At the end of the malaria transmission season in December 2017/January 2018 (t₁), a cross-sectional survey of children aged 8–15 years, residing in twenty-nine villages in the Upper River Region of The Gambia, was conducted to identify 1650 children in good general health and with no evidence of fever (body temperature <38°C) for inclusion in the study ( Figure 1 ). Additional exclusion criteria included participation in another ongoing research study; any signs of significant ill health (e.g. cardiovascular, pulmonary, renal, hepatic, neurological, dermatological, endocrine, malignant, infectious, immunodeficiency, psychiatric and other disorders); other known medical conditions (e.g., HIV infection, sickle cell disease or thalassaemia); recent antimalarial or antibiotic treatment (within the previous month). Height, weight, sex, age, and village of residence were recorded. Finger prick blood samples were obtained for malaria microscopy (Giemsa stained thick films), rapid diagnosis by lateral flow assay for P. falciparum histidine-rich protein II (PfHRP2) (SD BIOLINE Malaria Ag P.f, Abbott), preparation of dried blood spots for P. falciparum qPCR analysis (see below) and haemoglobin (Hb) estimation by Hemocue (Hb201+, Radiometer). In a follow up survey conducted in February/March 2018 (t₂), children identified as parasite positive by 18S PCR (n = 67) in the baseline survey were age, sex, and village matched to children with no detectable parasitaemia; anaemic children (Hb < 11 g/dL, n = 70) were similarly matched to children with Hb ≥11 g/dL, respectively. These children were invited for a second clinical examination and blood sample collection at Basse Regional Hospital. The study was approved by The Medical Research Council Gambia (MRCG) Scientific Coordinating Committee and by the Gambia Government/MRCG Joint Ethics Committee (reference 1545). Prior to enrolment, verbal assent was obtained from study participants and verbal or written consent was obtained from their parent or guardian. Stored plasma samples from 12 Gambian children with acute clinical malaria (11, 12) ( Table S1 ) were used as comparators in some assays.

P. falciparum diagnostic PCR was performed in two stages. For screening (t₁), PCR for 18S ribosomal RNA was performed as described previously (13, 14). For definitive diagnosis in the final cohort of children seen at both t₁ and t₂, qPCR against the var gene acidic terminal sequence (varATS) of P. falciparum was performed as described previously (15). Briefly, DNA from dried blood spots was extracted using the QIAamp 96 DNA QIAcube HT Kit (Qiagen). For varATS qPCR, samples were run in duplicate against a universal standard [NIBSC code 04/176 (16)]. Samples were deemed positive for P. falciparum DNA if both replicates were detectable at Ct<40 cycles. Discrepant samples (where only 1 of the 2 replicates were detected at Ct<40) were run again in duplicate; only samples which were positive in both replicates on the same plate were deemed infected. Reaction parameters for both 18S and varATS qPCR are described in the supplemental methods.

In the follow up survey (t₂), approx. 10 mL venous blood was collected into EDTA vacutainers (BD). Complete blood counts were performed using an automated haematology analyser (M series, Medonic). The remaining whole blood was layered onto Ficoll (Histopaque^®-1077; Hypaque) and centrifuged at 500 x g for 30 minutes (brake off). Plasma was removed and stored at -80°C. Cell pellets were cryopreserved in liquid nitrogen for future studies.

Enzyme-linked immunosorbent assays (ELISAs) were conducted according to manufacturers’ instructions to measure plasma concentrations of haemopexin (OKIA00066, Aviva Systems Biology) and erythropoietin (EPO, DY286-05, R&D Systems), at a dilution of 1:40,000 or undiluted, respectively. Colorimetric determination of haem in undiluted plasma samples was conducted according to manufacturers’ instructions (MAK316-1KT, Sigma-Aldrich). Luminex microbead-based suspension array (LXSAHM, R&D Systems) was used according to manufacturer instructions to detect plasma concentrations of IL-10, CD163, IFN-γ, IL-6, TNF-α, CXCL10, G-CSF, C5a, and S100a9 at plasma dilutions of 1:2. Finally, for detection of ferritin, transferrin, C reactive protein (CRP), LPS binding protein (LBP), myeloperoxidase (MPO), and matrix metallopeptidase 9 (MMP-9) by Luminex, plasma was diluted 1:100. Plasma protein concentrations were determined from standard curves after subtraction of background values, calculated using MS Excel. The upper and lower limits of quantification (ULOQ and LLOQ, respectively), and manufacturers’ codes for each analyte, are reported in Table S2 .

Field data were collected and stored on portable electronic devices using REDcap data management software (17). Electronic data were then exported to MS Excel for analysis. Comparisons between uninfected children and those with resolved or chronic P. falciparum parasitaemia were performed using Kruskal-Wallis/Dunn’s test with a Bonferroni adjustment for multiple testing. Correlations were assessed using Pearson’s correlation coefficient. The null hypothesis of zero correlation was tested using a Wald test for sample sizes >20 and a permutation test for sample sizes ≤20. To mitigate against the impact of potential outliers, the ROUT outlier test was applied to individual data points of concern. Percentages were compared using the chi-square test. All statistical analyses were performed using GraphPad Prism (v.9.1.0) or in R Studio (v4.0.4). A p value of <0.05 was considered statistically significant.

Of the 1650 healthy, afebrile children recruited at t₁, 920 were screened by 18S PCR and 67 were positive (“infected”). Of the remaining 730 children, who were not screened by 18S PCR at t₁, 70 had an Hb concentration <11g/dL (“anaemic”) ( Figure 1 ). As far as possible, these children were age (+/-1 year), sex and village matched to children who were either qPCR negative or who had a Hb ≥11 g/dL, respectively. This generated a cohort of 333 children who were recalled at t₂, of whom 259 attended for examination. Children whose height declined by >5 cm together with a weight decline of >15% (n = 14), or whose height increased by >10cm together with a weight increase of >20% (n = 12), or for whom reliable data on height and weight were not available (n = 7), could not be confirmed as being the same child and were excluded from further study ( Table S3 ).

Of the 226 children seen at t₁ and t₂, the status of 5 children who were qPCR negative at t₁ but qPCR positive at t₂ was deemed uncertain and they were omitted from the analysis, particularly as these may have represented new infections acquired during travel beyond the local area (as transmission is highly seasonal). Furthermore, 8 children who were microscopy positive at t₁ but negative by both qPCR and RDT, and 4 children for whom Hb concentration was not available at t₁, were also omitted ( Table S4 ) leaving a final t₂ cohort of 209 children ( Figure 1 ). Of these 209 children, 13 were varATS qPCR positive for P. falciparum at t₁ and t₂ (deemed “chronically infected”) and 44 were positive at t₁ but negative at t₂ (deemed “resolved” infections). Of the 152 uninfected children, 77 had an [Hb ≤11.5 g/dL, defined as anaemia in children (18)] at t₁ and were deemed “anaemic”, leaving 75 who were neither infected nor anaemic (healthy controls).

P. falciparum infections (both chronic and resolved) were more prevalent in the central and western part of the study area than in the eastern part ( Figure 2 ). However, the overall rate of subclinical parasitaemia at t₁ (~9%) was lower than the ~14% anticipated from previous surveys (19), and a high proportion of infections (77%) resolved in the approximately 2 months between t₁ and t₂, leaving the study underpowered for some analyses of persistent infections.

The median recall time (i.e., time between t₁ and t₂) was 64 days (IQR 62-66) and did not differ significantly among the groups. There were no significant differences between the groups in age or sex although uninfected, anaemic children were shorter and weighed less than non-anaemic controls ( Table 1 ). Parasitaemia at t₁ was significantly higher among children whose infections persisted at t₂ than among children whose infections had resolved at t₂ (median 74 parasites/µL vs. 2 parasites/µL; p = 0.003). Among children with persistent chronic infections, median parasitaemia did not differ significantly between t₁ and t₂ (74 parasites/µL and 120 parasites/µL, respectively; p = 0.59) ( Figure 2A ). Haemoglobin concentrations did not differ significantly between t₁ and t₂ for either healthy controls or persistently infected children, but haemoglobin concentrations increased significantly between t₁ and t₂ among uninfected anaemic children whilst remaining significantly lower than among healthy controls (median = 11.7 g/dL (IQR 11-12.3) vs. 12.2 g/dL (11.5-12.7), respectively; p < 0.0001) and than among those whose infections resolved ( Figure 2B ). Mean haemoglobin concentrations of persistently infected children were at the bottom of the normal range at t₁ and t₂ but small numbers precluded the drawing of any substantial conclusions.

The prevalence of moderate anaemia [Hb between 8 and 11.5 g/dL (18)] was similar among uninfected children (12%) and children with resolved infections (11%, p = 0.73), and similar to that in children with chronic infections (31%, p = 0.11) ( Figure 3 ). Severe anaemia (Hb ≤ 8 g/dL) was observed only among uninfected children, although this may be a chance finding given the much larger number of children in this group. Somewhat surprisingly, haematocrit (packed cell volume; PCV) and red blood cell (RBC) counts were significantly higher [and above the normal paediatric range (20)] among children with recently resolved malaria infections when compared with children with no evidence of recent infection, suggesting a rapid rebound in erythropoiesis once their malaria infections resolved. Although PCV and RBC counts did not differ significantly between children with persistent infections and the other groups of children, the power of these comparisons is limited by the small number of chronically infected children.

Interestingly, haematological indicators of uninfected children who were anaemic at t₁ improved somewhat by t₂ with median Hb being significantly higher (p < 0.001) ( Figure 2B ) and median RBC count being within the normal range ( Figure 3B ). Nevertheless, the uninfected anaemic group showed evidence of persisting in red cell abnormalities, with significantly increased red cell width, reduced PCV, reduced mean cell volume and reduced mean cell haemoglobin concentration when compared to healthy non-anaemic controls ( Figures 3C–F ).

To further explore haematological responses to subclinical malaria infections, plasma concentrations of haem, haemopexin, HO-1 and soluble CD163 (the high affinity scavenger for haptoglobin-haemoglobin complexes), all of which are markers of haemolysis ( Figures 4A–D ) and of erythropoietin (EPO), ferritin, transferrin (markers of iron status and mobilisation) ( Figures 4E–G ) were measured at t₂. In contrast to previous observations (8), the only evidence of ongoing hemolysis in the persistently infected children was a raised median concentration of CD163 ( Figure 4D ) and their iron status markers were within the normal range. Children with recently resolved malaria infections had HO-1 concentrations that were somewhat lower than the other children ( Figure 4C ) and median EPO concentrations that were significantly lower when compared to all other groups ( Figure 4E ). Meanwhile, uninfected, anaemic children had significantly lower ferritin concentrations and significantly higher transferrin concentrations when compared to healthy uninfected children ( Figures 4F, G ) suggesting that their anaemia may be due, in part, to iron insufficiency.

Overall, leucocyte counts fell within the normal paediatric range for the majority of children at t₂ ( Figure 5 ) with the only exception being that the median total leucocyte count of children with recently resolved malaria infections (4.8 x10⁶/µL; IQR 4-6.3) was slightly below the normal range of 5-14.5 x10⁶/µL (20). Children with recently resolved malaria infections also had significantly lower total white blood cell counts and lower granulocyte, monocyte and lymphocyte counts than the uninfected, anaemic group ( Figure 5 ). However, median leucocyte proportions concentrations did not differ significantly among the groups and were within the normal range ( Figure 5C ).

Systemic inflammation is a feature of symptomatic malaria infections (21); acquired immunity is associated with the ability to modulate this inflammation (21) such that raised concentrations of circulating anti-inflammatory cytokines – especially IL-10 – may be the only sign of ongoing immune disturbance (8, 22, 23). However, systemic inflammation has also been observed in subclinical, asymptomatic infections (24–26). We therefore measured plasma concentrations of a range of cytokines and inflammatory markers including IFN-γ, LPS binding protein (LBP), IL-6, TNF-α, and IL-10 ( Figure 6 and Figure S1 ).

Concentrations of pro-inflammatory markers did not differ significantly between persistently infected and uninfected children ( Figures 6A–D ) and were noticeably lower than in a historical cohort of children with acute, symptomatic malaria infection (Group 5). However, when compared to uninfected children, children with recently resolved malaria infections had modestly but significantly higher concentrations of IFN-γ and LBP ( Figures 6A, B ) and their concentrations of IL-6 and TNFα were significantly higher than among uninfected anaemic children ( Figures 6C, D ). No significant differences were observed between any of the groups in concentrations of CRP, C5a, S100a9, MPO, CXCL10, G-CSF or MMP-9 ( Figure S1 ). By contrast, IL-10 concentrations were modestly raised (albeit at much lower levels than in acutely infected symptomatic children) among the persistently infected children ( Figure 6E ). Moreover, in these chronically infected children, IL-10 concentrations were highly correlated with parasite density (r = 0.87, p = 0.005) ( Figure 6F ).

It was noticeable, however, that concentrations of pro-inflammatory markers varied considerably within the groups, including in apparently uninfected children. Indeed, concentrations of inflammatory proteins in some children in every group were as high as, or higher than, those of acutely malaria-infected children ( Table S7 ). For example, 22% of children without any evidence of malaria infection had C-reactive protein (CRP) concentrations above the normal threshold of 1mg/L (20) with values up to 3.9 mg/L ( Figure S2 ). Although these CRP concentrations are lower than those seen in children with acute clinical malaria (median 7.1, IQR 5.5 - 8.0), they may be indicative of a low level inflammatory state. Together, these data suggest that many apparently healthy Gambian children may be living with persistent, low-grade inflammation that may be indicative of cryptic infection (with malaria, or other co-endemic pathogens).