Lyophilised soluble S. haematobium adult worm antigen (SWAP) was obtained from the Theodor Bilharz Institute (Egypt) and reconstituted as previously described elsewhere 35. The parasite strain is one used for previous immuno-epidemiology studies 10.

The study was conducted in the Mashonaland East Province of Zimbabwe (31°30'E; 17°45'S) where S. haematobium is endemic. The study area is described in detail elsewhere 36 and the participants have been participating in an ongoing study of the immunoepidemiology of human schistosomiasis 3537. Permission to conduct the work in this province was obtained from the Provincial Medical Director (PMD) while ethical approval was received from the Medical Research Council of Zimbabwe. Following explanation of the study aims and procedures to the community, school children and their teachers, informed consent was obtained from participants or their parents/guardians. The villages were selected because health surveys regularly conducted in the region by the PMD showed little or no infection with other helminths and a low S. mansoni prevalence (<5%). Prior to our study the selected villages had not been included in the National Schistosome Control Programme (run by the Ministry of Health and Child Welfare in Zimbabwe) and therefore had not received treatment for schistosomiasis or other helminth infections meaning that we could study natural immune responses in the absence of drug-altered schistosome responses 1238. The main activity in these villages is subsistence farming and human water contact is frequent with at least 4 contacts/person/week due to insufficient safe water and sanitation facilities (see 36 for studies in neighbouring villages). Drinking water is collected from open wells while bathing and washing is conducted in two main rivers in the villages. Most families maintain a garden located near the river where water is collected for watering the crops. The schools surveyed (a secondary school and its feeder primary school (i.e. where the majority of the primary school children come from) Goromonzi and Shangure Schools in Goromonzi Village, and Chindenga and Nyambanje Schools in Mutoko Village) were all in close proximity to rivers. Stool and urine specimens were assayed for S. haematobium, S. mansoni and geohelminths using standard procedures 3940. In order to be included in the study, participants had to meet all the following criteria: 1) have provided at least 2 urine and 2 stool samples on consecutive days; 2) be negative for intestinal helminths including S. mansoni (no one was excluded on this criteria as everyone was negative for these infections), 3) have given a blood sample. 190 participants aged 6–40 years met these criteria and formed our study population. As is standard in all our studies after collection of all samples, all participants were offered treatment with the recommended dose of praziquantel (40 mg/kg of body weight).

Whole blood stimulations were set up following a protocol developed at the University of Cambridge, UK 41. 10 mls of heparinised venous blood were collected and immediately diluted 1:4 in RPMI 1640 medium (Sigma) with penicillin (50 U/ml), streptomycin (50 μg/ml), L-glutamine (2 mM) (Sigma) and 10% AB serum. 250 μl of this media was added to a 48-well culture plate which already contained 25 μl of antigen in RPMI media described above at 10 μg/ml (SWAP, or Con A) or media alone. The study focused on responses against whole worm antigen to provide a complete picture of antibody and cellular responses against worm antigens (see 3537 presenting results of antibody responses of participants from this population).

Additional cultures were set up to detect the effect of neutralising IL-10 using samples from a subgroup of the participants who had provided enough blood to allow additional assays (n = 17, age range 6–17 years, mean infection intensity range 0–111 eggs/10 ml urine). For these cultures, 50 μl/well of 1 μg/ml anti-human IL-10 (rat IgG1 anti-human, JES3-9D7) antibody was added to half the plate before addition of the whole blood and 50 μl/well of 1 μg/ml isotype control (rat IgG1 R3-34) was added to the remaining half, so that the same sample was run on the same plate in wells containing anti-IL-10 antibody and the isotype control. The plates were placed in a glass jar, together with a gas generating kit (Oxoid) and incubated at 37°C for 48 hrs. Supernatants were harvested into cryotubes and stored at -20°C in the field prior to transportation at -20°C to the United Kingdom for analysis. Samples were thawed and all 4 cytokines assayed at the same time for each sample. IFN-γ, IL-4, IL-5, IL10 cytokine assays were conducted using capture ELISAs with antibody sets from BD Pharmigen following previously published protocols 41. All assays were conducted in duplicate.

Initial analyses were to determine the relationship between cytokine levels and infection intensity, as well as between cytokine pairs. After exploratory plots showed that these relationships were non linear and were directional (either positive or negative), the analyses were conducted using a one-tailed non parametric Spearman correlation procedure 42.

The second step was to determine if the cytokines followed a distinct age profile. Because of the potential confounding relationships of infection intensity, host age and sex on cytokine levels (see 43), the effect of infection intensity and sex on net cytokine level was allowed for before testing if age had a significant association with net cytokine level. This was achieved by a multivariate analysis of variance (MANOVA). The dependent and independent variables were transformed or categorised as appropriate to satisfy the assumptions of the parametric tests. Cytokine level (square root transformed) was the dependent variable and sex (2 categories), age (6 categories as shown in Table 1) and infection intensity (log₁₀ (x+1) transformed) were the independent variables.

The next statistical tests were to determine the relationship of the cytokine responses (cumulatively) and infection status. This was achieved by factor analysis using principal components (PCA) 42. PCA is a standard technique for reducing multivariate data down to its main independent features 42. This procedure was used in this instance for two reasons (1) to avoid type I and type II errors in multiple tests using correlated independent variables 4344 and (2) to capture the effects of all cytokines in a single analysis since the responses are likely to be acting simultaneously (see 45). Following the factor analysis principal components were analyzed further with logistic regression to determine the risk factors associated with being infected (egg positive) or uninfected (egg negative). Independent variables included the principal components, sex and age group (categorised as above).

The final statistical test was a two-tailed paired t-test to determine the effect of blocking IL-10 on cytokine levels conducted on the square root transformed cytokine data. All statistical tests were conducted using the software package SPSS. There is an outlier in the IL-4 data, therefore all statistical procedures were repeated with the single outlier removed; this did not affect the outcome of any of the statistical tests, so the results reported here include the outlier. All significant p values (<0.05) were subjected to Bonferoni testing 46.

Mean infection intensity for the study population was 37 eggs/10 ml urine with a SE of the mean of 6.19 eggs/10 ml urine (range 0 eggs – 535 eggs/10 ml urine) while infection prevalence was 54.7%. Infection prevalence was similar across all age groups except in children aged 13–14 and 17–18 where males had higher prevalences than females. Although these infection levels are moderate relative to reports from other areas in Zimbabwe 10, the World Health Organisation denotes this prevalence level as high and infection intensity also as high as defined by having more than 10% of the population with >50 eggs/10 ml of urine (36 of the 190 participants had >50 eggs per 10 ml urine) 47. Infection rose to peak in childhood (11–12 years) followed by a sharp decline in infection intensity while prevalence fell more gradually as shown in Figure 1. Infection levels in this population peaked at lower levels and in older children compared to high infection areas in Zimbabwe 10. There were no other helminth species detected in the participants enrolled for this study.

We measured levels of IFN-γ, IL-4, IL-5 and IL-10 in vitro responses to mitogen (ConA) and parasite adult worm antigen (SWAP) challenge. In general, responses to ConA were higher than those against parasite antigens and spontaneous cytokine production was lower than parasite and mitogen-induced responses. Of the 190 participants, the following produced detectable parasite specific cytokines; 114 (60%) produced IFN-γ, 86 (45%) produced IL-4, 103 (54%) produced IL-5 and 144 (76%) produced IL-10, while 32 people (17%) produced all 4 cytokines simultaneously. The most distinct difference between age profiles of mitogen and parasite-specific responses was in IL-10 as shown in Figure 2. Levels of IL-10 rose to peak with infection intensity in 11–12 year olds, while levels of IL-4 and IL-5 rose more slowly peaking later in 15–16 year olds.

In order to compare age profiles we had to determine if the cytokines produced varied with host age after allowing for the effects of sex and infection intensity. MANOVA analyses showed that host age significantly affected levels of parasite-specific IL-4 and IL-5 and IL-10 but not IFN-γ as shown in Table 2.

Plots of the relationship between cytokine levels and infection intensity (Figure 3) indicate that

IFN-γ and IL-4 production did not show significant associations with infection intensity. People carrying high levels of infection (>50 eggs/10 ml urine) tended to make low levels of IL-5 (<10 ng/ml), while people with higher IL-5 levels had low levels of infection although this relationship was not statistically significant. IL-10 levels were significantly correlated to infection intensity and this correlation was positive as shown in Table 3.

The relationship between infection status (egg positive or negative) and cytokine response was explored using factor analysis and logistic regression. The factor analysis approach was essential in order to reduce the cytokine data to fewer variables thus minimising problems associated with multicolinearity 44 as well as to produce variables containing all the cytokines measured. Factor analysis gave 2 principal components with the strongest contributing variables being IFN-γ, IL-4, IL-5 for PCA1 and IL-10 for PCA2. The amount of variation explained by each of the components is given in Table 4 while the weightings of each cytokine in the two components is given in Table 5. Thus the strongest dichotomy in the immune responses was between the effector T cell cytokines in PCA1 and the immuno-modulatory cytokine IL-10 in PCA2. Logistic regression analyses showed that infection status was significantly affected by PCA 2 (made up largely of IL-10) (p = 0.039, β = 0.51 with a SE of 0.247, -2 Log likelihood of 6.522) and that the effect was positive. Therefore people producing high levels of parasite-specific IL-10 were significantly more likely to be egg positive than those producing lower levels of the cytokine.

The relationships between pairs of cytokine are plotted in Figure 4 and the results from the statistical tests are shown in Table 3. The most notable correlation was between parasite specific IL-10 and IL-5, which was significantly negative. Other cytokines which were significantly correlated were INF-y with both IL-4 and IL-5 and IL-4 and IL-5, which were all positively correlated. IL-4 and Il-10 showed no significant relationship with each other. People produced high levels of either IL-5 (>12 ng/ml) or IL-10 (>140 ng/ml) but not high levels of both. These significant positive correlations between INF-y, IL4 and IL-5 and the negative correlation between IL-10 and IL-5 indicate the relationships between the cytokines which underlie the PCA analyses.

Antibody to IL-10 was added to in vitro cultures during parasite antigen challenge. Blocking IL-10 in this way resulted in a significant change only in the level of parasite-specific IL-5, which increased from a mean of 5.15 ng/ml (SE = 1.39) in the control samples incubated in the presence of an isotype control to 6.56 ng/ml (SE 1.30) in the samples incubated in the presence of anti-IL-10 antibodies (T-Value = 3.12, df = 16, p = 0.007). IFN-γ and IL-4 production did not differ significantly between IL-10 blocked and normal control samples (T = 0.07, df = 16, p = 0.948; T = -0.62, df = 16, p = 0.542 respectively).