The current study adds to the available evidence on the wide-scale presence of submicroscopic parasitaemia. This study quantified submicroscopic parasites and concurrent gametocytes using highly sensitive molecular assays in asymptomatic schoolchildren. Whilst only 6 infections were detected by microscopy in 881 slides, 107 of 845 blood samples were parasite positive by qPCR. A higher prevalence of malaria infections by molecular methods is commonly reported and the proportion of infections that is submicroscopic is highest in low-endemic settings [
2,
3]. Nevertheless, the 18-fold higher parasite prevalence observed by qPCR compared to microscopy is remarkable [
3]. Median densities were <10 parasites/μL for both
P. falciparum and
P. vivax, below the density of 50–100 parasites/μL that is detectable by routine microscopy [
32]. Expert microscopy may detect lower parasite densities but even expert microscopy may only have a sensitivity of ~29% for detecting parasite densities in the range of 1–10 parasites/μL [
33].
The findings further demonstrated that approximately 10% of all
P. falciparum infections and 60% of all
P. vivax infections had concurrent detectable low-density gametocytaemia [
9,
11]. For both
Plasmodium species gametocyte densities were positively associated with total parasite densities [
11]. In
P. falciparum gametocytes typically comprise <5% of the total parasite biomass. With the very low parasite densities that were detected in this study, it is conceivable that concurrent gametocyte densities were too low to be detected by qRT-PCR. On the other hand, the finding that some individuals were gametocyte positive but negative for parasites by qPCR is likely to reflect the difference in sensitivity between DNA- and RNA-based assays [
4] and illustrates that some infections could have been missed by qPCR [
5]. This might partly explain infection of mosquitoes even when no parasites were detected by molecular methods [
34]. Higher blood volume PCR [
5] or more sensitive PCR targets [
35] may reduce the proportion of false-negative PCR samples although it is likely that a fraction of infections will remain undetected with currently available diagnostics [
5].
The current study identified heterogeneous transmission of
Plasmodium species in the five study sites. At four sites both
P. falciparum and
P. vivax were prevalent; the former was the dominant species. In one highland site, Ahuri at 2010 metres above sea level, only
P. falciparum infections were detected. Although parasite prevalence by qPCR was relatively low in Ahuri, it remains to be demonstrated whether infections were acquired locally. The finding may reflect a shift in altitude in
P. falciparum distribution. In warmer years, an increase in the altitude at which malaria is observed has been reported for Ethiopia and Colombia [
36]. The current findings may also be influenced by climatic changes, including the El Ǹino phenomenon that occurred during the study year and resulted in intermittent rain and increased temperature across the region. Intriguingly, antibodies to
P. vivax AMA-1 and MSP-1
19 were detected in approximately 9% of children in Ahuri. This may reflect past exposure to
P. vivax [
31], exposure outside the area of residence,
P. vivax exposure that is too infrequent to be detected in parasitological surveys [
17], or cross-reactivity of antibodies acquired following
P. falciparum exposure. For
P. falciparum, antibody responses were associated with
P. falciparum infections measured by qPCR. Individuals who were parasite positive in both surveys had higher prevalence and density of
P. falciparum antibodies [
37]. It is unclear whether this reflects boosting of antibody responses by circulating low densities of parasite antigens [
38] or the fact that current infections are strongly associated with cumulative past malaria exposure [
39]. Parasite prevalence and the incidence of new infections in the study populations was too low for meaningful assessments of serological markers of malaria exposure as reliable indicator of recent malaria exposure or explore the longevity of antibody responses in relation to time since last infection. However, the current findings of an association between antibody prevalence and qPCR-detected infections and qRT-PCR detected gametocyte carriage, suggest that serological markers might be useful to identify
P. falciparum exposed individuals or populations whose infections are undetectable by microscopy but may contribute to onward transmission. It is currently unclear why this association was not observed for
P. vivax and more detailed longitudinal studies are required to robustly demonstrate the ability of serological markers of malaria exposure to detect (recent or current) malaria infections. Recently described serological markers of recent malaria infection [
40] have the potential to perform considerably better in this respect since antibodies to AMA-1 and MSP-1
19 have a long half-life [
29] and may be less informative to indicate ongoing malaria transmission in settings where transmission has declined. The age-range of the current study population was too narrow to generate a full age-seroprevalence curve that may demonstrate whether recent changes in malaria exposure have occurred [
31,
41]. The association between antibody responses to
P. falciparum blood stage antigens and
P. falciparum gametocyte carriage is plausibly a consequence of a larger total parasite biomass in gametocyte carriers [
11] that forms a stimulus for antibody production.
The relevance of the low-density parasite reservoir that was described in the current study remains to be established. Whilst chronic low-density infections may have clinical consequences [
6], most attention goes to their possible role in the human infectious reservoir for malaria [
10,
13]. The current study demonstrates that a proportion of the low-density infections had detectable gametocyte densities. For
P. falciparum this proportion was only 10% which is lower than commonly reported in parasite carriers detected in cross-sectional or clinical surveys [
9]. It may be argued that these infections may produce detectable gametocytes at a later stage, but the dynamics of gametocyte production and infectivity of submicroscopic infections is currently unknown. By comparison, the prevalence of gametocytes in
P. vivax infections was higher, as has been described before, and may reflect the more rapid production and maturation of
P. vivax gametocytes [
9,
11,
42]. Efforts are currently ongoing to establish mosquito feeding facilities in the study area to determine the contribution of symptomatic, asymptomatic and submicroscopic infections to onward transmission to mosquitoes.