Infection History and Current Coinfection With Schistosoma mansoni Decreases Plasmodium Species Intensities in Preschool Children in Uganda

Abstract Malaria–schistosomiasis coinfections are common in sub-Saharan Africa but studies present equivocal results regarding the interspecific relationships between these parasites. Through mixed-model analyses of a dataset of Ugandan preschool children, we explore how current coinfection and prior infection with either Schistosoma mansoni or Plasmodium species alter subsequent Plasmodium intensity, Plasmodium risk, and S mansoni risk. Coinfection and prior infections with S mansoni were associated with reduced Plasmodium intensity, moderated by prior Plasmodium infections, wealth, and host age. Future work should assess whether these interactions impact host health and parasite control efficacy in this vulnerable age group.

Malaria and schistosomiasis are 2 of the most important tropical parasitic diseases, each infecting >200 million people per year with substantial morbidity and mortality [1,2]. Preschoolaged children (<6 years of age) can be at high risk for both infections [1,2]. Throughout sub-Saharan Africa, malaria and schistosomiasis are coendemic, and coinfections (ie, both species infecting the host simultaneously) are common [3]. Several research studies suggest that Plasmodium species and schistosomes interact during coinfection, yet none have explicitly considered the consequences of coinfection in preschool-aged children. This exposes a crucial knowledge gap in current options for effective control in younger children with no combined control strategy currently developed [1,2].
Prior studies of interactions between Plasmodium and schistosomes have provided equivocal outcomes, with some studies reporting increased Plasmodium intensity, prevalence, or incidence when in coinfection with schistosomes [3][4][5], while others suggest a decrease [3,6]. All studies to date have been conducted on older cohorts. Older study cohorts are likely to have had multiple cycles of infection and treatment and to have developed acquired immunity for both Plasmodium and schistosomes, biasing the immune system in ways that may contribute to disparities observed between studies [7].
To shed light on the dynamics of malaria and schistosomiasis in young children, we examined Plasmodium-Schistosoma mansoni coinfections in preschool-aged children and assessed whether prior infections and coinfections altered S mansoni and Plasmodium prevalence and Plasmodium intensity. We conducted an in-depth secondary analysis of the Schistosomiasis in Mothers and Infants (SIMI) dataset [8], which contains infection histories for each child, enabling us to tease apart the relationship between prior infections and current coinfections for Plasmodium and S mansoni. Previous work indicates that Plasmodium falciparum is the main species in our study regions [9]. The SIMI dataset does not, however, include species-specific measures and, therefore, we will use Plasmodium throughout since we cannot be certain of the Plasmodium species. In line with studies in older children, we hypothesize that S mansoni will increase the risk and intensity of Plasmodium infections but that S mansoni will be unaffected by Plasmodium.

MATERIALS AND METHODS
The SIMI project was approved by the London School of Hygiene and Tropical Medicine, United Kingdom (LSHTM 5538.09), and the Ugandan National Council of Science and Technology. In brief, for primary data collection informed consent was provided in writing, or by a thumbprint in cases of illiteracy, by mothers for themselves and on behalf of their children before enrollment (see further details in [8]).
The SIMI study took place in Uganda at 6 villages around Lake Albert (Piida, Bugoigo, Walukuba) and Lake Victoria (Bugoto, Bukoba, Lwanika), where infections in each child and mother were assessed at 6-month intervals. We used data from the initial baseline assessment and the 6-month follow-up due to participant dropout at later surveys. Extracted data detail S mansoni and Plasmodium infections of preschool-aged children, presence/absence of any soil-transmitted helminths (STHs), information on infection risk behaviors, age, sex, tribe, and mother's education level and occupation. Variables indicative of a family's socioeconomic status (Supplementary Table 1) were used to generate a wealth index using a principal component analysis (PCA) [10]. Variables were included in the PCA unless >95% of the participating families, or <5%, owned the asset to ensure a reasonable level of variation was included [10]. The first principal component of the PCA was extracted to form the wealth index, then divided into equal wealth quintiles.
The presence/absence of S mansoni was determined by a urine schistosome circulating cathodic antigen (CCA) rapid diagnostic test. The intensity of an S mansoni infection was inferred from Kato-Katz fecal egg counts as eggs per gram of feces and split into low (<100), moderate (100-399), and high (>399) intensity according to World Health Organization guidelines [11]. Parasitemia/µL blood, assessed from Giemsa-stained blood films, was used to determine both presence/absence and intensity of Plasmodium infection. Presence of STHs was determined using the Kato-Katz method. Praziquantel and albendazole were provided to all children at the baseline survey, regardless of infection status in line with mass drug administration protocols. Antimalarials (Lonart) were provided if Plasmodium was detected by microscopy or by a fingerpick rapid diagnostic test [9].
Of the 1211 children in the original dataset, 520 were excluded due to an incomplete sample profile (Supplementary Figure 1). Data from missing participants were representative of those individuals included for all variables in the analyses (Supplementary Table 2). The remaining 706 children provided a blood sample for Plasmodium diagnosis, a urine sample for schistosome CCA, and a stool sample to detect schistosome and STH egg patent infections, at both baseline and the 6-month survey. The participants' summary statistics are included in Supplementary Table 3.
All analyses were completed in R version 3.5.2 statistical software (R Core Team, 2018). The effect of infection history (infection at baseline) and coinfection (coinfection at the 6-month survey) on (1) Plasmodium intensity, log transformed to normalize the distribution (Ln[x + 1]), (2) Plasmodium risk, and (3) S mansoni risk were assessed in a series of general linear (Plasmodium intensity) and generalized linear (risk models) mixed models using the ASReml-R v4 mixed modeling package (VSN International Ltd, Hemel Hempstead, United Kingdom). In each assessment 2 models were created, with one set incorporating all infections as presence/absence data and the other including S mansoni and Plasmodium (Ln[x + 1]) data as inferred intensities. These latter models allowed the investigation of potential nonlinear relationships between the parasites. Plasmodium intensity models had a Gaussian error distribution and an identity link function. Risk models used a binomial error distribution with a logit link function.
Baseline prevalence of STH was included in all models; however, the 6-month follow-up STH prevalence was only 3.9% and therefore excluded. All starting models included a child's age, sex, family's wealth quintile, tribe, and mother's occupation and education level; behavioral variables associated with S mansoni infection (the number of times a child bathed and how long they spent in water) and Plasmodium infection (the use of bed nets, the presence of mosquitoes in the home, and whether a person sleeps outside); and the random terms of family identifier, the child's village, nested within lake, and age fitted with a cubic smoothing spline, as parasite-age profiles are often nonlinear [12]. Detailed model structures can be found in Supplementary  Table 3. For all models, the random terms were refined using maximum likelihood ratio tests, then stepwise deletion of insignificant fixed terms was conducted using the Wald test and assessment of the F-statistic (significance limit taken as P = .05).

RESULTS
Baseline Plasmodium infection (presence/absence), in interaction with an S mansoni coinfection, was associated with a significant reduction in intensity of Plasmodium (P = .001; Figure 1A). For children with no prior history of Plasmodium, current coinfection with S mansoni was associated with a lower intensity of Plasmodium infection (P < .001; Table 1). In children who had a baseline Plasmodium infection, there was no significant association with a current S mansoni coinfection (P = .126; Table 1). In S mansoni-coinfected children, Plasmodium infection intensity had a U-shaped relationship with wealth, with significant differences at moderate wealth quintiles (3-4; Table 1).
In the models where infection intensities were considered, the baseline S mansoni intensity, in interaction with age, had a nonlinear relationship with Plasmodium infection intensity (P = .011; Figure 1B). Plasmodium intensity fell as age increased except for children with a moderate (n = 34) S mansoni infection, whereas Plasmodium intensity increased with age. The agerelated decline in Plasmodium intensity was steeper for those children with prior low-intensity (n = 101) and high-intensity (n = 13) S mansoni infections. In children with moderate baseline infections, although Plasmodium intensity increased with age, it remained lower than in either infection category for the equivalent age group. Cumulative Plasmodium intensity is therefore lowest in children with moderate prior S mansoni infection (Supplementary Figure 3).
Coinfection was not associated with a change in infection risk of either parasite; however, each species was associated with an increase in its own subsequent risk of infection (Supplementary Results). Age and sex were also significantly associated with changes in Plasmodium and S mansoni risk (Supplementary Results).

DISCUSSION
In our analysis we detected, for the first time, an association between Plasmodium and S mansoni in preschool-aged children. In contrast to our hypothesis, we see that coinfection with S mansoni and Plasmodium is associated with a reduced Plasmodium infection intensity. This relationship between Plasmodium and S mansoni is, however, complex and moderated by host age, family wealth, and prior infections. There was no apparent effect of S mansoni on the risk of Plasmodium infection nor Plasmodium on the risk of S mansoni.
The body's response to S mansoni infection is generally considered to be biased toward a T-helper 2 (Th2) response that is largely directed against schistosome eggs [13]. As the Th2 response is co-downregulatory to the T-helper 1 (Th1) response, the primary response required for Plasmodium clearance [13,14], it might be expected that coinfection would lead to increased Plasmodium intensity, whereas we observed a reduction. Early-stage schistosome infections (ie, before egg production), however, stimulate a Th1 response [13]. Of the preschool-aged children in our study, only 29% had egg patent infections, suggesting that many S mansoni infections may have been in the Th1 phase of immune stimulation. An early S mansoni-induced Th1 response could negatively impact Plasmodium and explain the reduced intensity of Plasmodium we observe. Nevertheless, there is the possibility that the relationship between these parasite species is driven by resource competition. Plasmodium species and hookworms have been observed to compete for the same resource, blood, with hookworm coinfection reducing Plasmodium intensity [15]. Blood is also a resource for schistosomes; however, resource competition would not explain the increased Plasmodium intensity seen in schistosome coinfections in older children [4]. When we explored the relationship between Plasmodium and prior S mansoni infection in respect to different schistosome intensities, a child's age was an important moderator of this relationship. Age was associated with lower Plasmodium intensity at low and high prior S mansoni intensities; however, Plasmodium intensity increased with age for moderate infections. Nonlinear effects of schistosomes on Plasmodium have been observed elsewhere, with moderate schistosome infections reducing the incidence of Plasmodium infection while high-and low-intensity infections are associated with increased incidence [5]. It is feasible that the interactions between these parasites is both immune and resource mediated, and it may be that the interplay between these 2 mechanisms varies, with the effect of one becoming more dominant than the other at different stages of infection. The low number of high-intensity S mansoni infections is a limitation of our data. In older children most studies have observed positive associations between schistosome infection and Plasmodium intensity [4,6]. This shift to a positive association could be observed if more children of this age have moderate infection intensities, as the child age-Plasmodium intensity trajectory we observe is positive. The SIMI data do not contain complete information on all Plasmodium species present. Three species of Plasmodium can be found within the studied communities, but P falciparum dominates (75%) [11]. One study has reported differential outcomes of coinfection between hookworms and different Plasmodium species [15], and further work should assess whether our findings are robust to all Plasmodium species.
Assessing infections in preschool-aged children with known infection histories has enabled us to elucidate the complex relationship between Plasmodium and S mansoni. We show that prior infections can have long-lasting effects and suggest that Plasmodium-schistosome interactions are likely to be mediated by the host immune response since any infections have been removed or reduced below detection limits by chemotherapeutic interventions. Determining the order of infection was essential to understanding coinfection outcomes in this system. Our results also suggest that stage of infection (early or later-stage) may have an important influence on the relationship between Plasmodium and schistosomes, again likely mediated by the host immune response. Our work has focused on the effect parasites have on one another, but it will now be important to identify whether, and how, these interactions affect host health and the efficacy of the individual parasite control strategies, highlighting the potential need for an integrated Plasmodium-schistosome control strategy for this more vulnerable age group.