This study investigated the extent of low-density P. falciparum and P. vivax infections in moderate and low transmission settings in Thailand, Brazil and PNG, using ultra-sensitive detection assays. The LODs of the P. falciparum and P. vivax us-qPCR assays, which was 10-times lower than the standard assay Pf_18S rRNA qPCR [10,11] were in good agreement between the three laboratories. The inter-laboratory comparability of results was closely controlled by the use of external reference standards. To evaluate the usefulness of us-qPCR assays for epidemiological studies, age trends in prevalence rates were analysed. No differences were observed in age-dependent epidemiological pattern in prevalence nor density in any transmission setting, thus confirming the suitability of us-qPCR for epidemiological research.
Substantial differences in gains were observed across transmission settings. For example, at the PNG study site, P. falciparum prevalence in the community increased from 8.6% to 12.2% by us-qPCR, an increase of 30% in P. falciparum positivity. In Brazil P. falciparum prevalence was much lower and a two-fold gain was observed in the proportion of positive samples by use of the ultra-sensitive varATS assay. However, the absolute numbers of P. falciparum infections in this comparative study were too low (i.e. a rise from 6/651 to 11/651 P. falciparum-positive individuals) so that the difference between study sites in the proportional gains was not statistically significant. The question whether proportional gains are higher at very low P. falciparum transmission intensity remains to be clarified by a larger study with a sufficient P. falciparum sample size for statistical analysis. Gains that differ by transmission intensity could occur if low-density P. falciparum infections undetectable by standard Pf_18S rRNA qPCR, would be relatively more abundant in low compared to high transmission. Earlier meta-analyses of global prevalence data of both, PCR- and microscopy-based diagnosis [1,16] had revealed a negative relationship of the proportion of submicroscopic infections with transmission intensity.
The seemingly lower proportional gain in PNG compared to the Brazilian site (30% versus 55%) might be related to slightly higher overall P. falciparum parasite densities in community samples from PNG (1.9 parasites/µL) compared to Brazil (0.4 parasites/µL). The small difference in median P. falciparum densities between PNG and Brazil is in proximity to the LOD of the Pf_18S rRNA assay (1.57parasites/µL blood) and therefore could have caused a higher proportion of infections in PNG samples to be detectable by standard Pf_18S rRNA qPCR. Whether density differences between sites could explain the smaller gains in PNG by using us-qPCR remains unclear, as the limited number of P. falciparum cases from low endemic Brazil precludes clarification of reasons for differing proportional gains.
Among all three sites, the P. falciparum median density was highest in Thailand with 238 parasites/µL compared to 4.0 parasites/µL in Brazil. This was due to very few high density infections with >1000 parasites/µL in this community. In addition, 3 of these P. falciparum-infected cases showed evidence of fever within the preceding 2 days. These findings from Thailand could mirror declining transmission and indicate that less frequent exposure to P. falciparum may result in waning immunity against this parasite. This would suggest that P. falciparum-infected individuals at the Thai study site more often develop high parasite densities and become symptomatic and thus will receive treatment. It, therefore, can be speculated that the presence of malaria episodes has caused a lower proportional gain of infections detected by us-qPCR in Thailand compared to low-endemic Brazil.
Pv_mtCOX us-qPCR resulted in an analogical increase in the proportion of additionally detected P. vivax-infections across all study sites, ranging from 31% in PNG to 24% in Brazil and 23% in Thailand. Compared to the more pronounced differences in P. falciparum proportional gains, the Pv_mtCOX-based increase was similar across the 3 sites. When P. vivax median densities were consulted to explain the small difference in proportional gains between study sites, 11-16-fold higher P. vivax densities were observed in Brazil and Thailand compared to PNG. This might be responsible for the lower gain by more sensitive diagnostics in these two low endemic settings.
Benefits from us-qPCR consist in the improved detection of low-density P. falciparum and P. vivax infections. The question arises, whether these low-density infections carry gametocytes in sufficient numbers to contribute to onward transmission to mosquitoes. This question has been addressed by previous studies at all three sites, either by membrane feeding assays on symptomatic blood samples [20,21] or by using gametocyte density as surrogate for transmission potential, such as in a study recently performed in the same community in PNG, where gametocytes were quantified after enrichment from large volumes of blood [12]. In this recent study, more than half of the P. falciparum- and P. vivax-positive samples that were only detected by us-qPCR, carried gametocytes, (10 gametocyte carriers in 15 P. falciparum-infections and 11 gametocyte carriers in 19 P. vivax-infections) [12]. Gametocyte densities were 10-fold lower in infections detected by standard 18S rRNA qPCR compared to infections only detectable by us-qPCR and were below 1 male and 1 female gametocyte/µL [22]. Although the gametocyte sex-ratio suggested a very low likelihood of transmission [22], it cannot be excluded that parasite densities in untreated, chronic malaria infections in the community may fluctuate and transiently reach levels that could potentially be infective to mosquitoes [23].
A limitation in the quantification of P. vivax infections derives from differences in the numbers of mitochondria between different P. vivax blood stages, i.e. between a ring stage versus a multi-nucleated schizont stage, both present in peripheral blood. The number of target gene copies per parasite varies accordingly [10] [17]. Plasmodium vivax parasite densities calculated from qPCR copy numbers thus represent only an estimate. Yet, this equally applies to the variable number of nuclear genomes and thus to 18S rDNA copies circulating in P. vivax parasites of any stage. This explains our observation of a good correlation between both markers for quantification of P. vivax. For other human Plasmodium species occurring at the 3 study sites, us-qPCR assays have not yet been developed, thus the analyses performed inthis study were limited to P. falciparum and P. vivax.
Investigations of individuals at greatest risk of infection based on parasite detection by Pf_ and Pv_18S rRNA qPCR assays have been described previously for all three study sites [19,24] [W. Monteiro, personal communication]. These earlier analyses showed that main predictors of P. falciparum and P. vivax infections and density differed substantially between sites. To complement these former epidemiological analyses, age trends of P. falciparum and P. vivax infection in relation to parasite densities based on us-qPCR data were investigated. Parasite carriers with very low-density infections in PNG were primarily found in adolescents aged 10-20 years, whereas in Thailand and Brazil most additionally gained infections were found in older individuals (>40 years of age). Therefore, the additional P. vivax infections detected by us-qPCR coincided with the age group that had presented the highest risk of infection in earlier analyses from Thailand, Brazil and PNG [19,24] [W. Monteiro, personal communication].
Depending on the public health question to be addressed by molecular diagnosis, more precise prevalence data as achieved by us-qPCR may be of considerable relevance, particularly in low transmission settings, where each individual malaria case is investigated by elimination programmes [25]. Furthermore, results are likely to be more consistent across surveys, as stochastic amplification of low-density infections is reduced [2,11]. Because this study demonstrated a substantial gain of additionally detected low-density infections, particularly in low transmission settings, us-qPCR could be crucial for surveillance in elimination settings. Molecular diagnosis is increasingly used to determine presence of infections to guide control efforts [26]. The use of us-qPCR, and thus greater confidence in diagnostic metrics, could reduce the number of samples to be screened before declaring a region malaria-free.