Epidemiology, Evolution and Control of Infectious Diseases

The last century has seen the development and implementation of a range of antimalarials to treat infections caused by Plasmodium falciparum. The clinical efficacy of these antimalarials has repeatedly been compromised over time due to the evolution of resistant parasite strains (WHO, 2010). At present, artemisinin-based combination therapies (ACT) are recommended for the treatment of uncomplicated malaria. These drugs consist of an artemisinin component and a longer-lasting partner drug, so if the parasite acquires resistance to one component, it will likely be sensitive to the other component (Das et al., 2009; Phyo et al., 2012). Recent observations of ACT resistance in Southeast Asia (Amaratunga et al., 2016; Saunders and Lon, 2016), however, indicate that the emergence of clinical resistance is occurring, suggesting that parasites may be becoming resistant to both components of the ACT. The continued pattern of resistance emergence has highlighted a need to identify the best course of action in terms of balancing both the longevity of antimalarial efficacy and the immediate public health impact.

Previous work within the field of modelling and mapping antimalarial resistance has been categorised by the scale of evolution considered. Population demographic level approaches have utilised malaria epidemiological models to summarise the effect of treatment and malaria transmission intensity upon antimalarial resistance (Koella and Antia, 2003). The other class of models has aimed to describe and consider within-host dynamics and their impact on resistance evolution (Mackinnon, 2005), exploring the competition between multiple circulating strains (Antao and Hastings, 2011). More recent work has explored the need to consider both evolutionary scales (Kim et al., 2014), and substantial work has been carried out to identify the optimum ACT strategy at a population level in order to prolong the efficacy of ACT (Nguyen et al., 2015).

Previous work conducted by the researcher investigating the impact of PfHRP2-based rapid diagnostic tests upon pfhrp2 deletion involved the development of an individual-level stochastic model to simulate the transmission dynamics of Plasmodium falciparum. The model details multiple within-host immunity stages and considers the multiplicity of infection of each individual. The output of this work will be used heavily within the project, however with a number of improvements and extensions to create a model that captures both the epidemiological process and within-host bottlenecking.

A number of data collation and theoretical approaches will be undertaken in response to the research questions posed. In order to understand the factors contributing to ACT resistance emergence, the researcher will use computational models to capture the epidemiological and within-host features of malaria transmission that contribute to ACT resistance. This work will begin by extending the previously described transmission model to better incorporate the epidemiological outcomes of varying pharmacokinetic and pharmacodynamic drug profiles resulting from waning efficacy. The model will be further extended to consider the role of recombination and mutation in resistance evolution, allowing for de novo creation of resistant strains.

The added model complexity will require considerable parameterisation and model checking, and as such the study will initially utilise previous modelling efforts within the field of ACT resistance modelling. These efforts will form modelling priors that will be further parameterised through model fitting against available published longitudinal ACT resistance studies within Southeast Asia. This work will first involve the collation and characterisation of these datasets through the identification of relevant genomic markers of antimalarial resistance. The resistance data will then be geolocated, allowing for Bayesian model fitting against observed spatiotemporal patterns of ACT resistance.

Upon completion of model parameterisation, sensitivity analysis will be conducted to identify the major drivers of ACT resistance within the final model. These drivers will consider factors such as treatment regimens, transmission intensities, case management efficacy, and geographical covariates. Through these endeavours, the spatial patterns driving ACT resistance will be characterised and mapped within Southeast Asia.

To address how resistance emergence may be slowed, the parameter space within the model will be searched to identify strategies that reduce the spread of ACT resistance. Consideration will be required to restrict the parameter space to ranges that are biologically plausible and represent feasible treatment strategies. These strategies will be tailored to the geographical settings considered within Southeast Asia. The resultant treatment regimens will subsequently be used to forecast the potential impact of altering the rate of ACT resistance emergence and to identify geographical regions at highest risk of future ACT resistance.

This project will adopt novel approaches to characterise the spatial determinants of ACT resistance in Southeast Asia. Primarily, this will involve extending the previously described model to capture geographical variation in ACT resistance. The required modelling innovation will necessitate considerable algorithmic implementation to enable parallel regional simulations. To ensure the project aims are achievable, differential equation modelling of the scenario will be formulated in parallel to enable efficient error checking. The computational work will be tackled using agile software development approaches, drawing on the researcher’s previous modelling experience to ensure feasibility within the PhD time constraints. In addition, sufficient data availability for model fitting has already been ensured for Cambodia, southern Vietnam, eastern Myanmar, Thailand, and southern Laos through the Tracking Resistance to Artemisinin Collaboration (TRAC).

Both the collation and characterisation of available longitudinal ACT resistance studies and the modelling work will be conducted concurrently. Data collection is initially timetabled to take six months, with the expectation of continued publication of new datasets associated with the TRAC II follow-up study. The completed, parameterised model will represent the main body of work and is expected to be finalised by the end of year two. The substantial time allocated reflects the need to stratify treatment regimens, transmission intensities, case management efficacy, and geographical covariates across the regions studied.

Drug Resistance

Oct 2015 — Oct 2019

$220,101

United Kingdom