Mathematical modelling and analysis to understand interventions against malaria in humans

Background
Achieving further reduction of malaria burden world-wide requires not just increasing access to existing interventions, but also alternative deployments and new tools. Moreover, efficient clinical and field development of new tools is important to tackle increasing resistance and residual transmission. However, it will be impossible to test all combinations of existing or new tools in the field, or to bring tools to the field without substantial understanding of individual responses and their potential impact on overall mortality or other health metrics. Thus, this PhD research employed modelling and analysis for predictions at population level down to analysis of clinical studies and within host dynamics to understand interventions in humans, burden statistics, and thus the role of models and their assumptions in assessing new tools.
Methods
Several computational approaches were used throughout the thesis. Specifically, this thesis:
i) Estimated country specific access to in-patient care and incidence of severe cases by combining individual-based models of malaria transmission dynamics together with data, such as national reported statistics on burden. These models were further used to estimate the impact of improving access to in-patient care on mortality;
ii) Investigated the potential impact of mass interventions using drugs and anti-infective vaccines in elimination strategies via analysis of a large numbers of simulation results in order to support interpretation of likely impact before trials. Across a range of malaria settings, intervention coverages and deployment frequencies, simulation results were analysed for several epidemiological outcomes, namely prevalence reduction, chance of interruption of transmission, delay of resurgence, and synergism between drugs and vaccines;
iii) Assessed the antibody responses in malaria pre-exposed individuals immunized with the PfSPZ Vaccine in Tanzania via statistical analysis of whole protein microarray data from three cohorts in a clinical study. The humoral immune response pattern in regards single proteins or breadth of responses before and after immunisation in non-vaccinated and vaccinated were compared;
iv) Reviewed and analysed simulation outputs of several mechanistic within host mathematical models of parasite dynamics to understand their assumptions concerning blood stage parasite growth and host immune responses, which potentially impacts predictions of efficacy of within-host interventions.
Results and significance
Mortality estimates are sensitive to assumptions of access to treatment of clinical and severe-cases, and current reported data highlights large variation in levels of access to in-patient care across Africa. These results have implications on current approaches to estimate mortality statistics from reported data, which also form the basis of assessing new tools in policy decisions. In the context of elimination strategies, simulations suggest that potential synergism between drugs and vaccines can lead to a rapid decrease in malaria prevalence, delay malaria resurgence, and in some settings interrupt transmission. The substantial delay in resurgence predicted when a vaccine is delivered with a drug is a new observation from my study.
Personalized breadth of immune responses in pre-exposed individuals plays a role in vaccine take and protection which is not yet fully understood. This suggests further research on personalized responses to vaccines is required during clinical development. Finally, current mechanistic within host models of blood stage asexual parasite infections include large inter-individual variations, at the same time recent findings on parasite growth rates in vivo and on variant gene expression challenge model assumptions around both growth rates and parasite-immune effectors.
Conclusion
Modelling and statistical analysis plays an increasingly important role to understand and assess interventions within an individual, and across populations. Such approaches, only informed or combined with data, can help fill knowledge gaps before moving to costly clinical and field trials, generate hypothesis on alternative use of interventions and likely benefits to be observed, or estimate which outcomes could be monitored in trials. Credible use of these models and analyses depends on understanding their assumptions. Improved estimates of disease burden in malaria endemic countries are required to support targeted development of new tools and allow assessment of likely impact and cost-effectiveness for policy decisions.