As part of the Project Bioshield Act of 2003, the Department of Homeland Security (DHS) identified hemorrhagic fever viruses as a class A Material Threat to the population of the United States (US). Among the viruses in this class are the filovirus family, a class of filamentous RNA viruses including Ebola virus (EBOV), Sudan virus (SUDV), and Marburg virus (MARV) responsible for sporadic limited outbreaks and a high mortality rate [1
]. While major advances have been made in medical countermeasures against EBOV [1
] in the context of the recent outbreaks, medical countermeasures against MARV are much further behind. In the absence of a predictable or large outbreak, we expect any vaccine candidates against MARV will require non-traditional regulatory pathways for approval, such as the FDA Animal Rule, requiring well-characterized and reproducible animal models that are likely to be predictive of clinical benefit.
MARV was first identified in 1967 with cases in Germany and the former Yugoslavia when infected monkeys were imported from Uganda. Since that time, 12 different outbreaks have occurred with the number of cases per outbreak ranging from one to 252 [3
]. Outbreaks in humans are generally initiated after spillover events from animal hosts such as the fruit bat, Rousettus aegypticus
, as summarized with a characterization of MARV infections in a fruit bat model [4
]. MARV infection causes a mild disease in the R. aegypticus
model but causes severe disease in nonhuman primates (NHP) experimentally infected with MARV, and humans upon spillover events [4
A 2015 review provides the most comprehensive comparison between human MARV disease and results observed in different NHP challenge models [5
]. The human data are largely derived from 35 patients diagnosed with MARV infection and treated at modern medical facilities. Clinical signs generally appeared within 3 to 9 days post-exposure and progressively worsened to include diarrhea, nausea, vomiting, fever, and exanthema or enanthema. The condition of patients usually worsened during the second week, with hemorrhaging in some cases. Mortality ranged from 20−30 percent in individuals receiving intensive care, to up to 80–90 percent in rural African settings. Work from various groups has shown that the cynomolgus macaque model of MARV disease recapitulates what is seen in humans. In addition, 100% mortality is seen in animals that did not receive some type of treatment in animals challenged intramuscularly with 1000 plaque forming units (PFU). Death occurs in animals within 7−10 days (Table S3
) and is preceded by coagulopathy, fever, anorexia, changes in hematology and serum chemistry, and petechiation [6
]. Cynomolgus macaques have been the most common animal model for evaluation of vaccines against filovirus diseases. Importantly, efficacy against disease caused by EBOV in this model correlated closely with clinical efficacy observed, both in terms of levels of protection and onset of signs of disease [10
Here, we further characterized the cynomolgus macaque MARV intramuscular (IM) infection model by conducting a well-documented/controlled natural history study. The study was designed and executed to fulfill the FDA guidance for development of products under the Animal Rule which specifies that such studies are to be completed in advance of utilizing models for well-controlled pharmacokinetic/pharmacodynamics and efficacy studies that may be used to support product licensure.
2. Materials and Methods
2.1. Ethics Statement
This study complied with Final Rules of the Animal Welfare Act regulations (9 CFR Parts 1, 2, and 3) and Guide for the Care and Use of Laboratory Animals: Eighth Edition (Institute of Laboratory Animal Resources, National Academies Press, 2011; the Guide). This study was conducted in UTMB’s AAALAC (Association for the Assessment and Accreditation of Laboratory Animal Care)-accredited facility and was approved by UTMB’s Institutional Animal Care and Use Committee (protocol number 2006068, approved on 1 June 2020).
Nine male and nine female, experimentally naïve, cynomolgus macaques (Macaca fascicularis, Asiatic origin bred in Vietnam) weighing between 2.4−3.2 kg and 2.5−3.2 years of age were procured from Envigo (Alice, TX, USA). Prior to shipment the NHPs were verified negative for evidence of pre-existing immunity to Reston virus (RESTV), EBOV, SUDV, and MARV in addition to standard screening tests including tuberculosis, simian immunodeficiency virus, simian retrovirus 1 and 2, simian T-lymphotropic virus-1, macacine herpesvirus 1 (Herpes B virus), and Trypanosoma. All animals were housed in open stainless-steel standard NHP caging, provided with Certified Primate Diet (PMI, Inc., New York, NY, USA), and water was provided ad libitum through an automatic watering system. To promote and enhance the psychological well-being, both food and environmental enrichment were provided to each NHP.
Prior to placement on study, NHPs were surgically implanted with DST micro-T implantable temperature loggers (Star-Oddi, Gardabaer, Iceland). The temperature loggers were programmed to take measurements every 15 min. NHPs were anesthetized with Ketamine (5−20 mg/kg IM) prior to all procedures.
Animals were randomized by a biostatistician to two groups: MARV-exposed (N = 12) and mock-exposed (N = 6). Groups were stratified by sex and balanced by body weight.
Passage 2 of Marburg virus/H. sapiens-tc/AGO/2005/Angola-200501379 was obtained from BEI Resources (Lot number 200501379; Manassas, VA, USA) and was not passaged further. This well-characterized virus was diluted in Hank’s Balanced Salt Solution containing 2% heat-inactivated fetal bovine serum (HBSS/2% HI-FBS) such that animals received an intramuscular dose corresponding to 1000 PFU of MARV Angola in 0.5 mL. HBSS/2% HI-FBS was used for the mock-challenged controls. The administered challenge dose was confirmed by plaque assay of the challenge suspension collected before the challenge of the first animal and after the last animal was challenged.
2.5. Clinical Observation and Scoring
Animals were observed, at minimum, twice daily (6–8 h apart during the light cycle) and scored using the system shown in Table S1
. A score of 0–3 indicated that no intervention was needed; a score of ≥4 (or ≥3 in any single parameter) required additional monitoring of at least once in the evening, 4–6 h after the final late afternoon check; a score of 10 or greater is required euthanasia.
Any animals exhibiting signs consistent with significant distress/moribundity (score of 10 or greater) were evaluated for euthanasia by a veterinarian as per the approved IACUC protocol. Animals that survived until the scheduled study termination (28 days post-challenge) were humanely euthanized. Animals that required euthanasia were sedated as previously described and euthanized by administration of a pentobarbital-based euthanasia solution (e.g., Euthasol) via intravenous or intracardiac administration according to the AVMA Guidelines for the Euthanasia of Animals.
2.7. Blood Collection and Processing
Blood was collected from animals on Days −4, 0, 3, 5, 7, 10, 14, 21, and 28 relative to MARV challenge and at the time of euthanasia (terminal sample). The femoral vein was used for all scheduled biosampling events and intracardiac blood was collected at terminal time points or at the end of the study (Day 28). Blood was collected into serum separator tubes and tubes containing anticoagulant, specifically ethylenediaminetetraacetic acid (EDTA) and sodium citrate. Serum was aliquoted and stored at ≤−65 °C, while blood collected for hematology and coagulation was analyzed within 8 or 2 h, respectively.
2.8. Anti-MARV GP IgG ELISA
Serum collected for anti-MARV GP IgG ELISA analysis was inactivated by exposure with 5 MRads of gamma radiation using a previously validated method of inactivation prior to shipment to Battelle. The ELISA was conducted essentially as described in Rudge et al., 2019 [11
] using purified recombinant MARV GP with amino acid sequence corresponding to the GP from MARV Ci67 (Advanced Bioscience Laboratories, Rockville, MD, USA). Test samples were evaluated in the ELISA at a 1:50 starting dilution and reportable anti-MARV GP IgG concentrations were calculated using a qualified human serum reference standard with MARV GP-binding antibodies.
2.9. Clinical Chemistry
Clinical chemistry analysis was conducted on harvested serum using the Abaxis Piccolo Xpress Chemistry Analyzer in conjunction with the BioChemistry Panel Plus reagent discs to determine the levels of alanine aminotransferase (ALT), albumin (ALB), alkaline phosphatase (ALP), amylase (AMY), aspartate aminotransferase (AST), c-reactive protein (CRP), calcium (CA), creatinine (CRE), gamma glutamyltransferase (GGT), glucose (GLU), total protein (TP), blood urea nitrogen (BUN), and uric acid (UA).
Blood collected in sodium citrate tubes was used to measure Activated Partial Thromboplastin Time (aPTT) and Prothrombin Time (PT) using the IDEXX Coag DxTM Analyzer. Samples were analyzed within 2 h of collection using IDEXX Coag Dx PT and aPTT cartridges.
Hematology analysis was conducted on EDTA blood using the Abaxis VetScan HM5® Hematology Analyzer to measure the following parameters in whole blood collected in EDTA tubes: White blood cell concentration (WBC), lymphocyte concentration and percentage (LYM), monocyte concentration and percentage (MON), neutrophil concentration and percentage (NEU), basophil concentration and percentage (BAS), eosinophil concentration and percentage (EOS), red blood cell concentration (RBC), hemoglobin (HGB), hematocrit (HCT), mean corpuscular volume (MCV), mean corpuscular hemoglobin (MCH), mean corpuscular hemoglobin concentration (MCHC), red cell distribution width (RDW), platelet concentration (PLT), mean platelet volume (MPV), platelet hematocrit (PCT), and platelet distribution width (PDW).
2.12. Viral Load: Plaque Assay
Harvested serum was stored frozen (≤−65 °C) from the time of processing until plaque assay analysis. On the day(s) of plaque assay analysis, samples were thawed under ambient conditions, and assayed as described by Shurtleff et al. [12
]. Crystal violet stain was used to visualize plaques and results were reported as PFU/mL of serum.
2.13. Viral Load: qRT-PCR
On the day of collection, harvested serum (0.05 mL) was added to TRIzol® LS (5X volume; i.e., 0.25 mL) and stored at ≤−65 °C until RNA exaction. RNA was extracted from samples using the Zymo Research Direct-zol™ RNA MiniPrep kit (Zymo Research, Irvine, CA, USA). For sample quantification, each assay plate contained a standard curve prepared using MARV VP40 gene synthetic RNA (1.0 × 103 to 1.0 × 1010 genome equivalents/µL [GEq/µL] in duplicate wells). For the qRT-PCR, QuantiFast Probe RT-PCR Master Mix and QuantiFast RT Mix (Qiagen, Ilden, Germany) were used in conjunction with Forward primer: 5′- CCAgTTCCAgCAATTACAATACATACA-3′, Reverse primer: 5′- gCACCgTggTCAgCATAAggA-3′ and Probe: 5′-6FAM- CAATACCTTAACCCCC-MGBNFQ-3′. Primers and probe targeted the VP40 gene from MARV (GenBank accession no. DQ447660). The qRT-PCR was conducted on a Bio-Rad CFX96TM Real-Time PCR.
2.14.1. Necropsy and Gross Pathology
A gross necropsy was conducted on all animals that succumbed to disease or that lived until the end of study (Day 28). Gross necropsies included examinations of the external surface of the body, all external orifices, the thoracic and abdominal cavities, and their contents.
2.14.2. Tissue Collection for Histopathology
The following tissues from all animals were collected during necropsy and placed in 10% neutral buffered formalin for at least 21 days: lymph nodes (axillary from infected arm, mediastinal, mesenteric, and inguinal), challenge site (skin and underlying muscle), adrenal glands, stomach with pyloris, jejunum, duodenum, ileum, transverse colon, rectum, gall bladder, liver, spleen, kidneys, heart, lungs, and any gross lesions. All tissues were processed to slides and stained with hematoxylin and eosin and evaluated by a Board-Certified Veterinary Pathologist. Histopathologic grades were assigned according to the following scale: minimal, mild, moderate, and marked (Table S6
2.15. Statistical Analyses
All statistical analyses were conducted using SAS® (version 9.4) on the 64-bit platform or R. (version 3.6.3). All results are reported at the 0.05 level of significance. Mortality rates and exact 95% confidence intervals were calculated for each group using Clopper-Pearson 95% confidence intervals. A one-sided Boschloo’s test was performed to assess whether mortality outcome in the MARV-exposed group was greater than that of the mock-exposed control group. Time to death data were analyzed and compared between the MARV-exposed and the mock-exposed control groups. Kaplan–Meier curves were plotted, and median times to death were estimated with 95% confidence intervals. A log rank test was used to determine if there was a significant difference between the groups. All animals were included in the analysis; those surviving to end of study (termination on Day 28) were censored at the time of terminal sacrifice.
The maximum daily clinical scores for each study animal were used in the analysis. Summary statistics including means, standard deviations, minimums, medians, and maximums were calculated by group and study day. Median clinical scores were plotted by group and study day. In addition, the final clinical scores were plotted. For animals found dead, the final clinical score was the most recent clinical score, which may have been collected the previous day.
Analysis of variance (ANOVA) models were fitted separately to each clinical chemistry, hematology, and coagulation parameter with effects for group, study time point, and the interaction between group and study time point used to assess the model assumption of normality and to identify potential outliers. Deleted studentized residuals were calculated for each observation. If the absolute value of the deleted studentized residual was greater than 4, then the observation was considered a potential outlier. Each animal’s Day 0 value served as its own baseline in the shift from baseline analysis.
The development of effective countermeasures against rare diseases with high mortality and consequence is the responsibility of the Biomedical Advanced Research and Development Authority (BARDA). The development of such medical countermeasures relies on the evaluation of efficacy using animal models following the precepts of the FDA Animal Rule. The FDA Animal Rule was created to address the difficulty in developing medical countermeasures against diseases where clinical trials to demonstrate efficacy are either not feasible or not ethical. The central underpinning of efficacy evaluation is the use of reproducible, predictive animal models to demonstrate countermeasure efficacy and establish dose. This manuscript describes a BARDA-supported effort to determine if an infection of cynomolgus macaques can provide a model for evaluation of Marburg virus countermeasures. In this study, 12 animals were infected intramuscularly with a targeted 1000 PFU of MARV, and changes in clinical parameters were compared to 6 unchallenged time-matched controls as well as their own baseline data. There were two main aims with this experimental design. One aim was to determine, through the measure of physiological parameters (serum chemistry, hematology, coagulation times, body temperature) and viremia, if the infection was reproducible and could provide a faithful model for countermeasure evaluation. The second objective was to see if the kinetics of infection caused changes from baseline values that could serve as triggers for medical intervention. The data from this study confirm the cynomolgus macaque model as an appropriate model for evaluation of MARV countermeasures and the clinical presentation is reproducible when compared to published studies [5
]. The rapid time course of the disease in this model after onset of clinical signs makes the identification of triggers for medical intervention problematic, implying that time-based intervention, as used for the models to evaluate Ebola countermeasures, is most appropriate [13
]. Development of medical countermeasures against filoviruses and other high-consequence pathogens poses an ethical challenge to the traditional pathway to regulatory approval through demonstration of efficacy in well-designed and monitored clinical trials. The FDA Animal Rule provides the foundation for an alternative regulatory pathway by use of an appropriate and well-characterized animal model with the following requirements: (1) a reasonably well-understood mechanism of pathogenesis and its prevention or substantial reduction by the countermeasure, (2) demonstration of the effect of the countermeasure in one or more animal species sufficiently well characterized to predict the response in humans, (3) the animal study endpoint to be clearly related to the desired benefit in humans, and (4) sufficient data to select an efficacious dose in humans. Given the possibility that prophylactics and therapeutics developed against MARV may be licensed under the Animal Rule or the accelerated approval pathway, BARDA chose to develop a robust challenge model with uniform lethality that can provide conservative estimates of effectiveness for bridging of non-clinical immunogenicity and survival data to human immunogenicity data.
The development of products effective against EBOV used animal models with a 1000 PFU intramuscular challenge to demonstrate efficacy and establish the appropriate dose to establish a severe challenge model [20
]. Our primary goal in this study was to determine if an analogous model using MARV as the challenge agent was similarly adaptable for evaluation of countermeasures. Analogous to the observation in cynomolgus macaques infected with either Ebola or Sudan filoviruses [20
], a short refractory period with few signs of disease was followed by increasing clinical scores starting at Day 5 post challenge for some animals and increasing rapidly for all infected animals on subsequent days. The median time to death was 7.3 days post-challenge.
One of the traits of filovirus infection is a sustained fever. In the current study, elevated temperature was detected between day 3 and day 4 post-challenge in all infected animals and continued until very late in infection, where a loss of thermal control resulted in a dramatic temperature drop. This loss of thermal regulation and rapid drop in temperature was observed in 8 of 10 animals at time of euthanasia. The two animals that succumbed to disease had a precipitous temperature drop overnight, although it is difficult to separate the temperature drop that occurred prior to death due to loss of thermal control with the drop in temperature that occurred after death. In this study, the temperature was obtained post in-life phase by downloading the information from temperature loggers. The monitoring of temperature in real-time may provide additional data to establish appropriate euthanasia criteria in future studies.
The description of MARV infection in human and nonhuman primates can be found in Glaze et al. [5
]. In the study described in this manuscript, changes in clinical chemistry and hematologic and coagulation times were used to demonstrate that the infection of cynomolgus macaques with this preparation of MARV Angola at 1000 PFU faithfully recapitulates the disease characteristics observed in human infections. In every infected animal, evidence for severe Marburg disease became clear at later times post infection with the levels of enzymes indicative of systemic inflammation, liver damage, coagulopathy, and necrosis increasing at or later than 5 days after infection. These symptoms are observed in filovirus infection of humans, suggesting that the infection in the nonhuman primate resembles the human infection sufficiently well to provide a model which will be predictive of clinical benefit for the evaluation of medical countermeasures.
The presence of circulating MARV was determined in two ways: blood samples were screened for infectious virus by plaque assay and for the level of genomic RNA by qRT-PCR. A low level of infectious virus was detectable in all infected animals on Day 3 post-challenge with a rapid increase in viral titer until euthanasia. The presence of RNA copies of the MARV VP40 gene was measured by qRT-PCR on RNA extracted from blood draws. Interestingly, the first evidence for MARV RNA was detected only in two animals on Day 3 post-challenge and on Day 5 post-challenge for the other 10 infected animals. The Day 3 plaque and qRT-PCR results suggests the plaque assay is more sensitive in detecting viremia. It is important to note that neither assay has undergone a formal qualification or validation. Future work will investigate the sensitivity of the qRT-PCR assay for use on Animal Rule studies. However, both measures of viral load rapidly increased at each subsequent blood draw until euthanasia was reached.
The goal of this study was to demonstrate the feasibility of an NHP model for the evaluation of MARV countermeasures using cynomolgus macaques infected with 1000 PFU of MARV Angola via intramuscular injection. This study described the survival, clinical chemistry, hematology, pathology, coagulation, and viremia in infected animals while determining the time from exposure to onset of disease manifestations and the time course, frequency, and magnitude of the manifestations. Evidence of liver damage was apparent by the significant increases in ALT and AST in MARV-infected animals over the mock-exposed group by Day 3. Evidence of kidney damage was also apparent with increased levels of both BUN and CRE.
The portfolio of products to address a MARV outbreak should include both prophylactic vaccines and therapeutic antivirals to provide maximum outbreak control. The model as presently designed in this study is suitable for evaluation of vaccine efficacy with immunization and subsequent challenge. The data from this study provide many candidate signs of disease that can be monitored for evidence of viral infection. The primary endpoint will be survival but secondary endpoints measuring the amount of breakthrough can utilize many of the virus induced physiological changes described in this study. The evaluation of therapeutic countermeasures requires delayed medical intervention until an infection is well established to mimic the stage of infection representative of that encountered in the clinic. The evaluation of some therapeutic countermeasures using the Animal Rule has utilized biomarker detection as a sign for medical intervention [24
], while other models have relied on time-based intervention [25
]. In this model, the rapid disease progression and absence of an easily detectable biomarker unique for MARV infection prior to Day 5 after challenge, suggests a time-based intervention will be most feasible. The high dose and infectivity of MARV Angola [9
] contribute to the rapid disease course and high mortality in the model, but the evaluation of countermeasures in severe challenge models ensures that the countermeasures that succeed in such models will be effective in the field when used to treat this disease with high mortality in humans.