Spatio-temporal heterogeneity in transmission of Plasmodium falciparum and Plasmodium vivax
Malaria parasite prevalence in the eastern and western regions was substantially different, with
P. falciparum dominating in the east and
P. vivax dominating in the west. This is likely to be due to the ecological differences between the two areas, such as a more intensive deforestation and lower rainfall in the western region, but also to differences in educational levels and the use of bed nets (both of which were lower in the east).
P. vivax has been reported to become more prevalent as transmission decreases [
27] and some studies have suggested that individuals with detectable
P. falciparum infections may also have underlying
P. vivax infections [
20], which become detectable once
P. falciparum infections are cleared through treatment. In addition,
P. falciparum infections can sometimes mask other infections when diagnosis is performed by microscopy, meaning mixed or other malarial infections remain unrecorded. Importantly, the life cycle of
P. vivax differs from
P. falciparum. In
P. vivax infections dormant liver forms (hypnozoites) can remain present for long periods of time, causing relapses months or years later. Consequently,
P. vivax can be more difficult to eliminate and therefore may become more prominent in areas where
P. falciparum has been successfully reduced (i.e., the western region).
CART analysis indicated that individuals in different villages in the eastern region experienced varying levels of
P. falciparum exposure over the rainy season. Interestingly, the village OZ, characterized as a forested village, was highlighted as having little malaria exposure between August and November. It would be expected that a more forested village would be subject to higher malaria exposure. According to the entomological results obtained in the study villages, the vector densities in OZ were similar to the other villages in the eastern part of the country (Durnez
et al, Manuscript in preparation) however, there was a particularly high usage of bed nets in this village (71% reported sufficient bed net use compared to an average of 40% in the other villages), suggesting that this may have reduced exposure successfully. Fifty percent (55/110) of the participants in villages BY and BZ (both of which are forested) increased two or more
P. falciparum serological categories, suggesting high transmission in these villages between August and November. The serological differences between villages highlight the heterogeneity in transmission at a micro level, as has been reported previously [
5].
There is also a different temporal pattern for the two species. The simple model that has been used in this study has previously been utilized in areas where sero-reversion is thought to be low [
13], in which case a single survey is enough to produce an estimate of SCR for the whole year. However, in this study, substantial differences in force of infection were detected between August and November for
P. falciparum in the eastern sites. This may be a consequence of highly seasonal yet low transmission meaning that the rarity of exposure to malaria parasites results in the loss of antibodies out of season, and any new exposure causes a large boost of antibody responses. This is demonstrated by the substantial increase in overall sero-prevalence and age sero-prevalence curves between August and November (Table
2 and Figure
2). The rates of GLURP sero-reversion (ranging between 0.05 and 0.08 in the east) predicted by the model are noticeably higher than those reported by Drakeley and colleagues [
13] (where sero-reversion was fixed at 0.02), although it should be noted that different antigens are used in this study and it is likely that sero-reversion is antigen specific.
Interestingly, the parasitological measures suggested that transmission had decreased in the eastern region between August and November. This discrepancy can be explained through the delayed acquisition of antibodies and the likelihood of treatment with subsequent clearance of parasites, resulting in less parasite positive individuals by the end of the season, whilst antibody responses may still be increasing. The data is suggestive of a
P. falciparum malaria peak, possibly occurring after August but finishing before November, resulting in lower parasitological levels, but increased sero-prevalence (Table
2). Another, not mutually exclusive, explanation for the discrepancy between the two measures is that not all current infections are detected by microscopy (sub-patent infections) [
1]. A previous study in Rattanakiri in the east detected a large proportion of sub-patent infections, the majority found in adults [
28]. A similarly high level of sub-patent infections was seen in a study in Central Vietnam (AE, personal communication). Sub-microscopic parasites are more likely to be present in adults who have developed partial immunity over time and may continue to stimulate production of antibodies.
In the western region
P. falciparum SCR remained relatively steady in both surveys, although there was still evidence that adults experienced a higher force of infection. In addition,
P. vivax force of infection did not change substantially across season in either region. A previous study in Rattanakiri also detected fluctuations in cases of
P. falciparum throughout the year, whilst
P. vivax cases remained steady [
14]. This effect has also been reported in Vanuatu [
17] and is supported by a study in Peru showing differences in seasonality between the two species, with
P. falciparum being dominant during the wet season and
P. vivax becoming dominant during the dry season [
29]. This is likely to be a result of
P. vivax relapses occurring in the dry season and may be an explanation for the different patterns of transmission reported in this study.
Higher force of infection in adults
The sero-prevalence curves for
P. falciparum consistently demonstrated a higher force of infection in adults compared to children. The age at which the increase in force of infection occurred ranged between six and 10 years old and was relatively consistent across season. For the eastern region, CART also split the data at approximately these ages with one split at nine years old and another at six years old. This suggests there was a significant difference in responses (rather than a continuous increase) between adults and children. A step in the sero-prevalence curve can be indicative of a recent reduction in transmission intensity [
4]. However, in this scenario, it would be unlikely to see any large increases in antibody responses as seen in this study. Instead, the data is suggestive of adults experiencing a higher force of infection than children. In line with this, the CART analysis was able to find fewer homogenous groups in the younger age groups, suggesting the risk factors that are present for adults, are not as pertinent for children. However, in the western region,
P. falciparum seropositivity in children was extremely low and deforestation in this area has resulted in dramatic reductions in malaria transmission. In this area, the two forces of infection may represent current transmission intensity (extremely low) and previous transmission intensity (low). Future studies in the west would allow further interpretation depending on whether the step in the seroprevalence curves remains at the same age (indicating behavioural differences) or shifts with time (indicating a drop in the force of infection).
Occupational risk of malaria in South-east Asia has already been extensively documented [
14,
15,
22,
30‐
32]. This study was not specifically designed to assess the effect of working in the forest on exposure and as such, it was difficult to separate the effects of age (and the increase in sero-prevalence due to cumulative exposure or biological differences) and working in the forest. Sero-positivity to
P. falciparum in children in the western region was extremely low, suggesting they had experienced very little exposure. SCRs increased above the age of 10 years old, which is a plausible age at which children begin to accompany their parents to work in the forest. However, the sero-prevalence data indicated that SCR in children in the more forested eastern region also increased over season albeit to a lesser extent (Figure
2), suggesting transmission was not wholly attributable to occupational risk but also occurred at village level.
Another explanation for the differences in serological responses between adults and children is that immunological maturity may play a role. Children may lose their malarial specific antibodies faster than adults [
12,
33,
34] - it is not clear if this is related to a lower cumulative exposure, i.e. having fewer episodes of antigenic stimulus, or to the immaturity of the immune system, or a combination of both). A longitudinal study in a highland area of Kenya demonstrated that children tended to lose their antibodies over the dry season, whilst adults retained high levels of antibodies throughout the year [
35]. These differences between adult and child serological responses could be due to adults supporting low levels of infection through the dry season which boost immune responses, to adults having antibodies with longer half-lives, or as a result of higher levels of circulating memory cells or longer lived plasma cells [
36].
CART also indicated that ethnicity played an important role in malaria exposure in the eastern region (only Khmer were analysed in the western survey). Both the Tumpurn and Charay ethnic groups appear to be at higher risk of exposure, compared to Khmer individuals. Traditionally, the ethnic minorities in this area are poorer than Khmer individuals. In the population sampled in the eastern region, 60% of Charays were classified in the poorest income group, compared with 45% of Khmer and 42% of Tumpurn. Poorer status may mean these individuals spend longer periods in the forest, and are therefore exposed for longer time periods. However, CART highlighted that ethnicity was also important in the higher income groups. This may be a reflection of the type of work undertaken by the different ethnic groups, as Khmer traditionally spend less time in the forest. Moreover in this study, the use of bed nets is much lower in the minority groups with over 50% of Charays reporting no bed nets in their household, compared with less than 1% of Khmers.
It is important to acknowledge that the serological data used in this study is based on a single antigen for each species. Although both the antigens used have been demonstrated to be relatively immunogenic [
23,
24,
37‐
39], in areas of low transmission it is important to test as many antigens as possible in order to detect all possible responses. Some people do not make antibodies to particular antigens [
34] and although this may not be a large problem when looking at population-based data, it could have an effect when looking at the individual level.