Stunting Status and Exposure to Infection and Inflammation in Early Life Shape Antibacterial Immune Cell Function Among Zimbabwean Children

Abstract

Background: Children who are stunted (length-for-age Z-score<-2) are at greater risk of infectious morbidity and mortality. Previous studies suggest that stunted children have elevated inflammatory biomarkers, but no studies have characterised their capacity to respond to new infections (i.e., their immune function). We hypothesised that antibacterial immune function would differ between stunted and non-stunted children and relate to their health and environment during early life.

Methods: We enrolled a cross-sectional cohort of 113 HIV-negative children nested within a longitudinal cluster-randomised controlled trial of household-level infant and young child feeding (IYCF) and water, sanitation and hygiene (WASH) interventions in rural Zimbabwe (SHINE; Clinical trials registration: NCT01824940). Venous blood was collected at 18 months of age and cultured for 24 h without antigen or with bacterial antigens: heat-killed Salmonella typhimurium (HKST) or Escherichia coli lipopolysaccharide (LPS). TNFα, IL-6, IL-8, IL-12p70, hepcidin, soluble (s)CD163, myeloperoxidase (MPO) and IFNβ were quantified in culture supernatants by ELISA to determine antigen-specific immune function. The effect of stunting status and early-life exposures (anthropometry, inflammation at 18 months, maternal health during pregnancy, household WASH) on immune function was tested in logit and censored log-normal (tobit) regression models.

Results: Children who were stunted (n = 44) had higher proportions (86.4% vs. 65.2%; 88.6% vs. 73.4%) and concentrations of LPS-specific IL-6 (geometric mean difference (95% CI): 3.46 pg/mL (1.09, 10.80), p = 0.035) and IL-8 (3.52 pg/mL (1.20, 10.38), p = 0.022) than non-stunted children (n = 69). Bacterial antigen-specific pro-inflammatory cytokine concentrations were associated with biomarkers of child enteropathy at 18 months and biomarkers of systemic inflammation and enteropathy in their mothers during pregnancy. Children exposed to the WASH intervention (n = 33) produced higher LPS- (GMD (95% CI): 10.48 pg/mL (1.84, 60.31), p = 0.008) and HKST-specific MPO (5.10 pg/mL (1.77, 14.88), p = 0.003) than children in the no WASH group (n = 80). There was no difference in antigen-specific immune function between the IYCF (n = 55) and no IYCF groups (n = 58).

Conclusions: Antibacterial immune function among 18-month-old children in a low-income setting was shaped by their stunting status and prior exposure to maternal inflammation and household WASH. Heterogeneity in immune function due to adverse exposures in early life could plausibly contribute to infection susceptibility.

Keywords: Immune Cells; Immune Function; Inflammation; Malnutrition; Maternal and Child Health; Pregnancy; Stunting; Zimbabwe.

Figure 1

Selection of 18-month-old infants for inclusion in the immune function sub-study of SHINE.

Figure 2

Soluble mediator production by blood immune cells from 18-month-old Zimbabwean children in response to in vitro bacterial antigen challenge. Violin plots (median and interquartile range indicated) of (A) TNFα; (B) IL-6; (C) IL-8; (D) MPO; (E) IL-12p70; (F) IFNβ; (G) sCD163; and (H) hepcidin concentrations (pg/mL) present in supernatants from parallel 24h culture of whole blood samples from Zimbabwean children with culture media only (unstimulated; red), LPS (blue) and HKST (grey). Proportions indicate participants with mediator concentration > ELISA limit of detection; proportions in italics indicate participants with mediator concentration > unstimulated. Log-transformed mediator concentrations were compared between culture conditions using generalised estimating equations (n = 113); **p < 0.01, ***p < 0.001.

Figure 3

Unstimulated and bacterial antigen-stimulated immune-mediator production by blood immune cells from stunted versus non-stunted children. Violin plots (median and interquartile range indicated) of (A) unstimulated (red), (B) LPS-stimulated (blue) and (C) HKST-stimulated (grey) TNFα, IL-6, IL-8 and MPO concentrations in 24h culture supernatants. LPS- and HKST-specific concentrations (Δ) were calculated for each child by subtracting concentrations present in matched unstimulated culture supernatants. Proportions indicate participants with mediator concentration > ELISA limit of detection (A) and participants with antigen-stimulated mediator concentration > unstimulated (B, C). Proportions of children with detectable mediator levels and mediator concentrations were compared by stunting status (stunted n = 44 versus non-stunted n = 69) via multinomial logit regression and censored log-normal (tobit) regression, respectively (unadjusted analyses indicated; full results in Table 2 ); *p < 0.05.

Figure 4

Unstimulated and bacterial antigen-stimulated immune-mediator production by blood immune cells from children exposed versus not exposed to a household WASH intervention. Violin plots (median and interquartile range indicated) of (A) unstimulated (red), (B) LPS-stimulated (blue) and (C) HKST-stimulated (grey) TNFα, IL-6, IL-8 and MPO concentrations in 24h culture supernatants. LPS- and HKST-specific concentrations (Δ) were calculated for each child by subtracting concentrations present in matched unstimulated culture supernatants. Proportions indicate participants with mediator concentration > ELISA limit of detection (A) and participants with antigen-stimulated mediator concentration > unstimulated (B, C). Proportions of children with detectable mediator levels and mediator concentrations were compared by exposure to the SHINE WASH intervention (WASH n = 33, no WASH n = 80) via multinomial logit regression and censored log-normal (tobit) regression, respectively (unadjusted analyses indicated; full results in Table 4 ); *p < 0.05, **p < 0.01.

Figure 5

Unstimulated and bacterial antigen-stimulated immune-mediator production by blood immune cells from children exposed versus not exposed to a household IYCF intervention. Violin plots (median and interquartile range indicated) of (A) unstimulated (red), (B) LPS-stimulated (blue) and (C) HKST-stimulated (grey) TNFα, IL-6, IL-8 and MPO concentrations in 24h whole blood culture supernatants. LPS- and HKST-specific concentrations (Δ) were calculated for each child by subtracting concentrations present in matched unstimulated culture supernatants. Proportions indicate participants with mediator concentration > ELISA limit of detection (A) and participants with antigen-stimulated mediator concentration > unstimulated (B, C). Proportions of children with detectable mediator levels and mediator concentrations were compared between children exposed to the SHINE household IYCF intervention (n = 55) and those not exposed to the IYCF intervention (n = 58) by multinomial logit regression and censored log-normal (tobit) regression, respectively (unadjusted analyses indicated; full results in Table 5 ).