Non-invasive surveillance of Plasmodium infection by real-time PCR analysis of ethanol preserved faeces from Ugandan school children with intestinal schistosomiasis

Malaria continues to be a public health problem in 90 countries worldwide [1], with the impact of global control flat-lining over the past 2 years; the greatest disease burden remains in sub-Saharan Africa (SSA) where over 90% of deaths occur [2]. Here, weak health systems with restricted diagnostic repertoires and inadequate access to prompt treatment preside, alongside favorable conditions for Plasmodium transmission [2, 3]. Collectively, this places certain demographical groups, e.g. infants, more vulnerable than others [4] although infection prevalence and non-severe disease in other groups, such school children, should not be overlooked [4‐6]. A broader appraisal of infection surveillance is needed, especially in the future context of gaining transmission control and unveiling the importance of asymptomatic carriage [7‐9]. Indeed, with introduction of more sensitive diagnostics, over and above what current rapid diagnostic tests (RDTs) can provide, our understanding of the environmental importance of asymptomatics will change [10]. Exploration of non-invasive sampling, e.g. collection of saliva or urine, for detection of Plasmodium DNA has been explored but as of yet is not routinely used [7, 11, 12].

In Uganda, for example, control of malaria is a major public health challenge [13], especially in young children [6]. The geographical distribution of malaria also aligns with other mosquito-borne diseases such as lymphatic filariasis, making co-surveillance with neglected tropical diseases justifiable [14]. On the Lake Albert shoreline, however, intestinal schistosomiasis is hyperendemic [15] and ongoing school-based preventive chemotherapy control has taken place, with periodic disease surveillance, over the last 15 years [16, 17]. The local prevalence of Plasmodium falciparum in mothers and pre-school-aged children is very high [18‐20], with non-P. falciparum infections also reported [21] but the prevalence of infection in school children rarely measured [22]. As part of ongoing surveillance of intestinal schistosomiasis in this area, Al-Shehri et al. [15] revealed that over half of the sampled school children harboured RDT positive asymptomatic infections, a quarter being anaemic (< 115 g/L) and 11% had faecal occult blood in stool. The local prevalence of egg-patent intestinal schistosomiasis was 46.5%, ranging in excess of 80% in certain schools along the immediate shoreline [15].

Point-of-care (POC) and reference laboratory diagnostics tests play a critical role in monitoring and surveillance of disease control programmes, yet finding synergies for integrated surveillance is limited [23]. Indeed, current siloing of diseases and infection detection is unfortunate for there could be opportunities, where appropriate, to share programmatic costs and intervention repertoires for intestinal schistosomiasis and malaria [13, 24]. Faecal samples are widely collected for surveillance of intestinal schistosomiasis but have not been fully exploited to co-detect malaria, for example, by real-time PCR (rtPCR) analysis for Plasmodium DNA [25‐27]. By contrast, non-invasive faecal sampling has burgeoned the detection of Plasmodium DNA in non-human primates, developing our epidemiological understanding of natural and zoonotic malaria immensely [28‐31].

A total of 271 children were initially screened; 247/271 (91.1%) met the inclusion criteria of sample provision and obtaining verbal assent, were enrolled with mean age of 6.8 years (95% CI 6.6–6.9), with males accounting for 124/247 (50.2%), and females 123/247 (49.7%). Of 247 school-aged children that tested for malaria infection by RDTs using malaria Ag P.f/Pan test, 138 (55.8%; 95% CI 49.6–61.9) were positive for P. falciparum (PfHRP-2-detecting RDTs), and 45 samples (18.2%; 95% CI 13.9–23.5) were positive for Plasmodium species (PfHRP-II and pan-pLDH test lines). A singleplex rtPCR-based assay for DBS revealed that 198 (80.1%, 95% CI 74.7–84.6) children were positive for Plasmodium DNA whereas a singleplex rtPCR-based assay for EPF revealed that 158 (63.9%; 95% CI 57.8–69.7) children were positive, (Table 1).

Although absolute values differed, the trends of Plasmodium prevalence across the three diagnostic methods by “Low–High Altitude” and “distance from Lake Albert” were similar. A positive result by any test increased significantly relative to distance from the lake (p < 0.001 for all three diagnostic tools) (Fig. 1). The highest prevalence of malaria infection was by DBS analysis of 95.9% (95% CI 86.2–98.8) at Busingiro and, for EPF, 91.8% (95% CI 80.8–96.7) at Biiso. At low altitudes and the shoreline schools of Walakuba and Runga, prevalence by RDT was 32.5% (95% CI 20.4–47.4) and 58.0% (95% CI 44.2–70.6), by rtPCR for EPF was 34.8% (95% CI 22.4–49.8) and 38.0% (95% CI 25.8–51.8), and for DBS was 76.7% (95% CI 62.2–86.8) and 84.0% (95% CI 71.4–91.6). In Bugoigo school, the prevalence of malaria by rtPCR from EPF appeared raised, 76.7% (95% CI 64.2–85.9), upon comparison to the other two diagnostic tools.

The relationship between C_t value of rtPCR-based assays for DBS and C_t value for EPF was DBS (C_t) = 26.93 + 0.1763 EPF (C_t) and was statistically significant (p = 0.01), with R-squared 4.14% (p < 0.005). The positive correlation as shown r = 0.2 (Fig. 2), indicating that when EPF (C_t) increased, so did the DBS (C_t) (p < 0.05), although the magnitude of this change was not substantive. Using RDT results as convenient field-based comparison, the diagnostic score tabulations are presented (Table 2).

In the present study, amplification of Plasmodium DNA from non-invasive EPF and invasive DBS sampling has shown that 158 (63.9%; 95% CI 57.8–69.7) and 198 (80.1%; 95% CI 74.7–84.6) of 247 children tested positive for infection (Table 1). As might be expected given the greater sensitivity of rtPCR methods based in the laboratory [7, 36, 37], the prevalence estimated by either EPF or DBS was higher than that by RDTs where 138 children (55.8%; 95% CI 49.6–61.9) were judged infected. Using the RDT, inspecting lactate dehydrogenase (pLDH) reactions, 45 infected cases (18.2%; 95% CI 13.9–23.5) of pan-Plasmodium spp. co-infections were found. The general agreement of infection mean prevalence and prevalence by school for EPF and RDT was satisfactory and would infer that this non-invasive sampling protocol yields very similar school-level information. A further diagnostic evaluation might consider inclusion of point-of-contract microscopy [38, 39], particularly when parasite densities are below RDT detection thresholds [37].

Taking altitude and distance from Lake Albert into account, the highest prevalence of Plasmodium infection by any method was at Busingiro school, where 47 out of 49 pupils (95.9%; 95% CI 86.2–98.8) tested positive by DBS sampling (Table 1). In contrast, 73.4% (95% CI 65.2–79.8) of the same pupils tested positive by EPF sampling while prevalence by RDT was 65.3% (95% CI 51.3–77.0). The highest prevalence by EPF sampling was at Biiso school, and in all but one school (Runga), both DBS and EPF sampling provided higher prevalence of infection that RDTs (Fig. 1). These findings confirm those by Jirků et al. [25] in that Plasmodium DNA shed at detectable levels by rtPCR can be readily found within faecal material.

Unlike in previous studies where P. falciparum infection and parasite density was negatively associated with rising altitude [40‐43], this study found that the highest percentages of Plasmodium DNA were in higher altitude schools (Biiso and Busingiro) (Fig. 2b); in both schools, EPF sampling detected a higher percentage of infection than RDTs. At lower altitudes, the performance of EPF sampling was less predictable against the two other methods. It is worthy to note that egg-patent intestinal schistosomiasis is very common in these three schools [44] and may perhaps confound the detection of Plasmodium DNA. When determining the sensitivity and specificity of rtPCR-based assays; an initial evaluation of cycle threshold (C_t) values showed that there was a positive correlation, although not particularly impressive, between DBS- and EPF-derived samples (r = 0.2), with a mean EPF Ct value of 36.3, and a mean DBS Ct value of 33.3, (Fig. 2). When correlating these results to RDTs, rtPCR-based assay of DBS samples showed a greater sensitivity than EPF samples (94.2% versus 73.1%), with rtPCR of DBS samples having a specificity of 37.6% and a PPV of 65.6%. In contrast, the specificity of EPF sampling was 47.7%, with a PPV of 63.9% (Table 2).

It is known that schistosomes immuno-modulate host responses which interplay with Plasmodium parasitaemia, often reducing numbers of Plasmodium within the blood [45]. This may lower the amount of Plasmodium DNA ultimately shed into the bowel but as schistosome eggs also directly perforate the bowel, and some 11% of children also had faecal occult blood in stool, thus there may be a more complicated relationship with detected Plasmodium DNA in faecal material than first assumed [25]. It remains to be assessed if this phenomenon also occurs in EPF sampling of non-human primates as intestinal schistosomiasis can naturally occur in these animals [46]. Using rtPCR with Taqman^® probes for Schistosoma boosts the diagnostic repertoire for detection of human and zoonotic schistosomiasis [44, 47] and could be also applied in non-human primate sampling.

Although faecal sampling has been a routine activity of the national control programme for bilharzia and intestinal worms for over a decade [48, 49], an immediate drawback in the use of EPF sampling is the need for a laboratory setting. Future advances in handheld rtPCR equipment may provide an alternative avenue by which EPF sampling becomes more portable and affordable in field conditions [7, 50, 51]. What is evident from the results of this study is that EPF sampling of Plasmodium infections in school children in this high transmission setting provides information without invasive sampling. Similarly, it has been shown that EPF sampling can shed light on the occurrence of other parasites and pathogens whose nucleic acids may be shed in stool [23]. Such passive collection of stools, or targetted collection for faecal parasites, could aid in developing a general monitoring and surveillance platform. For example, EPF sampling may have future application in evaluating dynamics of malaria in schools where intermittent preventive treatment has been allocated [51]. However, in Ebola-risk areas, due care should be taken in stool collection and processing [52]. With regard to malaria control in Uganda, EPF analysis could contribute to general surveillance of infection in schools [13], encouraging further integrated mapping initiatives with neglected tropical diseases [53], and assessing the future impact of intermittent preventive treatment initiatives [51].