Novel machine-learning analysis of SARS-CoV-2 infection in a subclinical nonhuman primate model using radiomics and blood biomarkers.

Detection of the physiological response to severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection is challenging in the absence of overt clinical signs but remains necessary to understand a full subclinical disease spectrum. In this study, our objective was to use radiomics (from computed tomography images) and blood biomarkers to predict SARS-CoV-2 infection in a nonhuman primate model (NHP) with inapparent clinical disease. To accomplish this aim, we built machine-learning models to predict SARS-CoV-2 infection in a NHP model of subclinical disease using baseline-normalized radiomic and blood sample analyses data from SARS-CoV-2-exposed and control (mock-exposed) crab-eating macaques. We applied a novel adaptation of the minimum redundancy maximum relevance (mRMR) feature-selection technique, called mRMR-permute, for statistically-thresholded and unbiased feature selection. Through performance comparison of eight machine-learning models trained on 14 feature sets, we demonstrated that a logistic regression model trained on the mRMR-permute feature set can predict SARS-CoV-2 infection with very high accuracy. Eighty-nine percent of mRMR-permute selected features had strong and significant class effects. Through this work, we identified a key set of radiomic and blood biomarkers that can be used to predict infection status even in the absence of clinical signs. Furthermore, we proposed and demonstrated the utility of a novel feature-selection technique called mRMR-permute. This work lays the foundation for the prediction and classification of SARS-CoV-2 disease severity.

Deep-Learning-Based Whole-Lung and Lung-Lesion Quantification Despite Inconsistent Ground Truth: Application to Computerized Tomography in SARS-CoV-2 Nonhuman Primate Models.

RATIONALE AND OBJECTIVES: Animal modeling of infectious diseases such as coronavirus disease 2019 (COVID-19) is important for exploration of natural history, understanding of pathogenesis, and evaluation of countermeasures. Preclinical studies enable rigorous control of experimental conditions as well as pre-exposure baseline and longitudinal measurements, including medical imaging, that are often unavailable in the clinical research setting. Computerized tomography (CT) imaging provides important diagnostic, prognostic, and disease characterization to clinicians and clinical researchers. In that context, automated deep-learning systems for the analysis of CT imaging have been broadly proposed, but their practical utility has been limited. Manual outlining of the ground truth (i.e., lung-lesions) requires accurate distinctions between abnormal and normal tissues that often have vague boundaries and is subject to reader heterogeneity in interpretation. Indeed, this subjectivity is demonstrated as wide inconsistency in manual outlines among experts and from the same expert. The application of deep-learning data-science tools has been less well-evaluated in the preclinical setting, including in nonhuman primate (NHP) models of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection/COVID-19, in which the translation of human-derived deep-learning tools is challenging. The automated segmentation of the whole lung and lung lesions provides a potentially standardized and automated method to detect and quantify disease. MATERIALS AND METHODS: We used deep-learning-based quantification of the whole lung and lung lesions on CT scans of NHPs exposed to SARS-CoV-2. We proposed a novel multi-model ensemble technique to address the inconsistency in the ground truths for deep-learning-based automated segmentation of the whole lung and lung lesions. Multiple models were obtained by training the convolutional neural network (CNN) on different subsets of the training data instead of having a single model using the entire training dataset. Moreover, we employed a feature pyramid network (FPN), a CNN that provides predictions at different resolution levels, enabling the network to predict objects with wide size variations. RESULTS: We achieved an average of 99.4 and 60.2% Dice coefficients for whole-lung and lung-lesion segmentation, respectively. The proposed multi-model FPN outperformed well-accepted methods U-Net (50.5%), V-Net (54.5%), and Inception (53.4%) for the challenging lesion-segmentation task. We show the application of segmentation outputs for longitudinal quantification of lung disease in SARS-CoV-2-exposed and mock-exposed NHPs. CONCLUSION: Deep-learning methods should be optimally characterized for and targeted specifically to preclinical research needs in terms of impact, automation, and dynamic quantification independently from purely clinical applications.

Toward the determination of sensitive and reliable whole-lung computed tomography features for robust standard radiomics and delta-radiomics analysis in a nonhuman primate model of coronavirus disease 2019.

Purpose: We propose a method to identify sensitive and reliable whole-lung radiomic features from computed tomography (CT) images in a nonhuman primate model of coronavirus disease 2019 (COVID-19). Criteria used for feature selection in this method may improve the performance and robustness of predictive models. Approach: Fourteen crab-eating macaques were assigned to two experimental groups and exposed to either severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) or a mock inoculum. High-resolution CT scans were acquired before exposure and on several post-exposure days. Lung volumes were segmented using a deep-learning methodology, and radiomic features were extracted from the original image. The reliability of each feature was assessed by the intraclass correlation coefficient (ICC) using the mock-exposed group data. The sensitivity of each feature was assessed using the virus-exposed group data by defining a factor R that estimates the excess of variation above the maximum normal variation computed in the mock-exposed group. R and ICC were used to rank features and identify non-sensitive and unstable features. Results: Out of 111 radiomic features, 43% had excellent reliability ( ICC > 0.90 ), and 55% had either good ( ICC > 0.75 ) or moderate ( ICC > 0.50 ) reliability. Nineteen features were not sensitive to the radiological manifestations of SARS-CoV-2 exposure. The sensitivity of features showed patterns that suggested a correlation with the radiological manifestations. Conclusions: Features were quantified and ranked based on their sensitivity and reliability. Features to be excluded to create more robust models were identified. Applicability to similar viral pneumonia studies is also possible.

Single-Shot ChAd3-MARV Vaccine in Modified Formulation Buffer Shows 100% Protection of NHPs.

Marburg virus (MARV) is a virus of high human consequence with a case fatality rate of 24-88%. The global health and national security risks posed by Marburg virus disease (MVD) underscore the compelling need for a prophylactic vaccine, but no candidate has yet reached regulatory approval. Here, we evaluate a replication-defective chimpanzee adenovirus type 3 (ChAd3)-vectored MARV Angola glycoprotein (GP)-expressing vaccine against lethal MARV challenge in macaques. The ChAd3 platform has previously been reported to protect against the MARV-related viruses, Ebola virus (EBOV) and Sudan virus (SUDV), and MARV itself in macaques, with immunogenicity demonstrated in macaques and humans. In this study, we present data showing 100% protection against MARV Angola challenge (versus 0% control survival) and associated production of GP-specific IgGs generated by the ChAd3-MARV vaccine following a single dose of 1 × 1011 virus particles prepared in a new clinical formulation buffer designed to enhance product stability. These results are consistent with previously described data using the same vaccine in a different formulation and laboratory, demonstrating the reproducible and robust protective efficacy elicited by this promising vaccine for the prevention of MVD. Additionally, a qualified anti-GP MARV IgG ELISA was developed as a critical pre-requisite for clinical advancement and regulatory approval.

Bridging Animal and Human Data in Pursuit of Vaccine Licensure.

The FDA Animal Rule was devised to facilitate approval of candidate vaccines and therapeutics using animal survival data when human efficacy studies are not practical or ethical. This regulatory pathway is critical for candidates against pathogens with high case fatality rates that prohibit human challenge trials, as well as candidates with low and sporadic incidences of outbreaks that make human field trials difficult. Important components of a vaccine development plan for Animal Rule licensure are the identification of an immune correlate of protection and immunobridging to humans. The relationship of vaccine-induced immune responses to survival after vaccination and challenge must be established in validated animal models and then used to infer predictive vaccine efficacy in humans via immunobridging. The Sabin Vaccine Institute is pursuing licensure for candidate filovirus vaccines via the Animal Rule and has convened meetings of key opinion leaders and subject matter experts to define fundamental components for vaccine licensure in the absence of human efficacy data. Here, filoviruses are used as examples to review immune correlates of protection and immunobridging. The points presented herein reflect the presentations and discussions during the second meeting held in October 2021 and are intended to address important considerations for developing immunobridging strategies.

Vaccine Licensure in the Absence of Human Efficacy Data.

Clinical vaccine development and regulatory approval generally occurs in a linear, sequential manner: Phase 1: safety, immunogenicity; Phase 2: immunogenicity, safety, dose ranging, and preliminary efficacy; Phase 3: definitive efficacy, safety, lot consistency; and following regulatory approval, Phase 4: post-marketing safety and effectiveness. For candidate filovirus vaccines, where correlates of protection have not been identified, and phase 2 and 3 efficacy of disease prevention trials untenable, large and/or protracted, each trial may span decades, with full licensure expected only after several decades of development. Given the urgent unmet need for new Marburg virus and Ebola Sudan virus vaccines, the Sabin Vaccine Institute hosted a key stakeholder virtual meeting in May 2021 to explore the possibility of licensure by use of an "animal rule-like" licensure process, based on a risk/benefit assessment specific to regional needs and informed by epidemiology. This may be appropriate for diseases where there are no or limited treatment options, and those prone to sporadic outbreaks with high rates of transmission, morbidity, and mortality. The discussion focused on two contexts: licensure within the Ugandan regulatory environment, a high burden country where Ebola vaccine trials are ongoing, and licensure by the United States FDA-a well-resourced regulatory agency.

Medical imaging of pulmonary disease in SARS-CoV-2-exposed non-human primates.

Chest X-ray (CXR), computed tomography (CT), and positron emission tomography-computed tomography (PET-CT) are noninvasive imaging techniques widely used in human and veterinary pulmonary research and medicine. These techniques have recently been applied in studies of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)-exposed non-human primates (NHPs) to complement virological assessments with meaningful translational readouts of lung disease. Our review of the literature indicates that medical imaging of SARS-CoV-2-exposed NHPs enables high-resolution qualitative and quantitative characterization of disease otherwise clinically invisible and potentially provides user-independent and unbiased evaluation of medical countermeasures (MCMs). However, we also found high variability in image acquisition and analysis protocols among studies. These findings uncover an urgent need to improve standardization and ensure direct comparability across studies.

On-Demand Patient-Specific Phenotype-to-Genotype Ebola Virus Characterization.

Biosafety, biosecurity, logistical, political, and technical considerations can delay or prevent the wide dissemination of source material containing viable virus from the geographic origin of an outbreak to laboratories involved in developing medical countermeasures (MCMs). However, once virus genome sequence information is available from clinical samples, reverse-genetics systems can be used to generate virus stocks de novo to initiate MCM development. In this study, we developed a reverse-genetics system for natural isolates of Ebola virus (EBOV) variants Makona, Tumba, and Ituri, which have been challenging to obtain. These systems were generated starting solely with in silico genome sequence information and have been used successfully to produce recombinant stocks of each of the viruses for use in MCM testing. The antiviral activity of MCMs targeting viral entry varied depending on the recombinant virus isolate used. Collectively, selecting and synthetically engineering emerging EBOV variants and demonstrating their efficacy against available MCMs will be crucial for answering pressing public health and biosecurity concerns during Ebola disease (EBOD) outbreaks.

Formulation, Stability, Pharmacokinetic, and Modeling Studies for Tests of Synergistic Combinations of Orally Available Approved Drugs against Ebola Virus In Vivo.

Outbreaks of Ebola ebolavirus (EBOV) have been associated with high morbidity and mortality. Milestones have been reached recently in the management of EBOV disease (EVD) with licensure of an EBOV vaccine and two monoclonal antibody therapies. However, neither vaccines nor therapies are available for other disease-causing filoviruses. In preparation for such outbreaks, and for more facile and cost-effective management of EVD, we seek a cocktail containing orally available and room temperature stable drugs with strong activity against multiple filoviruses. We previously showed that (bepridil + sertraline) and (sertraline + toremifene) synergistically suppress EBOV in cell cultures. Here, we describe steps towards testing these combinations in a mouse model of EVD. We identified a vehicle suitable for oral delivery of the component drugs and determined that, thus formulated the drugs are equally active against EBOV as preparations in DMSO, and they maintain activity upon storage in solution for up to seven days. Pharmacokinetic (PK) studies indicated that the drugs in the oral delivery vehicle are well tolerated in mice at the highest doses tested. Collectively the data support advancement of these combinations to tests for synergy in a mouse model of EVD. Moreover, mathematical modeling based on human oral PK projects that the combinations would be more active in humans than their component single drugs.

Development and Characterization of a cDNA-Launch Recombinant Simian Hemorrhagic Fever Virus Expressing Enhanced Green Fluorescent Protein: ORF 2b' Is Not Required for In Vitro Virus Replication.

Simian hemorrhagic fever virus (SHFV) causes acute, lethal disease in macaques. We developed a single-plasmid cDNA-launch infectious clone of SHFV (rSHFV) and modified the clone to rescue an enhanced green fluorescent protein-expressing rSHFV-eGFP that can be used for rapid and quantitative detection of infection. SHFV has a narrow cell tropism in vitro, with only the grivet MA-104 cell line and a few other grivet cell lines being susceptible to virion entry and permissive to infection. Using rSHFV-eGFP, we demonstrate that one cricetid rodent cell line and three ape cell lines also fully support SHFV replication, whereas 55 human cell lines, 11 bat cell lines, and three rodent cells do not. Interestingly, some human and other mammalian cell lines apparently resistant to SHFV infection are permissive after transfection with the rSHFV-eGFP cDNA-launch plasmid. To further demonstrate the investigative potential of the infectious clone system, we introduced stop codons into eight viral open reading frames (ORFs). This approach suggested that at least one ORF, ORF 2b', is dispensable for SHFV in vitro replication. Our proof-of-principle experiments indicated that rSHFV-eGFP is a useful tool for illuminating the understudied molecular biology of SHFV.

Animal models for COVID-19.

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is the aetiological agent of coronavirus disease 2019 (COVID-19), an emerging respiratory infection caused by the introduction of a novel coronavirus into humans late in 2019 (first detected in Hubei province, China). As of 18 September 2020, SARS-CoV-2 has spread to 215 countries, has infected more than 30 million people and has caused more than 950,000 deaths. As humans do not have pre-existing immunity to SARS-CoV-2, there is an urgent need to develop therapeutic agents and vaccines to mitigate the current pandemic and to prevent the re-emergence of COVID-19. In February 2020, the World Health Organization (WHO) assembled an international panel to develop animal models for COVID-19 to accelerate the testing of vaccines and therapeutic agents. Here we summarize the findings to date and provides relevant information for preclinical testing of vaccine candidates and therapeutic agents for COVID-19.

Characteristic and quantifiable COVID-19-like abnormalities in CT- and PET/CT-imaged lungs of SARS-CoV-2-infected crab-eating macaques (Macaca fascicularis).

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is causing an exponentially increasing number of coronavirus disease 19 (COVID-19) cases globally. Prioritization of medical countermeasures for evaluation in randomized clinical trials is critically hindered by the lack of COVID-19 animal models that enable accurate, quantifiable, and reproducible measurement of COVID-19 pulmonary disease free from observer bias. We first used serial computed tomography (CT) to demonstrate that bilateral intrabronchial instillation of SARS-CoV-2 into crab-eating macaques (Macaca fascicularis) results in mild-to-moderate lung abnormalities qualitatively characteristic of subclinical or mild-to-moderate COVID-19 (e.g., ground-glass opacities with or without reticulation, paving, or alveolar consolidation, peri-bronchial thickening, linear opacities) at typical locations (peripheral>central, posterior and dependent, bilateral, multi-lobar). We then used positron emission tomography (PET) analysis to demonstrate increased FDG uptake in the CT-defined lung abnormalities and regional lymph nodes. PET/CT imaging findings appeared in all macaques as early as 2 days post-exposure, variably progressed, and subsequently resolved by 6-12 days post-exposure. Finally, we applied operator-independent, semi-automatic quantification of the volume and radiodensity of CT abnormalities as a possible primary endpoint for immediate and objective efficacy testing of candidate medical countermeasures.

Reporter Assays for Ebola Virus Nucleoprotein Oligomerization, Virion-Like Particle Budding, and Minigenome Activity Reveal the Importance of Nucleoprotein Amino Acid Position 111.

For highly pathogenic viruses, reporter assays that can be rapidly performed are critically needed to identify potentially functional mutations for further study under maximal containment (e.g., biosafety level 4 [BSL-4]). The Ebola virus nucleoprotein (NP) plays multiple essential roles during the viral life cycle, yet few tools exist to study the protein under BSL-2 or equivalent containment. Therefore, we adapted reporter assays to measure NP oligomerization and virion-like particle (VLP) production in live cells and further measured transcription and replication using established minigenome assays. As a proof-of-concept, we examined the NP-R111C substitution, which emerged during the 2013‒2016 Western African Ebola virus disease epidemic and rose to high frequency. NP-R111C slightly increased NP oligomerization and VLP budding but slightly decreased transcription and replication. By contrast, a synthetic charge-reversal mutant, NP-R111E, greatly increased oligomerization but abrogated transcription and replication. These results are intriguing in light of recent structures of NP oligomers, which reveal that the neighboring residue, K110, forms a salt bridge with E349 on adjacent NP molecules. By developing and utilizing multiple reporter assays, we find that the NP-111 position mediates a complex interplay between NP's roles in protein structure, virion budding, and transcription and replication.

A Novel Ebola Virus VP40 Matrix Protein-Based Screening for Identification of Novel Candidate Medical Countermeasures.

Filoviruses, such as Ebola virus and Marburg virus, are of significant human health concern. From 2013 to 2016, Ebola virus caused 11,323 fatalities in Western Africa. Since 2018, two Ebola virus disease outbreaks in the Democratic Republic of the Congo resulted in 2354 fatalities. Although there is progress in medical countermeasure (MCM) development (in particular, vaccines and antibody-based therapeutics), the need for efficacious small-molecule therapeutics remains unmet. Here we describe a novel high-throughput screening assay to identify inhibitors of Ebola virus VP40 matrix protein association with viral particle assembly sites on the interior of the host cell plasma membrane. Using this assay, we screened nearly 3000 small molecules and identified several molecules with the desired inhibitory properties. In secondary assays, one identified compound, sangivamycin, inhibited not only Ebola viral infectivity but also that of other viruses. This finding indicates that it is possible for this new VP40-based screening method to identify highly potent MCMs against Ebola virus and its relatives.

Pregnancy level of estradiol attenuated virus-specific humoral immune response in H5N1-infected female mice despite inducing anti-inflammatory protection.

Estradiol, a major female steroid produced during pregnancy, has been reported to protect ovariectomized animals against H1N1 influenza infections via its anti-inflammatory effects. However, it remains unclear why pregnant women with high gestational estradiol levels are highly susceptible to influenza infections. This study was aimed to investigate the effects of pregnancy level of estradiol on female immunity against H5N1 infection in Balb/c mice. A sex-dependent susceptibility to H5N1 infection (higher morbidity and higher mortality) was observed in both pregnant and non-pregnant female mice as compared to male mice. Subcutaneous implantation of estradiol pellets increased serum estradiol concentrations of non-pregnant female mice to the pregnancy level. These mice were protected from H5N1 infection through downregulation of pulmonary pro-inflammatory cytokines. However, the production of virus-specific antibodies after infection was significantly delayed in estradiol-implanted mice when compared to placebos. Virus-specific IgG-secreting and IL-4-secreting cells were also reduced in estradiol-implanted mice. Similarly, lower antibody titers to seasonal vaccine antigens were found in pregnant women as compared to non-pregnant females without hormone usage. Our results indicate that estradiol levels equivalent to those found during pregnancy have divergent effects on female immunity against influenza, highlighting the importance of vaccination during pregnancy to prevent severe influenza infections.

Human, Nonhuman Primate, and Bat Cells Are Broadly Susceptible to Tibrovirus Particle Cell Entry.

In 2012, the genome of a novel rhabdovirus, Bas-Congo virus (BASV), was discovered in the acute-phase serum of a Congolese patient with presumed viral hemorrhagic fever. In the absence of a replicating virus isolate, fulfilling Koch's postulates to determine whether BASV is indeed a human virus and/or pathogen has been impossible. However, experiments with vesiculoviral particles pseudotyped with Bas-Congo glycoprotein suggested that BASV particles can enter cells from multiple animals, including humans. In 2015, genomes of two related viruses, Ekpoma virus 1 (EKV-1) and Ekpoma virus 2 (EKV-2), were detected in human sera in Nigeria. Isolates could not be obtained. Phylogenetic analyses led to the classification of BASV, EKV-1, and EKV-2 in the same genus, Tibrovirus, together with five biting midge-borne rhabdoviruses [i.e., Beatrice Hill virus (BHV), Bivens Arm virus (BAV), Coastal Plains virus (CPV), Sweetwater Branch virus (SWBV), and Tibrogargan virus (TIBV)] not known to infect humans. Using individual recombinant vesiculoviruses expressing the glycoproteins of all eight known tibroviruses and more than 75 cell lines representing different animal species, we demonstrate that the glycoproteins of all tibroviruses can mediate vesiculovirus particle entry into human, bat, nonhuman primate, cotton rat, boa constrictor, and Asian tiger mosquito cells. Using four of five isolated authentic tibroviruses (i.e., BAV, CPV, SWBV, and TIBV), our experiments indicate that many cell types may be partially resistant to tibrovirus replication after virion cell entry. Consequently, experimental data solely obtained from experiments using tibrovirus surrogate systems (e.g., vesiculoviral pseudotypes, recombinant vesiculoviruses) cannot be used to predict whether BASV, or any other tibrovirus, infects humans.

Differential Effects of Prior Influenza Exposures on H3N2 Cross-reactivity of Human Postvaccination Sera.

BACKGROUND: Effectiveness of seasonal influenza vaccines mainly depends upon how well vaccine strains represent circulating viruses; mismatched strains can lead to reduced protection. Humans have complex influenza exposure histories that increase with age, which may lead to different postvaccination responses to emerging influenza variants. Recent observational studies also suggest that prior vaccination may influence the performance of current seasonal vaccines. METHODS: To elucidate the effects of age and influenza preexposures on cross-reactivity of vaccination-induced human antibodies, we generated antigenic maps based on postvaccination hemagglutination inhibition titers against representative H3 variants circulating during the 2015-2016, 2014-2015, and 2012-2013 influenza seasons. RESULTS: Antigenic maps determined using sera from subjects 18-64 and ≥65 years of age correlated well with each other but poorly with those determined using sera from children. Antigenic maps derived from human postvaccination sera with H1 influenza preexposure also correlated poorly with those derived from sera with neither H1 nor type B influenza preexposure, and the correlation lessened considerably over time. In contrast, antigenic maps derived from human postvaccination sera with only type B influenza preexposure consistently showed good correlation with those derived from sera with neither H1 nor type B influenza preexposure. CONCLUSIONS: Our results suggest an age-specific difference in human postvaccination responses. Our findings also suggest that prior exposure to H1 or type B influenza may differentially affect cross-reactivity of vaccination-induced H3-specific hemagglutination inhibition antibody responses, and consequently might affect vaccine effectiveness. Our study highlights the need to study the impact of prior exposure on influenza vaccine performance.