Access the full text.
Sign up today, get DeepDyve free for 14 days.
Downloaded from https://academic.oup.com/jid/article/218/5/716/4975541 by DeepDyve user on 19 July 2022 The First Norovirus Longitudinal Seroepidemiological Study From Sub-Saharan Africa Reveals High Seroprevalence of Diverse Genotypes Associated With Host Susceptibility Factors Lucy Thorne, Angela Nalwoga, Alexander J Mentzer, Alexis de Rougemont, Myra Hosmillo, Emily Webb, Margaret Nampiija, Allan Muhwezi, Tommy Carstensen, Deepti Gurdasani, Adrian V Hill, Manj S Sandhu, Alison Elliott, Ian Goodfellow Downloaded from https://academic.oup.com/jid/article/218/5/716/4975541 by DeepDyve user on 19 July 2022 The Journal of Infectious Diseases MAJOR ARTICLE e Fir Th st Norovirus Longitudinal Seroepidemiological Study From Sub-Saharan Africa Reveals High Seroprevalence of Diverse Genotypes Associated With Host Susceptibility Factors 1,2,a 3,a 4 5,6 1 7 3 Lucy Thorne, Angela Nalwoga, Alexander J. Mentzer, Alexis de Rougemont, Myra Hosmillo, Emily Webb, Margaret Nampiija, 3 8 8 4,9 8 3,10 1,11 Allan Muhwezi, Tommy Carstensen, Deepti Gurdasani, Adrian V. Hill, Manj S. Sandhu, Alison Elliott, and Ian Goodfellow 1 2 Division of Virology, Department of Pathology, University of Cambridge, UK; Division of Infection and Immunity, University College London, Medical Research Council Centre 3 4 for Medical Molecular Virology, UK; Medical Research Council/Uganda Virus Research Institute, Uganda Research Unit, Entebbe, Uganda; Wellcome Trust Centre for Human Genetics, University of Oxford, UK; Centre National de Référence des Virus Entériques, Laboratoire de Virologie-Sérologie, Centre Hospitalier Universitaire de Dijon, France; 6 7 L'Unité Mixte de Recherche Procédés Alimentaires et Microbiologiques, Université de Bourgogne Franche-Comté/AgroSup Dijon, France; Department of Infectious Disease 8 9 Epidemiology, London School of Hygiene and Tropical Medicine, UK; Wellcome Trust Sanger Institute, Hinxton, Cambridge, UK; The Jenner Institute, University of Oxford, UK; 10 11 Department of Clinical Research, London School of Hygiene and Tropical Medicine, UK; and School of Public Health, University of Makeni, Sierra Leone (See the Editorial Commentary by Mans, on pages 676–8.) Background. Human noroviruses (HuNoVs) are a prominent cause of gastroenteritis, yet fundamental questions remain regard- ing epidemiology, diversity, and immunity in sub-Saharan African children. We investigated HuNoV seroprevalence and genetic and sociodemographic risk factors in Ugandan children. September Methods. We randomly screened 797 participants of a longitudinal birth cohort (Entebbe, EMaBS) and 378 from a cross-sec- tional survey (rural Lake Victoria, LaVIISWA), for antibodies against HuNoV genotypes by ELISA. We used linear regression mod- eling to test for associations between HuNoV antibody levels and sociodemographic factors, and with the human susceptibility rs601338 FUT2 secretor SNP and histo-blood group antigens (A/B/O). Results. Of EMaBS participants, 76.6% were seropositive by age 1, rising to 94.5% by age 2 years. Seroprevalence in 1 year olds of the rural LaVIISWA survey was even higher (95%). In the birth cohort, 99% of seropositive 2 year olds had responses to multiple −52 HuNoV genotypes. We identified associations between secretor status and genogroup GII antibody levels (GII.4 P = 3.1 × 10 ), as −12 well as ABO and GI (GI.2 P = 2.1 × 10 ). Conclusions. HuNoVs are highly prevalent in Ugandan children, indicating a substantial burden of diarrhea-associated mor- bidity with recurrent infections. Public health interventions, including vaccination, and increased surveillance are urgently needed. Keywords. norovirus; seroepidemiology; Uganda; histo-blood group antigens. OA-CC-BY Gastroenteritis remains the second leading cause of infection-re- responsible for the majority of outbreaks since 1996, with new lated deaths in children under 5 years old globally; sub-Saha- pandemic variants emerging every few years, although greater ran Africa bears the greatest burden . Human noroviruses diversity has been reported in some low-income settings . (HuNoVs) are a leading cause of viral gastroenteritis across all In low- and middle-income countries, HuNoVs are respon- age groups; however, information regarding their impact and sible for >200 000 deaths/year in children <5 years old , epidemiology in African children is extremely limited [2, 3]. although this is likely to be an underestimate given the lack of The Norovirus genus of positive-strand RNA viruses is divided surveillance in most low-resource countries. Recent studies, into 7 proposed genogroups (GI–GVII), each subdivided into including 2 pioneering multicenter surveys of childhood diar- genotypes, on the basis of the major capsid protein and poly- rheal diseases, have begun to address the gaps in knowledge of merase sequences. Strains within genotype GII.4 have been pediatric HuNoV infections in low- and middle-income coun- tries [6–9]. The geographical distribution of data from these Received 1 February 2018; editorial decision 5 March 2018; accepted 17 April 2018; published and other studies highlights the paucity of information on online April 18, 2018. HuNoV prevalence in sub-Saharan Africa [8–10]. The major - L. T. and A. N. contributed equally to this study. Correspondence: L. Thorne, PhD, Infection and Immunity, UCL, Cruciform Building, Gower ity of studies of HuNoV in Africa have focused on incidence Street, London, WC1E 6BT, UK (firstname.lastname@example.org). reporting in diarrhea cases for small cohorts using reverse tran- The Journal of Infectious Diseases 2018;218:716–25 scription polymerase chain reaction (RT-PCR) based methods. © The Author(s) 2018. Published by Oxford University Press for the Infectious Diseases Society of America. This is an Open Access article distributed under the terms of the Creative Commons Serological data can complement incidence reporting to provide Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted an overall picture of infections in a particular population. The reuse, distribution, and reproduction in any medium, provided the original work is properly cited. DOI: 10.1093/infdis/jiy219 last study of HuNoV seroprevalence from an African country 716 • JID 2018:218 (1 September) • Thorne et al Downloaded from https://academic.oup.com/jid/article/218/5/716/4975541 by DeepDyve user on 19 July 2022 was published in 1999 ; since then the global prevalence of genotypes, some associated with pediatric infections [20, 21], HuNoV is thought to have increased due to the emergence of were selected to represent a diverse range that would detect pandemic strains belonging to the GII.4 genotype. responses to recent and older strains. As the genotype GII.4 dis- Susceptibility to HuNoV-induced diarrhea is at least partially plays the greatest antigenic variability and rates of evolution [22, governed by expression of histo-blood group antigens (HBGAs) 23], we selected a GII.4 strain that was identified prior to the in the gastrointestinal tract, which serve as viral attachment fac- study sampling dates (Hu/GII.4/FRA/Dijon/1996/Dijon171, tors . Intestinal HBGA expression is controlled by the α(1,2) AF472623). Non-GII.4 genotypes are more antigenically static fucosyltransferase 2 (FUT2) enzyme. Individuals homozygous [21, 23] and the following strains were selected: GI.1 (Hu/ for a FUT2 nonsense mutation (G428A) are termed “nonsecre- NV/I/Norwalk/68/US, GenBank accession no. AAA59229), tors” (Se−) and have been shown to be less susceptible to infec- GI.2 (Hu/GI.2/FRA/Dijon/2008/E2818, GenBank accession tion by certain genotypes of HuNoV, although the relationship no. KP064095), GII.3 (Hu/GII.3/ FRA/Dijon/2008/E2419, is complex . There has not yet been a large-scale extensive GenBank accession no. KP064097), GII.6 (Hu/GII.6/FRA/ investigation of the association between HBGA status and Dijon/2011/E5915, GenBank accession no. KP064098), and HuNoV infections in an African population. GII.12 (Hu/GII.12/FRA/Dijon/2010/E5152, GenBank acces- Recent breakthroughs in the development of long-sought sion no. KP064099). systems for HuNoV in vitro cultivation [14, 15], mean that ELISA for HuNoV IgG Levels targeted antivirals and a multivalent vaccine for HuNoVs are To determine IgG responses to a pool of HuNoV VLPs, enzyme- within reach, with a vaccine currently advancing in phase 2b linked immunosorbent assays (ELISAs) were performed as clinical trials . Understanding HuNoV prevalence and nat- previously reported  (Supplementary Methods). For the ural immunity in children in sub-Saharan Africa, where any EMaBS cohort, we began by screening plasma samples from future vaccine could have the greatest impact, is therefore essen- children aged 1 year; any that tested negative were then fol- tial. To address this, we investigated the seroprevalence, age of lowed up at each subsequent age group until a positive response seroconversion, genotype diversity, and genetic and social risk was detected. We estimated the cumulative seroprevalence for factors for HuNoVs among 2 cohorts of Ugandan children. each year group, making the assumption that the missing sam- METHODS ples seroconverted at the same rate as the rest of the age group (Supplementary Methods). To screen for individual genotype Ethical Approval responses, we randomly selected another 800 seropositive sam- This study was approved by the Research and Ethics Committee ples from 2-year-old children. Uganda Virus Research Institute, Uganda National Council for Science and Technology, and London School of Hygiene and Statistical Analysis Tropical Medicine Research and Ethics Committee. To estimate the cumulative seroprevalence in the EMaBS cohort (Supplementary Table 2) for ages 2 years and onwards; we made Study Populations the assumption that children whose samples were not available e En Th tebbe Mother and Baby Study (EMaBS) (ISRCTN32849447, at follow-up ages would have seroconverted at the same rate as http://emabs.lshtm.ac.uk/) is a birth cohort that originated as a the ones for whom samples were available in that age group. trial of anthelminthics . In total, 2507 pregnant women in their er Th efore we calculated the rate of seroconversion for each age third trimester were enrolled between 2003 and 2005. Blood sam- group (as number seropositive/number available to screen) and ples collected from the children every year aer b ft irth (between multiplied it by the total number that should have been screened 2004 and 2011) with informed parental consent. The study catch- for that age group (the number that had tested seronegative at ment area was comprised periurban, fishing, and rural commu- the previous age) to give the estimated positive number. This nities around Entebbe and Lake Victoria. From this cohort, 797 value was added to the estimated positive values of each previ- children were randomly selected for this study. The Lake Victoria ous year and expressed as a percentage out of the total number Island Intervention Study on Worms and Allergy-related dis- of children included in the study (n = 797). HuNoV seroprev- eases (LaVIISWA), included a baseline survey performed in alence and 95% confidence intervals for the LaVIISWA survey 2012–2013 among 26 island fishing communities in Koome sub- were obtained using “svy” commands in Stata. county, Mukono District, Uganda . All available samples from Linear regression modeling of risk factors was performed 1–5 year olds in the LaVIISWA baseline survey were studied. using Stata-13 software (Stata 13.0, Statacorp, College Station). Virus-Like Particle Preparation For both EMaBS and LaVIISWA, IgG levels (OD450 values) Virus-like particles (VLPs), which are structurally and antigen- to HuNoV were log10 transformed to become approximately ically indistinguishable from native virions, were generated for normally distributed. For EMaBS, linear regression with each strain as previously published . Strains from different bootstrapped confidence intervals was used to determine the Norovirus Seroepidemiology in Africa • JID 2018:218 (1 September) • 717 Downloaded from https://academic.oup.com/jid/article/218/5/716/4975541 by DeepDyve user on 19 July 2022 Genotype Data for Determining FUT2 Secretor Status and ABO Blood Type association between HuNoV antibodies and various risk fac- Whole-genome genotype data were available for 651/797 tors: sex, geographical location, mother’s occupation, house- selected children from EMaBS generated on the Illumina hold socioeconomic status (hhSES), human immunodeficiency Omni2.5M array and quality controlled using standard pipe- virus (HIV) exposure status, and diarrhea events. Mother’s lines, including removing individuals and variants with high occupation was categorized into 3 groups based on relatedness levels of missingness or deviations from expected levels of het- of occupation. Group 1 included farming/fishing, unskilled erozygosity or Hardy-Weinberg equilibrium. Untyped genetic manual labor, and bar/hotel attendants; group 2 included variants, including the FUT2 secretor status SNP (rs601338 at housewives; while group 3 included business, students, and pro- position 49 206 674 on chromosome 19 in genome build 37) and fessionals. A composite value was obtained for hhSES derived ABO SNPs (rs8176747 position 136 131 315 and rs8176719 from the home building materials, number of rooms, and items position 136 132 908 on chromosome 9) were imputed using collectively owned at the mother’s home, using principal com- a merged 1000 Genomes and African-specific reference panel ponent analysis. Geographical location was categorized into 3  using SHAPEIT2  and IMPUTE2  with default groups: urban (Entebbe), periurban (Kingungu and Manyago), settings applicable for African populations, and all imputed and rural (Katabi). The mother’s HIV status was used to group variants were filtered using an info score threshold of 0.3 and children as either exposed or unexposed, as the number of chil- a minor allele frequency of 0.01 or greater. Imputed dosages of dren that tested positive for HIV at 1 year was very low (7/773). secretor A alleles were used to determine secretor status. Se sta- For mother’s occupation and geographical location, the groups tus was confirmed by PCR amplification and Sanger sequenc- were analyzed as nominal variables aer t ft esting for departure ing of the rs601338 SNP at position 49 206 674 on chromosome from linear trend. In the crude analysis we reported the P value 19 for 80 randomly selected EMaBS participants. For ABO, we from the Wald test and in the adjusted analysis we reported the used phased genotypes to determine genetic blood groups using P value from the likelihood ratio test. Supplementary Table 1 (adapted from ). The corresponding For LaVIISWA, linear regression analyses employed the “svy” phenotypic blood groups used for downstream analyses were survey commands in Stata to allow for clustering of respondents determined using these genetic blood groups. within villages using linearized standard errors and for variable village sizes using weights. Exposures considered were sex, age, RESULTS and child’s school status. Because the outcome variable (IgG antibody levels to HuNoV) was log transformed prior to lin- In participants of the EMaBS longitudinal birth cohort ear regression modeling, geometric mean ratios were derived (Figure 1A) seroprevalence to the pool of HuNoV was 76.6% by back-transforming regression coefficients and their corre- (611/797) by age 1 year, and 94.5%, 95.2%, 96.1%, and 96.6% at sponding 95% confidence intervals (CI). During univariate 2, 3, 4, and 5 years, respectively (Figure 1B and Supplementary analysis, each risk factor was analyzed separately with the out- Table 2). Of the sociodemographic variables considered, resi- come variable for each study. A linear trend assumption was dence in a rural location (within the EMaBS catchment area) made and tested using the likelihood ratio test when analyzing was the only factor significantly associated with having higher age and hhSES, mother’s occupation, and geographical location. antibody levels at age 1 (P = .006; Table 1). During multivariate analysis all risk factors were analyzed with To corroborate the high seroprevalence in the EMaBS cohort the outcome variable using 1 model for each study. and investigate whether it differed in more rural geographical Associations between secretor status and ABO phenotypic locations, we used a cross-sectional survey (LaVIISWA) con- blood group were assessed using the Student t test implemented ducted in remote island-shore fishing villages in Lake Victoria, in R. Differences in log-transformed distributions of geno- only accessible from the mainland in 3 hours by powered canoe type-specific antibody levels were tested between 2 groups and (Figure 1A) . Seropositivity to the pool of HuNoVs was the final P value reported for each comparison. To test for evi- higher in 1 year olds in the LaVIISWA survey (94.7%) than in dence of interaction between ABO and FUT2 loci, the likeli- the EMaBS cohort, but was comparable from age 2 onwards hood of an interaction model was compared to the likelihood of (Figure 1B and Supplementary Table 3). Age was the only factor a multivariable model (with no interaction) including secretor significantly associated with increased antibody levels in this status (Se+ or Se−) and blood group (O or B), using the likeli- survey (P < .0001; Table 2). For every 1-year increase in age, hood ratio test. Sensitivity analyses were performed to ensure we observed an approximately 5% increase in OD450 readings, that the statistics generated from the genetic association analy- suggesting an increase in HuNoV antibody levels that is most ses were not false positives originating from population stratifi- likely due to repeat infections (Table 2). cation (Supplementary Methods). Associations between genetic To gain insight into the diversity of HuNoVs circulating in variants and diarrhea incidence were assessed using Cox regres- Ugandan children, we determined specific genotype seropreva- sion with robust standard errors to allow for the fact that some lence in 2 year olds from the EMaBS cohort. The seroprevalence children experienced multiple diarrhea episodes. of the GII.4 virus was comparable to the genotype GI.1 virus 718 • JID 2018:218 (1 September) • Thorne et al Downloaded from https://academic.oup.com/jid/article/218/5/716/4975541 by DeepDyve user on 19 July 2022 Kajjansi Uganda Kisubi Mpunge Kampala Kenya Lyamutundwe Nakuru Olenguruone Entebbe Nairobi Lwaji Island Damba Island Rwanda Arusha Burundi Mombasa Kome island Dar es Salaam Tanzania EMaBS LaVIISWA (Cumulative) (Cross-Sectional) 12 34 5 12 345 Age (years) Figure 1. Human norovirus (HuNoV) seroprevalence in Ugandan children between 1 and 5 years old. A, Study catchment areas for the longitudinal birth cohort around Entebbe (Entebbe Mother and Baby Study [EMaBS], shown in red) and a cross-sectional study in the islands of Lake Victoria (Lake Victoria Island Intervention Study on Worms and Allergy-related diseases [LaVIISWA] , shown in blue). B, Cumulative (EMaBS) and cross-sectional (LaVIISWA) seroprevalence of a pool of HuNoVs in children between 1 and 5 years old. Error bars indicated 95% confidence limits. at 92.1% and 92.6% respectively (Figure 2A), higher than for mutation), in line with the frequency previously reported for all other genotypes (Figure 3). All combinations of genotype populations with African ancestry . PCR amplification and responses were observed for different individuals (Figure 3A). Sanger sequencing confirmed the accuracy of secretor status Overall, the genotype GII.12 virus had the lowest seropreva- imputation in 80 individuals selected at random. Se− individ- lence at 69.4% (Figure 2A). In total, 99.1% of seropositive 2 year uals exhibited significantly lower antibody levels against all olds displayed multiple genotype-specific serological responses. genotypes except GI.1 (Figure 2B), and the effect was most As susceptibility to HuNoV-induced diarrhea is at least par- prominent for GII genotypes, especially GII.4 (geometric mean −52 tially governed by FUT2 secretor (Se) status and gastrointesti- ratio 1.54 (95% CI, 1.49–1.59); P = 3.12 × 10 ). In contrast, nal HBGA expression , we investigated whether there were an effect was only observed at the ABO locus for GI genotypes, −12 associations between levels of genotype-specific antibody and GI.2 in particular (P = 2.1 × 10 ). Given the strong omnibus Se− or ABO type in the EMaBS population. Of the 651 EMaBS concurrent associations at the ABO and FUT2 loci for GI.2 participants for whom whole-genome genotype data were avail- and the known mechanisms of HBGA expression involving able, 23.0% (150/651) were Se− (homozygous for the G428A FUT2, we expected to see evidence of epistasis between these Norovirus Seroepidemiology in Africa • JID 2018:218 (1 September) • 719 Seroprevalence (%) Downloaded from https://academic.oup.com/jid/article/218/5/716/4975541 by DeepDyve user on 19 July 2022 Table 1. Association Between Human Norovirus (HuNoV) Antibody DISCUSSION Levels (OD450 Values) and Sociodemographic Factors Among 797 1-Year- Our findings of high seroprevalence in 1-year-old children in Old Children From the Entebbe Mother and Baby Study Cohort the 2 cohorts of Ugandan children exceeds that reported for the Crude GMRs and P Values Adjusted GMRs and P Values same age group from several European countries (for exam- a a ple, 47.3% in Finland, 42.7% in The Netherlands), but becomes Factor GMR P Value GMR P Value comparable by 5 years old [30–32], suggesting greater risk of Sex infection during infancy in Uganda. These findings are consist- Male Female 0.99 (0.94–1.04) .67 0.99 (0.94–1.05) .78 ent with the high incidence reported by RT-PCR–based fecal Location detection of HuNoVs in community birth cohort studies from Urban India, Ecuador, Peru, and the multicentre MAL-ED study [6, Periurban 1.06 (1.00–1.13) .04 1.08 (1.01–1.14) 20, 33, 34]. Such high infection rates may not be captured by Rural 1.09 (1.01–1.17) 1.10 (1.02–1.19) .02 incidence reporting in health care settings . Taken as a whole, Mother’s occupation these community-based studies paint a picture of widespread Group 1 Group2 1.00 (0.93–1.07) .22 1.01 (0.94–1.09) .19 HuNoV infections occurring early in life in low- and mid- Group 3 0.94 (0.86–1.03) 0.95 (0.87–1.04) dle-income countries, where children are particularly vulner- Household SES able to diarrheal diseases. This suggests that implementation of any future vaccination schedules would need to begin early in life, such as through a birth dose and/or early infant schedule, as per the World Health Organization Expanded Program on Immunization schedule that already exists for routine vacci- 6 1.00 (0.98–1.02) .99 (trend) 1.00 (0.98–1.02) .99 (trend) nations in low-income settings. An effective HuNoV vaccine HIV would also need to confer broad-range protection across geno- Exposed types, as is evident from the high percentage of individuals with Unexposed 0.94 (0.85–1.03) .14 0.93 (0.85–1.02) .13 responses to multiple genotypes by age 2 years. Diarrhea events 0 e s Th eroprevalence was higher in 1-year-old children in the 1–2 1.05 (0.99–1.12) 0.12 1.05 (0.99–1.12) LaVIISWA survey of remote island-shore fishing villages than >2 1.00 (0.92–1.07) 1.00 (0.92–1.07) .14 in the EMaBS cohort. It is possible that this reflects the nature All factors were adjusted for each other. Results obtained using linear regression of log10 of cross-sectional sampling in LaVIISWA, where those classi- HuNoV antibody levels and back-transforming to obtain GMRs. fied as 1 year olds could be closer to age 2, in comparison to the Abbreviation: GMR, geometric mean ratio. Values in parentheses indicate bootstrapped 95% confidence intervals. Urban is Entebbe; periurban is Kigungu and Manyago; and rural is Katabi. Group 1 (farmer/fishing, unskilled manual, and bar/hotel); Group 2 (housewife); and Table 2. Association Between Human Norovirus Antibody Levels and Group 3 (business, student, professional). Sociodemographic Factors Among 375 Children From the Lake Victoria SES, household socioeconomic status, 1– 6 (low–high). The GMR for household SES is for Island Intervention Study on Worms and Allergy-Related Diseases Survey each unit increase in SES. Crude GMRs and P Values Adjusted GMRs and P Values a a 2 loci. We found that Se− individuals had similar distributions Factor GMR P Value GMR P Value of GI.2 specific antibodies irrespective of ABO status and the Sex protective B type effect was only observed in Se+ individuals Male −4 Female 1.05 (0.98–1.14) .17 1.08 (0.99–1.16) .08 (P = 5.3 × 10 ) (Figure 4), as reported in outbreak stud- interaction Age ies cohorts . Given the high seroprevalence of HuNoV in the Ugandan population, we used the clinical data available within EMaBS to determine the impact of FUT2 and ABO variation on episodes of diarrhea during early childhood. We identified a significantly 5 1.05 (1.03–1.07) <.0001 (trend) 1.05 (1.03–1.08) <.0001 (trend) Child’s school status lower rate of doctor-diagnosed diarrhea in Se− participants Student <1 year old (hazard ratio 0.89; 95% CI, 0.79–1.00); however, the Stay at 1.09 (1.03–1.19) .04 1.00 (0.92–1.08) .93 difference was not sustained to age 5 years. There was statistical home evidence for a difference in the association between the 2 age All factors were adjusted for each other. Results obtained using linear regression analyses using the “svy” survey commands in Stata to allow for clustering of respondents within groups (P = .008; Table 3). We observed no effect on episodes villages using linearized standard errors and for variable village sizes using weights. of diarrhea at the ABO locus (Table 3), and the P value for the Abbreviation: GMR, geometric mean ratio. interaction between age and genotype was 0.77. Values in parentheses indicate bootstrapped 95% confidence intervals. 720 • JID 2018:218 (1 September) • Thorne et al Downloaded from https://academic.oup.com/jid/article/218/5/716/4975541 by DeepDyve user on 19 July 2022 GI.1 GI.2 GII.3 GII.4 GII.6 GII.12 100% 75% 50% 25% 0% GI.2 GII.3 GII.4 GII.6 GII.12 GI.1 1.5 1.0 0.5 0.0 –0.5 NS *** *** *** *** *** –1.0 Se– Se+ Se– Se+ Se– Se+ Se– Se+ Se– Se+ Se– Se+ Secretor status GI.1 GI.2 GII.3 GII.4 GII.6 GII.12 1.5 1.0 0.5 0.0 ** *** NS NS NS NS –0.5 OA AB BO AAB BO AABB O AABB OA AB BO AABB Blood group Figure 2. Human genotype analysis of the relationship between human norovirus (HuNoV) antibody levels and α(1,2) fucosyltransferase 2 (FUT2) secretor status and or histo-blood group antigens (HBGA) type. A, Seroprevalence of different HuNoV genotypes in 2-year-old participants in the Entebbe Mother and Baby Study. B, Association between FUT2 secretor status and genotype-specific antibody levels. C, Association between HBGA type and genotype-specific antibody levels. ** P < .01; *** P < .001; NS, nonsignificant. in urban areas, suggesting they have a greater chance of recur- EMaBS where the samples are taken close to each child’s birth- rent infection. day. However the increased seroprevalence was also expected In estimating the cumulative seroprevalence for the EMaBS given the lower standards of living in the island communities, longitudinal birth cohort, we had to make a number of assump- including reduced access to proper sanitation and running tions, whilst minimizing the potential of these to lead to over- water . This is in line with our finding that 1-year-old chil- estimation. Firstly, we assumed that samples were missing from dren in rural regions of the EMaBS catchment area had signifi- cantly higher antibody levels to the pool of HuNoVs than those the database at random and that the rate of seroconversion for Norovirus Seroepidemiology in Africa • JID 2018:218 (1 September) • 721 Log (genotype-speciﬁc Log (genotype-speciﬁc 10 10 Seroprevalence antibody) antibody) Downloaded from https://academic.oup.com/jid/article/218/5/716/4975541 by DeepDyve user on 19 July 2022 Relative OD450 value vs. positive threshold <1 >1–5 5–10 >10 GI.1 GI.2 GII.3 GII.4 GII.6 GII.12 40 **** **** **** **** **** GI.1 GI.2 GII.3 GII.4 GII.6 GII.12 Figure 3. Seropositivity to different human norovirus (HuNoV) genotypes in the 2-year-old age group of a longitudinal clinical cohort of Ugandan children (Entebbe Mother and Baby Study). A, The results are represented as a heat map, where each column represents a single sample and are ordered based on GI.1 responses from high to low. Colors indicate the relative OD450 value for each sample against each genotype. B, Comparison of the relative OD450 values for each genotype. Boxplot was generated with the Tukey method. Statistically significant differences were analyzed by 1-way ANOVA, using a Dunn multiple comparison test to compare all responses to the highest, GII.4. **** P < .0001. these would be the same as for those which were available in fecal or salivary IgA responses in any future serological studies that age group. This would not hold true if the children had will be essential to further understand the duration of protec- different genetic or social risk factors; however, there were no tive immune responses. differences in the recorded characteristics for individuals whose Surprisingly, considering GII.4 global dominance , the samples were present or missing. Secondly, we assumed that seroprevalence of the GII.4 virus was comparable to the gen- once a child had seroconverted they would remain seropositive. otype GI.1 Norwalk virus. GII.4 antibody levels were, however, This was based on the results of a pilot screen in which we tested significantly higher than for all other genotypes (Figure 3), sug- consecutive yearly samples from EMaBS participants and found gesting either repeat GII.4 infections that boost intragenotype that antibody levels to the pool of HuNoVs remained high each responses, or that the GII.4 virus is more immunogenic, which year aer s ft eroconversion (Supplementary Figure 1). In support could be linked to prolonged infections reported for some GII.4 of this, we found that higher antibody levels were significantly viruses compared to GI viruses . Se− individuals have been associated with increasing age in the LaVIISWA cross-sectional shown to be resistant to infection by GI.1 Norwalk virus , survey, indicative of repeat infections that may boost antibody yet the percentage seropositive for GI.1 exceeded the percent- titers. Additional studies are required to determine whether the age of Se− individuals in the EMaBS cohort and Se− individu- IgG responses measured here are protective and capable of neu- als did not have significantly lower antibody levels to Norwalk tralizing infection, which could now be tested using the recently virus. As sensitivity to FUT2 status and HBGA expression can developed cell culture systems for HuNoVs [14, 15]. Mucosal vary between strains , it is possible that we are measuring IgA has also recently been strongly correlated with protection intragenotype responses to a related GI.1 virus, which may be in human volunteer challenge studies  and so inclusion of able to infect Se− individuals. Previous volunteer studies have 722 • JID 2018:218 (1 September) • Thorne et al Relative OD450 value Downloaded from https://academic.oup.com/jid/article/218/5/716/4975541 by DeepDyve user on 19 July 2022 1.5 NS *** 1.0 0.5 0.0 –4 P = 1.0 ×10 interaction Se– / O Se– / A Se– / AB Se– / B Se+ / O Se+ / A Se+ / AB Se+ / B Secretor status and blood group Figure 4. Relationship between α(1,2) fucosyltransferase 2 (FUT2) secretor status, histo-blood group antigens (HBGA) type, and GI.2-specific antibodies. Significance values are provided for tests of differences between distributions in individuals of O or B HGBA type, stratified by secretor status. The final significance value testing for interaction between secretor status and O/B HGBA type is also shown. Distributions of A and AB HGBA type individuals are provided for comparison although not included in the final interaction model. *** P < .001; NS, nonsignificant. also reported a detectable IgG response against Norwalk virus e fin Th ding that nearly all seropositive 2 year olds display in 78% of Se− individuals, who were otherwise protected from multiple genotype-specific serological responses, indicates the symptomatic infection . Although GI viruses are thought to occurrence of repeat infections and the circulation of diverse circulate at low levels worldwide , the seroprevalence of GI HuNoV genotypes, as reported elsewhere [8, 23]. Importantly, viruses in some African countries is higher than the global aver- the possibility of cross-reactive antibody responses, particularly age , in line with our serology data. Multiple GI strains have between genotypes of the same genogroup, such as GII.4 and been detected in surface water and sewage in different African GII.6, cannot be excluded. However, we observed all combina- countries . Taken altogether, this suggests that GI viruses are tions of genotype responses, strongly supporting assay specific- endemic in Uganda and unsampled in the community, and that ity, and limited cross-reactivity has been reported for children local circulation patterns exist, which may not reflect the global <5 years old . Furthermore, the genotypes used here have circulation of all strains. all recently been proposed to fall into different “immunotypes,” Table 3. Association Between (1,2) Fucosyltransferase 2 (FUT2) Secretor Status and ABO Phenotypic Blood Group and Diarrhea in 0 to 5-Year-Old Children Age Group Genotype Person-Years (× 100) Number of Diarrhea Events Rate per 100 person-years Hazard Ratio (95% CI) P value Secretor versus nonsecretor (rs601338) <1 year Secretor (AG or GG) 10.9 1657 152.6 1 Nonsecretor (AA) 3.1 412 135.1 0.89 (0.79–1.00) 1–5 years Secretor (AG or GG) 39.6 1877 47.4 1 .21 Nonsecretor (AA) 11.3 581 51.5 1.09 (0.95–1.25) ABO blood group (rs8176747, rs8176719) <1 year O 7.2 1083 151.3 1 .89 A 3.6 525 144.2 0.95 (0.84–1.08) AB 0.4 65 144.4 0.95 (0.73–1.24) B 2.7 396 148.9 0.98 (0.87–1.11) 1–5 years O 26.3 1289 49.1 1 .77 A 13.3 639 47.9 0.97 (0.86–1.10) AB 1.6 84 52.8 1.07 (0.76–1.49) B 9.7 446 46.0 0.93 (0.80–1.09) Norovirus Seroepidemiology in Africa • JID 2018:218 (1 September) • 723 Log (GI.2-speciﬁc antibody) 10 Downloaded from https://academic.oup.com/jid/article/218/5/716/4975541 by DeepDyve user on 19 July 2022 reducing the likelihood of cross-reactive responses . Several the editors consider relevant to the content of the manuscript African case-control studies have described relatively high detec- have been disclosed. tion rates of HuNoV secretion in healthy controls . It remains References unclear whether this is due to asymptomatic infection or recent resolution of symptomatic infection; however, these studies sug- 1. Liu L, Oza S, Hogan D, et al. Global, regional, and national gest that the appearance of symptoms may vary with genotype, causes of under-5 mortality in 2000-15: an updated sys- host susceptibility (potentially Se status), and acquired immu- tematic analysis with implications for the Sustainable nity. Further incidence studies are therefore required to confirm Development Goals. Lancet 2016; 388:3027–35. which circulating genotypes cause the most substantial disease 2. Desselberger U, Goodfellow I. Noroviruses: a global cause in Uganda and how this links to Se status in this population. of acute gastroenteritis. Lancet Infect Dis 2014; 14:664–5. In the context of numerous reports of incidence in health care 3. Lopman BA, Grassly NC. Editorial commentary: pediatric settings [8–10], this dataset provides essential insight into the norovirus in developing countries: a picture slowly comes high community prevalence of HuNoVs in children in sub-Sa- into focus. Clin Infect Dis 2016; 62:1218–20. haran Africa and suggests that HuNoV infections are ubiqui- 4. Yassin MA, Kirby A, Mengistu AA, et al. Unusual norovi- tous, recurrent, occur within the first year of life, and may be rus and rotavirus genotypes in Ethiopia. Paediatr Int Child a substantial cause of endemic childhood diarrhea in Uganda. Health 2012; 32:51–5. Other birth cohort studies have shown that multiple symp- 5. Patel MM, Widdowson MA, Glass RI, Akazawa K, Vinjé J, tomatic HuNoV infections at an early age are associated with Parashar UD. Systematic literature review of role of noro- growth stunting . This raises concerns as to whether the viruses in sporadic gastroenteritis. Emerg Infect Dis 2008; commonality of multiple HuNoV infections could be impact- 14:1224–31. ing the growth and development of young Ugandan children 6. Rouhani S, Peñataro Yori P, Paredes Olortegui M, et al. and further studies will also be important to address this. Our Norovirus infection and acquired immunity in 8 coun- data support the need for public health interventions coupled tries: results from the MAL-ED study. Clin Infect Dis 2016; with routine HuNoV surveillance and molecular diagnostics 62:1210–7. across Africa, the latter of which could be achieved by expan- 7. Kotloff KL, Nataro JP, Blackwelder WC, et al. Burden and sion of the African Rotavirus Surveillance Network to include aetiology of diarrhoeal disease in infants and young children HuNoVs. This is particularly important given the high preva- in developing countries (the Global Enteric Multicenter lence of unsampled community HuNoV infections suggests Study, GEMS): a prospective, case-control study. Lancet that emergence of novel variants in Africa is likely. 2013; 382:209–22. 8. Munjita SM. Current status of norovirus infections in chil- Supplementary Data dren in Sub-Saharan Africa. J Trop Med 2015; 2015:309648. Supplementary materials are available at e Th Journal of Infectious 9. Mans J, Armah GE, Steele AD, Taylor MB. Norovirus epide- Diseases online. Consisting of data provided by the authors to miology in Africa: a review. PLoS One 2016; 11:e0146280. benefit the reader, the posted materials are not copyedited and 10. Kabue JP, Meader E, Hunter PR, Potgieter N. Human noro- are the sole responsibility of the authors, so questions or com- virus prevalence in Africa: a review of studies from 1990 to ments should be addressed to the corresponding author. 2013. Trop Med Int Health 2016; 21:2–17. 11. Smit TK, Bos P, Peenze I, Jiang X, Estes MK, Steele AD. Notes Seroepidemiological study of genogroup I and II calicivirus Acknowledgments. We thank Edward Emmott for his tech- infections in South and southern Africa. J Med Virol 1999; nical assistance in preparing VLPs. 59:227–31. Financial support. This work was supported by a Sir Henry 12. Hutson AM, Atmar RL, Marcus DM, Estes MK. Norwalk Wellcome Fellowship to L. T., and senior fellowships to I. G. and virus-like particle hemagglutination by binding to h his- A. E.; a European Research Council Advanced Grant to A. H. to-blood group antigens. J Virol 2003; 77:405–15. (grant number 294557); and a Wellcome Trust Clinical Research 13. Huang P, Farkas T, Zhong W, et al. Norovirus and his- Training Fellowship to A. M. (grant number 106289). Entebbe to-blood group antigens: demonstration of a wide spectrum Mother and Baby Study and Lake Victoria Island Intervention of strain specificities and classification of two major bind- Study on Worms and Allergy-related diseases are funded by the ing groups among multiple binding patterns. J Virol 2005; Wellcome Trust (grant numbers 064693, 079110, 95778). 79:6714–22. Potential coni fl cts of interest. All authors: No reported con- 14. Jones MK, Watanabe M, Zhu S, et al. Enteric bacteria pro- flicts of interest. All authors have submitted the ICMJE Form mote human and mouse norovirus infection of B cells. for Disclosure of Potential Conflicts of Interest. Conflicts that Science 2014; 346:755–9. 724 • JID 2018:218 (1 September) • Thorne et al Downloaded from https://academic.oup.com/jid/article/218/5/716/4975541 by DeepDyve user on 19 July 2022 15. Ettayebi K, Crawford SE, Murakami K, et al. Replication of 28. Currier RL, Payne DC, Staat MA, et al. Innate susceptibil- human noroviruses in stem cell-derived human enteroids. ity to norovirus infections influenced by FUT2 genotype in Science 2016; 353:1387–93. a United States pediatric population. Clin Infect Dis 2015; 16. Baehner F, Bogaerts H, Goodwin R. Vaccines against nor- 60:1631–8. ovirus: state of the art trials in children and adults. Clin 29. Rockx BH, Vennema H, Hoebe CJ, Duizer E, Koopmans Microbiol Infect 2016; 22(Suppl 5):136–9. MP. Association of histo-blood group antigens and sus- 17. Ndibazza J, Mpairwe H, Webb EL, et al. Impact of anthel- ceptibility to norovirus infections. J Infect Dis 2005; minthic treatment in pregnancy and childhood on immuni- 191:749–54. sations, infections and eczema in childhood: a randomised 30. Nurminen K, Blazevic V, Huhti L, et al. Prevalence of nor- controlled trial. PLoS One 2012; 7:e50325. ovirus GII-4 antibodies in Finnish children. J Med Virol 18. Nampijja M, Webb EL, Kaweesa J, et al.; LaVIISWA trial 2011; 83:525–31. team. The Lake Victoria Island Intervention Study on Worms 31. Menon VK, George S, Aladin F, et al. Comparison of and Allergy-related diseases (LaVIISWA): study protocol for age-stratified seroprevalence of antibodies against norovi- a randomised controlled trial. Trials 2015; 16:187. rus GII in India and the United Kingdom. PLoS One 2013; 19. Caddy S, Emmott E, El-Attar L, et al. Serological evidence 8:e56239. for multiple strains of canine norovirus in the UK dog pop- 32. Son H, Jeong HS, Cho M, et al. Seroepidemiology of pre- ulation. PLoS One 2013; 8:e81596. dominant norovirus strains circulating in Korea by using 20. Saito M, Goel-Apaza S, Espetia S, et al.; Norovirus Working recombinant virus-like particle antigens. Foodborne Pathog Group in Peru. Multiple norovirus infections in a birth Dis 2013; 10:461–6. cohort in a Peruvian periurban community. Clin Infect Dis 33. Lopman BA, Trivedi T, Vicuña Y, et al. Norovirus infection 2014; 58:483–91. and disease in an ecuadorian birth cohort: association of 21. Sakon N, Yamazaki K, Nakata K, et al. Impact of geno- certain norovirus genotypes with host FUT2 secretor status. type-specific herd immunity on the circulatory dynamism J Infect Dis 2015; 211:1813–21. of norovirus: a 10-year longitudinal study of viral acute gas- 34. Menon VK, George S, Sarkar R, et al. Norovirus gastro- troenteritis. J Infect Dis 2015; 211:879–88. enteritis in a birth cohort in Southern India. PLoS One 22. Bull RA, Eden JS, Rawlinson WD, White PA. Rapid evo- 2016; 11:e0157007. lution of pandemic noroviruses of the GII.4 lineage. PLoS 35. Ramani S, Neill FH, Opekun AR, et al. Mucosal and cellu- Pathog 2010; 6:e1000831. lar immune responses to norwalk virus. J Infect Dis 2015; 23. Parra GI, Squires RB, Karangwa CK, et al. Static and evolv- 212:397–405. ing norovirus genotypes: implications for epidemiology and 36. Siebenga JJ, Vennema H, Zheng DP, et al. Norovirus illness immunity. PLoS Pathog 2017; 13:e1006136. is a global problem: emergence and spread of norovirus 24. Gurdasani D, Carstensen T, Tekola-Ayele F, et al. The GII.4 variants, 2001-2007. J Infect Dis 2009; 200:802–12. African Genome Variation Project shapes medical genetics 37. Lindesmith L, Moe C, Marionneau S, et al. Human suscep- in Africa. Nature 2015; 517:327–32. tibility and resistance to Norwalk virus infection. Nat Med 25. O’Connell J, Gurdasani D, Delaneau O, et al. A general 2003; 9:548–53. approach for haplotype phasing across the full spectrum of 38. Hoa Tran TN, Trainor E, Nakagomi T, Cunliffe NA, relatedness. PLoS Genet 2014; 10:e1004234. Nakagomi O. Molecular epidemiology of noroviruses asso- 26. Marchini J, Howie B, Myers S, McVean G, Donnelly P. A ciated with acute sporadic gastroenteritis in children: global new multipoint method for genome-wide association stud- distribution of genogroups, genotypes and GII.4 variants. ies by imputation of genotypes. Nat Genet 2007; 39:906–13. J Clin Virol 2013; 56:185–93. 27. Fry AE, Ghansa A, Small KS, et al. Positive selection of a 39. van Beek J, de Graaf M, Xia M, et al. Comparison of norovi- CD36 nonsense variant in sub-Saharan Africa, but no asso- rus genogroup I, II and IV seroprevalence among children ciation with severe malaria phenotypes. Hum Mol Genet in the Netherlands, 1963, 1983 and 2006. J Gen Virol 2016; 2009; 18:2683–92. 97:2255–64. Norovirus Seroepidemiology in Africa • JID 2018:218 (1 September) • 725
The Journal of Infectious Diseases – Oxford University Press
Published: Jul 24, 2018
Keywords: antigens; seroprevalence; africa south of the sahara; diarrhea; seroepidemiologic studies; uganda; enzyme-linked immunosorbent assay
Access the full text.
Sign up today, get DeepDyve free for 14 days.