Ethical Approval
Informed consent was obtained in the appropriate local language. Ethical approval was obtained from the University Hospitals Cleveland Medical Center IRB (08-07-09), Colorado Multiple IRB (15–1277), and Kenya Medical Research Institute (KEMRI) Scientific and Ethical Review Unit (NON SSC 089).
Study Site and Participants
The study was conducted in Kisumu County, Kenya, at the Chulaimbo Sub-County Hospital, from 2011–2016. Malaria transmission in this area is intense and year-round, with peaks coinciding with rainy seasons (17). Chulaimbo Hospital serves a primarily rural population and is an Academic Model Providing Access to Healthcare (AMPATH) site. Clinical services and medications for HIV + patients and their families are supported by USAID and the Indiana University–Kenya partnership.
Participants were enrolled in a prospective observational study in which HIV + and HIV- pregnant women were enrolled at their first prenatal visit (typically during the second trimester) and followed through pregnancy. Infants were followed from birth to 24 months of age. All HIV + women received ART therapy (Lamivudine + Zidovudine (Hb > 8g/dL) or Tenofovir (Hb < 8g/dL) + Nevirapine (CD4 < 250 cells/µL) or Lopinavir/ritonavir (CD4 > 250 cells/µL)). HEU newborns were treated with one dose of Nevirapine after delivery and 6 weeks of Zidovudine, with zero HIV + infants in this cohort.
Eighty-five HEU and 168 HUU children with complete clinical data and samples were included in this study. Study visits occurred at birth and at 6, 10, 14, and 18 weeks and 6, 9, 12, 15, 18, 21, and 24 months of age. The Kenya Expanded Programme on Immunization schedule included diphtheria-pertussis-tetanus (DPT), Haemophilus influenzae type b (Hib), hepatitis B, oral polio (OPV) and pneumococcus (PCV) vaccines at 6, 10, and 14 weeks and measles vaccine at 9 and 18 months of age. According to Kenyan Ministry of Health guidelines, HEU infants received cotrimoxazole prophylaxis from 6 weeks of age until cessation of breastfeeding and definitive exclusion of HIV infection (typically at 18 months of age).
Clinical Events
Data collected during each visit included clinical history, physical exam and anthropometric measurements, and any concurrent diagnoses (e.g., upper respiratory tract infection, malaria, gastroenteritis). Participants were seen at the study clinic for any interim sick visits, where data were collected regarding physical exam findings, diagnoses, treatments given, and severity of illness (any illness that required hospitalization was classified as severe). Major diagnostic categories from interim or concurrent sick visits included clinical malaria (defined as febrile illness with Plasmodium falciparum (Pf) parasitemia by blood smear or rapid diagnostic test), upper respiratory tract infection (URTI), any respiratory illness, gastroenteritis/diarrhea, and any severe illness.
Blood Sample Collection and Processing
At delivery, venous cord blood samples were collected in heparinized syringes. At all other visits, heparinized finger prick or venous blood samples were obtained. Aliquots of 200 µL whole blood were stored at − 20°C. Plasma was separated and stored at − 80°C. All sample processing occurred within 1–5 hours of collection at the laboratory facilities at the Center for Global Health Research of KEMRI. All assays were conducted at the KEMRI laboratories.
Detection of Pf infection by PCR
To determine the prevalence of asymptomatic Pf infection in the cohort, Pf PCR was performed on all available samples at 6, 9, 12, 15, 18, 21, and 24 month visits. DNA was extracted from whole blood using Qiagen QIAmp DNA Mini Kits. Pf PCR was performed as previously described (18).
Measurement of Cytokines
Immunological assays were performed for a subset of children with plasma samples available at all timepoints. We measured levels of 12 cytokines in plasma samples from 59 HEU and 58 HUU children at birth, 6, 10, 14, 18, 26, 39, and 52 weeks of age. All plasma samples were assayed immediately after initial thawing. A multiplexed bead-based immunoassay was used to measure plasma concentrations of IFN-γ, IL-1β, IL-6, IL-10, IL-12P70, IL-17A, IL-17E, IL-17F, IL-21, IL-22, IL-23, and TNF (Human Th17 Magnetic Bead Panel, EMD Millipore).
Vaccine-Specific IgG
We measured IgG antibodies to diphtheria, tetanus, hepatitis B, and measles in plasma from 61 HEU and 54 HUU children by ELISA, as previously described (16, 19, 20). Time points included birth, 6, 10, 14, and 18 weeks, and 6, 9, 12, 15, 18, 21, and 24 months. Serial dilutions of samples were compared with 5-point standard curves made with serial dilutions from World Health Organization–approved antigen-specific reference sera; diphtheria Ig, human (NIBSC 10/262, 2 IU/mL), tetanus Ig, human (NIBSC TE-3, 120 IU/ mL), hepatitis B Ig, human (NIBSC 07/164, 100 IU/mL), and measles Ig, human (NIBSC 97/648, 3 IU/mL).
Pf Antigen-Specific IgG
We measured IgG antibodies to 14 recombinant Pf proteins in plasma samples from 69 HEU and 76 HUU children using Luminex MagPix assays (MagPlex, Luminex). Time points included birth, 6, 10, 14, and 18 weeks, and 6, 9, 12, 15, 18, 21, and 24 months. The assays were performed as previously described (16, 21). Supplementary Table 1 contains sources and quantities of conjugated antigens. For each assay, plasma was diluted 1:100 and 1:1000. The secondary antibody was R-Phycoerythrin-conjugated AffiniPure F(ab’) Fragment Goat Anti-Human IgG Fcγ Fragment Specific (Jackson ImmunoReaserch, West Grove, PA). Seven malaria-naïve North American adult plasma samples were tested on all plates as negative controls. Mean fluorescent intensity (MFI) values were divided by average MFIs of negative controls. Data are expressed as the fold-increase of the sample MFI relative to the negative control MFI (reported as fold over North American), as previously described (21).
Statistical Analysis
World Health Organization (WHO) child growth standards were used to generate Z scores for weight-for-age (WAZ), length-for-age (LAZ), head circumference for age (HAZ), and body mass index (BMI) for age (BAZ) (WHO Anthro Survey Analyser, R package “anthro” v0.9.4) (22). Bayesian hierarchical regression analysis was used to fit WAZ, LAZ, HAZ, and BAZ linear growth curve models for the HEU vs. HUU groups, which accounts for both individual-specific effects and group-level effects. All models were fit using JAGS software via the R package “rjags” (23, 24). Missingness in the data was assumed to be completely at random. Prior distributions for group means for intercept and slope were weakly informative and specified as normal distributions with mean 0 and variance 100. Standard non-informative uniform distributions were specified for the residual variance and the standard deviations of intercept and slope. Posterior distributions were checked for convergence graphically and numerically using the \(\widehat{R}\) statistic; convergence criteria with \(\widehat{R}\) < 1.1 were met for all parameters for all models. For each growth curve model, posterior means for intercepts and slopes, with 95% highest probability density interval (HDI), were contrasted between HEU vs. HUU groups. We selected a region of practical equivalence (ROPE) with upper and lower limits of -0.05 and 0.05; contrast terms for group intercepts and slopes were deemed to be meaningfully different if the 95% HDI excluded this interval.
The prevalence of asymptomatic Pf infection among HEU and HUU infants was calculated at 6, 9, 12, 15, 18, 21, and 24 months of age. The relative risk of Pf infection in the HEU group compared to the HUU group was determined at each time point.
Clinical event rates for the major diagnostic categories (malaria, URTI, any respiratory illness, gastroenteritis/diarrhea, and severe illness) were calculated by the number of diagnoses per 1,000 person-months of follow up in each group. Rate ratios were calculated as the ratio of clinical event rates in HEU children divided by the clinical event rates in HUU children.
Time fixed effects regression models were used to analyze plasma cytokines, vaccine antibodies, and antimalarial antibodies in HEU vs. HUU children. We first tested whether trajectories differed among HEU vs. HUU children (“group x time” interaction effects). Then if the trajectories did not differ, we tested whether values for HEU children differed from values for HUU children (“group” main effect) and whether the values changed over time within a group (“time” main effect).