of Macaques and their Use in the Tuberculosis Model
Rhesus macaques were initially the dominant species used to establish the NHP model of tuberculosis (Barclay et al. 1970, 1973; Good 1968; Janicki et al. 1973), largely due to their relative ease of maintaining and breeding in captivity and use in alternative modeling including polio (Sabin 1957). In the first studies, rhesus macaques were experimentally infected with the H37Rv strain of Mtb, causing the majority of animals to progress to active disease within 50 to 60 days postinfection. Since then, with lower doses and differing strains of Mtb, rhesus macaques have been shown to establish latent, asymptomatic infection determined by tuberculin skin test conversion and assessment of lungs for persistent bacteria months after infection (Foreman et al. 2016; Mehra et al. 2011). However, when challenged with more virulent strains such as Erdman K01, rhesus macaques not only develop higher bacterial burdens but increased pathology over time (Sharpe et al. 2016, 2009). Vaccination with BCG provides increased protection in cynomolgus macaques when challenged with a high dose of Erdman. These animals in general exhibit lower bacterial burdens relative to BCG-vaccinated rhesus macaques (Langermans et al. 2001). This increased susceptibility to highly virulent strains is not detrimental to the rhesus macaque model, however, as it provides an accurate model for rapid testing of vaccine candidates wherein protection in a more susceptible species may provide greater justification for clinical testing of the preclinical vaccine.
The first studies utilizing cynomolgus macaques established a chronic tuberculosis model when using very low doses (10 colony forming units) of the virulent Erdman strain (Walsh et al. 1996). This model was further tested up to 20 months post-low-dose infection with Erdman, where researchers described latent, active, and spontaneously reactive (latently infected progressing to active disease) as synonymous to human tuberculosis disease (Capuano et al. 2003; Lin et al. 2009). The cynomolgus model has since been rapidly exploited for vaccine studies, which used moderate doses of Erdman (Mehra et al. 2013; Okada et al. 2007; Reed et al. 2009). Similarly, the rhesus macaque model, using both high doses of low-virulence strains like CDC1551 or low doses of high-virulence strains like Erdman, has been used in vaccine development efforts (Darrah et al. 2014; Kaushal et al. 2015). As studies delineate the subtle differences in the progression of Mtb infection in these two species, it is becoming increasingly clear that both provide insights into the different aspects of human disease syndrome that must be protected against.
Routes of Vaccination and Infection
Throughout the NHP model there have been two primary routes of challenge or vaccination. The vast majority of studies have used intra-tracheal or intra-bronchial instillation of bacteria, which allows for tighter control of the presented dose but only allows deposition of bacteria in isolated pulmonary lobes. However, recent studies have established an aerosol infection route, which provides inherent variables but is more representative of the natural route of infection. These variables include the rate and volume at which the animal is breathing; however, this route exposes bacteria to all lobes of the lung. Controlling for the time (and thus the volume of infectious air the animal breathes) allows for increased control of low infectious doses such as 10 to 20 CFU (Foreman et al. 2016).
It has been shown in lower animal models, as well as in rhesus macaques more recently, that aerosol-delivered BCG provides increased protection when compared to intradermal delivery (Kaushal et al. 2015; Sharpe et al. 2016; White et al. 2015). This may point to the value in eliciting localized adaptive responses. As clinicians continue to advocate for mucosal-delivered vaccines, the NHP model must continue to also test, and even compare, route of vaccination to hone in on optimal dose and route of protective vaccines. It is accepted that aerosol delivery of vaccines contains inherent difficulties in logistics and implementation, especially in the regions of the world where vaccination is most urgently needed. Therefore, the NHP model serves as an optimal model for testing alternative vaccine strategies such as intranasal, sublingual, oral, or adjuvanted intramuscular/intradermal routes. Alternative vaccine delivery routes have already been tested in rhesus macaques albeit to induce immunity to SIV (Veazey et al. 2015).
The Use of Different Strains of Mtb in the NHP Model
Due to genetic differences between mice and macaques, at least some immune responses to Mtb are divergent in these model systems. Recent work in the NHP model has not only demonstrated that macaques recapitulate the human tuberculosis syndrome (Flynn et al. 2015; Kaushal and Mehra 2012; Kaushal et al. 2012; Lin et al. 2006, 2009; Pena and Ho 2016), but also established key differences in the pathogenesis of disease with rodent models. This could be due to differential induction of key virulence and adaptation programs in the bacilli such as hypoxia, which is found in well-organized necrotic granulomas. As previously described, no major pathological differences were observed when BALB/c and C57Bl/6 mice (Rustad et al. 2008) were infected with Mtb strains deficient in hypoxia response genes (dosR, dosS, dosT). However, when these same strains were used to infect rhesus macaques, there were significant differences in pathology, bacterial burden, as well as the immune response (Mehra et al. 2015). Furthermore, a mutant of Mtb deficient in its ability to scavenge oxidative stress (MtbΔsigH) was not restricted in bacterial replication in the lungs of C57Bl/6 mice (Kaushal et al. 2002), but failed to grow in rhesus macaque lungs (Mehra et al. 2012). These results strongly suggest that Mtb bacilli experience greater stress in primate tissues relative to their murine counterparts and demonstrate that the NHP model can additionally be used to screen mutant strains of Mtb, particularly in genes that may not be particularly necessary in the mouse model. The NHP model can further be utilized to assay multiple gene knockouts via infection with transposon mutant libraries (Dutta et al. 2010). When such infections were compared between macaques and mice, it was found that the growth of a significantly larger frequency of Tn-mutants was interrupted in the former, relative to the latter model system, again underscoring greater and/or wider variety of stress in macaques. There are some advantages of the mouse model, however, especially in reagent optimization, which may not be possible to perform using macaques. One such contrasting attribute that mice permit is the ability to test novel PET tracers more effectively in mice (Jain 2017; Weinstein et al. 2014). Such screening experiments are impractical to perform in NHPs, because these molecules are synthesized in minute amounts and leave the animal radioactive (generally requiring sacrifice and radioactive waste disposal).
With each bacillus capable of establishing a single granuloma, the macaque lung can be assayed for bacterial persistence, pathogenicity and attenuation, and novel immune phenotypes such as development of inducible bronchus associated lymphoid tissue (Gopal et al. 2013; Kaushal et al. 2015; Slight et al. 2013). Using barcoded strains of Mtb and low-dose, intrabronchial infection, the NHP model was used to track the intralobar, intrapulmonary, and extrapulmonary migration of individual bacilli. This study was the first to show that phagocytic cells traffic from the individual granulomas to the draining lymph nodes where multiple barcodes can often be found within a single bronchial lymph node (Lin et al. 2014). Complementing these approaches, this model has recently been leveraged to study the transcriptome of the pathogen in its intra-granulomatous niches (Hudock et al. 2017) using a microdissection coupled with microarray analysis approach. This result suggests that the genes of the DosR regulon are expressed in virtually all lesion types. These would include both early lesions as well as lesions during the acute infection of lungs. However, the extent of DosR-regulon expression was higher in discrete, caseating lesions from animals with asymptomatic, latent infection. The results parallel with both the recently uncovered phenotype of this regulon in models that generate human-like lung pathology (Gautam et al. 2014; Malhotra et al. 2004; Mehra et al. 2015), but also recent knowledge of Mtb gene expression in humans (Commandeur et al. 2013; Walter et al. 2016). These studies open the possibility that in the near future, the field may be able to leverage recent progress in bar coding, high-throughput sequencing, and RNAseq to better understand single cell Mtb transcriptomics within individual compartments of the various pathological niches. These fundamental experiments should extend our understanding of the physiology of the pathogen in its natural environment, eventually leading to proteome- and lipidome-wide studies.
Modeling Comorbidities
One highly unique aspect of the NHP model of tuberculosis is the ability to co-infect macaques with SIV, as a model of human HIV co-infection (Diedrich et al. 2010; Foreman et al. 2016). The macaque model of SIV infection has been well established since the emergence of AIDS in the 1980s. Much of the initial work in this direction was performed in the “pathogenic” model involving either rhesus or pigtailed macaques (Macaca nemestrina) infected with strains of SIV that resulted in rapid immunodeficiency. The rhesus and cynomolgus model clearly showed that SIV infection results in a significant depletion of CD4+ T cells from the blood, but even more profoundly impacts lymphoid tissues, specifically the gut. This work had a significant impact on the field and has since led to the testing of remarkable therapeutic and vaccination approaches (Apetrei et al. 2012; Haase 2011; McNicholl 2016; Shedlock et al. 2009). There has now emerged a specific interest in the nonpathogenic model of SIV co-infection, using sooty mangabeys (Cercocebus atys), African green monkeys (Cercopithecus aethiops), or other such species that do not exhibit signs of rapid pathogenic infection; however, co-infection with Mtb has not been modeled using the species less susceptible to SIV infection. Mtb/SIV co-infection studies have primarily used rhesus (Foreman et al. 2016; Mehra et al. 2011) or cynomolgus (Diedrich et al. 2010; Mattila et al. 2011) macaques, where results have highlighted that while CD4+ T-cells play a critical role in mediating protection from SIV-induced reactivation, especially by regulating the timing of cytokine signaling (Coleman et al. 2014a), other CD4+ independent mechanisms such CD8+ T-cells and B-cells may compensate for the loss of CD4+ T-cells (Foreman et al. 2016). Work in this area also suggests that treatment of mycobacteria/SIV co-infection by antiretroviral therapy can control reactivation by restoring Mtb-specific T-cells (Shen et al. 2001), although it is possible that all T-cells with all antigen-specificities may not be completely restored. It is expected that in near future, this model will be extended to include mechanistic studies on viral-mediated dissemination to extrathoracic regions, to model immune reconstitution inflammatory syndrome, and to further understand the role of CD4+-independent mechanisms of protection from tuberculosis.
In addition, the NHP model of tuberculosis lends itself well to the study of various other co-infections and comorbidities. In preliminary studies, macaques that received long-term alcohol exposure exhibited reduced levels of key correlates of immunity to tuberculosis and progressed more rapidly to acute disease (Kaushal D, unpublished observations). Alcoholism has long been associated with the progression of human tuberculosis (Lonnroth et al. 2008). It is possible that the impact of chronic cigarette smoke exposure on Mtb infection can also be studied in this model. Recent data indicate that cigarette smoke exposure has a profound effect on macrophage function in the context of Mtb infection (O'Leary et al. 2014), and long-term exposure to cigarette smoke can be modeled in macaques (Polverino et al. 2015). Moreover, it has recently been demonstrated that lysosomal storage impairment impairs macrophage migration causing granuloma breakdown and disease expansion (Berg et al. 2016). It is tempting to speculate that that comparable phenomenon may be at play in smokers. Finally, it is possible that the control of SIV (HIV)-mediated reactivation of LTBI may also similarly depend on macrophage dysfunction (D. Kaushal, M. Kuroda, unpublished observations).
Our understanding of the role played by central metabolic processes in regulating immunity in general and particularly Mtb-specific immune responses has increased. Thus, we understand that diabetes mellitus can impact the progression of tuberculosis via different mechanisms. These data have prompted the use of Metformin, a key DM drug, as an adjunctive therapy for tuberculosis (Singhal et al. 2014). Similarly, the role of key cellular metabolic processes and pathways like sirtuins (Singhal et al., unpublished data), mTOR, poly (ADP-ribose) polymerases, DAMP inhibitors, etc. is now being studied in this model. In this vein, in vivo inhibition of indoleamine dioxygenase (IDO1) activity in rhesus macaques infected with Mtb, using a safe and approved compound, resulted in reduction of disease, pathology, and bacterial levels, concomitant with enhanced Th1 responses (S Mehra et al, unpublished data). IDO1 encodes a tryptophan catabolizing activity as an innate strategy to starve pathogens of this essential amino acid. IDO1 is also a powerful immunosuppressor of T-cell activity, since Trp depletion can impact the development of rapidly proliferating Th1 effectors (Munn et al. 1999). Mtb can, however, synthesize its own Trp (Zhang and Rubin 2013; Zhang et al. 2013), and thus the induction of IDO1 in Mtb-infected lesions is unable to contain bacillary replication; rather it dampens productive immune responses. A IDO1-/- knockout strain in B6 mice did not exhibit any differences in terms of Mtb replication and pathology relative to parental mice. These results again underscore the value of studying Mtb-host interplay in the setting of human-like lung granulomas.
Technological Advances
Recently, technologies have emerged that permit big data acquisition. Similarly, multiple platforms for high-throughput data analysis have become available. NHPs are a valuable model for vetting determining the value of these technologies for diagnostics, therapeutics, and preventative measures. One such technology, fluorine-18 fluorodeoxyglucose (18F-FDG) positron emission tomography with computer tomography (PET/CT), has emerged as a pertinent tool for measuring granuloma dynamics, determining study data output, and consistent multisite comparison of data. Tubercular lesions were originally misdiagnosed as small cell carcinomas or primary lung cancer in human patients being screened by 18F-FDG PET/CT (Hofmeyr et al. 2007). This finding suggested an exciting potential to monitor the progression of tuberculosis disease. 18F-FDG incorporates into cells with increased glycolysis such as inflammatory macrophages and lymphocytes, and this finding led to the first Phase I trial to test the efficacy of PET/CT as a noninvasive biomarker of therapeutic responses (Martinez et al. 2012) and was rapidly followed up with a confirmatory study using the cynomolgus model and a four-drug regimen where the animals were serially imaged using PET/CT to track both disease progression and chemotherapeutic responses (Lin et al. 2013). PET/CT was further established to accurately measure pathology findings when the marmoset, Callithrix jacchus, model of tuberculosis was developed (Via et al. 2013). These initial studies spurred a movement to incorporate PET/CT analysis into all preclinical trials using the NHP model. However, this imaging platform comes with inherent challenges such as establishing facilities compliant for both radiographic material (e.g., 18F-FDG) and biosafety level 3 animal housing (Scanga et al. 2014). Furthermore, data analysis of PET/CT imaging from multiple studies at multiple locations must be standardized to conform to human clinical trial standards. Once implemented, PET/CT imaging can be used as a worldwide standard readout of prophylactic and therapeutic efficacy in vaccine and drug trials using the NHP model.
Due to the integrity of the NHP model, PET/CT imaging can also be utilized to track natural disease progression. Using the cynomolgus model of SIV co-infection, researchers demonstrated that early changes in PET/CT-imaged granulomas predict the long-term clinical outcomes (Coleman et al. 2014b) and established a correlative readout for biomarker discovery (Gideon et al. 2016). The dose and route of infection can further be tailored to address specific questions PET/CT imaging can analyze such as intrapulmonary spread of bacilli, individual granuloma plasticity, and perturbation of granuloma integrity by comorbidities. As more radioactive reagents are made with differing specificities (e.g., 18F-FDG for glycolysis, α-CD3 radioactive-labeled antibody for T-cells), PET/CT imaging can be implemented not only as a standard readout for large-scale drug or vaccine studies, but also to produce data for hypothesis generating, exploratory, or mechanistic NHP studies.
Translating from the NHP Model to Clinical Vaccine Efficacy Trials
The NHP model has served to advance our understanding about the many key aspects of tuberculosis disease and infection, and through the established integrity of the model will continue to serve as a prominent checkpoint of preclinical candidates. While BCG was already 50 years old, the first NHP vaccine study tested intravenous delivery of BCG and monitored the primates after aerosol challenge with the virulent strain H37Rv. Since then macaques have served as the premier model system to study immune responses and efficacy of vaccine candidates. Several of these candidates then moved to human clinical trials (Table 1). The H56 vaccine, a subunit fusion protein of Ag85B, ESAT-6, and Rv2660c delivered with IC31 adjuvant, is one such vaccine candidate that progressed through mouse models into the NHP model. Animals were primed with BCG and subsequently boosted with H56:IC31 and compared to unvaccinated and unboosted animals. H56:IC31-vaccinated animals survived longer and had reduced pathology and bacterial burden after challenge (Lin et al. 2012). Subsequent studies further examined the stimulation of adjuvant by priming with BCG and boosting with H56 mixed with four different adjuvants (Billeskov et al. 2016). These studies provided significant justification for progression into clinical trials where the findings of these NHP studies could be validated. H56:IC31 has now progressed through Phase I clinical trials (Luabeya et al. 2015) and continuing into Phase II prevention of tuberculosis infection trials with extremely promising, albeit preliminary, results. Additionally, ID93, a subunit vaccine comprised of Mtb proteins Rv3619, Rv1813, Rv3620, and Rv2608, adjuvanted with GLA-SE, a synthetic TLR4 agonist, has also been tested in multiple preclinical animal models including macaques (Bertholet et al. 2010). Instead of utilizing the macaque model to test the prevention of tuberculosis disease, the study was designed to use the vaccine therapeutically after antibacterial chemotherapy drug use. After high-dose infection, macaques were treated with a suboptimal chemotherapy regimen for 4 weeks and subsequently vaccinated with ID93:GLA-SE. Animals that were vaccinated therapeutically displayed significantly less bacterial burden and radiological evidence of disease compared to treated but unvaccinated animals (Coler et al. 2013). Subsequently this vaccine candidate has progressed into the first human trial to therapeutically vaccinate infected humans (Clinical trials ID NCT02465216) (Table 1). Another vaccine currently being considered is MTBVAC (Martin et al. 2006; Verreck et al. 2009), which is based on the deletion of the phoP allele (Perez et al. 2001) and fadD26 (Arbues et al. 2013). This strain has proven to be safe in a Phase I trial (Spertini et al. 2015) and is currently being further evaluated in Phase II (Table 1).
Vaccines in clinical trials . | Type . | Phase . | NHP safety study . | NHP efficacy study . | Challenge . |
---|---|---|---|---|---|
M. vaccae (SRL 172) | Whole-cell M. vaccae | III | − | − | |
M. indicus pranii | Whole-cell M. indicus pranii | III | − | − | |
M72/AS01 | Protein/adjuvant | IIb | + | + | 500 CFU Erdman |
Aeras-402 | Viral vector | IIb | + | + | 275 CFU Erdman |
VPM1002 | Live-recombinant BCG | IIa | + | − | |
H4:IC31 | Protein/adjuvant | IIa | − | − | |
H1/H56:IC31 | Protein/adjuvant | IIa | + | + | 500/25 CFU Erdman |
ID93 + GLA-SE | Protein/adjuvant | IIa | + | − | |
MTBVAC | Live-attenuated M. tuberculosis | IIa | + | + | 1000 CFU Erdman |
RUTI | Whole-cell fractionated M. tuberculosis | IIa | − | − | |
Dar-901 | Whole-cell M. obuense | I | − | − | |
Ad5 Ag85A | Viral vector | I | − | − | |
Crucell Ad35 + MVA85A | Viral vector | I | − | − | |
ChAdOc1.85A + MVA85A | Viral vector | I | − | − | |
MVA85A-Aerosol | Viral vector | I | + | − | |
MVA85A-IMX313 | Viral vector/adjuvant | I | + | − | |
TB/FLU-04L | Viral vector | I | |||
Vaccines previously in clinical trials | |||||
BCG-MVA85A boost | Viral vector boost | − | + | + | 1000 CFU Erdman |
Vaccines in clinical trials . | Type . | Phase . | NHP safety study . | NHP efficacy study . | Challenge . |
---|---|---|---|---|---|
M. vaccae (SRL 172) | Whole-cell M. vaccae | III | − | − | |
M. indicus pranii | Whole-cell M. indicus pranii | III | − | − | |
M72/AS01 | Protein/adjuvant | IIb | + | + | 500 CFU Erdman |
Aeras-402 | Viral vector | IIb | + | + | 275 CFU Erdman |
VPM1002 | Live-recombinant BCG | IIa | + | − | |
H4:IC31 | Protein/adjuvant | IIa | − | − | |
H1/H56:IC31 | Protein/adjuvant | IIa | + | + | 500/25 CFU Erdman |
ID93 + GLA-SE | Protein/adjuvant | IIa | + | − | |
MTBVAC | Live-attenuated M. tuberculosis | IIa | + | + | 1000 CFU Erdman |
RUTI | Whole-cell fractionated M. tuberculosis | IIa | − | − | |
Dar-901 | Whole-cell M. obuense | I | − | − | |
Ad5 Ag85A | Viral vector | I | − | − | |
Crucell Ad35 + MVA85A | Viral vector | I | − | − | |
ChAdOc1.85A + MVA85A | Viral vector | I | − | − | |
MVA85A-Aerosol | Viral vector | I | + | − | |
MVA85A-IMX313 | Viral vector/adjuvant | I | + | − | |
TB/FLU-04L | Viral vector | I | |||
Vaccines previously in clinical trials | |||||
BCG-MVA85A boost | Viral vector boost | − | + | + | 1000 CFU Erdman |
Vaccines in clinical trials . | Type . | Phase . | NHP safety study . | NHP efficacy study . | Challenge . |
---|---|---|---|---|---|
M. vaccae (SRL 172) | Whole-cell M. vaccae | III | − | − | |
M. indicus pranii | Whole-cell M. indicus pranii | III | − | − | |
M72/AS01 | Protein/adjuvant | IIb | + | + | 500 CFU Erdman |
Aeras-402 | Viral vector | IIb | + | + | 275 CFU Erdman |
VPM1002 | Live-recombinant BCG | IIa | + | − | |
H4:IC31 | Protein/adjuvant | IIa | − | − | |
H1/H56:IC31 | Protein/adjuvant | IIa | + | + | 500/25 CFU Erdman |
ID93 + GLA-SE | Protein/adjuvant | IIa | + | − | |
MTBVAC | Live-attenuated M. tuberculosis | IIa | + | + | 1000 CFU Erdman |
RUTI | Whole-cell fractionated M. tuberculosis | IIa | − | − | |
Dar-901 | Whole-cell M. obuense | I | − | − | |
Ad5 Ag85A | Viral vector | I | − | − | |
Crucell Ad35 + MVA85A | Viral vector | I | − | − | |
ChAdOc1.85A + MVA85A | Viral vector | I | − | − | |
MVA85A-Aerosol | Viral vector | I | + | − | |
MVA85A-IMX313 | Viral vector/adjuvant | I | + | − | |
TB/FLU-04L | Viral vector | I | |||
Vaccines previously in clinical trials | |||||
BCG-MVA85A boost | Viral vector boost | − | + | + | 1000 CFU Erdman |
Vaccines in clinical trials . | Type . | Phase . | NHP safety study . | NHP efficacy study . | Challenge . |
---|---|---|---|---|---|
M. vaccae (SRL 172) | Whole-cell M. vaccae | III | − | − | |
M. indicus pranii | Whole-cell M. indicus pranii | III | − | − | |
M72/AS01 | Protein/adjuvant | IIb | + | + | 500 CFU Erdman |
Aeras-402 | Viral vector | IIb | + | + | 275 CFU Erdman |
VPM1002 | Live-recombinant BCG | IIa | + | − | |
H4:IC31 | Protein/adjuvant | IIa | − | − | |
H1/H56:IC31 | Protein/adjuvant | IIa | + | + | 500/25 CFU Erdman |
ID93 + GLA-SE | Protein/adjuvant | IIa | + | − | |
MTBVAC | Live-attenuated M. tuberculosis | IIa | + | + | 1000 CFU Erdman |
RUTI | Whole-cell fractionated M. tuberculosis | IIa | − | − | |
Dar-901 | Whole-cell M. obuense | I | − | − | |
Ad5 Ag85A | Viral vector | I | − | − | |
Crucell Ad35 + MVA85A | Viral vector | I | − | − | |
ChAdOc1.85A + MVA85A | Viral vector | I | − | − | |
MVA85A-Aerosol | Viral vector | I | + | − | |
MVA85A-IMX313 | Viral vector/adjuvant | I | + | − | |
TB/FLU-04L | Viral vector | I | |||
Vaccines previously in clinical trials | |||||
BCG-MVA85A boost | Viral vector boost | − | + | + | 1000 CFU Erdman |
While the NHP model of tuberculosis has served a positive predictor and provides clear justification for progression into clinical trials, the model has also served as a checkpoint, limiting unjustified vaccine candidates from progression. Thus, Aeras-402 could not protect rhesus macaques from a lethal 200-cfu Mtb Erdman challenge (Darrah et al. 2014) despite the generation of robust and sustained T cell responses in the lung (Darrah et al. 2014; Hokey et al. 2014). Limiting progression from preclinical status into clinical trials can save significant financial and material resources to be directed to more promising vaccine candidates. MVA85A, a virus vector vaccine, is a vaccine candidate that progressed into Phase I/II clinical trials in the absence of NHP efficacy data. The lack of efficacy of MVA85A in clinical trials (Sharpe et al. 2010; Verreck et al. 2009) has highlighted the value in using animal models to appropriately model clinical trials, and based on the data currently available, NHPs are the most appropriate such system. Such studies in the past could have resulted in a better use of clinical trial resources. While macaque experiments are themselves expensive, these have the potential to result in significant downstream fiscal savings. The authors would like to suggest that efficacy must be demonstrated in the NHP model of tuberculosis before vaccine candidates are approved for clinical trials.
Conclusions
The NHP model of tuberculosis has evolved into an accurate preclinical model of Mtb infection. Similarities in genetics and immune responses allow the prognosis of disease to parallel human infection. There remain subtle differences between the differing species of NHPs, strains of Mtb, and routes of infection/vaccination that all serve a purpose to allow the model to be optimized for testing the hypothesis of the study. While resources, the need for specialty facilities, and financial costs often impede on the widespread use of the NHP model, the accuracy yet plasticity of the model justifies the necessary use for many aspects of tuberculosis research. The NHP model further serves to perform preclinical tests on the advent and repurposing of novel technologies. These technologies can and have already been used to increase our clinical analyses of NHP studies, and may even shorten the time needed to demonstrate efficacy in vaccine or drug trials. Utilization of the NHP model as a checkpoint prior to clinical implementation of vaccines and therapeutics may also serve to save significant financial resources and time invested necessary for clinical trials.
Acknowledgments
The authors would like to acknowledge funding support from PHS (OD011104, RR026006, GM110760, AI089323, AI111943, AI127160, AI127222 and AI058609). This article is dedicated to the memory of Andrew A. Lackner, DVM, PhD, DACVP, long-serving Director of the TNPRC, who passed away on April 2, 2017, after revisions to this article were complete.