Background
Plasmodium falciparum malaria remains a pressing global health emergency. Encouraging progress in its control has been made in some areas of Africa [
1], but elimination does not appear realistic in many areas. The current lead vaccine candidates are based on the circumsporozoite protein (CSP) and have been shown to be protect against clinical manifestations of
P. falciparum disease in children [
2,
3]. Higher vaccine efficacy against clinical manifestations might be achievable through inducing immune responses against antigens from the asexual blood-stages [
4]. The clinical development pathway for any one candidate vaccine is expensive and lengthy. None of the blood-stage candidate vaccines subjected to field trials have progressed to Phase III trials [
5,
6]. The need to understand and interrogate naturally acquired immunity to malaria is fundamental to antigen selection and vaccine design. A common approach is to use immuno-epidemiological studies in malaria endemic regions, where immunological responses from cross-sectional surveys of children are linked with the risk of subsequent malaria episodes [
7‐
11]. A limitation of this approach has been the reliance on uncontrolled natural but heterogenous exposure to malaria [
8,
9,
12] as well as exposure to genetically diverse parasites [
13] in the field.
Controlled Human Malaria Infection (CHMI) studies have the potential to accelerate the selection of antigens for vaccine development by controlling for malaria exposure, including parasite strain, as well as level of infectious dose. For ethical reasons, CHMI requires adult volunteers rather than children. In endemic areas immunity is acquired with age and adults usually have high levels of immunity to the consequences of infection [
14]. Nevertheless, even among adults’ levels of immunity may be variable. We have recently described clinical outcomes and safety of CHMI in Kenyan adults after infection with cryopreserved viable, aseptic, and purified
Plasmodium falciparum sporozoites (PfSPZ Challenge) at a dose of 3200 injected by syringe [
15]. We showed that using CHMI in this population of 142 pre-exposed adults, 26 (18.3%) had febrile symptoms and were treated; 30 (21.1%) reached ≥ 500 parasites/µl and were treated; 53 (37.3%) had parasitaemia without meeting thresholds for treatment and; whilst 33 (23.2%) remained qPCR negative (in a subset of volunteers, some of those qPCR negative between days 8 and 10 post-infection had low parasitaemia in comparison to two other qPCR methods) [
15]. These findings are consistent with other CHMI studies in volunteers from endemic areas [
16,
17]. However, the outcomes of CHMI that are most strongly associated with naturally acquired immunity have not yet been determined. Furthermore, categorizations into multi-level descriptive outcomes do not maximize analytical power for correlates of immunity, and either a binary classification or a continuous variable would be analytically optimal.
We therefore in this study conducted an analysis using anti-schizont antibody responses and location of residence as surrogates of immunity. We examined various parameters from the patterns of parasite growth during CHMI. This was to determine which parameters most closely associated with these two surrogates of immunity, and to identify whether any were more discriminatory of host immunity than the standard immuno-epidemiological studies conducted in the field.
Discussion
We used serial qPCR to determine the outcomes most strongly associated with anti-schizont antibodies and location of residence (low vs. high transmission), in order to define the outcomes for CHMI in exposed adults that are most strongly associated with surrogates of immunity. We used anti-schizont antibody levels and location of residence at varying prior exposure to malaria as surrogates for immunity to malaria. We examined several potential parameters based on the qPCR monitoring done for CHMI for their association with anti-schizont antibodies and location. Time to treatment and time to 250 parasites/µl were strongly associated with anti-schizont antibodies and with location. We preferred time to treatment rather than time to 250 parasites/µl, as the former included the full set of volunteer data and avoids the potential bias of missing data from volunteers who were treated before reaching 250 parasites/µl.
After adjusting for time to treatment, there were no other independent predictors of anti-schizont antibodies or of location of residence. We therefore developed a survival analysis based on time to treatment. The combination of location of residence and anti-schizont antibodies as a continuous variable explained 35% of the variability in time to treatment in CHMI. Since prior residence and anti-schizont antibodies only offer limited information on the true extent of host immunity, this implies that a very significant proportion of the variability in outcome in CHMI is due to host immunity.
We examined whether the analysis of CHMI for naturally acquired immunity was a significant advance over previous studies conducted in the field based on natural exposure to malaria. Adults have higher levels of immunity than children, different endpoints are used for adults participating in CHMI compared with children in field observational studies, but nevertheless these study designs both share the aim to define potential correlates of immunity. In order to make comparisons, we used logistic regression with febrile malaria as the outcome in the field studies, and with treatment criteria as the outcome in CHMI. We used anti-schizont antibodies as the predictor variable. Levels of anti-schizont antibodies higher among adults than children, so we divided antibodies into high or low categories based on the median antibody level in each study. In the field-based observational study analysed here in a cohort of children, anti-schizont antibody responses explained less than 1% of the observed variability, but anti-schizont antibodies explained 17% of the variability in CHMI outcomes. This is not surprising given the variability in exposure to malaria seen in the field [
8,
9], whereas in CHMI exposure is controlled and does not vary between participants.
This analysis, here, shows how, adjusting and accounting for heterogeneity of exposure and infection, and given that anti-schizont antibodies in field-based studies account for a small fraction of the variability, CHMI in an adult pre-exposed population has a larger discriminatory power to study immunity in relation to past exposure. Furthermore, in non-immune CHMI studies, a large proportion of the volunteers develop illness and require treatment at relatively low parasitaemia thresholds (between 5 and 50 parasites/ml) whilst in our study, individuals were often asymptomatic and parasite-free and this could largely be as a result of differences in responses in non-immunes with semi-immunes [
27,
28]. With the exception of innate factors such as sickle cell trait [
17], this resistance in previously malaria exposed individuals thus might be as a result of acquired adaptive immunity which is confirmed by anti-schizont antibody responses.
Thus, these findings presented here, provide a unique opportunity to advance the field of vaccine antigen discovery utilizing the CHMI platform with characterization and better understanding of the development of immunity to infection in the context of past malaria exposure. A comprehensive analysis of signatures or correlates of immunity as has been recent detailed utilizing systems serology approaches (both qualitative and quantitative antibody-based approaches) [
29] will significantly advance the field.
This analysis, not relying on one or two parameters of the outcome measure (PCR) especially in the context of undertaking these studies in populations with varying past exposures to malaria is warranted. Traditionally, studies have relied on the parasite growth kinetics/rate as an important measure of outcome in CHMI studies—including as an assessment of vaccine or drug efficacy [
30]. CHMI studies enrolling volunteers with a range of parasite exposures, to date in Africa, have taken the approach of endpoint measurement largely based on thick-blood microscopy at a particular threshold for diagnosis [
17,
22,
31] to explain parasite growth rates. Achan et al. [
16] despite using PCR as a criteria for endpoint, did not have the same breadth of past exposures as presented here. Hence, it is important for studies particularly utilizing PCR to undertake a detailed analysis of the most reliable parameter that would account for diversity in parasite growth.
Acknowledgements
We would like to thank all the study volunteers who participated in the CHMI-SIKA study and contributed data to this analysis. We are also very thankful to the larger study teams in Kilifi and Ahero specifically all the fieldworkers and health community workers who recruited volunteers; data entry clerks; clinical, pharmacy, and laboratory teams; and the collaborating manufacturing, quality systems, regulatory, pharmaceutical operations and clinical teams at Sanaria Inc. without whom this work would not have been possible. This manuscript is published with permission and/or approval of the Director KEMRI. Members of the CHMI-SIKA Study Team are as follows: Abdirahman I. Abdi, Centre for Geographic Medicine Research (Coast), Kenya Medical Research Institute-Wellcome Trust Research Programme, Kilifi, Kenya; Yonas Abebe, Sanaria Inc., Rockville, MD, USA; Philip Bejon, Centre for Geographic Medicine Research (Coast), Kenya Medical Research Institute-Wellcome Trust Research Programme, Kilifi, Kenya and Centre for Tropical Medicine and Global Health, Nuffield Department of Medicine, University Oxford, Oxford, UK; Peter F. Billingsley, Sanaria Inc., Rockville, MD, USA; Peter C. Bull, Department of Pathology, University of Cambridge, Cambridge, UK; Zaydah de Laurent, Centre for Geographic Medicine Research (Coast), Kenya Medical Research Institute-Wellcome Trust Research Programme, Kilifi, Kenya; Stephen L. Hoffman, Sanaria Inc., Rockville, MD, USA; Eric R. James, Sanaria Inc., Rockville, MD, USA; Silvia Kariuki, Centre for Geographic Medicine Research (Coast), Kenya Medical Research Institute-Wellcome Trust Research Programme, Kilifi, Kenya; Sam Kinyanjui, Centre for Geographic Medicine Research (Coast), Kenya Medical Research Institute-Wellcome Trust Research Programme, Kilifi, Kenya and Pwani University, Kilifi, Kenya; Cheryl Kivisi, Centre for Geographic Medicine Research (Coast), Kenya Medical Research Institute-Wellcome Trust Research Programme, Kilifi, Kenya and Pwani University, Kilifi, Kenya; Johnstone Makale, Centre for Geographic Medicine Research (Coast), Kenya Medical Research Institute-Wellcome Trust Research Programme, Kilifi, Kenya; Kevin Marsh, Centre for Geographic Medicine Research (Coast), Kenya Medical Research Institute-Wellcome Trust Research Programme, Kilifi, Kenya and Centre for Tropical Medicine and Global Health, Nuffield Department of Medicine, University Oxford, Oxford, UK; Khadija Said Mohammed, Centre for Geographic Medicine Research (Coast), Kenya Medical Research Institute-Wellcome Trust Research Programme, Kilifi, Kenya; Moses Mosobo, Centre for Geographic Medicine Research (Coast), Kenya Medical Research Institute-Wellcome Trust Research Programme, Kilifi, Kenya; Janet Musembi, Centre for Geographic Medicine Research (Coast), Kenya Medical Research Institute-Wellcome Trust Research Programme, Kilifi, Kenya; Jennifer Musyoki, Centre for Geographic Medicine Research (Coast), Kenya Medical Research Institute-Wellcome Trust Research Programme, Kilifi, Kenya; Michelle Muthui, Centre for Geographic Medicine Research (Coast), Kenya Medical Research Institute-Wellcome Trust Research Programme, Kilifi, Kenya; Jedidah Mwacharo, Centre for Geographic Medicine Research (Coast), Kenya Medical Research Institute-Wellcome Trust Research Programme, Kilifi, Kenya; Kennedy Mwai, Centre for Geographic Medicine Research (Coast), Kenya Medical Research Institute-Wellcome Trust Research Programme, Kilifi, Kenya and Epidemiology and Biostatistics Division, School of Public Health, University of the Witwatersrand, Johannesburg, South Africa; Joyce M. Ngoi, Centre for Geographic Medicine Research (Coast), Kenya Medical Research Institute-Wellcome Trust Research Programme, Kilifi, Kenya; Omar Ngoto, Centre for Geographic Medicine Research (Coast), Kenya Medical Research Institute-Wellcome Trust Research Programme, Kilifi, Kenya; Irene Nkumama, Centre for Geographic Medicine Research (Coast), Kenya Medical Research Institute-Wellcome Trust Research Programme, Kilifi, Kenya and Centre for Infectious Diseases, Heidelberg University Hospital, Heidelberg, Germany; Francis Ndungu, Centre for Geographic Medicine Research (Coast), Kenya Medical Research Institute-Wellcome Trust Research Programme, Kilifi, Kenya; Dennis Odera, Centre for Geographic Medicine Research (Coast), Kenya Medical Research Institute-Wellcome Trust Research Programme, Kilifi, Kenya and Centre for Infectious Diseases, Heidelberg University Hospital, Heidelberg, Germany; Bernhards Ogutu, Centre for Clinical Research, Kenya Medical Research Institute, Kisumu, Kenya and Center for Research in Therapeutic Sciences, Strathmore University, Nairobi, Kenya; Fredrick Olewe, Centre for Clinical Research, Kenya Medical Research Institute, Kisumu, Kenya and Center for Research in Therapeutic Sciences, Strathmore University, Nairobi, Kenya; Donwilliams Omuoyo, Centre for Geographic Medicine Research (Coast), Kenya Medical Research Institute-Wellcome Trust Research Programme, Kilifi, Kenya; John Ong’echa, Centre for Clinical Research, Kenya Medical Research Institute, Kisumu, Kenya and Center for Research in Therapeutic Sciences, Strathmore University, Nairobi, Kenya; Faith Osier, Centre for Geographic Medicine Research (Coast), Kenya Medical Research Institute-Wellcome Trust Research Programme, Kilifi, Kenya and Centre for Infectious Diseases, Heidelberg University Hospital, Heidelberg, Germany; Thomas L. Richie, Sanaria Inc., Rockville, MD, USA; Jimmy Shangala, Centre for Geographic Medicine Research (Coast), Kenya Medical Research Institute-Wellcome Trust Research Programme, Kilifi, Kenya; Juliana Wambua, Centre for Geographic Medicine Research (Coast), Kenya Medical Research Institute-Wellcome Trust Research Programme, Kilifi, Kenya; and Thomas N Williams, Centre for Geographic Medicine Research (Coast), Kenya Medical Research Institute-Wellcome Trust Research Programme, Kilifi, Kenya and Department of Medicine, Imperial College, London, UK.