Schistosoma haematobium infection is associated with lower serum cholesterol levels and improved lipid profile in overweight/obese individuals

Infection with parasitic helminths has been reported to improve insulin sensitivity and glucose homeostasis, lowering the risk for type 2 diabetes. However, little is known about its impact on whole-body lipid homeostasis, especially in obese individuals. For this purpose, a cross-sectional study was carried out in lean and overweight/obese adults residing in the Lambaréné region of Gabon, an area endemic for Schistosoma haematobium. Helminth infection status, peripheral blood immune cell counts, and serum metabolic and lipid/lipoprotein levels were analyzed. We found that urine S. haematobium egg-positive individuals exhibited lower serum total cholesterol (TC; 4.42 vs 4.01 mmol/L, adjusted mean difference [95%CI] -0.30 [-0.68,-0.06]; P = 0.109), high-density lipoprotein (HDL)-C (1.44 vs 1.12 mmol/L, -0.24 [-0.43,-0.06]; P = 0.009) and triglyceride (TG; 0.93 vs 0.72 mmol/L, -0.20 [-0.39,-0.03]; P = 0.022) levels than egg-negative individuals. However, when stratified according to body mass index, these effects were only observed in overweight/obese infected individuals. Similarly, significant negative correlations between the intensity of infection, assessed by serum circulating anodic antigen (CAA) concentrations, and TC (r = -0.555; P<0.001), HDL-C (r = -0.327; P = 0.068), LDL-C (r = -0.396; P = 0.025) and TG (r = -0.381; P = 0.032) levels were found in overweight/obese individuals but not in lean subjects. Quantitative lipidomic analysis showed that circulating levels of some lipid species associated with cholesterol-rich lipoprotein particles were also significantly reduced in overweight/obese infected individuals in an intensity-dependent manner. In conclusion, we reported that infection with S. haematobium is associated with improved lipid profile in overweight/obese individuals, a feature that might contribute reducing the risk of cardiometabolic diseases in such population.

The authors do not mention how the sample size was calculated. Please include this information in the Methods section. One of the main limitations of this study is the sample size. This information is crucial to determine if the sample size is appropriate, and therefore the conclusions can be supported.

Authors' reply:
We agree with the reviewer that one of the limitation of our study, as also underlined in the discussion section (page 15), is its rather small sample size. To roughly determine the sample size, we used the average value for total cholesterol levels in a previous small cohort study performed in Lambaréné. For this primary outcome, we aimed to be able to detect a mean difference of ~12,5% between Sh-and Sh+ group, with alpha = 0.05 and a power of 80%. The number of volunteers to be recruited was calculated to be 33 per group. Taking into account the infection prevalence in the study area, a compliance rate of ~80% at screening, and a 5-10% drop-out rate after inclusion (e.g. P. falciparum infection), ~110 individuals were intended to be screened, among which 71 were finally included. This information has been added to the Methods section (line 97-99, Page 7). Individuals Sh+ were negative for STH and Plasmodium? It is not clear in the Methods if coinfections were excluded.
Authors' reply: All the individuals found to be positive for Plasmodium falciparum, in both Sh-and Sh+ groups, were excluded (see Figure S1). We agree that this was not crystal clear in the method section, so we have adjusted the sentence accordingly (line 108-111, Page 7). Concerning the co-infection with STH, we have deliberately decided not to exclude the 7 Sh+ individuals found to be infected with other helminths (see response to Reviewer 2 above). As underlined, the impact of infection with other helminths is negligible and removing them from our analyses only reduces the statistical power but does not affect our conclusions.
Line 112: for treatment of Sh+ individuals, parasitological (presence of eggs in urine) and CAA results were considered?
Authors' reply: For antihelminthic treatment of Sh+ individuals, the presence of urine eggs was used as readout for infection (CAA detection was not done during the field study but months later in the whole samples collection, together with other serum parameters). We have now added this information (line 115-116, page 8).
I strongly recommend including the reference values of the biochemical parameters for your study population as supplemental material.

Authors' reply:
Reference values for biochemical parameters at the whole population level are usually wellestablished in Westernized countries but are not always easily available for African individuals. We provided to the reviewer (see Table 2 for reviewers below) some of the available information obtained from local/national Gabonese health care system but we think it will not be of crucial interest to add them as supplementary data. Of note, the average values for all the biochemical parameters were within the 'normal' physiological ranges for the different groups.

in healthy Caucasian adults
Quantitative insulin sensitivity check index (QUICKI) = 1 / (log(fasting insulin μU/mL) + log(fasting glucose mg/dL)) should be calculated and included as an additional measurement of insulin resistance.

Authors' reply:
We agree that calculating the QUICKI index might be an alternative to HOMA-IR for assessing whole-body insulin resistance. However, we do not find any differences using both methods, whatever the conditions (see Table 3 for reviewers as example below). Taking into account that HOMA-IR is by far the most common index used and that it has also been previously used in the few publications investigating the impact of helminths on metabolic homeostasis in humans, we decided not to include this redundant calculated parameter in our already large tables.

Reviewer #1:
Results are clearly and completely presented. One aspect not addressed is the relationship of the immune response (as determined by eosinophils numbers and or %) to the lipid profile for each individual. The authors do mention in the discussion that the possible effect of IL4/IL13 on hepatocyte function may be one of the mechanisms for lowering TG. Eosinophil levels do reflect the immune response to the parasite by an individual and although the authors are correct to use CAA levels to measure intensity of parasite infection it would also be worth it I thought to measure intensity of the immune response to the Sh and lipid levels.

P-value
Age (

Authors' reply:
We do agree with the reviewer that his/her suggested analysis would have make sense and nicely complement the one done using CAA. Unfortunately, as acknowledged, part of the eosinophil data are missing due to lost samples during field analysis and, as such, we do have a reduced statistical power (especially when the data are stratified according to BMI), preventing to draw reliable and firm conclusions. Of note, when doing this analysis for the whole population (see Figure 1 for reviewers below), some similar trends than the ones observed with CAA are still observed but none of them reached statistical significance due to low sample size.

Figure 1 for reviewers: Associations between intensity of S. haematobium infection assessed by blood eosinophil levels and serum lipid parameters in the whole population
The 'Statistical Analysis' paragraph is well written. These are a good choice of tests, consideration of confounding factors, and multiple testing corrections. Tables S1 & S2 is where they show the raw Odds Ratio and confounder adjusted OR for egg and serum levels. Tables S1, S2 are very informative, and are well presented by showing both raw and adjusted results (that consider multiple important confounding factors). Tables S3, S4, S5 are mislabelled.

Authors' reply:
The labels of Tables S3-5 have been corrected (see new Tables).   Table S3 summarises factors stratified by CAA range; the Eosinophil response appears to show incredibly strong correlation to stratification level, it is unfortunate they lost some measurements for Eosinophils as stated, but their choice of statistical test is well suited to different sample sizes. Table S4 shows the disparity between gender between BMI >/< 25 groups, showing the importance of showing adjusted OR -It is important that the presentation of raw data is included. Fig. 1 It is interesting when whole population the HDL-C has a significant p-value but when you split it into lean & obese there is no significant p-value in either, despite the same trend. The overall trend is clear however that serum CAA levels are associated with many cholesterol measurements in the obese category. Fig. 2 a and b are skillfully plotted -and a statement is needed with respect to what p-values are associated to (#/*/#*) on the heatmap.

Authors' reply:
The definition of the p-value labels has been added to the legend of Figure 2 (Page 23)

Reviewer #2:
The results are well presented and clear We fully agree with the reviewer and, although we speculated in the discussion on some possible underlying mechanism(s), further studies are definitely required for improving our understanding.

Reviewer #3:
I recommend the authors to include a short conclusion paragraph at the end of the discussion. Paragraph 273-292: The article by Cortes-Selva D et al. (Frontiers in Immunology 12;9:2580) should be included since it reinforces the hypothesis that Schistosoma induced Th2 response confers protection from hyperlipidemia, atherosclerosis, and glucose intolerance.
Authors' reply: Together with a short sentence, we have added the suggested publication in the discussion section (line 297-298, Page 15). It might indeed support changes in tissue-resident immune cell lipid/cholesterol metabolism in response to helminth infection.

Reviewer #1:
The 'Statistical Analysis' paragraph is well written. These are a good choice of tests, consideration of confounding factors, and multiple testing corrections. Tables S1 & S2 is where they show the raw Odds Ratio and confounder adjusted OR for egg and serum levels. Tables S1, S2 are very informative, and are well presented by showing both raw and adjusted results (that consider multiple important confounding factors). Tables S3, S4, S5 are mislabelled. Table S3 summarises factors stratified by CAA range; the Eosinophil response appears to show incredibly strong correlation to stratification level, it is unfortunate they lost some measurements for Eosinophils as stated, but their choice of statistical test is well suited to different sample sizes. Table S4 shows the disparity between gender between BMI >/< 25 groups, showing the importance of showing adjusted OR -It is important that the presentation of raw data is included. Fig. 1 It is interesting when whole population the HDL-C has a significant p-value but when you split it into lean & obese there is no significant p-value in either, despite the same trend. The overall trend is clear however that serum CAA levels are associated with many cholesterol measurements in the obese category. Fig. 2 a and b are skillfully plotted -and a statement is needed with respect to what p-values are associated to (#/*/#*) on the heatmap.