SNPs in IL4 and IFNG show no protective associations with human African trypanosomiasis in the Democratic Republic of the Congo: a case-control study

Background: Human African trypanosomiasis (HAT) is a protozoal disease transmitted by tsetse flies. Infection with trypanosomes can lead directly to active HAT or latent infection with no detectable parasites, which may progress to active HAT or to spontaneous self-cure. Genetic variation could explain these differences in the outcome of infection. To test this hypothesis, polymorphisms in 17 candidate genes were tested ( APOL1 [ G1 and G2], CFH, HLA-A, HPR, HP, IL1B, IL12B, IL12RB1, IL10, IL4R, MIF, TNFA , IL6, IL4, IL8, IFNG, and HLA-G). Methods: Samples were collected in Democratic Republic of the Congo. 233 samples were genotyped: 100 active HAT cases, 33 from subjects with latent infections and 100 negative controls. Commercial service providers genotyped polymorphisms at 96 single nucleotide polymorphisms (SNPs) on 17 genes. Data were analyzed using Plink V1.9 software and R. Loci, with suggestive associations (uncorrected p < 0.05) validated using an additional 594 individuals, including 164 cases and 430 controls. Results: After quality control, 87 SNPs remained in the analysis. Two SNPs in IL4 and two in IFNG were suggestively associated (uncorrected p<0.05) with a differential risk of developing a Trypanosoma brucei gambiense infection in the Congolese population. The IFNG minor allele (rs2430561, rs2069718) SNPs were protective in comparison between latent infections and controls. Carriers of the rs2243258_T and rs2243279_A alleles of IL4 and the rs2069728_T allele of IFNG had a reduced risk of developing illness or latent infection, respectively. None of these associations were significant after Bonferroni correction for multiple testing. A validation study using more samples was run to determine if the absence of significant association was due to lack of power. Conclusions: This study showed no evidence of an association of HAT with IL4 and IFNG SNPs or with APOL1 G1 and G2 alleles, which have been found to be protective in other studies.


Introduction
Human African trypanosomiasis (HAT), or sleeping sickness, is a neglected tropical disease caused by infection with extracellular blood protozoan parasites, which are transmitted by the bite of the tsetse fly (Glossina sp) (Headrick, 2014).
The disease is present in more than 250 foci in Africa and nearly 70 million people are at risk of infection with the two human-infective forms. A chronic form is found in Central and Western Africa and is caused by Trypanosoma brucei gambiense (Tbg); an acute form is caused by Trypanosoma brucei rhodesiense (Tbr) and is found in East Africa. About 97% of all HAT cases are due to Tbg infection, which causes a chronic disease with a long latency period (Kennedy, 2013).
There is currently no vaccine against HAT. The control of Tbg HAT is mainly based on the active detection of infected cases followed by their treatment and also vector control. After infection, the parasites may be undetectable by microscopy for months or years; these cases are known as latent infections and are detected by serology. Active HAT with microscopically detectable parasites evolves in two successive stages: stage 1 (haemolymphatic) and stage 2 (meningoencephalitis, also known as neurological). The transition from stage 1 to stage 2 is a consequence of the parasites crossing the blood-brain barrier. For a long time, HAT was considered fatal if untreated; however, the dogma that HAT is 100% fatal has been questioned (Jamonneau et al., 2012). It is now recognised that some individuals self-cure after developing a latent infection.
The Democratic Republic of the Congo (DRC) is the country with the highest incidence of HAT. More than half of districts are endemic for HAT, with more than 36 million people at risk (Report of a WHO Expert Committee, 2013). There was an increase in the number of new cases at the end of the 20th century in both rural and urban areas; however, the number of cases is currently decreasing, with fewer than 3,000 cases detected in 2015(Büscher et al., 2017. There are several HAT foci in DRC with very different clinical presentations and outcomes, which could be explained by the presence of genetically different subpopulations of humans and possibly parasites, as well as environmental factors (Kande Betu Kumesu, 2012). There is also variation in disease phenotype within populations, which may be partly due to individual genetic variation.
Several genes have been implicated in the control of other infectious diseases (AIDS, tuberculosis, malaria, etc.), but genetic data on HAT are more recent and incomplete. For example, differences in cytokine levels and other immunological markers may be a consequence of differences in disease states or of genetic differences between individuals in these cytokine genes or in genes that regulate them.
Polymorphisms in IL6, IL10, IL1A and TNFA have been associated with risk of developing HAT (Bucheton et al., 2011;Courtin et al., 2006;Courtin et al., 2007) in the DRC, although only the association with IL6 remained significant after Bonferroni correction. An HLA-G haplotype has also been associated with HAT susceptibility in a Congolese population (Courtin et al., 2013).
It has also been reported that levels of interleukins IL8 and IL10 increase during infection and then decrease after treatment (Lejon et al., 2002) and that the stages of the disease are also correlated with levels of IL6 and IL10 (Ilboudo et al., 2012;Sternberg et al., 2005). Our study is designed to investigate polymorphisms in candidate genes that have previously been implicated in the outcome of HAT, in an active focus in DRC.

Ethical statement
The study protocol was submitted to the ethics committee of the National Ministry of Public Health (DRC) and approved under number 1/2013 and by the Ethics Advisory Committee (EACC) of the Institute for Research and Development (IRD), Marseille.
All enrolled subjects signed an informed consent form written in French which was translated into the local language if necessary. Parents or guardians signed consent forms on behalf of minors when they couldn't sign themselves and they had given verbal consent.

Study design
This study was one of six studies of populations of HAT endemic areas in Guinea, Cameroon, Ivory Coast, DRC, Malawi and Uganda (Ahouty et al., 2017;Kabore et al., 2017;Kamoto et al., 2019;Kimuda et al., 2018;Ofon et al., 2017). The studies were designed to have 80% power to detect odds ratios (OR) >2 for loci with disease allele frequencies of 0.15 -0.65, with the 96 single nucleotide polymorphisms (SNPs) genotyped for 100 cases and 100 controls.

Study site
Our study was conducted in the district of Masi-Manimba, in the province of Kwilu, which was formerly a part of Bandundu province, in the south-west of the DRC, more than 388 km from Kinshasa.

Recruitment
The study was undertaken in two stages: 1) a screen of 96 candidate loci in 233 subjects and 2) a validation study of two loci that were suggestively positive in the first round in 594 additional subjects.
Subjects were aged at least 12 years old and were enrolled in this study from July 2013 to March 2016 during the active screening campaigns of the mobile team of the National Program of Control of Human African Trypanosomiasis (PNLTHA). Participants were recruited in their respective villages during active screening campaigns. Simultaneous serological (Card Agglutination Test for Trypanosomiasis; CATT) testing and parasitological examination were performed. Samples that were CATT positive and lymph node fluid or Buffy coat positive for trypanosomes by microscopy were subject to the immune trypanolysis test, which is highly specific for T. b. gambiense (Van Meirvenne et al., 1995).
The immune trypanolysis test allows to confirm or not the presence of T. b. gambiense specific antibodies in an individual. Individual's plasma was collected in a 2 ml cryotube in the field, kept in a cryocontainer filled with liquid nitrogen and brought to the laboratory where plasma was stored in deep freezer -80°c. Live particular strains of T. b. gambiense, LiTat 1.3 and LiTat 1.5, were incubated with plasma as well as with guinea pig serum, serving as a source of complement. If the T. b. gambiense specific antibodies are present in the plasma, they bind to trypanosomes that will be lysed. Absence of trypanosome mobility, observed under a microscope, implied a positive result.
Cases (stages 1 and 2) were recruited from the positive test subjects for CATT, for whom parasitological evaluation confirmed the presence of the parasites in blood or lymph node fluid (for stage 1) and/or cerebrospinal fluid (CSF) (for stage 2).
Controls were CATT negative individuals living in the same area as cases and for whom the trypanolysis test was negative.
The subjects with latent infections were individuals who were negative by microscopy despite positive CATT and positive trypanolysis test results.
Controls and subjects with latent infections were enrolled as a comparison group. The first have never been in contact with T.b.gambiense , the latter have been in contact with T.b.gambiense but in this group, the parasite is no longer found in the absence of any treatment.
Individuals who could not provide a sufficient volume of blood (less than 9 ml) or were under 12 years old were excluded. Previously treated patients were not included as controls.
Sample collection and processing At inclusion, blood was collected through venepuncture by a skilled technician of PNLTHA using two heparin tubes of 5 ml. The first tube was used for serological (CATT) and parasitological examinations (capillary tube centrifugation [CTC], or mini-Anion Exchange Column Trypanosoma on Buffy Coat [mAECT-BC]). The second tube was used to collect plasma and buffy coat in 2 ml cryotubes. These two last samples were taken after centrifugation at 3500 rpm of the blood in the heparinized tube for 5 minutes. In addition, if the individual had shown enlargement of cervical ganglia, they were punctured and lymph node fluid examined by direct examination for trypanosomes. Once the trypanosome was isolated either in lymph node fluid or in the blood, the cerebrospinal fluid (CSF) was collected by lumbar puncture and examined for stage determination of the disease according to PNLTHA instructions. Plasma and Buffy coat were kept in containers filled with liquid nitrogen until they were sent to INRB where plasma was stored in the freezer at -20 ° C and Buffy coat in liquid nitrogen for subsequent DNA extraction for genotyping. DNA was extracted from Buffy Coat using the Maxwell16 Promega kit following the manufacturer's instructions (Maxwell®16 Tissue DNA Purification kit, cat# AS1030), quantified by Nanodrop (Thermo Scientific Nanodrop 2000, ISOGEN, Life Science) and then stored at -20°C. All DNA were sent to Makerere University, Uganda where they were quantified by Qubit ® 3.0 Fluorometer (Invitrogen by Life Technologies) prior to shipment for genotyping. DNAs were shipped at room temperature. Transportation lasted less than 48 hours.

Data and statistical analysis
The results were analyzed using Plink V1.9 software (Chang et al., 2015;Purcell et al., 2007) and R (R Core Team, 2008) for viewing. Markers with >40% missing data (--geno 0.4) were removed and individuals with >30% (--mind 0.3) missing data were removed Power calculations were undertaken using the genetics analysis package (GAP) in R (R Core Team, 2008;Zhao, 2007).
Fisher's exact test was used to identify associations between phenotypes and SNP loci.
The Bonferroni correction was used to correct for multiple testing. A p value below 0.05/n was considered statistically significant, where n is the number of SNPs in a given comparison after quality control.

Quality control of data
In the candidate gene study, 233 DNA samples were genotyped: 100 active HAT cases, 33 from subjects with latent infections and 100 negative controls with a male/female sex ratio of 0.63. Data were filtered to remove loci and individuals with excessive missing data. After examining the distribution of the data, markers with >40% missing data were removed and individuals with >30% missing data were removed. These filters removed nine SNP loci, one case, four controls and four seropositives. All remaining polymorphisms were in Hardy-Weinberg equilibrium.
For the validation study, 594 DNA samples were genotyped: 164 active HAT cases and 430 negative controls with a male/ female sex ratio of 0.58. There was no significant difference in sex ratio between cases and controls (p=0.56) or between discovery and validation groups (p=0.43).
Association analysis in candidate gene study Data for the three possible comparisons between the three phenotypes (cases versus controls; latent infections versus controls; and cases versus latent infections) were analysed separately.
Cases and controls. Two SNPs in IL4 (rs2243258 and rs2243279) and two in IFNG (rs2430561, rs2069718) had minor alleles that appeared to be protective against HAT in the comparison between cases and controls (odds ratios <1) before Bonferroni correction (Table 1). After Bonferroni correction, neither remained significant (p>0.05).

Latent infections and controls.
The minor alleles of IFNG (rs2430561 and rs2069718) SNPs were also protective in the comparison between latent infections and controls. The minor (T) allele of IL6 (rs2069830) was not found at all in the 29 subjects with latent infections but was found at 7% frequency in the controls ( Table 2), suggesting that it might be protective against seroconversion.  Cases and latent infections. Three SNPs in IL4 (rs2243258, rs2243279 and rs71889624) were associated with protection against progression from latent infection to case (Table 3). Two of these were also just as strongly associated with cases and controls, despite less than a third of the number of samples of latent infections being available. This was a consequence of a higher frequency of the minor allele in latent infections than in controls. Carriers of the T allele of rs2243258 and those of the A allele of rs2243279 of IL4 have a reduced risked of developing illness; as do carriers of the T allele of rs2069728 of IFNG, for asymptomatic carriers in relation to the controls.
The complete absence of the T allele of rs2069830 of IL6 in latent infections also suggests that this SNP is associated with risk of progression to active HAT. None of the associations with the latent infections remained significant after Bonferroni correction.
Validation study. IL4 and IFNG SNPs that were marginally significant in the candidate gene study were not significant in the validation study (Table 4).

Discussion
In the candidate gene study, 87 SNPs in 17 candidate genes remained after quality control. We found suggestive associations with HAT at SNP loci in IL4, IFNG, IL6 and MIF, but none of these remained significant after a Bonferroni correction. It was particularly notable that no association was found  with APOL1 despite the APOL1 protein being lytic to T. brucei brucei and some alleles being lytic to T. b. rhodesiense and T.b. gambiense in vitro (Genovese et al., 2010). Associations have been found with the APOL1 G1 allele in comparisons with latent infections in Guinea (Cooper et al., 2017). In that study, latent infections had been followed for two years, during which time they remained serologically positive but parasitologically negative. In our study, latent infections were defined as serologically positive but parasitologically negative at a single visit. This different case definition and the small sample size of latent infections (n=33) may account for the difference in outcomes.
Associations have also been found with APOL1 G2 frequencies between cases and controls in T.b.rhodesiense infections in Uganda and Malawi (Cooper et al., 2017;Kamoto et al., 2019), but another study in Uganda found no association with APOL1 G2.
Although there is strong evidence for a role for APOL1 in HAT, the role of the different alleles in disease response is much less clear and may depend on the immunological context.

IL4
The minor alleles of rs2243258 and 2243279 were suggestively associated with protection against development of HAT in the case-control comparison and with protection against progression from latent infection to HAT, with odds ratios of 0.28±0.06 in cases versus controls, indicating a potentially large effect on the outcome of infection. However, rs2243258 showed no sign of association with HAT in the validation study (Table 4).
A similar study in the same Bandundu focus published in 2007 tested four SNPs in IL4 and found no association with disease, but none of those SNPs were the ones with suggestive p values in the present study (Courtin et al., 2007). Another TrypanoGEN study in Ivory Coast using the same SNP set as in the present study found five IL4 SNPs suggestively associated with risk of developing a latent infection (uncorrected p values <0.05) (Ahouty et al., 2017). None of those five loci were associated with disease in our study. If these are genuine associations, two different haplotypes with different modes of action may be regulating responses to infection in the two countries.
IL4 is an anti-inflammatory cytokine that is more highly expressed in late stage or chronic infections in mice and may be involved in preventing an excessive inflammatory response, causing tissue damage (Bakhiet et al., 1996;Namangala et al., 2009). However, the only study of IL4 in humans with HAT, which we are aware of, found no differences in IL4 levels in the cerebrospinal fluid of T. rhodesiense-HAT patients (Maclean et al., 2001).

IL6
IL6 (rs2069830, T allele) was suggestively associated with protection against HAT. This result adds to the evidence from previous studies for a role for IL6 in protection against Tbg HAT (Bucheton et al., 2011;Sternberg et al., 2005).
A previous study in the same province of Bandundu found a significant association at IL6rs2069849 (Courtin et al., 2007). This SNP was not included in the present study, but this finding supports a role for IL6 variants in the development of HAT. A recent study in Guinea found that IL6rs1818879 was associated with a lower risk of progressing from latent infection to active disease (Kabore et al., 2017) and an Ivorian study found that rs62449495 and rs2069830 have a protective effect against developing active Tbg HAT (Ahouty et al., 2017). The minor allele of this last SNP (rs2069830) was also found in the same study to be protective against progressing from latent infection to cases (Ahouty et al., 2017).
IL6 is an inflammatory protein that plays an important role in host immunity, particularly during the acute phase of the infection. Its production increases considerably in the late phase of infection and decreases dramatically after treatment, suggesting that it plays an important role in humans infected with Tbg (Lejon et al., 2002).

IFNG
The minor alleles of rs2069718, rs2430561 and rs2069728 were associated with protection against progression in Tbg HAT in the case versus control, latent infection versus control and latent infection versus case comparisons, respectively, although the results were not significant after Bonferroni correction. Individuals with the T allele of rs2069728 had about half the risk of developing HAT in the latent infection versus case comparison, as compared to carriers of the C allele. Two of the three SNPs within the same gene were tested in an Ivorian population and there was no protective effect in a comparison of cases versus controls (p>0.05) (Ahouty et al., 2017). A previous study of cases and controls in the same focus in Bandundu (Courtin et al., 2007) revealed no association with HAT at four IFNG SNPs but did not include those that were suggestively associated in this study. SNP rs2430561 was included in the validation study and no evidence of any association was found (Table 4). A parallel study in Guinea with the same SNP (rs2430561) as the present study also found that IFNG was not associated with HAT in both case versus latent infection and latent infection versus control comparisons (Kabore et al., 2017). Recent studies on humans suggest that IFNG can be associated with neurological symptoms in Tbr HAT (Kato et al., 2016) but it is not known if this is also true of Tbg HAT. IFNG is involved in parasite control in mice (Hertz et al., 1998) and recently, its production was found to be elevated above background in the plasma of individuals with latent infections after stimulation with Tbg lysates but not in controls or cases of active HAT (Ilboudo et al., 2016).

Latent infections
It is important to note that more suggestive associations were observed in comparisons of latent infections with cases or controls than were observed in the case versus control comparison, despite the small number of latent infection samples available (only 29 after quality control). This may be the most important observation of this study and suggests that latent infections are a genetically distinct group, as has been found in Guinea and Ivory Coast (Ilboudo et al., 2016;Kabore et al., 2017). Latent infections might be an important reservoir of infection, but their role and significance in the epidemiology of HAT is poorly understood. If people with latent infections have distinct genetic characteristics, it may be possible to develop genetic approaches to estimating the size of this group.

Validation study
Despite more samples being available in the validation study, including 164 cases and 430 controls, IL4 and IFNG SNPs were not significantly associated with HAT. Random sample variation may account for the difference in outcome of the candidate gene and validation studies that used two different sets of samples. The negative result of the validation study shows how important it is to validate observations from candidate gene studies using fresh datasets.

Conclusions
In conclusion, our survey reveals that two SNPs in each of IL4 (rs2243258, T allele, and rs2243279, A allele) and IFNG (rs2430561, A allele, and rs2069718, G allele) appeared to be associated with a low risk of developing symptomatic Tbg infection in the Congolese population; however, these associations were not significant after Bonferroni correction or validation. The genetic factors of susceptibility to infection by Tbg in DRC were not replicated in other populations, suggesting that genetic risk factors may vary according to the population.
Access to the data will be granted to life science researchers for research purposes only. Users should write to Prof. Dieudonné Mumba Ngoyi (mumbadieudonne@yahoo.fr) who will authorise the EGA to release the data to the named researchers after completion of a data access agreement.  A, HPR, HP, IL1B, IL12B, IL12RB1, IL10, IL4R,  MIF, TNFA, IL6, IL4, IL8, IFNG, and HLA-G). This research is of interest and will contribute towards our understanding of gHAT clinical spectrum and thus guide the design of sustainable control approaches. However, a number of improvements as highlighted below are required so as to make the work more interesting and conclusive. Throughout the write out, including the conclusion, the authors should clearly indicate that the results are for gHAT since these findings might not directly apply to rHAT that was not considered in this manuscript.
○ I would have liked to see improvements in the background section. The study concludes as "no evidence of an association of HAT with IL4 and IFNG SNPs or with APOL1 G1 and G2 alleles, which have been found to be protective in other studies". However, the introduction elaborates more about other markers and does not give evidence of studies that have reported protective infects of IL4 and IFNg.

○
Within the Introduction section, the authors need to provide references for this statement "It is now recognized that some individuals self-cure after developing a latent infection".
○ "For the validation study, 594 DNA samples were genotyped". Validation of a marker is a critical step in association studies. However, the connection between the discovery and validation steps are not well articulated. The authors say they validated with 594 samples but it is not clear about which population and sampling period that these samples were ○ collected. Were these samples collected from the same locality and matched accordingly? My other concern is why the validation step did not include latent infection. If this was an issue of getting samples then it should be indicated.
The last sentence in the result section of the abstract "A validation study using more samples was run to determine if the absence of significant association was due to lack of power" is hanging and appears incomplete.

Reviewer Expertise: Bioinformatics and Computational Biology, Population genetics and Genome-Wide Association Studies
I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.