Background
Plasmodium vivax is the most widely distributed species of human malaria, with an estimated 40% of the world's population being at risk of
P. vivax infection [
1]. The majority of
P. vivax infections occur in Central and South-East Asia, and there are approximately 80 to 300 million clinical cases of
P. vivax malaria each year [
1]. Despite the large burden of disease,
P. vivax has traditionally been neglected because it has been considered a relatively benign form of malaria. Now it is recognized that
P. vivax can cause severe disease (similar to that of
Plasmodium falciparum) [
2] and, together with increased recognition of the burden, there are renewed efforts in the development of
P. vivax-specific interventions (that is, vaccines) and surveillance tools (diagnostics, serosurveillance) to expedite the goal of malaria elimination and eradication [
3],[
4].
Currently, only two
P. vivax vaccine candidates (
Pv Duffy binding protein (
PvDBP) and
Pv circumsporozoite protein
PvCSP) are in clinical trials (Phase I) compared with 23
P. falciparum vaccine candidates (including one in Phase III trials: RTS,S) [
5],[
6]. This may reflect the previous neglect of
P. vivax, the difficulty in maintaining
P. vivax in culture, and the limited animal models of infection currently available. Such technical challenges have hindered the ability to prioritize
P. vivax candidate antigens against pre-clinical selection criteria, including knowledge of protein function and antigenic diversity, and demonstrations that antibodies against an antigen inhibit growth
in vitro or function in other ways, or are protective in animal models of infection [
7]. In the absence of an
in vitro system,
P. vivax antigens can be selected based on
P. falciparum homologues and an additional pre-clinical selection criterion, namely, that the antigen induces naturally acquired immunity in individuals living in malaria-endemic areas [
7].
In
P. vivax-endemic areas, the prevalence and density of
P. vivax infection and the incidence of
P. vivax symptomatic malaria decrease with age [
8]. This epidemiological pattern reflects the acquisition of natural immunity that develops after repeated exposure [
9]. This immunity is non-sterilizing and does not protect against infection, but acts by reducing parasite numbers in the blood and the subsequent clinical symptoms. Antibodies are thought to be an important component of naturally acquired immunity, and are considered to be biomarkers of both immunity and exposure. Potential antibody targets include
P. vivax antigens expressed on sporozoites (the pre-erythrocytic liver stage), the invading merozoite and the surface of infected erythrocytes (erythrocytic stage) and the gametocyte (sexual stage) [
8].
P. vivax also has an additional dormant stage in the liver, the hypnozoite, which is believed to be responsible for relapses in
P. vivax infection [
10].
There have been numerous studies investigating associations between
P. vivax immune responses and
P. vivax infection, but there is considerable heterogeneity between studies, both in terms of methodology and presentation of results, making cross-study comparison problematic. Here, we aimed to review and synthesize the literature, by standardizing analyses and identifying targets of naturally acquired immunity to
P. vivax, which we have previously done similarly for
P. falciparum[
11]. There were two key objectives of this study: to determine antigen-specific antibody responses associated with infection, and to determine antibody responses associated with protective immunity. We included cross-sectional and case-control studies in order to identify markers of
P. vivax infection, and also included cohort studies, which provide the highest level of evidence to detect causal effect in observational research, in order to identify antibody responses that protect against
P. vivax malaria. The overarching aim of the study was to provide a more comprehensive understanding of antibody-mediated immunity to
P. vivax and, more specifically, to help inform the development of vaccines and serosurveillance tools to facilitate the control, elimination and eradication of
P. vivax.
Discussion
In this systematic review, we aimed to identify immunological biomarkers of
P. vivax infection and protective immunity by standardizing estimates of the association between
P. vivax antibodies and
P. vivax outcomes across populations. We found a paucity of studies investigating associations between antibody responses to
P. vivax antigens and risk of
P. vivax, particularly cohort studies, and studies conducted in the Asia-Pacific [
1]. Although there was considerable heterogeneity between studies, antibody responses to several antigens were associated with
P. vivax infection and protective immunity to
P. vivax. However, this review highlights the need for additional studies, and identifies several issues in the interpretation and reporting of data from epidemiological studies investigating immunity to
P. vivax.
Studies included in the review represented diverse geographical populations living in areas of varying
P. vivax endemicity. However, the geographical regions and countries represented were limited. Half of the studies provided data from the Asia-Pacific region, which represents 91% of the population at risk of
P. vivax malaria [
1], but only four countries were represented (Indonesia, Turkey, Thailand, and Papua New Guinea). The remaining half of the studies provided data from South America, representing only 6% of the population at risk of
P. vivax malaria [
1], but all were performed in Brazil. This predominance of data from Brazil has implications for the generalizability of findings to other
P. vivax-endemic regions in South America and the Asia-Pacific. However, despite the population heterogeneity and the considerable heterogeneity in estimates observed, immunological markers of
P. vivax infection could be identified: IgG responses to
PvCSP,
PvMSP-1
19,
PvMSP-9
NT, and
PvAMA1 were associated with increased odds of
P. vivax in geographically diverse populations. Other antigens were also shown to be markers of
P. vivax infection, but only in single populations (
PvMSP-3α,
PvMSP-9
RIRII,
PvDBP, and
PvRBP1). Serosurveillance using
PvCSP in Korea [
63]-[
67] and
PvMSP-1
19 and
PvAMA1 in Vanuatu [
68], Cambodia [
69], and Somalia [
70] has been employed to successfully map
P. vivax transmission, and data from this review support their use in serosurveillance campaigns. However, this review highlights that further studies, conducted in diverse geographical settings and including additional antigens, are needed to ensure the generalizability of results across different populations with variable
P. vivax transmission.
Protective immunity could only be examined in a handful of cohort studies, all of which showed evidence for protective blood-stage antibodies targeting
PvMSP-1
19,
PvMSP-1
NT,
PvMSP-3α and
PvMSP-9
NT antigens but only in single geographical locations. This was also the case for
PvDBP, a prime vaccine target (because of its essential role in invasion) [
44] that is currently in Phase I trials [
5].
PvDBP was examined in only two cohort studies (which looked at different regions) and only Cole-Tobian
et al.[
34] showed evidence of allele-specific
PvDBPII protective immunity against high-density parasitemia. Interestingly, no cohort study examined the protective effect of antibody responses to either the pre-erythrocytic antigen
PvCSP or the gametocyte antigen
Pvs25. Both of these have previously been assessed in Phase I trials [
6],[
71]-[
73], and
PfCSP comprises the current Phase III
P. falciparum vaccine RTS,S, which has demonstrated around 50% efficacy in young children and around 30% efficacy in infants [
15],[
16]. This review shows that very few antigens meet the pre-clinical criteria for prioritizing candidate antigens (targets of protective immunity in humans) for vaccine development, which is particularly pertinent given the difficulties in meeting other
in vitro pre-clinical criteria (demonstrating essential/important function, abundance, limited genetic diversity, inhibition of parasite growth, protection in animal models of infection) [
7] because of difficulties in maintaining
P. vivax in culture. In order to prioritize antigens for
P. vivax vaccine development, further studies including additional antigens and established, clinically relevant end-points (for example, allele-specific responses with allele-specific end-points) are needed to provide valuable evidence for the role of particular
P. vivax antigens in protective immunity.
The considerable heterogeneity observed in the estimates of association, which meant that the magnitude and the direction of effect estimates from different studies varied considerably, was a major issue in the meta-analyses, such that study estimates could not be reliably combined in some instances. Methodological diversity between studies may have contributed to the heterogeneity: antibody responses were measured in different ways (alleles, antigen preparation);
P. vivax infection was determined using detection methods of varying sensitivities (PCR is more sensitive than light microscopy); and statistical methodology varied. Furthermore, the estimates from the majority of studies were unadjusted for potential confounders, and within-study bias may also have contributed to the heterogeneity observed. Transmission micro-epidemiology within study sites may be an important confounder, biasing the direction of effect in either way: individuals living in areas with the highest
P. vivax exposure will acquire both biomarkers of exposure and protective immunity, but will also be at increased risk of future
P. vivax infections. Study design may also be an important source of heterogeneity. The majority of studies were cross-sectional or case-control studies in which antibody responses and
P. vivax outcomes had been determined at a single time point, in those with or without
P. vivax outcome. Although we used this study design to identify immunological markers of
P. vivax exposure, using data from a single time point has the potential to also capture a degree of protective immunity in the population. Indeed, we observed these types of divergent associations for several antigens, including studies by the same authors using the same methodology both in different populations [
22] and within the same population [
24],[
26]. These findings highlight the limitations of using cross-sectional data, particularly when interpreting and comparing data across populations with varying degrees of
P. vivax endemicity and immunity.
Differences in
P. vivax transmission and exposure history will result in differential acquisition of immunity, which will influence associations between
P. vivax antibody responses and clinical outcomes. To reduce bias in the systematic review, we excluded studies on transmigrants and studies in which the majority of the population resided in a malaria-endemic area for a short time. This bias was highlighted in two studies, which met the respective inclusion criteria, both by Lima-Junior
et al. and performed in the same region of Brazil [
24],[
26].
PvMSP-9 IgG responders were found to have increased odds of
P. vivax infection in 2008 [
24], but decreased odds of infection in 2012 [
26]. However, the population composition changed between the two studies: in 2008, 82% of participants were indigenous to the malaria-endemic area, compared with only 59% in 2012 (J. Ferreira, personal communication). This may explain, in part, the differences observed, because in both studies, time of residence in the malaria-endemic area was positively correlated with the anti-
PvMSP-9 response [
24],[
26]. Differential effects according to transmission were also anecdotally observed: one study by Yidez-Zeyrek
et al. in Turkey [
39] showed greater magnitudes of effect with IgM than with IgG responses (
PvMSP-1
19 and
PvSERA4), indicating that individuals living in this
P. vivax-endemic area had limited exposure to
P. vivax. Interestingly, the ability of IgG to serve as a marker of exposure in this study was more than twice that of estimates from areas of higher
P. vivax transmission (in Brazil and PNG), highlighting the potential for transmission intensity to influence results. Future studies should be aware of the potential confounding introduced by variations in
P. vivax exposure and transmission intensity, particularly those conducted in areas in which
P. vivax epidemiology is complicated by the presence of migrant workers or transient communities, which is common in
P. vivax-endemic areas in South America and South-East Asia.
This review aimed to be as comprehensive as possible, and to identify all data by which an association between P. vivax responses and P. vivax outcomes could be examined. By contacting authors directly, we were able to obtain data from a further 11 studies for which data was not originally published. Commonly, these studies were descriptive in nature, comparing antibody prevalence in P. vivax infected versus uninfected individuals, with no quantification of the magnitude of effect. Consequently, many included studies were not sufficiently powered to detect a statistically significant association between antibody responses and P. vivax outcomes. Publication bias may also be an issue in the P. vivax immunity literature, which could not be assessed in this review because of the small number of studies in each analysis.
In this review, we also included total IgG subclasses, as well as IgG and IgM, to infer potential functional mechanisms, with similar associations seen with subclasses as to total IgG (see Additional file
5). IgG1 and IgG3 were the predominant subclasses to
P. vivax antigens, and may function by opsonic phagocytosis [
74], or by fixing complement. We found only one study that utilized a functional assay: King
et al.[
35] showed that binding inhibitory antibodies to
PvDBPII were associated with protection from
P. vivax infection. The lack of a continuous culturing system for
P. vivax currently prohibits the use of most types of functional assays, but will clearly be important in future studies to determine the relative role of various immune mechanisms in protection against
P. vivax.
Acknowledgments
We thank Robyn Damary-Homan and Cameron McPherson for pilot literature searches and data extraction. We also thank John Adams, Mercia Arruda, Sukla Biswas, Tom Burkot, Luzia Carvalho, Cevayir Coban, Giuseppe Del Giudice, Yagya Dutta Sharma, Annette Erhart, Marcelo Ferreira, Daniel Gordon, Patricia Graves, Stephen Hoffman, Trevor Jones, Flora Kano, Chris King, Peter Kremsner, Kee-Hyoung Lee, George Lewis, Josue Lima-Junior, Ivo Mueller, Paulo Noguira, Ruth Nussenzweig, Joseli de Oliveira-Ferreira, Danielle Stanisic, Tuan Tran, Eileen Villasante, Chansuda Wongsrichanalai, and Sedigheh Zakeri for providing information and/or data for the systematic review.
Financial disclosure
This work was supported by the National Health and Medical Research Council of Australia (project grant, fellowships to FJIF and JGB, and an Infrastructure for Research Institutes Support Scheme Grant), the Australian Research Council (Future Fellowship to FJIF and JGB), and a Victorian State Government Operational Infrastructure Support grant. The funders had no role in study design, data collection or analysis, decision to publish, or preparation of the manuscript.
Competing interests
The authors declare that no competing interests exist.