Immune responses to oral poliovirus vaccine in HIV-exposed uninfected Zimbabwean infants

It remains uncertain whether HIV-exposed uninfected (HEU) infants have impaired responses to oral vaccines. We performed a cross-sectional study of 6-month-old infants recruited at birth to the ZVITAMBO trial in Zimbabwe between 1997–2001, before introduction of prevention of mother-to-child transmission interventions. We measured poliovirus-specific IgA to type 1–3 polio strains by semi-quantitative capture ELISA in cryopreserved serum samples collected from 85 HEU and 101 HIV-unexposed infants at 6 months of age, one month after their last immunisation with trivalent OPV. Almost all infants were breastfed, with the majority in both groups mixed breastfed (70.6% HEU versus 71.3% HIV-unexposed). Median (IQR) vaccine titers for HEU and HIV-unexposed infants were 1592 (618–4896) vs. 1774 (711–5431) for Sabin 1 (P = 0.46); 1895 (810–4398) vs. 2308 (1081–4283) for Sabin 2 (P = 0.52); and 1798 (774–4192) vs. 2260 (996–5723) for Sabin 3 (P = 0.18). There were no significant differences in vaccine titers between HEU and HIV-unexposed infants, suggesting that vertical HIV exposure does not impact oral poliovirus vaccine immunogenicity. This study utilised archived samples from the Zimbabwe Vitamin A for Mothers and Babies (ZVITAMBO) trial, which has been described previously.24 Briefly, 14110 mother-infant pairs were enrolled within 96 hours of delivery in Harare, Zimbabwe between 1997 and 2001. Mother-infant pairs were eligible if neither had an acutely life-threatening condition and the infant was a singleton with birth weight >1500 g. Written informed consent was obtained. Socioeconomic and demographic information was collected by maternal interview. Follow-up was conducted at 6 weeks, 3 months and then 3 monthly to 12–24 months of age. The trial was conducted in a peri-urban setting and preceded availability of antiretroviral therapy in Zimbabwe or use of cotrimoxazole prophylaxis for HIV-exposed infants. Anthropometry was conducted at each visit, using methods and WHO reference standards as described previously.25 Blood was collected by venipuncture from all mothers and infants at baseline (≤96 hours after delivery) and at all follow-up visits. Samples were centrifuged and plasma removed within 2 hours of collection. Samples were stored in −80°C freezers with automatic generator backup Mothers underwent HIV testing at baseline using 2 parallel ELISA assays. Women testing HIV-negative were re-tested at every visit to detect HIV seroconversion. The last available sample from each child was tested for HIV by GeneScreen ELISA on plasma if aged ≥ 18 months, or by DNA polymerase chain reaction (Roche Amplicor version 1.5; Roche Diagnostic Systems, Alameda, CA) in cell pellets if aged <18 months. If the last available sample was negative, the child was classified as HIV-negative. Children were classified as HIV-unexposed if the mother tested HIV-negative at baseline and did not seroconvert during follow-up; children were classified as HIV-exposed uninfected (HEU) if the mother tested HIV-positive at baseline and the last available infant sample was HIV-negative. Infants of mothers who seroconverted during follow up were not included. For the current cross-sectional study, we retrieved samples for all HEU and HIV-unexposed infants fulfilling the following criteria: 1) Mother and infant both received placebo in the original trial; 2) infants were followed to 6 months with available feeding and anthropometry data; and 3) a sufficient sample of cryopreserved serum was available at 6 months of age. All mothers, irrespective of HIV status, were encouraged to exclusively breastfeed their infants for 6 months. Data on feeding practices at 6 weeks and 3 months were used to categorise infants as exclusively, predominantly or mixed breastfed, as previously defined.26 Infants followed the routine Expanded Programme of Immunisation schedule in Zimbabwe at the time, which included trivalent OPV at 3, 4 and 5 months of age. Specific vaccination data were not collected as part of the trial, but OPV3 immunisation coverage (collated by WHO using data from routine reporting systems) at that time in Zimbabwe was 70–81%.27 Although neutralising antibody titer is the best correlate of protection for OPV,28,29 measurements of mucosal (salivary/stool secretory IgA) and circulating (serum) IgA responses to poliovirus are useful in the detection and control of poliovirus infection,30-35 particularly in settings without access to a WHO-accredited polio reference laboratory capable of performing neutralization assays. In all 6-month serum samples (i.e. one month after receipt of final OPV), we measured poliovirus-specific IgA to type 1–3 polio strains in a capture ELISA developed by collaborators at the Centers for Disease Control and Prevention (CDC) as described previously.36 First, 96-well microplates were coated with goat anti-human IgA (SeraCare, Massachusetts) for 60 minutes at 37°C. Next plates were washed with buffer containing phosphate buffered saline (PBS) and 0.05% Tween 20 and incubated for a further 60 minutes with dilution buffer containing bovine serum albumin dissolved in PBS (all purchased from Sigma, St Louis). We then made 3-fold serial dilutions of serum samples beginning at a dilution of 1 in 100 and added 50 uL of diluted sample to each well (one patient sample per column). After a further wash step, the plates were incubated overnight in a moist chamber at room temperature with Sabin antigen, cultured from Hep2C cells at the University of Vermont. The next day, the plates were washed followed by addition of monoclonal Sabin antibody (Merck Millipore, Massachusetts) to the corresponding Sabin antigen and incubated at 37°C for 60 minutes. After another wash step, goat anti-mouse IgG conjugated to enzyme (SeraCare) was added to the plates for a further 60 minutes at 37°C. We then developed the reactions by addition of substrate for 15 minutes at room temperature, with the output being optical densitiy (O.D.) at 450 nm. Pooled serum from poliovirus-vaccinated healthy donor volunteers was used as a positive control; each plate also included a blank column with no test sample as a negative control. Using 3-fold serial dilutions of patient samples run out in a single column allowed for a semi-quantitative measure of poliovirus-specific IgA, an adaptation of the original method. Endpoint titers were calculated by subtracting the plate background (i.e., the average O.D. of the negative control wells), then identifying the dilution of the final well in each sample column with an O.D. >0.07 (a 95% confidence value cut-off used to distinguish “negative” from “positive” absorbance values and calculated according to the original method36). The intra- and inter-plate coefficients of variation for Sabin 1, 2 and 3 strains, derived from the mean O.D.s and standard deviations from 7 successive preliminary experiments run before including study samples, were 8.2%, 4.0% and 4.8% and 20.1%, 17.7% and 16.8%, respectively. As a means of continued quality assurance, the end point titer of the positive control sample in all subsequent assays had to fall within an acceptable range for the experiment to be deemed valid. The range was pre-determined based on the upper and lower limit O.D.s in the 7 preliminary experiments. Laboratory scientists were blinded to infant HIV exposure status when conducting the assays. If an end point titer could not be derived at the first attempt (e.g. bottom well O.D. ≥0.08), the assay was repeated using a 10-fold higher or lower concentration of patient sample where sufficient serum volume was available. If an end point titer could not be derived using the new sample concentration, the lowest or highest dilution factor was taken as the final end point. Extreme low and high end point titers were subsequently truncated and assigned a value equivalent to the 5th and 95th centile within the data set respectively. Baseline characteristics were compared between groups using Fisher’s exact tests for categorical variables, and Mann-Whitney or 2-sample t-tests for continuous variables, depending on data distribution. A regression model was used to calculate adjusted differences between median vaccine titers between groups, using breastfeeding, birth weight and infant sex as covariates. Statistical analyses were performed using STATA 13 (Stata-Corp, College Station, TX, USA) and Prism version 6 (GraphPad Software Inc., La Jolla, CA, USA). The original ZVITAMBO trial and this laboratory substudy were approved by the Medical Research Council of Zimbabwe, John Hopkins School of Public Health Committee on Human Research and the Montreal General Hospital Ethics Committee.