Background
In recent years, there has been a decline in malaria transmission in many regions, leading to optimism that malaria elimination might be achieved in numerous countries [
1‐
8]. As transmission declines, monitoring changes in malaria transmission intensity and disease prevalence through surveillance systems becomes increasingly important to allow the evaluation of health services and control programs [
9,
10]. The latest World Health Organization (WHO) malaria surveillance manual confirms that improved surveillance is a major component of the WHO strategy [
11]. However, conventional measures such as entomological estimates and parasitaemia point prevalence become less sensitive and relatively more expensive as transmission declines [
12,
13]. Disease surveillance is further compounded by difficult access to remote and isolated communities, increased risks in forest workers and other highly mobile populations and the difficulties of tracking cross-border movements [
14‐
20].
An additional approach to measure malaria transmission is to detect anti-malarial antibodies, which provide a marker for exposure to malaria [
9]. Malaria infections generate antibodies which can be detected for several months and years after the infection has been resolved. Although serology is unlikely to be useful for diagnosing actively infected individuals because antibodies take days to develop and then persist after infection [
9,
13], detection of these antibodies indicates previous exposure and offers an additional, more sensitive measure of infection and transmission, particularly in low endemic settings where the sensitivity of parasitological tools is inadequate [
21‐
25] and gold standard tests like the parasite rate and the entomological inoculation rate (EIR), may have insufficient statistical power unless the sampling is intensively done [
26‐
28]. This approach has been utilized in several countries and reported as a more sensitive tool to assess population-level malaria exposure in low-transmission settings [
9,
13].
Seroconversion rate (the proportion of people in the population who are expected to seroconvert each year) is a serological parameter used to understand malaria transmission dynamics. Previous studies found that seroconversion rate (SCR) provides a proxy measure for estimating the transmission intensity in a community as it was strongly correlated with the EIR and annual parasite incidence collected by the malaria surveillance programme [
10,
14]. Serological estimates of transmission have been utilized in many low endemic settings, including Indonesia [
29,
30], and have additionally been used to identify populations at higher risk of malaria exposure [
9,
31], foci of transmission [
32,
33] and to describe historical changes in disease burden [
25]. While there is great promise in this approach, it needs further refinement.
Recent studies have reported the potential use of recombinant Merozoite Surface Protein 1 (PfMSP-1
19) and Apical Membrane Antigen 1 (PfAMA-1) as serological parameters to assess malaria transmission intensity in Indonesia. First, a population-based cross-sectional study conducted in three different endemicity areas showed the potential application of these methods for detecting changes in transmission exposure, particularly in lower transmission settings and with less immunogenic antigens (such as PfMSP-1
19) [
30]. Second, a cohort study of Indonesian schoolchildren found that it is possible to assess the interruption of transmission by measuring seroconversion rates from individual-level longitudinal data on antibody titres [
29]. These studies suggested serological analysis has the potential to assess malaria burden and heterogeneity of infections in the Indonesian population. As antibodies to AMA-1 and MSP-1
19 antigens have been reported to persist for several years after infection and in the absence of reinfection, any antibodies detected in younger children would reflect more recent infection in low transmission settings [
10]. Therefore, as Indonesia aims to eliminate malaria by 2030, further implementation and evaluation of sero-epidemiological analysis in areas moving towards elimination would garner valuable information for malaria control programmes. This study explores the use of sero-epidemiological analysis for assessing the intensity and heterogeneity of malaria transmission as well as factors associated with malaria exposure in an area conducting elimination in Indonesia.
Discussion
This study describes the analysis of community-based serological data to investigate malaria transmission dynamics in a low transmission setting, Sabang, Indonesia. The seroprevalence and SCR data represent exposure to infection and demonstrate that the population level of transmission intensity were similarly very low for both
P. falciparum and
P. vivax. The seroprevalence in children under 15 years old was negligible, 1.2% and 0.5% for
P. falciparum and
P. vivax, respectively. The spatial analysis of household-level data on antibody responses to any of the antigens tested describe the heterogeneity of both
P. falciparum and
P. vivax exposure in the study area. These results supported previous utilization of sero-epidemiological analysis in assessing population–level transmission intensity and differentiating between areas of different endemicity in Indonesia [
30]. Moreover, multivariable analysis utilizing serological and epidemiological data collected through community-based survey identified that age,
P. vivax seropositivity and LLIN use were significantly associated with
P. falciparum seropositivity. These associations are likely related to historical exposure as
P. falciparum seroprevalence was estimated to be low and parasite screening found no active infections detected by microscopy. Although sub-microscopic infections might present in the community, a previous study suggested that the proportion of sub-microscopic infections detected via PCR (polymerase chain reaction) was very low 0.07% (11/16,229) in the region [
34].
The
P. falciparum SCR estimates suggest that there was no exposure seen in children under 5 years old in both sub-districts in Sabang municipality. These results could represent a step change in
P. falciparum transmission due to the successful impact of malaria control programme implemented in the study area, evidenced by lower antibody prevalence in children born after the intervention scale-up. This finding was supported by a previous study documenting a significant drop in malaria cases after the launch of the control program in 2004. Malaria cases in Sabang declined from 88 cases per 1000 population in 2004 to 1 per 1000 by 2010. The decline in malaria transmission in Sabang is likely related to an extensive IRS programme immediately following the tsunami in 2004, large scale LLIN distribution, and a change in malaria treatment policy to artemisinin-based combined therapy as first-line treatment for uncomplicated malaria [
34]. Sabang was certified as a malaria-free region by the Indonesian Government as a result of successfully maintaining zero cases since the last locally transmitted case reported in 2011. Since then, the surveillance system detected 12 imported cases consisting of 6
P. vivax,
4 P. falciparum and 2 mixed
P. vivax and
P. falciparum infections from 2011 to 2013, with no local transmission. However, the surveillance system detected 15 PCR confirmed
Plasmodium knowlesi infections that classified as an outbreak in 2014 [
38].
Consistent with the higher
P. falciparum SCR estimates in people over 5 years old, multivariable analysis revealed that adults were more likely to be seropositive compared to children under 15 years old. This is likely the result of higher exposure by staying overnight in high-risk areas. A recent study revealed that the clusters of malaria (
P. knowlesi) infections in Sabang was associated with people who had a history of staying overnight in the forest, without protection from mosquitoes, in an area where macaques are common [
38]. Unfortunately, data on travel behaviour and occupation in these surveys were not recorded to enable testing of these hypotheses. Future research would need to include more detailed questions regarding travel behaviour, occupation and other essential risk factor data such as travel history to high-risk areas, night outdoor activities, sleeping in plantation or forest, housing, personal protection, etc. Several programme initiatives, for example a multi-country study on vector control tools to address outdoor transmission and project management quality improvement for national malaria program workforce carried out under the Asia Pacific Malaria Elimination Network would be beneficial for the malaria elimination effort in the region. In addition, the use of LLIN was almost two times higher in area where
P. falciparum seroprevalence was higher. Consistent with previous report suggesting high coverage of LLINs (over 75%) in six malaria focal villages in Sabang, this finding suggests that people living in higher risk of exposure were aware of the importance of LLIN to prevent malaria transmission in those areas [
34].
The estimated age-seroprevalence curves and SCR value suggested that age was not associated with
P. vivax transmission in either sub-district in Sabang.
Plasmodium vivax seroprevalence was very low (2.0%) and, therefore, the absence of any associations is likely due to the statistical limitations of the low number of seropositive samples. The other possible explanation is that
P. vivax infections may induce lower antibody responses or shorter-lived responses which the current assay may miss. Work is ongoing to identify
P. vivax antigens that elicit short-term responses for easy identification of very recent exposure [
39,
40]. The need for testing more potential
P. vivax antigens is supported by a previous study showing that the number of
P. vivax cases tend to be higher than the number of
P. falciparum cases in Sabang [
34].
The spatial analysis of age-adjusted antibody responses to either antigen (AMA-1 or MSP-1
19) identified significant clusters of higher exposure (hotspots) for both
P. falciparum and
P. vivax exposure across the study areas. Although multivariable risk factors analysis found there was no significant association between residence and higher seroprevalence to
P. falciparum and
P. vivax, the spatial analysis suggested that the risk of malaria transmission in the study setting is heterogeneous with people experiencing higher exposure in Sukajaya sub-district. The spatial analysis also suggest that the clusters identified for
P. falciparum and
P. vivax were seen in the same areas. Being able to characterize the micro-epidemiology of malaria exposure could assist malaria control programme to better allocate resources and target the intervention to achieve their goal of elimination. Targeting hotspots could be a highly efficient way to reduce malaria transmission at all levels of transmission intensity [
41]. Although this study identified potential high-risk areas using historical data collected in 2013, being able to identify areas which had the most recent exposure is useful for malaria surveillance. A recent study suggested that one of two clusters of
P. knowlesi infections in Sabang were identified in similar high-risk areas identified in this study [
38]. As suggested in the latest WHO malaria surveillance manual [
11], maintaining surveillance activities in the most receptive areas could be useful to prevent potential reintroduction or resurgence of the disease in the future. Therefore, utilizing antibody responses data to identify recent or historical hotspots of transmission could be a powerful alternative approach where gaining direct evidence of an increased exposure to infectious mosquito bites is no longer ideal in low transmission settings.
Finally, people who were seropositive to any
P. vixax antigen were 3 times more likely to be
P. falciparum seropositive, after controlling for age, gender, residence, employment, education, IRS, fever status, and altitude. In addition, clusters of high antibody responses suggest that
P. falciparum and
P. vivax receptive areas were seen in the same areas. As there was no cross-reactivity evident from the serological data (Additional file
3), these findings could suggest that people were historically exposed to both infections, potentially due to the presence of efficient vectors in those identified areas.
Findings in this study are based on community-based samples and data collected during the malaria transmission season. Although this study describes the potential use of serological data analysis in estimating malaria transmission intensity, heterogeneity and factors associated to disease exposure, the results generated would need to be carefully interpreted. Previous studies suggested that malaria transmission in other areas of Indonesia was affected by seasonality [
30,
34,
42‐
44] and behavioural factors such as farm or forest-related night outdoor activity (e.g. sleeping in forest gardens) [
45,
46] and domestic travel to higher endemic areas [
47]. However, due to limited data collected, our study could not examine the effect of behavioural factors such as forest-related activities or recent travel history to high-risk areas outside Sabang. Therefore, future studies measuring population level antibody responses coupled with collecting more data that could describe behavioural factors associated to higher risk of exposure would be more epidemiologically informative to assist malaria surveillance and control programme to achieve elimination in the region.
Conclusion
In conclusion, these data add to the body of evidence that sero-epidemiological analysis of community-based surveys are an important additional tool to investigate malaria transmission dynamics in area aiming for elimination in Indonesia. Recent identification of alternative antigens associated with short-lived antibody responses suggests a potentially key indicator of very recent exposure which would be a very important information for public health surveillance [
48]. The addition of a novel panel of
P. knowlesi antigens [
49] would enhance understanding of malaria transmission dynamics as recent studies reported that although laboratory identification of
P. knowlesi in Indonesia is challenging [
50], surprisingly, there were two clusters of
P. knowlesi cases detected in Sabang after the municipality successfully eliminated
P. falciparum and
P. vivax cases [
38]. Moreover, another recent study also reported there was a considerable proportion of
P. knowlesi infection in another western part of Indonesia, in North Sumatera province [
51]. Exploratory work employing techniques such as multiplex fluorescent magnetic bead-based serological assay to investigate and validate a panel of potential antigens for these applications is underway [
40,
52]. The development and validation of a standardized serological sample and data collection methods utilizing existing public health surveillance system, for example as described in [
53] will also facilitate the optimization of serological surveillance in understanding transmission dynamics to support malaria control programme in achieving elimination.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.