Search

Background: Neglected tropical diseases remain a significant burden to global health, despite substantial efforts among global governments and partnerships to control these diseases. In recent years, there has been considerable attention directed towards integrating control platforms across these geographically co-endemic diseases due to demonstrated resource efficacy and sustainability moving forward. As surveillance is a key component of disease control, effective surveillance tools are needed to support these integrated disease platforms. Serological multiplex bead assays are capable of monitoring numerous pathogens simultaneously by using antigens of specific pathogens. However, gaps in knowledge pertaining to the interpretation of serological data from these assays and their utility in public health surveillance necessitates further research to determine their capabilities in supporting concurrent surveillance of multiple diseases. This thesis aims to determine appropriate methods of interpreting serological data from multiplex bead assays and assess their utility in the context of public health surveillance of neglected tropical diseases. Methods: Three datasets from Haiti (2015 Tracking Results Continuously Survey, n=4438 ; 2017 Artibonite Easy Access Group Survey, n= 6004 ; and 2017 Artibonite Community Household Survey; n=21222 ) and one dataset from Malaysia (2015 Sabah Household Cluster Survey, n=10100 ) were used in this thesis. Collectively, these datasets included antigens of twelve different pathogens that were analysed using serological multiplex bead assays . The first objective of this thesis was to identify existing methods of characterising serological data from neglected tropical diseases using multiplex bead assays. This was done through a systematic review of literature which examined existing and potential methods of characterizing antibody responses from these assays. Several of these methods were then applied to two case studies to evaluate potential implications pertaining to method choice on public health interpretation. The second objective was to assess the utility of multiplex bead assays for multi-disease surveillance. To do this, panels of diverse tropical disease antigens (encompassing twelve different pathogens collectively) from three datasets in Haiti and one dataset from Malaysia were used to analyse concurrent monitoring of multiple diseases and to assess demographic and spatial disease risk factors for co-endemic pathogens. The third 3 objective of this thesis was to determine the capacity of serological multiplex bead assays to support different sampling strategies. This was done by formally comparing outputs from serological multiplex bead assays from a community household active sampling strategy and an easy-access group convenient sampling strategy in Haiti. Results: The review of literature revealed that serological data from multiplex bead assays are typically converted to seroprevalence for programme interpretation, however there is currently an absence of a standard approach to determine serological prevalence estimates. Instead, seven different approaches were identified in the literature. Comparing three different approaches resulted in varying disease prevalence estimates in both Haiti and Malaysia, suggesting potential impacts of classification approaches on postliminary programmatic interpretation in both case studies. Multiplex bead assays provided concurrent estimates of exposure to various pathogens simultaneously at the national and subnational levels of surveillance within Haitian and Malaysian study populations. Demographic and spatial data collected alongside serologic surveys determined several consistent risk factors across antigens assessed, including age, wealth, gender, and also allowed for visualization of any spatial trends in disease exposure in both settings. Using multiplex bead assays in two different sampling approaches demonstrated its capacity to support multi-disease surveillance in different sampling approaches. In comparing prevalence estimates between surveys, observed differences may be attributable to inherent biases in sampling populations and design. Conclusions: The research in this thesis contributes to the understanding of the utility of serological multiplex bead assays to support multi-disease surveillance and provides a foundation in the broader study of applying these platforms to neglected tropical disease control. This thesis also highlights the need to develop standardized approaches in sampling, laboratory protocols, and analysis for these platforms to ensure consistent and confident disease estimates and data reporting.

Introduction: Travel is an underappreciated risk factor for malaria in residents of malaria-endemic countries. In Uganda, malaria transmission is heterogeneous, in part due to expansion of malaria control interventions and increased urbanization. As a result, individuals who travel may be at higher risk of malaria infection. However, our understanding of the association between travel and malaria infection in Uganda is limited. Methods: This thesis aimed to address this evidence gap through: 1) case-control study in urban Kampala, 2) cohort study in three sites of varied malaria transmission, and 3) cohort study in rural Tororo, which is under intense malaria control with indoor residual spraying (IRS). For the case control study, 5 controls were selected for every 2 cases, matching on age. Data were collected in July and August 2019 on recent overnight travel out of Kampala (past 60 days), destination and duration of travel, and behavioural factors, including sleeping under a long-lasting insecticidal net (LLIN) during travel. For the cohort study at three sites in Uganda (PRISM 1), information on overnight travel was collected between 2015 and 2016 from children aged 0.5-10 years and one adult living in 266 randomly selected households. Malaria, defined as fever with parasites detected by microscopy, was measured using passive surveillance. For the cohort in rural Tororo (PRISM 2), data on overnight travel and behaviour during travel were collected from residents of 80 households between 2017 and 2019. Behaviour while at home was assessed using a similar questionnaire during two-weekly home visits. Results: In the case-control study, 162 cases and 405 controls were enrolled. Overall, 158 (27.9%) participants reported recent overnight travel. Travellers were far more likely to be diagnosed with malaria than those who did not travel (80.4% vs 8.6%, odds ratio 58.9, 95% 4 confidence interval [CI] 23.1-150.1, p<0.001); travelling to a non-IRS district, not using LLINs during travel, and engaging in outdoor activities were associated with increased odds of malaria. In the PRISM 1 cohort study, at least one overnight trip was reported by 120 of 906 (13.3%) participants. Among individuals who travelled, the incidence of malaria was higher in the first 60 days after travelling, compared to periods without recent travel at all sites (overall 1.15 vs 0.33 episodes per person-year, incidence rate ratio 3.53, 95% CI 1.85-6.73, p<0.001). In the PRISM 2 cohort study, 527 participants were enrolled and 123 (23.2%) reported taking at least one overnight trip. Overall, participants were less likely to use LLINs when travelling than at home (41.0% vs. 56.2%, relative risk [RR] 0.73, 95% CI: 0.60-0.89, p=0.002). In the analysis adjusted for gender and age, significantly lower LLIN use during travel was found for female participants (38.8% when travelling vs 59.2% at home, risk ratio [RR] 0.66, 95% CI 0.52-0.83, p=0.001) but not for males (48.3% vs 46.6%, RR 0.96, 95% CI 0.67-1.40, p=0.85), and those > 15 years (33.9% travel vs 61.3% home, RR 0.55, 95% CI 0.41-0.74, p<0.001). Conclusions: Residents of malaria endemic countries who travel are a high-risk group that should be targeted for malaria prevention. For these travellers, personal protection measures, including sleeping under LLINs when traveling, application of creams or sprays to prevent outdoor mosquito bites, and administration of chemoprophylaxis, should be advocated.

Background: After standard effective antimalarial treatment with artemisinin-based combination therapy (ACT), a proportion of individuals remain infectious to mosquitoes, enabling onward transmission. This is due to persisting gametocytes, the sexual stage of the malaria parasite. Primaquine, an 8-aminoquinoline drug, sterilizes and clears gametocytes. The World Health Organization recommends that, in areas where malaria elimination is targeted, a single dose of primaquine should be given in addition to standard antimalarial treatment. Despite its recommendation in WHO guidelines since the 1970s, the optimal dose of primaquine for this purpose had not been determined. Primaquine is associated with haemolytic toxicity in individuals with glucose-6-phosphate dehydrogenase (G6PD) deficiency, a condition that is prevalent in malaria-endemic regions. This thesis presents the first dose ranging trial to assess the safety and efficacy of reducing doses of primaquine in combination with ACT to treat children with uncomplicated falciparum malaria infection. Methods: A literature review was conducted to inform a novel, evidence-based trial design. Based in Jinja, Uganda, this randomised, double-blind, and placebo-controlled trial had four parallel treatment groups of reducing doses of primaquine plus ACT. Results: For trial participants, a single dose of 0.4mg/kg primaquine base had non-inferior efficacy (measured by gametocyte clearance) to the WHO-recommended dose of 0.75mg/kg, whereas a dose of 0.1mg/kg was not non-inferior. There was no significant haemolysis in any of the treatment arms and the fall in haemoglobin was not associated with the dose of primaquine. However, a sub-analysis showed a dose-dependent reduction in haemoglobin in participants who were G6PD deficient (heterozygous or hemi-/homozygous genotype). Subsequently, the WHO reduced the recommended dose to 0.25mg/kg primaquine base. Additional trial analyses that were not included in the published manuscript are presented in the thesis, together with descriptions of the trial's impact and data sharing. Conclusion: The findings of this trial have contributed to changes in malaria elimination policy and to the prioritisation of primaquine evaluation in research agendas. This thesis puts the trial in the context of the body of evidence that has amassed since the trial results were published and highlights priorities for further research.

Serological surveillance involves the detection of Plasmodium species-specific antibodies as biomarkers for monitoring recent and historical malaria transmission dynamics at population-level. These methods are particularly useful in low transmission settings where standard surveillance such as parasitological and entomological approaches are inefficient. This thesis explores the use of serological surveillance to estimate the magnitude and heterogeneity of malaria transmission using different sampling strategies, mapping techniques and serological assays in three areas of differing endemicity in Indonesia. Findings suggest that: 1) Analysis of community-based serological data can confirm the discontinuation of transmission and be used to identify high-risk areas where malaria is most likely to be reintroduced, 2) Mobile technology-based participatory mapping approaches can be used to quickly obtain spatial residential information for individuals presenting at health facilities in resource poor areas where formal addresses are typically not used and internet connectivity is limited, 3) the combination of facility-based sampling, multiplex serological assays and participatory mapping can be used as an additional surveillance method to better identify and target areas still receptive to malaria in very low transmission area conducting elimination, 4) seropositivity to Etramp5.Ag1 is sensitive and specific in predicting Plasmodium falciparum PCR positivity in children in a high transmission setting, suggesting its potential use as a marker of recent exposure in elimination setting. In conclusion, this thesis demonstrates the various applications of serological surveillance at different levels of malaria endemicity. Further implementation research is needed to enable the integration of these methods to the existing surveillance systems.

Since identification of a large number of human infections in 2004, the zoonotic malaria species Plasmodium knowlesi is currently the main cause of human malaria in Malaysian Borneo. Carried by long and pig-tailed macaques, deforestation and associated environmental and population changes have been hypothesised to be the main drivers of this emergence. This thesis aims to describe the epidemiology of P. knowlesi in Northern Sabah, Malaysia at a range of spatial scales and evaluate how environmental change and behaviour affect human infection risks. Satellite-based and aerial remote sensing technologies were utilised with GPS tracking and field surveys to characterise dynamic interactions between the environment, human, macaque and mosquito populations (Chapter 4). A retrospective analysis of reported P. knowlesi cases found marked spatial heterogeneity in village-level P. knowlesi incidence, with village-level numbers of cases positively associated with both forest cover and forest loss in surrounding areas (Chapter 5). To explore how these people used these environments, mobility patterns and resource utilisation were mapped using GPS tracking devices; these data were integrated with predictions of mosquito biting rates to estimate individual and location-specific exposure risks (Chapter 6). Additional surveys were conducted in households and villages of symptomatic cases and within case study communities to evaluate levels of infection and exposure within the wider population. These identified a substantial proportion of asymptomatic infections not detected by hospital-based surveillance systems (Chapter 7) and found positive associations between knowlesi sero-positivity and environmental factors at a fine spatial scale (Chapter 8). Results from these studies were used to design an ecologically-stratified cross-sectional survey across four districts in Northern Sabah; this study was used to identify individual and environmental variables associated with exposure and infection risk across a wider geographical area (Chapter 9). Together, these studies indicate a wide distribution of P. knowlesi infection and exposure in demographic groups underrepresented in clinical reports and highlight the role of local environmental change in P. knowlesi risk. Further research is needed to refine diagnostic methods and understand the longer-term impacts of ecological changes on disease dynamics.

Global declines in malaria transmission in recent years have refocused efforts towards elimination. An essential part of this effort is adequate detection of transmission patterns. However, meta-analyses have shown that microscopy detects about half of all polymerase chain reaction (PCR)-confirmed infections. To date, the performance of rapid diagnostic tests (RDTs) relative to microscopy and PCR has not been evaluated in a comprehensive meta-analysis. Therefore, the relationship between PCR, RDT and microscopy prevalence estimates in asymptomatic populations was determined using data from cross-sectional surveys in endemic settings (Chapter 3). Overall, RDTs detected 41% of all PCR-confirmed infections, and RDT-undetected (i.e. low-density) infections increased with age and decreasing transmission intensity. Another approach to estimate transmission is to use malaria-specific immune responses of resident populations as a measure of exposure to infection. Antimalarial antibodies, in combination with age, reflect both historical and recent exposure. Until recently, the majority of sero-surveillance data were based on a few well-characterised antigens using enzyme-linked immunosorbent assays (ELISA). However, which antibodies most accurately reflect exposure to recent low-density infections remains largely unknown. To address this, Chapter 4 examined antibody responses in previously naïve, controlled human malaria infections (CHMI) participants using protein microarray. Nearly all participants showed measurable antibody responses to a subset of four antigens one month post-CHMI. In addition to protein microarray, multiplex bead assays (MBAs) enable multiplex detection of antibodies. However, MBA protocols may require further adaptation in scenarios where results must be readily available to inform control and elimination policies. The precision of an adapted MBA protocol with improved throughput and ease-of-use was determined in Chapter 5 using data collected in three large-scale transmission surveys. For some antigens, inter-plate variability seemed to increase during the third survey, possibly due to long-term storage of reagents. Commercially available ELISAs are standardised in their production and have been used to test blood products prior to donation to reduce the risk of transfusion-transmitted malaria. However, their performance in an epidemiological context has not been investigated. In Chapter 6, one of five commercially available ELISAs evaluated, accurately described transmission in a low transmission and pre-elimination setting. A low-cost, high-throughput assay for which results are readily interpretable may help to directly inform control activities targeting areas with remaining transmission.

As malaria endemic countries progress towards elimination, accurately measuring community-level transmission is critical for monitoring control strategies and the design of efficacy trials. However, measuring malaria transmission often faces challenges in pre-elimination settings due to the complexity of human immunity and its interaction with vector and parasite dynamics. This study evaluates current and emerging epidemiological measures of malaria transmission, and explores the use of novel serological markers of malaria infection as metrics in surveillance and cluster-randomised trials (CRTs). The relative sensitivity of commonly used surveillance diagnostics - polymerase chain reaction (PCR), rapid diagnostic tests (RDTs), and microscopy - are cross-compared with respect to their accuracy in quantifying cluster-level prevalence of malaria infection. These are further evaluated against immunological measures of transmission based on antibody responses to two malaria parasite antigens - PfMSP119 and PfAMA1 - used extensively in serological surveillance for the last decade. To investigate novel serological markers of malaria infection, a multiplexed immunoassay was used to characterise post-infection antibody dynamics to 20 Plasmodium falciparum antigens. This was based on a subset of 192 individuals from an all-age longitudinal cohort study in The Gambia. Antibody responses against several antigens showed accuracy in detecting infection in the preceding six months. These may have potential utility in measuring time since infection or short-term changes in transmission. However, variations in immune response by age and transmission intensity were observed and should be taken into consideration for future optimisation of serological assays. Antigens identified as the most promising biomarkers of recent infection were used to estimate cluster-level transmission in four villages in The Gambia. Serological responses are compared between dry and wet season and geographical regions of low and high transmission intensity. Their application was also extended to compare study arms of a cluster-randomised trial in the Zambezi Region, Namibia, comparing the effectiveness of reactive focal case detection, reactive focal mass drug administration, and reactive vector control. These findings may help to inform the development of new serological diagnostic assays for malaria, their use in future malaria surveillance and elimination strategies, and the design of cluster-randomised trials in low transmission settings.

Malaria control strategies are often targeted at individuals who suffer most morbidity, not at those driving transmission. This is in part due to our limited understanding, described in Chapter 1, of the human infectious reservoir of falciparum malaria – i.e. individuals responsible for human-to-mosquito transmission of Plasmodium falciparum parasites in endemic areas. This work, whose objectives are listed in Chapter 2, assessed the prevalence of infectiousness in naturally exposed human populations during dry and wet seasons. Exposure to Anopheles mosquitoes, a measure of transmission opportunities, was also quantified; and to determine the value of infectiousness-reducing interventions, the use of primaquine to block transmission from infectious individuals was investigated. Experimental infections of mosquitoes were performed to determine malaria infectivity of randomly selected individuals in two villages in Burkina Faso. Molecular assays were used to quantify parasite, including gametocyte, densities. Less than 10% of the population was infectious to mosquitoes. These results are presented in a manuscript that included data from other study sites (Chapter 3). To assess exposure to malaria vectors, bloodfed mosquitoes were collected indoors in one of the study villages in Burkina Faso. A multiplex PCR assay targeting nine human microsatellites and a gender-specific marker was used to identify the human sources of mosquito blood meals. Although there was substantial variation in the number of mosquito bites each individual received (Chapter 4), on average adults received more mosquito bites than children. This suggests that, despite their lower infectiousness, adults are major contributors to malaria transmission in endemic areas. An efficacy trial of single low dose primaquine was performed in Burkina Faso and pre- and post-treatment infectiousness were quantified by mosquito feeding experiments to assess primaquine’s infectiousness-reducing activity (Chapter 5). Individuals receiving primaquine cleared gametocytes faster than individuals who received artemether-lumefantrine alone. Feeding assays, however, suggest that artemether-lumefantrine blocks most parasite transmission after treatment administration. In Chapter 6, these findings, and how they can inform future control strategies, are discussed.

Heterogeneous malaria exposure may result in distinct clusters of higher malaria burden, or hotspots, across space and time. Targeting control programs to these areas may be highly efficient, however, operationally attractive approaches for identifying hotspots are needed for any such program to be sustained by local malaria control programs. The principal aim of this project was to investigate the ability of convenient sampling to identify hotspots of malaria transmission in a low endemic transmission setting in the western Kenyan highlands: 1) The boundaries of hotspots, and associated uncertainties, was determined using a large community survey; 2) The value of convenience sampling to estimate transmission in the community was assessed using cross-­‐sectional surveys of 4964 children in 46 government primary schools and 3042 individuals in five rural-­‐health facilities; 3) The value of compound-­‐level screening of sentinel age groups that are likely to have patent level infections was determined and; 4) The potential use for convenience sampling in hotspot targeted approaches was assessed using spatial information on residences collected during the school and health-­‐facility surveys. The community-­‐based approach was able to detect 77% of the parasite infections in selected hotspots of malaria exposure using field-­‐based tools in sentinel age groups. Both convenience-­‐sampling approaches tested produced similar estimates of malaria transmission to the community when restricted to those residing in the same catchment areas and those testing positive for malaria were more likely to reside in a hotspot. The findings suggest that operationally attractive approaches provide reliable information on malaria transmission and may play an important role in targeted malaria control strategies. Future research on ascertaining what coverage of the hotspot is needed to see sustainable reductions in transmission would provide a threshold with which to gauge the utility of this strategy.

Northern Pakistan, Khyber Pakhtunkhwa (KP) is low malaria endemic area characterised by seasonal transmission with predominantly vivax malaria. Migration of high number of Afghan Refugees in 1978 into KP led to concerns for an increase in malaria, as the malaria incidence in this group was reportedly high compared to the local Pakistani population. Considerable progress has been made in controlling malaria through operational research in the camps where the Afghan refugees reside. However, this process requires effective, repeatable active surveillance tools for monitoring malaria control as availability of accurate data is the major challenge at present. The aim of this PhD project was to generate current information on malaria infection rates through parasite prevalence and malaria exposure using antimalarial antibody responses. The project also investigated the risk factors of malaria and heterogeneity in the geographic distribution of malaria in the camps by using GIS data with serological responses and parasite prevalence data. As an ancillary objective the project aimed to determine the prevalence of G6PO deficiency in the study population. A cross-sectional survey was conducted in five Afghan refugee camps of KP between June and September in 2010. Blood samples were obtained on filter paper from 2526 individuals and tested by rapid diagnostic test, paraSite species specific PCR and ElISA for antibody responses to Plasmodium vivax and Plasmodium falciparum. A questionnaire was administered to collect household and individual based information to determine the potential risk factors of malaria. Heterogeneity in malaria was observed between the studied camps based on seroprevalence, which ranged from 17%-45% for P. vivax and 3% to 11% for P. falciparum. Variation in P. vivax infection prevalence was also detected between the camps, which ranged from 0.4-9% (ROT) and 5-15% (peR). Variation in the distribution of malaria was also found within the camp using spatial/GIS data with clear foci of infection identified in 4 of 5 camps. The results showed that as expected parasite based prevalence measures (ROT and peR) are significantly lower than serological measure of exposure. P. falciparum infection prevalence (ROT and PCR) and seroprevalence was found to be extremely low with P. vivax infections predominant. Age seroprevalence changes were more pronounced for P. vivax than P. falciparum and seroconversion rate was strongly associated with parasite rate. Increasing age .and poorly built houses were associated with increasing risk, while staying in the same camp for the last 6 months and using measures to reducing vector biting such as repellents repellent, coils or insecticide spraying were associated with reduce risk of falciparum malaria. The risk of vivax malaria was observed to increase with increasing age, sharing house with cattle and having fever within 24 hours or two weeks and a reduction in the risk was seen in the individuals who reported use of Insecticide treated Bed Nets (ITN) night prior to surveyor used self protection measures from vector. The 563C-T polymorphism of G6PD gene was observed in only 2 unrelated individuals out of 505 individuals tested (O.4%). In conclusion, both parasitological and serological measures were able to detect spatial variation in infection and exposure to malaria at the micro epidemiological level within the camp. This data will help to provide beneficial and up-to-date information to manage control activities in the study area.