Patterns of inflammatory responses and parasite tolerance vary with malaria transmission intensity

Three outpatient hospitals at locations (Kintampo, Navrongo and Accra) representing distinct malaria transmission intensities in Ghana were selected for this study. Kintampo is holo-endemic for malaria with year-round transmission, and an entomological inoculation rate (EIR) of >250 infective bites/person/year [35]. Navrongo is hyperendemic for malaria with seasonal rainfall and transmission (high transmission from May to November, low transmission from December to April) and EIR of 50–250 infective bites/person/year [36]. Accra is the capital city and has a relatively low transmission intensity (<50 infective bites/person/year) that peaks between June and August annually [37]. Samples were collected from 2011 to 2013 during the peak transmission seasons at the respective study sites.

Ethical approvals were obtained from the ethics committees of the Ghana Health Service, Navrongo Health Research Centre, Kintampo Health Research Centre and Noguchi Memorial Institute for Medical Research, University of Ghana, Accra, Ghana. Participation was voluntary, and written informed consent was obtained from parents/guardians of the children. Study participants were children aged 2–14 years who were showing signs of clinical malaria, and had been referred for malaria tests by the attending physician. Parasitaemia was detected by malaria rapid diagnostic tests (RDTs) and confirmed by microscopic examination of thick and thin blood smears. Parasite density was estimated by counting the number of parasites per 200 white blood cells as previously described [38, 39]. Haemoglobin levels were quantified using an automated haematology analyzer. Before delivery of anti-malarial and/or any other treatment interventions, 5 mL of venous blood was obtained from each child. Plasma samples were separated from whole blood by centrifugation at 2500 rpm (Eppendorf, model: 5810 R) for 10 min and aliquoted into Eppendorf tubes for storage at −80 °C until further experiments. Sample collection, storage and analysis were done using the same protocols and procedures to ensure uniformity and comparability of data from the different hospitals.

Plasma concentrations of cytokines were measured using the highly sensitive xMAP technology (Luminex Corporation), which allows the simultaneous quantification of several biological analytes in a 96-well format. The MILLIPLEX^® MAP 13-Plex Kits from Millipore (Merck Group, magnetic beads) were used because of their higher detection accuracy and reproducibility of results compared to other vendors [40]. These kits were used to quantify eight pro-inflammatory cytokines (TNF-α, IFN-γ, IL-1β, IL-2, IL-8, granulocyte monocyte colony stimulating factor (GM-CSF), IL-6 and IL-12p70) and four anti-inflammatory cytokines (IL-4, IL-7, IL-10, IL-13) in duplicate wells for each plasma sample. These analytes were selected based on relevance and association with malaria. The assays were conducted strictly following the manufacturer’s instructions without any modification. The kits used were from the same lot, and the samples were randomly distributed across plates. Prior to assay, samples were thawed and clarified by centrifugation (2000 rpm for 10 min). There were no readings from the background wells while the quality control and the Standards wells were within the specified range of the kits. Samples with percentage coefficient of variation (%CV) >15% were excluded from further analysis. Cytokine detection limits are found in Additional file 1.

Data analyses and graphs were done using GraphPrism version 6.01 (GraphPad Software, Inc.) and Minitab version 17.1.0.0 (Minitab Inc.). After initial normality tests, patients’ demographics and clinical parameters were compared across the three sites either by Pearson’s Chi square (χ²) test (to compare proportions in categorical variables) or One-way ANOVA or Kruskal–Wallis H (K–W) test (for continuous data sets), depending on normality of data. An across-site comparison of cytokine levels was performed with the K–W test, while Dunn’s multiple comparison test was used to reveal between-site pairwise significant differences. Spearman’s correlation analyses were performed for associations of cytokines levels with age and parasite density. In addition, a Spearman’s correlation matrix was built to detect associations between cytokines. Multiple linear regression analyses were conducted to detect the variable(s) that is/are the best predictor(s) of cytokine levels. For the regression models, cytokine levels served as the outcome variables while parasitaemia, age, gender and transmission intensity served as the predictor variables. Statistical significance was generally set at P < 0.05, however, after Bonferroni’s procedure, the critical value (α) of the regression models was adjusted to 0.004.

To investigate the role of inflammatory responses in the development of malaria parasite tolerance in endemic areas, this study examined the relationship between transmission intensity and the patterns of cytokine production in children with malaria. Using a cross-sectional approach, a total of 173 children who tested positive for malaria by RDTs and microscopy were recruited from three community hospitals in three areas with varying transmission intensities: Accra (N = 71) < Navrongo (N = 44) < Kintampo (N = 58). The proportions of both sexes were comparable across the study sites (P = 0.270; Table 1). Children in Accra were relatively older than those in Navrongo (P = 0.010) and Kintampo (P = 0.025), however, the ages of children from Navrongo and Kintampo did not differ (P = 0.999). Parasitaemia reflected the intensity of transmission, with children from Kintampo having higher parasitaemia compared with those in Accra (P = 0.005) and Navrongo (P = 0.070), although these differences were not statistically significant for Navrongo (Table 1; Additional file 2). Haemoglobin levels decreased as transmission intensity increased (P = 0.007), with children from Accra having significantly higher haemoglobin levels compared with those from Kintampo (P = 0.008). Although haemoglobin levels in children in Accra were also higher than those in Navrongo, this difference was not statistically significant (P = 0.076). Children from Navrongo and Kintampo had comparable haemoglobin levels (P = 0.545). The median temperature at clinic also decreased with increasing transmission intensity (P < 0.001), with children from Accra having significantly higher median temperature compared to Navrongo (P < 0.001) and Kintampo (P < 0.001). Children in Navrongo also had higher median temperature than those from Kintampo (P < 0.001). Since these children were recruited at presentation to hospital with symptoms of malaria, the pattern of parasitaemia suggests that the parasitaemia threshold for clinical manisfestation of infection seems to increase with increasing transmission intensity.

The secretion of pro-inflammatory cytokines during malaria infection has been shown to culminate in the clinical manifestations of disease [13, 14]. Consequently, the role of cytokine levels in mediating parasite tolerance was investigated in children exposed to different malaria transmission intensities. Quantification of levels of pro-inflammatory cytokines, including TNF-α, IFN-γ, IL-1β, IL-2, IL-8, IL-6, IL-12, and GM-CSF in children with malaria in the three transmission areas revealed a pattern of decreasing cytokine levels with increasing transmission intensity (Accra > Navrongo > Kintampo; Fig. 1). Levels of all pro-inflammatory cytokines were significantly lower in children from Kintampo compared to those in Accra (P < 0.005 for all cytokines; Fig. 1a–h). In addition, levels of all pro-inflammatory cytokines except IFN-γ (P = 0.834), IL-8 (P = 0.056) and IL-6 (P = 0.260) were lower in the Kintampo group compared to the Navrongo group (P < 0.05 for all comparisons; Fig. 1). Given that parasitaemia levels increased with increasing transmission intensity (Table 1), the reverse pattern of pro-inflammatory cytokine suggest that lower cytokine responses appear to favour increased parasite tolerance.

Since pro-inflammatory responses are usually followed by the secretion of anti-inflammatory mediators to balance their effects [13, 14, 29, 30], the levels of four key anti-inflammatory cytokine including IL-4, IL-7, IL-10, and IL-13 were also examined. The pattern observed was similar to that of the pro-inflammatory cytokines, with cytokine levels decreasing with increasing transmission intensity across the three areas (Accra > Navrongo > Kintampo; Fig. 2). Specifically, levels of all four cytokines were significantly lower in children recruited in Kintampo compared to those in Accra (P < 0.01 for all comparisons; Fig. 2). Taken together, these data buttress the patterns observed for pro-inflammatory responses, and further confirm an association between transmission intensity and the magnitude of inflammatory responses induced during clinical malaria, and suggest a likely association with parasitaemia levels.

Given the strong relationship between cytokine levels and transmission intensity, the correlations between levels of pro-inflammatory mediators and parasitaemia were directly examined in each of the three sites. There were significant positive correlations between parasite density in children with malaria and levels of key pro-inflammatory cytokines, including TNF-α, IFN-γ and IL-6 (Fig. 3). However, these correlations were observed in the Accra group only, with none showing significant correlation in the Navrongo group, and only IL-6 showing a significant correlation in the Kintampo group (Fig. 3). Since malaria parasite antigens induce the production of pro-inflammatory cytokines [14, 16, 41, 42], these results are further evidence of increased parasite tolerance in the higher transmission areas, where further increases in parasite levels above a certain threshold no longer augment cytokine production.

Subsequently, the correlations between anti-inflammatory cytokines and parasite burden were examined in children with malaria across the different transmission areas. Unlike the patterns observed with the pro-inflammatory cytokines, the anti-inflammatory cytokines did not generally show significant correlation with parasitaemia in any of the endemic areas (Fig. 4). However, IL-10, which is considered a critical anti-inflammatory mediator in regulating the pro-inflammatory response during malaria [29, 43, 44], showed significant correlation with parasite density in children residing in Accra and Kintampo (Fig. 4). This correlation was particularly strong in the Accra group, which is consistent with the correlation observed for the key pro-inflammatory cytokines. In addition, IL-4 showed a significant association with parasitaemia in children from Navrongo, however, this correlation was negative (Fig. 4).

Cytokine levels during parasitic infections, including malaria, have been shown to vary with age [45]. Since age significantly differed between sites (Table 1), the study subsequently determined whether the age difference affected cytokine responses during acute malaria infection. The results showed limited associations between age and cytokine levels in this cohort, with IFN-γ correlating negatively with age in Navrongo and Kintampo (Additional file 3), while IL-8 and IL-4 showed positive correlations with age in Navrongo and Accra, respectively (Additional file 3). Thus, contrary to previous reports where age was found to significantly affect levels of cytokines during malaria infection [6, 13, 45, 46], age did not seem to be a major determinant of plasma levels of cytokines across the sites.

Using multiple linear regression analyses, the relative contributions of age, gender, parasite density, and transmission intensity as predictors of cytokine levels was examined. These analyses revealed that study site (coded in order of increasing transmission intensity) was the strongest predictor of all cytokine levels, except GM-CSF, for which sex was the best predictor (see F values in Table 2). Transmission intensity had negative regression coefficients in all the regression models (negative β-weights, Additional file 4); indicating an inverse relationship with cytokine levels. In addition, the interrelationships among the cytokines across the sites were investigated, and this showed that cytokines generally correlated positively with each other (Fig. 5). Most significantly, the key pro-inflammatory cytokines, including IFN-γ, IL-12 and IL-2 strongly correlated with each other, and a strong association was also found between TNF-α and IL-6 (Fig. 5).