Forecasting the risk of pathogen spillover from reservoir populations of wild or domestic animals is essential for the effective deployment of interventions such as wildlife vaccination or culling. Due to the sporadic nature of spillover events and limited availability of data, developing and validating robust, spatially explicit, predictions is challenging. Recent efforts have begun to make progress in this direction by capitalizing on machine learning methodologies. An important weakness of existing approaches, however, is that they generally rely on combining human and reservoir infection data during the training process and thus conflate risk attributable to the prevalence of the pathogen in the reservoir population with the risk attributed to the realized rate of spillover into the human population. Because effective planning of interventions requires that these components of risk be disentangled, we developed a multi-layer machine learning framework that separates these processes. Our approach begins by training models to predict the geographic range of the primary reservoir and the subset of this range in which the pathogen occurs. The spillover risk predicted by the product of these reservoir specific models is then fit to data on realized patterns of historical spillover into the human population. The result is a geographically specific spillover risk forecast that can be easily decomposed and used to guide effective intervention. Applying our method to Lassa virus, a zoonotic pathogen that regularly spills over into the human population across West Africa, results in a model that explains a modest but statistically significant portion of geographic variation in historical patterns of spillover. When combined with a mechanistic mathematical model of infection dynamics, our spillover risk model predicts that 897,700 humans are infected by Lassa virus each year across West Africa, with Nigeria accounting for more than half of these human infections.
This code is used to generate the figures displayed in the Journal of Applied Ecology paper, "When to vaccinate a fluctuating wildlife population: is timing everything?".
Transmissible vaccines have the potential to revolutionize infectious disease control by reducing the vaccination effort required to protect a population against a disease. Recent efforts to develop transmissible vaccines focus on recombinant transmissible vaccine designs (RTVs) because they pose reduced risk if intra-host evolution causes the vaccine to revert to its vector form. However, the shared antigenicity of the vaccine and vector may confer vaccine-immunity to hosts infected with the vector, thwarting the ability of the vaccine to spread through the population. We build a mathematical model to test whether a RTV can facilitate disease management in instances where reversion is likely to introduce the vector into the population or when the vector organism is already established in the host population, and the vector and vaccine share perfect cross-immunity. Our results show that a RTV can autonomously eradicate a pathogen, or protect a population from pathogen invasion, when cross-immunity between vaccine and vector is absent. If cross-immunity between vaccine and vector exists, however, our results show that a RTV can substantially reduce the vaccination effort necessary to control or eradicate a pathogen only when continuously augmented with direct manual vaccination. These results demonstrate that estimating the extent of cross-immunity between vector and vaccine is a critical step in RTV design, and that herpesvirus vectors showing facile reinfection and weak cross-immunity are promising.
Abstract The black rat ( Rattus rattus ) is a globally invasive species that has been widely introduced across Africa. Within its invasive range in West Africa, R. rattus may compete with the native rodent Mastomys natalensis , the primary reservoir host of Lassa virus, a zoonotic pathogen that kills thousands annually. Here, we use rodent trapping data from Sierra Leone and Guinea to show that R. rattus presence reduces M. natalensis density within the human dwellings where Lassa virus exposure is most likely to occur. Further, we integrate infection data from M. natalensis to demonstrate that Lassa virus zoonotic spillover risk is lower at sites with R. rattus . While non-native species can have numerous negative effects on ecosystems, our results suggest that R. rattus invasion has the indirect benefit of decreasing zoonotic spillover of an endemic pathogen, with important implications for invasive species control across West Africa.
Anti-PD-1 therapies have achieved durable clinical responses in a wide range of malignancies, but responses are limited to a small subset of patients. Expression of PD-L1 on tumor cells by immunohistochemistry (IHC) has been applied as a companion diagnostic for anti-PD-1 therapy. However, recent studies have called in to question the reliability of this method to predict response.
Methods
Here we developed a novel platform that integrates in vitro pharmacogenomic and functional data with clinical pharmacodynamic responses to immunotherapy using proprietary in silico approaches. The data originate from a long-term co-culture of primary antigen-specific T cells and cancer cells which drives T cells to a terminally dysfunctional, PD-1 refractory state. T cell effector functions and gene expression changes were monitored in the presence or absence of anti-PD-1 antibody or genetic knockouts. RNA expression signatures were refined with a randomized sliding window approach to generate a deep learning neural network for PD-1 response prediction.
Results
We defined five T cell states associated with distinct phenotypic and molecular features - naïve, active, effector, transition and dysfunction. Among the genes that were selectively expressed in the dysfunction state, we identified a 96-gene signature that is closely associated with clinical outcomes to anti-PD-1 therapy. In PD-1 treated patients across multiple solid tumor indications, this signature correlates with objective response rate and outperforms traditional metrics such as tumor mutation burden or PD-L1 IHC signal. Moreover, this signature combines with tumor sequencing data to generate a powerful machine-learning model that predicts anti-PD-1 responses in metastatic melanoma patients with significantly higher accuracy than PD-L1 IHC. Having established that the T cell states in our co-culture relate to clinical outcomes, we leveraged the system to investigate the molecular basis for PD-1 responses. Single cell mapping of transition state T cells in the presence of anti-PD-1 revealed an expanded population of T cells that co-expresses PD-1, TIGIT and activation markers. Likewise, PD-L1 knockout on cancer cells identified the TIGIT ligand, CD155, as a potential tumor escape mechanism to anti-PD-1 therapy. Consistent with this, the combination of PD-1 and TIGIT blockade enhanced T cell cytotoxicity of tumor cells relative to monotherapies.
Conclusions
Agenus' T cell dysfunction platform combines deep in vitro profiling and AI-based approaches to predict clinical outcomes. Here, we defined a predictive biomarker signature that outperforms standard PD-L1 IHC. Further, we identified known (TIGIT) and potentially novel combination partners predicted to enhance the durability of anti-PD-1 responses.
Transmissible vaccines may provide a promising solution for improving the control of infectious disease, particularly zoonotic pathogens with wildlife reservoirs. Although it is well known that heterogeneity in pathogen transmission impacts the spread of infectious disease, the effects of heterogeneity on vaccine transmission are largely unknown. Here we develop and analyze a mathematical model that quantifies the potential benefits of a transmissible vaccine in a population where transmission is heterogeneous between two subgroups. Our results demonstrate that the effect of heterogeneity on the benefit of vaccine transmission largely depends on the vaccine design and the pattern of vaccine administration across subgroups. Specifically, our results show that in most cases a transmissible vaccine designed to mirror the transmission of the pathogen is optimal. If the vaccination effort can be preferentially biased towards a given subgroup, a vaccine with a pattern of transmission opposite to that of the pathogen can become optimal in some cases. To better understand the consequences of heterogeneity on the effectiveness of a transmissible vaccine in the real world, we parameterized our model using data from Sin Nombre virus in deer mice (Peromyscus maniculatus). The results of this analysis reveal that when a vaccination campaign is limited in vaccine availability, a traditional vaccine must be administered primarily to males for the spread of Sin Nombre virus to be prevented. In contrast, a transmissible vaccine remains effective even when it cannot be preferentially administered to males.
Abstract Zoonotic diseases threaten human health worldwide and are often associated with anthropogenic disturbance. Predicting how disturbance influences spillover risk is critical for effective disease intervention but difficult to achieve at fine spatial scales. Here, we develop a method that learns the spatial distribution of a reservoir species from aerial imagery. Our approach uses neural networks to extract features of known or hypothesized importance from images. The spatial distribution of these features is then summarized and linked to spatially explicit reservoir presence/absence data using boosted regression trees. We demonstrate the utility of our method by applying it to the reservoir of Lassa virus, Mastomys natalensis , within the West African nations of Sierra Leone and Guinea. We show that, when trained using reservoir trapping data and publicly available aerial imagery, our framework learns relationships between environmental features and reservoir occurrence and accurately ranks areas according to the likelihood of reservoir presence.
Transmissible vaccines offer a revolutionary approach for controlling infectious disease and may provide one of the few feasible methods for eliminating pathogens from inaccessible wildlife populations. Current efforts to develop transmissible vaccines use recombinant vector technology whereby pathogen antigens are engineered to be expressed from innocuous infectious viral vectors. The resulting vaccines can transmit from host to host, amplifying the number of vaccine-protected individuals beyond those initially vaccinated directly through parenteral inoculation. One main engineering challenge is the potential for natural selection to favor vaccine mutants that eliminate or reduce expression of antigenic inserts, resulting in immunogenic decay of the vaccine over time. Here, we study a mathematical model of vector mutation whereby continuous elimination of the antigenic insert results in reversion of the vaccine back into the insert-free vector. We use this model to quantify the maximum allowable rate of reversion that can be tolerated for a transmissible vaccine to maintain a critical threshold level of immunogenicity against a target pathogen. Our results demonstrate that even for transmissible vaccines where reversion is frequent, performance will often substantially exceed that of conventional, directly administered vaccines. Further, our results demonstrate the feasibility of designing transmissible vaccines that yield desired levels of immunogenicity, yet degrade at a rate sufficient for persistence of the recombinant vaccine within the environment to be minimized.
Abstract Lassa virus is a significant burden on human health throughout its endemic region in West Africa, with most human infections the result of spillover from the primary rodent reservoir of the virus, the natal multimammate mouse, M. natalensis . Here we develop a Bayesian methodology for estimating epidemiological parameters of Lassa virus within its rodent reservoir and for generating probabilistic predictions for the efficacy of rodent vaccination programs. Our approach uses Approximate Bayesian Computation (ABC) to integrate mechanistic mathematical models, remotely-sensed precipitation data, and Lassa virus surveillance data from rodent populations. Using simulated data, we show that our method accurately estimates key model parameters, even when surveillance data are available from only a relatively small number of points in space and time. Applying our method to previously published data from two villages in Guinea estimates the time-averaged R 0 of Lassa virus to be 1.658 and 1.453 for rodent populations in the villages of Bantou and Tanganya, respectively. Using the posterior distribution for model parameters derived from these Guinean populations, we evaluate the likely efficacy of vaccination programs relying on distribution of vaccine-laced baits. Our results demonstrate that effective and durable reductions in the risk of Lassa virus spillover into the human population will require repeated distribution of large quantities of vaccine. Author Summary Lassa virus is a chronic source of illness throughout West Africa, and is considered to be a threat for widespread emergence. Because most human infections result from contact with infected rodents, interventions that reduce the number of rodents infected with Lassa virus represent promising opportunities for reducing the public health burden of this disease. Evaluating how well alternative interventions are likely to perform is complicated by our relatively poor understanding of viral epidemiology within the reservoir population. Here we develop a novel statistical approach that couples mathematical models and viral surveillance data from rodent populations to robustly estimate key epidemiological parameters. Applying our method to existing data from Guinea yields well-resolved parameter estimates and allows us to simulate a variety of rodent vaccination programs. Together, our results demonstrate that rodent vaccination alone is unlikely to be an effective tool for reducing that public health burden of Lassa fever within West Africa.
Reducing spillover of zoonotic pathogens is an appealing approach to preventing human disease and minimizing the risk of future epidemics and pandemics. Although the immediate human health benefit of reducing spillover is clear, over time, spillover reduction could lead to counterintuitive negative consequences for human health. Here, we use mathematical models and computer simulations to explore the conditions under which unanticipated consequences of spillover reduction can occur in systems where the severity of disease increases with age at infection. Our results demonstrate that, because the average age at infection increases as spillover is reduced, programs that reduce spillover can actually increase population-level disease burden if the clinical severity of infection increases sufficiently rapidly with age. If, however, immunity wanes over time and reinfection is possible, our results reveal that negative health impacts of spillover reduction become substantially less likely. When our model is parameterized using published data on Lassa virus in West Africa, it predicts that negative health outcomes are possible, but likely to be restricted to a small subset of populations where spillover is unusually intense. Together, our results suggest that adverse consequences of spillover reduction programs are unlikely but that the public health gains observed immediately after spillover reduction may fade over time as the age structure of immunity gradually re-equilibrates to a reduced force of infection.
