Reduced plasma levels of GM-CSF is a common feature of Schistosoma mansoni-infected school-aged children

Abstract

Background: Currently available schistosomiasis diagnostic and monitoring tools are limited, and the development of novel technologies is necessary to enhance disease diagnostic and surveillance by supporting elimination efforts. Novel disease-specific biomarkers can facilitate the development of these technologies. Through the comparison of parasite burden and host factors, we assessed whether host plasma cytokines could be used as robust biomarkers for intestinal schistosomiasis and associated pathology in school-aged children (SAC) living in endemic areas.

Methods: Levels of host plasma cytokines were measured in SAC from a low-to-moderate burden region five months deworming with praziquantel, using Luminex assay for exploration analysis and ELISA for validation.

Results: The concentration of GM-CSF, IL-2, and VEGF in plasma was significantly lower in schistosome-infected compared to non-infected children, as determined by Luminex assay. Further evaluation by ELISA revealed a negative correlation between GM-CSF plasma levels, but not those of IL-2 or VEGF, and S. mansoni egg burdens in infected individuals. Common coinfections in the study area such as geohelminths, hepatitis or malaria failed to alter plasma GM-CSF levels arguing in favor of a potential specific effect of S. mansoni infection on this cytokine. Receiver operating characteristic analysis confirmed GM-CSF as an acceptable predictive marker of S. mansoni infection, with an area under the curve (AUC) of 75%. Finally, the adjunct use of plasmatic GM-CSF thresholds for screening S. mansoni at-risk children and identify S. mansoni-infected ones increased the sensitivity of a single Kato-Katz test by averagely 15%.

Conclusions: Our findings highlight the potential of using plasma GM-CSF levels to biomark S. mansoni infection and improve the sensitivity of single Kato-Katz based diagnostic for low- to-moderate burden infections.

Keywords: GM-CSF; adjunct diagnostic; biomarker; cytokine; schistosomiasis.

Figure 1

Flow diagram describing the enrolment strategy, examination process and designing of participant groups for the discovery and validation cohorts. For the selection of participants for the Luminex experiment (n=40) and enzyme-linked immunosorbent assay (ELISA) (n=80), infection was defined as the presence of S. mansoni eggs in the participant’s stool, as observed by KK technique; Liver fibrosis was defined using the liver image pattern (LIP) score. No hepatic fibrosis: Final score of 0, consisting of participants with only LIP A. Hepatic fibrosis: LIP score ≥ 2/(LIP C – LIP E). Participants were grouped into four groups based on the presence or absence of eggs in the stool (KK+ versus KK-) and the presence or absence of conclusive signs of liver fibrosis (US+ versus US-). Excluded participants: Degraded samples (n =13) or participants positive for HBV and/or HCV (n=43). This resulted in 4 phenotypic groups of patients, i.e., KK-US-, KK-US+, KK+US- and KK+US+. KK: Kato Katz; US: Ultrasonographic examination; (+): Positives; (-): Negatives. All participants with available blood samples were considered for further analyses and screened for concomitant infections such as geohelminths in stool samples, hepatitis B or C in blood and malaria thick smear from blood samples. All participants with another reported infection of the above was excluded from the further analyses to strictly enable S. mansoni mono-infections to be probed comparatively against non-infected controls. Two cohorts were constituted from the retained participants to perform first a discovery run of 10 participants per group (total, n = 40). Plasma samples selected for the discovery run were tested by Luminex for 27 cytokines to assess differential expression between groups with different infection and/or liver fibrosis statuses. The validation run was then performed with another 20 age-, gender-, BMI (body mass index)- and FCW (frequency of contact with infested water)- matched participants per group (total, n = 80) by cytokine-specific ELISA for confirmation of the biomarking potential of any candidate cytokine deemed interesting from the discovery run results. Maquisard study: An EDCTP and UK Royal Society funded study on school children from rural Cameroon.

Figure 2

Cytokines profiles across participants infected with S. mansoni and/or hepatic fibrosis from the discovery run (A). Heatmap after unsupervised hierarchical clustering based on cytokine expression similarity and showing the variation of the 27 cytokines expression in individual participant and per clusters. The R package “ComplexHeatmap” was used to generate the heatmap. (B) Radar plot showing the variation of the 27 cytokine’ expressions between the four groups of participants from the discovery run with indicated p-value of multiple comparison. For statistical comparison, using graph pad prism, Kruskal-Wallis test followed by Dunn test was performed to assess significant differences between 4 groups. (n = 40).

Figure 3

Differential cytokine expression during S.mansoni infection and/or liver fibrosis by ELISA across validation run samples. Four analytical univariate schemes are presented here i.e. overall: to unveil any suggestive (though not definitive) differences as a result of S mansoni infection and/or liver fibrosis. Scheme 1: to unveil any suggestive (though not definitive) differences as a result of S. mansoni infection before the onset of liver fibrosis. Scheme 2: to unveil any suggestive (though not definitive) differences as a result of liver fibrosis onset following S. mansoni infection. Scheme 3: to unveil any suggestive (though not definitive) differences as a result of any stage of S. mansoni infection. (A) <Comparison of plasma levels of GM-CSF across the 4 groups KK-US-, KK-US+, KK+US-, KK+US+. (B) Plasma levels of GM-CSF in KK-US- Vs KK+US-. (C) Plasma level of GM-CSF in KK+US- Vs KK+US+. (D) Plasma levels of GM-CSF in KK-US- versus KK-US-/+ (i.e. both KK+US- and KK+US+). (E) Comparison of plasma levels of IL-2 across the 4 groups KK-US-, KK-US+, KK+US-, KK+US+. (F) Plasma levels of IL-2 in KK-US- Vs KK+US-. (G) Plasma level of IL-2 in KK+US- Vs KK+US+. (H) Plasma levels of IL-2 in KK-US- versus KK-US-/+ (i.e. both KK+US- and KK+US+). (I) Comparison of plasma levels of VEGF across the 4 groups KK-US-, KK-US+, KK+US-, KK+US+. (J) Plasma levels of VEGF in KK-US- Vs KK+US-. (K) Plasma level of VEGF in KK+US- Vs KK+US+. (L) Plasma levels of VEGF in KK-US- versus KK-US-/+(i.e. both KK+US- and KK+US+). For statistical comparison, using graph pad prism, Kruskal-Wallis test followed by Dunn test was performed to assess significant differences between the 4 groups (A, E, I). Mann Whitney U test was used to perform preliminary univariate comparison between groups in schemes 1,2,3. GM-CSF, Granulocyte Monocyte Colony Stimulating Factor; IL, Interleukin; VEGF, Vascular endothelial growth factor; KK-US-, negative for both S. mansoni eggs and hepatic fibrosis; KK-US+, negative for S. mansoni eggs and positive for hepatic fibrosis; KK+US-, positive for S. mansoni eggs and negative for hepatic fibrosis; KK+US+, positive for both S. mansoni eggs and hepatic fibrosis. ns= not significant; *=p<0.05; **=p<0.01.

Figure 4

Correlation of levels of plasma cytokine candidate biomarkers with S. mansoni infection burden. (A) Correlation between plasma levels of GM-CSF and S. mansoni infection burden (B) Correlation between plasma levels of IL-2 and S. mansoni infection burden. (C) Correlation between plasma levels of VEGF and S. mansoni infection burden. For statistical comparison, using graph pad prism, Kruskal-Wallis test followed by Dunn test was performed to assess significant differences between the 3 egg burden groups. p< 0.05 was considered significant. GM-CSF, Granulocyte Monocyte Colony Stimulating Factor; IL, Interleukin; VEGF, Vascular Endothelial Growth Factor; N, Negative for Sm eggs; L, light infection (1–99 EPG); M, moderate infection (100–399 EPG); EPG, eggs per gram of stool. *=p<0.05.

Figure 5

Receiver Operating Characteristic (ROC) curves to assess the diagnostic potential of S. mansoni-infected individuals using plasma levels of identified candidate cytokine biomarkers. (A) ROC curves showing the potential of GM-CSF in the diagnosis of S. mansoni infection. (B) ROC curves showing the potential of IL-2 in the diagnosis of S. mansoni infection. (C) ROC curves showing the potential of VEGF in the diagnosis of S. mansoni infection. GM-CSF, Granulocyte Monocyte Colony Stimulating Factor; IL, Interleukin; VEGF, Vascular Endothelial Growth Factor; AUC, Area Under the Curve; ROC, Receiving Operating Curve.

Figure 6

Adjunct diagnostic potential of plasma GM-CSF levels in better identifying S. mansoni-infected individuals after a single Kato-katz. (A) Plasma levels of GM-CSF threshold design for the separation of non-infected and S. mansoni-infected children. Plasma levels of GM-CSF were used as a diagnostic tool, with the geometric mean minus standard deviation (GM-SD) of non-infected individuals representing the lower cut-off (red line) separating non-infected (above) from infected individuals (below). (B) Summary table of the performance of each KK, plasma levels of GM-CSF, combination of single KK and GM-CSF-based identification versus our study gold standard (i.e. two repeated KK, Kato 1 and Kato 2, with each performed in duplicate). The sensitivity and specificity of the several diagnostic approaches were used against Kato final (the gold standard of mean of Kato1 and Kato2). Kato1 was performed after the first sample stool was collected, whereas Kato2 was performed after the second sample stool was collected from the same individual 5 days later. The sensitivity of the combination of plasma levels of GM-CSF with either Kato1 or Kato2 were also determined (in red) and compared to the original sensitivity of the single KK-based detection (in blue).

Figure 7

Study pathway to impact. Biorender.com software was used to design this workflow. (A) Detection of eggs in stools samples was used to diagnose schistosomiasis infection. (B) Kato Katz method is recommended for egg detection in stool samples and usually conducted on stool from a singlecollection day. (C) Analysis of stool from a single collection day by the KK method, even in duplicate, presents a low sensitivity in the diagnosis of schistosomiasis infection. (A’) Assessment of the role of host cytokine level changes during schistosomiasis. (B’) Large screening of host plasma cytokines using the Luminex method in a discovery run. (C’) Validation of cytokine candidates obtained from the discovery run after statistical analyses on other cohorts (with or without coinfections) using cytokine-specific ELISAs. (D) Identification of plasma GM-CSF as a robust plasma biomarker and use as adjunct diagnostic tool for the common single day stool-based KK achieving a 15% increase in sensitivity (E).