Kaposi’s sarcoma-associated herpesvirus seropositivity is associated with parasite infections in Ugandan fishing communities on Lake Victoria islands

We investigated the impact of helminths and malaria infection on Kaposi’s sarcoma associated herpesvirus (KSHV) seropositivity, using samples and data collected from a cluster-randomised trial of intensive versus standard anthelminthic treatment. The trial was carried out in 2012 to 2016 among fishing communities on Lake Victoria islands in Uganda. Plasma samples from 2881 participants from two household surveys, the baseline (1310 participants) and the final (1571 participants) surveys were tested for KSHV IgG antibody responses to K8.1 and ORF73 recombinant proteins using ELISA. The baseline survey was carried out before the trial intervention while the final survey was carried out after three years of the trial intervention. Additionally, a subset sample of 372 participants from the final survey was tested for IgE, IgG and IgG4 antibody concentrations to S. mansoni adults worm antigen (SWA) and S. mansoni egg antigen (SEA) using ELISA. Infection by helminths (S. mansoni, N. americanus, T. trichiura and S. stercoralis) was diagnosed using real-time PCR, urine circulating cathodic antigen (CCA) and stool microscopy (Kato-Katz method) while malaria infection was diagnosed using microscopy. We analysed the relationship between helminth and malaria infections and KSHV seropositivity using regression modelling, allowing for survey design. At baseline, 56% of the participants were male while 48% of the participants were male in the final survey. The most prevalent helminth infection was S. mansoni (at baseline 52% and 34% in the final survey by microscopy, 86% by CCA and 50% by PCR in the final survey). KSHV seropositivity was 66% (baseline) and 56% (final survey) among those 1–12 years and >80% in those 13+ years in both surveys; malaria parasitaemia prevalence was 7% (baseline) and 4% (final survey). At baseline, individuals infected with S. mansoni (detected by microscopy) were more likely to be KSHV seropositive (aOR = 1.86 (1.16, 2.99) p = 0.012) and had higher anti-K8.1 antibody levels (acoefficient = 0.03 (0.01, 0.06) p = 0.02). In the final survey, S. mansoni (by microscopy, adjusted Odds Ratio (aOR = 1.43 (1.04–1.95), p = 0.028) and malaria parasitaemia (aOR = 3.49 (1.08–11.28), p = 0.038) were positively associated with KSHV seropositivity. Additionally, KSHV seropositive participants had higher S. mansoni-specific IgE and IgG antibody concentrations in plasma. Furthermore, HIV infected individuals on cART were less likely to be KSHV seropositive compared to HIV negative individuals (aOR = 0.46 (0.30, 0.71) p = 0.002). Schistosoma species skew the immune response towards Th2 and regulatory responses, which could impact on KSHV reactivation if co-infected with both organisms.

Co-infection with helminths has been shown to modulate immune responses to other infections and vaccines [12][13][14]. Chronic infection with Schistosoma is characterised by the production of IL4, IL5 and IL13 cytokines, typical of a T helper (Th) type 2 response and IL10, a regulatory cytokine [15,16]. The skewed immune response to a Th2 and regulatory response may impair the T helper (Th) 1 response, vital for control of viral infections [17][18][19]. The impact of Schistosoma co-infection on herpesviruses and other viruses has been demonstrated in animal models, where Schistosoma mansoni infection led to IL4-mediated reactivation of murine herpesvirus 68 and M2 macrophage polarization [17,18].
Our group has documented associations between KSHV antibodies and parasite infections including P. falciparum and helminths (hookworm and Mansonella perstans) in rural [4] and peri-urban [20][21][22] populations in Uganda. The Lake Victoria island communities in East Africa are characterised by poor sanitation and a high prevalence of infectious diseases, including schistosomiasis [23][24][25][26]. No study to date has documented the burden of KSHV or KS in these unique communities. This study aimed to determine the seropositivity of KSHV in the Lake Victoria island communities of Koome sub-county, Uganda and the association between KSHV seropositivity and parasite co-infections.

Statistical analysis
Statistical analysis was carried out using STATA version 13 (StataCorp, College Station, Texas USA). The survey study design of the main trial was non self-weighting (because the number of households selected from each village were not dependant on village size, therefore households from smaller villages were more likely to be included in the survey than households from larger villages). To allow for this non self-weighting design and to ensure that our analyses are representative of the study area, we therefore took into account clustering within villages and applied village-level weights to allow for the different village sizes for all observational analyses [36]. Logistic regression (allowing for the survey design) was used to determine associations between risk factors and KSHV seropositivity. Linear regression (allowing for the survey design) was used to determine associations between risk factors and KSHV antibody levels. Although KSHV antibody levels did not attain normal frequency distributions, they were log 10 transformed prior to linear regression modelling. For assessing the effect of intensive versus standard treatment, the analysis was done at the cluster level. The proportion of KSHV seropositive participants was calculated for each village, and the mean of these taken for the two trial arms. The risk ratio (RR) was then calculated by dividing the mean KSHV prevalence in the intensive arm by that in the standard arm, and a Taylor approximation was used to calculate a 95% confidence interval for this RR. The p-value was generated from a t-test comparing the village-level prevalences between the two arms. A similar approach was used to assess the effect of intensive versus standard treatment on KSHV antibody levels.
A p-value of less than 5% was considered statistically significant. Multivariable models included age (grouped), sex, HIV status, S. mansoni, hookworm and malaria parasite infection. Participants not tested for HIV were also included in the analyses.

Participants characteristics
Baseline and final survey results were analysed separately. The median age at baseline was 25 with an interquartile range (IQR) of 3 to 33 years. In the final survey, the median age was 24 years with an interquartile range (IQR) of 9 to 33 years. The overall proportion of males was 56% at baseline and 48% in the final survey. Details of the socio-demographic characteristics of the study population are shown in Table 1. At baseline, the HIV prevalence was 13% overall and 17% in those aged 13 years and above. In the final survey, around a quarter of participants (26%), mainly children, were not tested for HIV. There were 201 HIV seropositive individuals among 1229 tested (17% prevalence), with 103 participants confirmed to be on antiretroviral treatment (ART). At baseline, the malaria prevalence was 7% overall and 14% in children below 12 years. In the final survey, malaria infection prevalence was lower, at 4% overall, and 8% in children aged 1 to 12 years. Among the helminth infections tested, Schistosoma mansoni was the most prevalent in both surveys, as expected, due to the close proximity of the study sites to the waters of Lake Victoria. The prevalence was 86% by CCA, 50% by PCR and 34% by microscopy in the final survey (Table 1). At baseline, the prevalence was 72% by CCA in a subset of 569 participants and 52% by microscopy in 1137 participants. Hookworm prevalence was 7% by microscopy and 26% by PCR at baseline while in the final survey, it was 2% by microscopy and 8% by PCR. The prevalence of other helminths at baseline was 14% for Strongyloides stercoralis (using PCR), 11% for Trichuris trichiura (using KK), 0.1% for Ascaris lumbricoides (using KK) and 3% for Mansonella perstans (Table 1). While in the final survey, the prevalence of other helminths was 6% for Strongyloides stercoralis (using PCR), 9% for Trichuris trichiura (using KK), 0.04% for Ascaris lumbricoides (using KK) and 0.9% for Mansonella perstans (Table 1). At baseline, 20% of the S. monsoni infected individuals had light intensity infections, 15% moderate and 16% heavy infections. On the other hand, the majority of the infected individuals in the final survey had light to moderate helminth infections; 8% had a heavy S. mansoni infection based on microscopy (Table 1).
In the final survey, the prevalence increased steeply with age (overall p<0.0001), rising from 56% in the 1-12-year age group to 84% in the 13-30-year age group, and plateaued thereafter (Fig 2).

Associations between KSHV seropositivity and risk factors
We investigated associations of KSHV seropositivity with parasite infections and other factors at baseline and in the final survey. Overall, KSHV prevalence was higher in males compared to females (adjusted Odds Ratio (aOR) = 1.72 (1.29, 2.30), p = 0.001) in the final survey (Table 2). HIV seropositive individuals on ART were less likely to be KSHV seropositive compared to HIV seronegative individuals in the final survey (Table 2). Individuals infected with malaria parasites (aOR = 3.49 (1.08, 11.28), p = 0.038) were more likely to be KSHV seropositive in the final survey (Table 2). Although in the final survey, hookworm infection was positively associated with KSHV seropositivity in the unadjusted analysis (OR = 2.15 (1.18, 3.94), p = 0.015), this association was lost after adjusting for age group, sex, HIV serostatus, S. mansoni infection status and malaria infection status ( Table 2). Helminth infections including Trichuris trichiura and Strongyloides stercoralis showed no association with KSHV seropositivity both at baseline and in the final survey (Table 3 & Table 2). Other helminth infections such as Ascaris lumbricoides and Mansonella perstans were not analysed using regression modelling due to the small numbers of infected participants.
Individuals who were microscopy positive for S. mansoni at baseline (aOR = 1.86 (1.16, 2.99) p = 0.012 (Table 3) and in the final survey (aOR = 1.43 (1.04, 1.95), p = 0.028) ( were more likely to be KSHV seropositive. KSHV seropositivity increased with increasing S. mansoni infection intensity at baseline (p for trend = 0.013) ( Table 4). In the final survey, the seropositivity of KSHV among individuals heavily infected with S. mansoni was 82% compared to 80% among lightly or moderately infected individuals, and 73% among uninfected individuals, although evidence for an increasing prevalence with increasing intensity was borderline after adjusting for possible confounders (P value for trend = 0.068) ( Table 5). We did not observe significant associations between S. mansoni detected by CCA or PCR and KSHV seropositivity (S1 Text and S2 Text). There was no effect of intensive versus standard anthelminthic treatment on either KSHV seropositivity or antibody levels ( Table 6).

Association between KSHV antibody levels and microscopy status of S. mansoni infection
We further investigated the association between S. mansoni infection and infection intensity detected by microscopy and KSHV IgG antibody levels to K8. 1 and ORF73 separately. At baseline, microscopy positive individuals for S. mansoni had higher antibodies to K8.1 compared to microscopy negative individuals (adjusted linear regression coefficient = 0.03 (0.01, 0.05) p = 0.02). Furthermore, this association increased with increasing infection intensity (p value for trend = 0.005; S3 Text). Although antibody levels to ORF73 were not associated with S. mansoni infection at baseline, there was an association between ORF73 antibodies and S. mansoni infection intensity (p-value for trend = 0.026). Antibody levels to both K8.1 and ORF73 were neither associated with S. mansoni infection nor with S. mansoni infection intensity in the final survey (S3 Text).

Associations between KSHV seropositivity and Schistosoma mansoni antibody concentrations
IgE (n = 364), IgG (n = 372) and IgG4 (n = 370) antibody concentrations against S. mansoni egg and adult worm antigens were measured from the final survey in a subset of individuals with sufficient plasma for the analysis. Participants whose samples were used for this analysis were on average older than participants whose samples were not used; other participant characteristics were comparable (S4 Text). After adjusting for age group, sex and HIV status, increased levels of IgE to SWA (aOR = 55.03 (3.14, 963.65), p = 0.008) and SEA (aOR = 8.20 (1.53, 44.05), p = 0.016) as well as IgG to SEA (aOR = 2.57 (1.17, 5.68), p = 0.02) were associated with an increased risk of being KSHV seropositive (Table 7).

Discussion
KSHV prevalence can vary, even between geographically proximate areas [3]. We have previously reported a high KSHV seroprevalence of >95% in adults in the General Population Cohort in rural southwestern Uganda [5], and a prevalence of 61% amongst mothers in a periurban cohort [37]. This current study shows a high seropositivity of KSHV (>80% in 13 +-year-olds) amongst Lake Victoria island communities with seropositive participants as young as one year. Additionally, we show that males were more likely to be KSHV seropositive compared to females. These findings are similar to those documented in other studies carried out in sub-Saharan Africa [4,5,38]. The HIV prevalence in the studied communities was high (17%). We observed a lower risk of being KSHV seropositive among individuals treated for HIV compared to HIV negative individuals. Since our study was cross-sectional and given the fact that there was missing HIV data, this finding should be treated with caution. However, ART has been shown to lead to tumour regression among AIDS-KS patients [39][40][41]. Additionally, others have shown a decline in KSHV viral load following HAART initiation [42][43][44], perhaps suggesting a direct effect of ART on KSHV replication. The high untreated HIV prevalence, coupled with other factors including parasite infections, may contribute to the high prevalence of KSHV in this area. The burden of S. mansoni in these island communities is very high. We showed that being infected with S. mansoni, detected by microscopy, was associated with an increased risk of being KSHV seropositive. Others have reported no association between S. mansoni and KSHV infections, possibly due to the low prevalence of KSHV or low infection intensity of S. mansoni in the study areas [45,46]. In vitro reactivation of the model gammaherpesvirus MHV68 by S. mansoni was demonstrated by Reese et al., mediated through IL4 production [18]. Our human data are consistent with this model, as we observed an association with S. mansoni and KSHV seropositivity. Furthermore, at baseline (before antihelminthic treatment), anti-K8.1 antibody levels were higher in S. mansoni infected individuals compared to uninfected individuals, and S. mansoni infection intensity was associated with higher antibody levels to K8.1 and ORF73. These associations were not observed in the final survey after antihelminthic treatment. Nonetheless, higher antibody levels to KSHV antigens, and particularly the lytic antigen K8.1, has been associated with viral reactivation and KS disease progression [47][48][49][50]. Our findings may imply KSHV reactivation in individuals with a heavy S. mansoni infection, although a major limitation in interpretation is that we are unable to assign causation. An additional limitation is that we did not measure KSHV viral load in blood due to the unavailability of appropriate samples, and this would reaffirm the role of S. mansoni infection in KSHV reactivation. We also showed that increasing IgE and IgG, but not IgG4 antibodies to S. mansoni are associated with an increased risk of being KSHV seropositive in older individuals. The (Th) 2 immune response to S. mansoni, characterised by IgE, IL4 and IL5 production has been linked to protection from S mansoni reinfection, while IgG4 production has been linked to susceptibility to reinfection [51]. This protective immunity tends to increase with age and requires repeated exposure to develop. This would then suggest that S. mansoni, through (Th) 2 immune response upregulation, may reactivate KSHV latently infected cells.
Alternatively, the association between KSHV seropositivity and S. mansoni infection could be caused by increased inflammation due to new S. mansoni infections, leading to an increase in KSHV antibody levels in individuals already infected with KSHV. However, the association with anti-K8.1 (a lytic antigen) but not anti-ORF73 (a latent antigen) antibodies with S. mansoni infection at baseline may imply specific effects of S. mansoni on KSHV reactivation as opposed to non-specific inflammatory effects. Furthermore, higher anti-Ag85A (a Mycobacteria tuberculosis antigen unrelated to S. mansoni) IgG4, but not IgG, has been reported in S. mansoni infected individuals compared to S. mansoni uninfected individuals [14]. Taken   Table 6. Effect of helminths treatment on KSHV seropositivity and antibody levels. together, this might suggest a specific effect of S. mansoni on immunity to other unrelated chronic infections. The association between S. mansoni and KSHV seropositivity was only observed if S. mansoni was detected using microscopy (KK) but not PCR or CCA, in this study. Microscopic examination of faecal samples is the WHO recommended diagnostic test for S. mansoni [52,53], and has a documented specificity of 100%, although varying sensitivity (depending on infection intensity and studied population) has been observed [53,54]. Detection of S. mansoni by CCA may be influenced by cross-reactivity, leading to false positives, and may explain the larger numbers diagnosed by this method in our study [53,54]. The PCR technique has been shown to be sensitive and specific for detecting S. mansoni [52][53][54]. The ability of the PCR technique to detect low-intensity S. mansoni infection, as well as the reduced infection intensity in the final survey due to the three years antihelminthic treatment, may have obscured the effect that S. mansoni has on KSHV antibody levels. Heavy S. mansoni infections might have a larger effect which could easily be observed with the current sample size. Furthermore, a light infection intensity could have effects on KSHV incident cases, as opposed to prevalent cases. Because of the observational study design, incident cases were not specifically detected, and our study likely picked up prevalent cases.
We did not see any effect of intensive versus standard anthelminthic treatment on KSHV seropositivity. This was not surprisingly as such an intervention would likely affect incident KSHV infections, rather than prevalent KSHV infections and antibody titres against KSHV would be unlikely to change over the short follow-up period. Migration rates in these Island communities were very high. Therefore very few people were tested both at baseline and in the final survey. Consequently, the majority of the participants analysed in the final survey might have been new immigrants. The original trial intended to obtain community/population-wide effects as opposed to individual participant effects.
We also found malaria parasitemia to be associated with KSHV seropositivity, consistent with our previous findings [4,21], and these data reinforce the need to investigate the mechanism through which malaria infection impacts on KSHV and its subsequent role in KSHV transmission. Infection with Plasmodium falciparum, the main cause of malaria disease in Africa, induces inflammation which is normally regulated through induction of regulatory T cells and production of IL10 and TGF-β. Increased IL10 levels, a cytokine mainly produced by regulatory cells, has been reported in disseminated KS [55]. Plasmodia have also been shown to cause macrophage and dendritic cell dysfunction [56]. The immunosuppression caused by P. falciparum infection has also been shown to lead to the reactivation of some herpesviruses such as EBV, HSV-1 and VZV [57][58][59][60][61]. We therefore hypothesize that the immune dysregulation caused by malaria infection contributes to frequent reactivation of KSHV from latency. Since KSHV is transmitted by salivary exchange [3,62,63], studies examining parasite coinfections and KSHV viral load in saliva are warranted.