Background
Globally an estimated 3.4 billion people in 107 malaria-endemic countries are at risk of malaria, of which 1.2 billion at high risk [
1] live mostly in the African region (47 %) and the Southeast Asian region (37 %). To control and eliminate malaria, WHO recommends a multi-pronged strategy, which includes vector control interventions, preventive therapy, diagnostic testing, treatment with quality-assured artemisinin-based combination therapy (ACT) as well as strong epidemiological surveillance [
1]. Through upscaling of several elements of this strategy, many countries are on the verge of reaching pre-elimination. The low transmission rates in these areas pose considerable challenges for epidemiological surveillance [
2], hindering the evaluation of new (vector) control tools necessary to reach elimination [
1].
Detection of malaria-infected persons by microscopy, rapid diagnostic tests (RDTs) and even PCR lacks sensitivity because of low numbers of positive samples [
3], representing a formidable logistical challenge due to the very big sample sizes required in surveillance and evaluation surveys. While parasite-prevalence (measured by microscopy, RDTs or PCR) provides a snapshot of the exposure to malaria, the use of serological markers can provide a picture of the malaria transmission over a prolonged period [
3,
4]. Serology is based on the detection of antibodies (Abs) against antigens (Ags) of malaria parasites, which offers an advantage as anti-
Plasmodium Abs can persist for months after infection. Therefore, these Abs have been suggested as indicators of malaria transmission [
5‐
8], and are believed to be a better approach to determine past, recent and present malaria exposure [
9]. Previous studies performed in low transmission settings, such as Cambodia, also suggest that serological assays are promising for indicating malaria transmission [
3,
10‐
12].
Since the 1960s, serological markers detected by indirect immunofluorescence antibody tests (IFAT) were used to assess malaria transmission intensity and reductions in transmission [
13]. This has proven to be a reliable and useful serological test for malaria in epidemiological surveys [
5,
6,
14]. However, variation in source of Ags and the subjectivity of IFAT has led to this method falling out of favour [
4]. Standardized tests based on recombinant Ags used in an enzyme-linked immunosorbent assay (ELISA) were therefore developed [
4,
7‐
9]. However, an ELISA can only assess one marker at a time, making it labour intensive and time consuming when interested in multiple Ab responses. In the context of malaria elimination it will become essential to take into account individual variations in Ab responses, the occurrence of multiple malaria parasites [
15], as well as to increase the probability of measuring changes in Ab responses by combining different markers. Recently, several multiplex assays that were testing for different serological markers in the same blood sample, were developed by different research teams based on the Luminex technology [
15‐
18].
In this context, the general objective of this study was to implement an existing assay based on the Luminex technology for detection of Abs against malaria parasites in blood samples from Ratanakiri Province, Cambodia. This is the first and most extensive multiplex assay in malaria serology executed in the Southeast Asian region, including 20 Ags (recombinant proteins and peptides) directed against different specific malaria parasites. Furthermore, this study includes a detailed analysis on the stability of coated beads over time and the reproducibility of the beads coupling and immunoassay.
Discussion
This is the first implemented multiplex assay for malaria serology in Southeast Asia. Moreover, with the use of 20 specific Ags against different
Plasmodium parasites (Table
1), this assay is the most extensive multiplex assay published on malaria serology, besides Helb et al. [
32]. In contrast to most studies, the Abs that can be detected are not only directed against one or two different
Plasmodium parasites [
5,
10,
17,
24,
25,
33], but responses against
P. falciparum,
P. vivax,
P. malariae, and
An. gambiae are simultaneously detected. This is important in malaria elimination programmes, and especially in Southeast Asia where all four human malaria parasites as well as
Plasmodium knowlesi are co-occurring [
34]. Unfortunately at this stage, Ags specific for
Plasmodium ovale and
P. knowlesi were not available and thus not included in the assay. As the final aim of the study was to analyse 2000 field blood samples, the stability of the coated beads over time and the reproducibility of the assay were first thoroughly assessed and proved to be satisfactory.
Previously it has been shown that multiplex-based serological studies are useful for detecting anti-
Plasmodium Abs [
5,
10,
15,
17,
18,
24,
25] as well as Abs against other pathogens [
31,
35‐
38]. Also in the present study, the technique was shown to be reliable and fast in detecting several serological markers at once. All samples were analysed in less than 8 weeks of time, and more recently, over 8000 samples were analysed in duplicate in the same timeframe, showing the possibility to upscale the sample throughput. However, complexity, high investment costs in purchasing a Luminex Analyzer and need for special training, prevent widespread use as compared to ELISA [
5,
6]. A major restriction of this study is the use of the pool of control serum, as this has to be the same during the entire study, adequate quantities should therefore be prepared and stored in advance. In previous studies, satisfactory evidence has been presented on the validation of bead-based in comparison to ELISA [
5,
39], and this was, therefore, not further explored in the present study.
Several features were considered to assess the performance of the assay. In accordance to previous observations [
17], this study confirmed that a reduction of the beads to 1000 beads/Ag/well leads to similar results as compared to the amount of beads recommended by the standard protocols for Luminex-based multiplex immunoassays (2000 or 5000 beads/Ag/well). A rough cost estimation indicates that the use of 1000 instead of 2000 beads/Ag/well reduces the assay costs by one-third. Moreover, the results of previous studies [
5,
15,
17] in which monoplex assays resulted in highly comparable ΔMFI values as multiplex assays were confirmed. This refutes the concern on cross-reactivity between the different Ag-coupled beads or blocking of Ab responses [
15].
The obtained results of the performance assessment revealed that most of the coated beads were stable for up to 2 months, except for Ag PvCSP. This is important as screening approximately 2000 blood spot samples in duplicate took about 7 weeks from the first coupling until the last immunoassay (based on the results 1143 samples can be analysed per week). This stability is lower compared to previous studies, stating that beads are stable for up to 3 or 7 months [
5,
33], except for Ag STARP-R which was previously shown to be stable for only 1 month [
5].
Good reproducibility is an important aspect of an assay as it guarantees that results obtained with the assay performed on different plates and at different time points are comparable to each other. In the experimental set-up used in this study, there is no big difference in reproducibility between and within different plates as shown by similar RSDs. This is in line with other studies that showed accurate and repeatable results according to the interplate, intraplate [
5,
15,
27,
31], and intercoupling [
40] reproducibility of multiplex assays. In this study, the reproducibility was also assessed by looking at the results obtained from the positive control pool dilutions analysed in duplicate on each of the plates used for screening 2000 blood samples from Ratanakiri Province (total of 50 plates). A much smaller intraplate variability (RSD <3 %) was observed compared to the experimental set-up, which was probably due to the fact that samples were only tested in duplicate instead of six times in the experimental set-up. A higher interplate variability was observed, which was still acceptable (RSD <25 %) [
27], and most probably due to the use of only two positive controls per plate.
In the present assay the use of peptides and recombinant proteins was combined. Fourteen Plasmodium-specific peptides (CSP, STARP-R, SALSA1, SALSA2, SR11.1, LSA1-41, LSA1-J, LSA3-NR2, LSA3-RE, GLURP, GLURP-P3, PvlikeCSP, PvVK210CSP, and PmCSP), one peptide specific for the An. gambiae saliva protein (SALIV2), and five Plasmodium-specific recombinant proteins (Pf13, PfMSP1-19, PfGLURP-R2, PvDBP, and PvMSP1-19) were included in the immunoassay.
After assessment of its performance, the multiplex assay was applied to blood spot samples collected in Ratanakiri Province in Cambodia to detect Ab responses against 20 different Ags. The seroconversion rate (SCR) was then estimated per Ag by constructing age-seroprevalence curves based on a threshold approach. This is currently the most frequently used way of analysing serological data in malaria epidemiology [
4,
7], but which until the present was not applied to data from multiplex assays. By fitting models allowing two SCRs [
10], age groups exhibiting similar SCRs can be defined. For six out of 20 Ags (CSP, SALSA2, LSA3-RE, PfGLURP-R2, PvVK210CSP, and PmCSP), the models allowing two SCRs fitted better, and a higher SCR was observed in the older age group. This observation corroborates results obtained in previous studies in Cambodia using GLURP and MSP1-19 as serological markers [
9]. A lower SCR in the younger age groups may indicate a lower malaria exposure in the age group. This can be attributed to reductions in malaria transmission over time because of intensified malaria control interventions in the last years, or to different risk behaviour linked to the age groups [
9]. Moreover, another explanation would be that children might lose their malaria-specific Abs faster [
9]. Indeed, previous studies on malaria serology have shown that Ab responses in adults fluctuate to a lesser extent than in children [
41,
42]. The statistical model fit used to obtain the estimates for the serological parameters was constrained to have two constant (i.e. age-independent) SCRs and one constant sero-reversion rate (SRR). More elaborate model fitting incorporating age-dependent SCRs and SRRs fell outside the scope of this study.
Until the present very limited information was available on the half-life of the Ab responses to the Ags used and on the individual variation in those responses. This is crucial for interpreting the results obtained with this assay. Serological markers with a long half-life can be used for assessing past exposure, while a short half-life will be useful for assessing recent or even present exposure. The present study has obtained information on the Ab half-life estimated from the SRR as modelled in the reverse catalytic model [
43]. Based on this information, fluctuations in Ab half-life were observed per Ag, but all of them seem to have a relatively long half-life (ranging from 5 to >100 years). Most studies assume a fixed SRR independent of age and transmission rate [
30]. A study based on one Ag (PfMSP1-19) [
10] compared longitudinal and cross-sectional data (children ages ≤11 years), showing a high concordance of SCRs between both study designs, while large discrepancies were seen between the SRRs. A follow-up study based on a selected combination of Ags (10 out of 655) that compared Ab-responses at individual and population level also shows reliable rates of exposure [
32]. The present study is based on cross-sectional sampling and the obtained SRRs are probably not reliable. More extensive analysis on the half-life of these Ab responses is planned by using sequentially collected blood samples from the same individuals, as well as by exploring which of the Ab responses can best capture the fluctuations in malaria prevalence and incidence observed over time. SCR estimates based on a single serological marker can give a good indication of the force of infection [
4,
7]. However, combining the signal from multiple markers representing different life-stages of the parasite in an individual could help reducing individually heterogeneous responses to one particular marker, thereby providing more accurate estimation of the force of infection. A recent follow-up study conducted on a cohort of children demonstrated that measuring Ab-responses to a selection of few Ags provide an accurate estimate of exposure for individuals and communities [
32]. Therefore, further data analysis performed with multivariate analysis should provide added value. Moreover, presently it is not possible to exploit the full potential of the assay to take into account individual variation in Ab responses, as analysis of the SCR is still carried out Ag by Ag. Data analysis methods to combine Ab responses for estimation of SCRs exist [
44] and are currently being extended in the context of this multiplex assay for detection of anti-
Plasmodium Abs.