SARS-CoV-2 infections in infants in Haiti 2020-2021; evidence from a serological cohort

59 This study was conducted under review of the University of Florida (UF) IRB and the Haitian 60 Comite National de Bioethique of the Ministere De La Sante Publique Et De La Population. 61 Mothers in the study were recruited at the time of their first antenatal visit; basic demographic, 62 epidemiologic, and clinical data were obtained; and efforts were made to follow all infants born 63 to enrolled mothers, with visits scheduled at 0, 4, 12, and 18-30 months of age. During visits, 64 infants were screened for developmental delays, and DBS samples were collected by heel-stick. 65 Mothers were also asked to report illness in their infants since the infant was last seen by study 66 staff. Travel within Haiti was severely limited during the study period because of substantial 67 political unrest and the COVID epidemic. Many mothers were also hesitant about allowing a 68 heel-stick for blood collection. Because of this, it was not possible to obtain samples from all 69 infants at all scheduled time points. 70 71 Detailed laboratory methods are provided in Supplemental Materials. In brief, a single 6mm 72 biopsy punch was used to capture the area covered with each blood spot. Because the blood spots 73 varied in size and intensity, the amount of total protein in each sample was determined by using a 74 Bradford protein microassay. A research ELISA, adapted from a previously published protocol,8 75 was developed and used to target the SARS-CoV-2 Receptor Binding Domain (RBD). 76 77 We estimated that with 50 samples collected before the pandemic and 50 after, we had 80% 78 power to detect a difference in seroprevalence assuming 15% of babies would be seropositive 79 after the beginning of the pandemic and 0% before (with 95% confidence). We fit regression 80 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 20, 2022. ; https://doi.org/10.1101/2022.03.17.22272561doi: medRxiv preprint 5 models to average ODs to determine factors associated with larger OD values. A mixture model 81 was fit to the average OD’s across all samples, assuming that observations came from two 82 distributions, those that had not been infected and those that had in the past. A threshold in OD 83 value was identified from the mixture model that indicated a 95% probability of belonging to the 84 distribution associated with higher values (OD>0·21). This threshold was used to define 85 individual measurements as indicating seropositivity. Interval censored survival analysis was 86 used to investigate the association of the hazard of SARS-CoV-2 seropositivity possible risk 87 factors. 88 89

90 We obtained 388 samples from 257 children; numbers of samples by year and time point are 91 shown in Supplemental Table 1. Longitudinal samples were obtained from 107 of these children 92 consisting of two or more samples: this includes 84 infants for whom two samples were 93 available, 22 with three samples, and one for whom samples were available for all four time 94 points. The average protein concentration for all samples was 5,194 ug/mL with 18 measuring 95 below 100 ug/mL and the highest measuring 13,209 ug/mL. There were no differences in mean 96 protein levels when analyzed for age or year. 97 98 Using the dataset of all samples (388), a spline investigating the relationship between time of 99 sampling and OD indicated that average ODs increased ~0·200 in value at 400 days past January 100 1, 2019 or approximately March 2020 (Supplemental Figure 1). Increases in ODs that occurred 101 in 2020 and 2021 reflected the timing of waves of reported for SARS-CoV-2 cases and deaths in 102 Haiti over the same time periods (Figure 1). No anti-SARS-CoV-2 IgG responses above the 103 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 20, 2022. ; https://doi.org/10.1101/2022.03.17.22272561doi: medRxiv preprint 6 predicted cut-off occurred before July, 2020 and no infant was seropositive at birth. Forty-three 104 (16·7% [95% CI 12·7%-21·8%]) of 257 unique children tested seropositive at some point during 105 the study. 106 107 We conducted a survival analysis to account for differing amounts of person time for each 108 individual and over time (Figure 2, Supplemental Figure 2). We fit piecewise constant hazard 109 survival models to estimate the hazard of anti-SARS-CoV-2 seropositivity over time in our study 110 population. Hazards estimated for intervals before and after March 1, 2020 found that hazards 111 were significantly larger in the later interval (Hazard ratio of 39·1 95% CI 13·1, 2.8e7). Age of 112 child, mother’s age, educational attainment, marital status, urban/rural status and parity of birth 113 were not associated with the hazard of infection. Trends in OD were similar across age groups 114 (Supplemental Figure 3). 115 116 Twenty-one of the consecutively tested children tested above the cut-off in 2020 and 2021, with 117 seven testing above the cut-off on two of the samples. One child (of 8 total) who had a sample 118 collected after an initial seropositive sample became seronegative upon follow up. Only limited 119 data were available on clinical illnesses experienced by children during the study. However, no 120 child was diagnosed at a medical facility as having COVID-19, based either on clinical 121 presentation or laboratory testing, and mothers did not spontaneously report occurrence of more 122 severe illnesses in their infants that might have been consistent with COVID-19. 123 124 125 126 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 20, 2022. ; https://doi.org/10.1101/2022.03.17.22272561doi: medRxiv preprint 7 Comment 127 The overall seropositivity rate of 16.7% among our infant cohort was lower than the 39% 128 seroprevalence previously reported by our group for adult populations in Haiti during a 129 comparable time period.7 This may be a reflection of lower exposure of neonates and very young 130 children to the general population. The adult data were collected from predominantly urban 131 populations in Port-au-Prince. Given that studies in Peru and South Africa have demonstrated 132 higher seroprevalence rates in more densely populated urban areas as compared with rural 133 areas,4,5 caution should be used in directly comparing the data from our infants (the majority of 134 whom came from a rural origin) with these adult data. 135 136 In work performed in middle and upper income countries, SARS-CoV-2 seropositivity in 137 children generally identified as below the age of five has been highly variable, from less than 1% 138 in German and Switzerland to 6% in Spain after during the first surge of the pandemic.9-11 139 Several U.S. studies examined antibody prevalence in residual blood samples of children during 140 the late spring and early fall of 2020, with seroprevalence rates of 9·5 - 16·3%.12,13 Recently, a 141 cross sectional investigation of SARS in Virginia reported 8·5% seropositivity in a study of 142 1,038 children; seroprevalence was highest (13·7%) in children in the 0-5 year age group.14 Our 143 data are consistent with the observation that rates of seropositivity are higher in LMIC settings 144 and underscore the widespread distribution of infection in very young children in these areas, 145 with no risk factors identified other than occurrence in the midst of an epidemic wave. 146 147 In keeping with data showing little evidence of serious clinical illness in infected infants, we did 148 not obtain a history of a serious preceding illness among our seropositive infants, nor were any 149 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 20, 2022. ; https://doi.org/10.1101/2022.03.17.22272561doi: medRxiv preprint 8 of the children in the study formally diagnosed as having COVID-19. However, our data 150 document that infections do occur in infants and elicit an immune response. These studies were 151 conducted at a time when B.1 lineage strains were predominant in Haiti and need to be repeated 152 with the successive waves of Gamma, Delta, and Omicron strains which have occurred;7 the 153 health impact of these later lineages also remains to be determined. However, given that infants 154 have been shown to have high SARS-CoV-2 viral loads,5 our findings underscore the potential 155 importance of very young children in facilitating virus transmission within LMIC households 156 and communities, and the need to consider these youngest members of society in developing 157 models and prevention strategies for COVID-19. 158 159 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 20, 2022. ; https://doi.org/10.1101/2022.03.17.22272561doi: medRxiv preprint 9 Contributors 160 Conceptualization (RL, VMBDR, DATC, MTL, JGM); investigation (RL, RC, VMBDR); data 161 curation and analysis (RL, RC, DATC, MTL, JGM); methodology (RP, TDL, LT-S, AG, EN, 162 MTL); writing and reviewing and editing (RL, VMBDR, EN, DATC, MTL, JGM) 163 Data Sharing 164 Contingent on adherence to IRB requirements, the complete de-identified participant dataset will 165 be available on request. Data will be available with publication. Additional documents (study 166 protocol, statistical analysis plan, informed consent form) will also available if needed with 167 publication. 168

169 This project has received funding from the European Union’s Horizon 2020 research and 170 innovation programme under grant agreement No 734857 through the Penta Foundation (the 171 ZIKAction Project); and by University of Florida Emerging Pathogens Institute, the Clinical and 172 Translational Research Institute, and the Fern Audette Endowment, College of Veterinary 173 Medicine, Gainesville FL. We thank and are grateful the provision of reagents and protocols for 174 the rbdELISA from Aaron Schmidt, Ragon Institute/Harvard MedicalSchool, Boston MA and 175 Jason Harris at Massachusetts General Hospital/Harvard Medical School. 176 Declaration of conflicting interests 177 The authors declared no competing interests. 178 179 180 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 20, 2022. ; https://doi.org/10.1101/2022.03.17.22272561doi: medRxiv preprint 10

181 1. Bobrovitz N, Arora RK, Cao C, et al. Global seroprevalence of SARS-CoV-2 antibodies: A 182 systematic review and meta-analysis. PLoS One 2021; 16(6): e0252617. 183 2. Bloomfield M, Pospisilova I, Cabelova T, et al. Searching for COVID-19 Antibodies in Czech 184 Children-A Needle in the Haystack. Front Pediatr 2020; 8: 597736. 185 3. Reyes-Vega MF, Soto-Cabezas MG, Cardenas F, et al. SARS-CoV-2 prevalence associated to 186 low socioeconomic status and overcrowding in an LMIC megacity: A population-based 187 seroepidemiological survey in Lima, Peru. EClinicalMedicine 2021; 34: 100801. 188 4. Kleynhans J, Tempia S, Wolter N, et al. SARS-CoV-2 Seroprevalence in a Rural and Urban 189 Household Cohort during First and Second Waves of Infections, South Africa, July 2020-March 190 2021. Emerg Infect Dis 2021; 27(12): 3020-9. 191 5. Ochoa V, Díaz FE, Ramirez E, Fentini MC, Carobene M, Geffner J, Arruvito L, Remes 192 Lenicov F; INBIRS COVID-19 Study Group. Infants Younger Than 6 Months Infected With 193 SARS-CoV-2 Show the Highest Respiratory Viral Loads. J Infect Dis. 2022 Feb 1;225(3):392-194 395. doi: 10.1093/infdis/jiab577. PMID: 34850028; PMCID: PMC8690165. 195 6. Ades AE, Brickley EB, Alexander N, et al, including EC Zika Consortia Vertical 196 Transmission Study Group. Zika virus infection in pregnancy: a protocol for the joint analysis of 197 the prospective cohort studies of the ZIKAlliance, ZikaPLAN and ZIKAction consortia. BMJ 198 Open 2020;10:e035307. doi: 10.1136/bmjopen-2019-035307 199 7. Tagliamonte MS, Mavian C, Zainabadi K, et al. Rapid emergence and spread of SARS-CoV-2 200 gamma (P.1) variant in Haiti. Clin Infect Dis 2021 Sep 2:ciab736. doi: 10.1093/cid/ciab736. 201 Online ahead of print. 202 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 20, 2022. ; https://doi.org/10.1101/2022.03.17.22272561doi: medRxiv preprint 11 8. Roy V, Fischinger S, Atyeo C, et al. SARS-CoV-2-specific ELISA development. J Immunol 203

2020; 484-485: 112832. 204 9. Pollán M, Pérez-Gómez B, Pastor-Barriuso R, et al. SARS-CoV-2 seroprevalence in Spain - 205 Authors' reply. Lancet 2020; 396(10261): 1484-5. 206 10. Stringhini S, Wisniak A, Piumatti G, et al. Seroprevalence of anti-SARS-CoV-2 IgG 207 antibodies in Geneva, Switzerland (SEROCoV-POP): a population-based study. Lancet 2020. 208 396:313-319. doi: 10.1016/S0140-6736(20)31304-0. 209 11. Tönshoff B, Müller B, Elling R, et al. Prevalence of SARS-CoV-2 Infection in Children and 210 Their Parents in Southwest Germany. JAMA Pediatr 2021; 175(6): 586-93. 211 12. Hobbs CV, Drobeniuc J, Kittle T, et al. Estimated SARS-CoV-2 Seroprevalence Among 212 Persons Aged <18 Years - Mississippi, May-September 2020. MMWR Morb Mortal Wkly Rep 213 2021; 70(9): 312-5. 214 13. Bahar B, Simpson JN, Biddle C, et al. Estimated SARS-CoV-2 Seroprevalence in Healthy 215 Children and Those with Chronic Illnesses in the Washington Metropolitan Area as of October 216 2020. Pediatr Infect Dis J 2021; 40(7): e272-e4. 217 14. Levorson RE, Christian E, Hunter B, et al. A cross-sectional investigation of SARS-CoV-2 218 seroprevalence and associated risk factors in children and adolescents in the United States. PLoS 219 One 2021; 16(11): e0259823. 220 221 222 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 20, 2022. ; https://doi.org/10.1101/2022.03.17.22272561doi: medRxiv preprint 12 223 224 225 Figure 1. Reported cases (blue) and deaths (red) due to SARS-CoV-2 in Haiti (top) and optical density 226 from ELISA that measures the antibody to SARS-CoV-2 Receptor Binding Domain IgG (bottom) by date 227 of sampling for Haitian infants born between March 2019 and May 2021 and followed for up to 30 mos. 228 229 230 231 232 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 20, 2022. ; https://doi.org/10.1101/2022.03.17.22272561doi: medRxiv preprint 13 233 234 Figure 2. Times of follow up for each infant with serostatus indicated from 2019 to 2021. Seropositive 235 samples are shown in red and the seronegative are without fill. Lines connect samples from the same 236 infant. 237 238 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 20, 2022. ; https://doi.org/10.1101/2022.03.17.22272561doi: medRxiv preprint 14 Supplemental Figure 1: Density of observed OD values with distributions identified by a mixture model 239 that assumed the values were drawn from two distributions, one that we associated with negative response 240 (red) and positive response (green). 241 242 Supplemental Figure 2: Kaplan Meier survival plot showing estimated probabilities of individuals in our 243 cohort remaining seronegative 244 245 Supplemental Figure 3: Times of follow up for each infant with serostatus indicated from 2019 to 2021. 246 Seropositive samples are shown in red and the seronegative are without fill. Shape of points indicate age 247 of participant (circles, at birth, squares ~4 months, diamonds ~12 months, triangles ~24 months). Lines 248 connect samples from the same infant. 249 250 Supplemental Table 1. Age of infants and year of sampling of children tested for antibodies to SARS-251 CoV-2 sampled in the Gessier Region between June 2019 and March 2021. 252 253 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 20, 2022. ; https://doi.org/10.1101/2022.03.17.22272561doi: medRxiv preprint