Seasonal variation in the non-specific effects of BCG vaccination on neonatal mortality: three randomised controlled trials in Guinea-Bissau

Abstract

The BCG vaccine protects non-specifically against other diseases than tuberculosis. Three randomised controlled trials of early BCG in Guinea-Bissau found a 38% reduction in all-cause neonatal mortality. Little is known about the underlying mechanisms. In Guinea-Bissau, prevalent infectious diseases display distinct seasonality. Revisiting the three trials (>6500 infants) comparing early BCG versus no early BCG in low weight infants on all-cause neonatal mortality over 12 consecutive years, we explored the seasonal variation in BCG's effect on mortality. In a subgroup of participants, adaptive and innate cytokine responses were measured 4 weeks after randomisation. Consistently over the course of the three trials and 12 years, the effect of BCG on all-cause neonatal mortality was particularly beneficial when administered in November to January, coincident with peaking malaria infections. During these months, BCG was also associated with stronger proinflammatory responses to heterologous challenge. Recent studies have suggested a protective effect of BCG against malaria. BCG may also ameliorate immune-compromising fatal effects of placental malaria in the newborn.

Trial registration: ClinicalTrials.gov NCT00146302 NCT00625482.

Keywords: immunisation; malaria; maternal health; paediatrics; vaccines.

Figure 1

HR for neonatal death comparing early BCG with delayed BCG, per calendar month and by trial. For Trial II, no HR estimate for January and July due to no deaths in the early BCG group in January and delayed group in July, respectively. HR<1 indicates lower mortality in the BCG group; HR>1 indicates lower mortality in the delayed BCG group.

Figure 2

HR for neonatal death comparing early BCG with delayed BCG, stratified by sex, trial and season. Estimated HR (with 95% CI) comparing early BCG to delayed BCG, stratified by season and trial. Season of BCG administration is defined as November to January versus February to October or dry (December to May) versus rainy (June to November). A HR of 1 is indicated with a vertical hatched line. Δp<0.05; ΔΔp<0.01; ΔΔΔp<0.001 for interaction between season and BCG.

Figure 3

Interferon (IFN)-γ responses to purified protein derivative (PPD) and medium alone. Geometric means of IFN-γ concentrations in whole blood in vitro stimulated with PPD (10 μg/mL) 4 weeks after randomisation to early BCG or nothing (control), stratified by season (November to December vs March to October). Error bar is 95% CI.

Figure 4

The immunological effects of BCG on in vitro cytokine responses (collective test). Whole blood from infants randomised to early BCG or nothing was collected 4 weeks after randomisation and stimulated with a panel of innate agonists and recall antigens (Jensen et al, J Infect D is 2015 211: 956–967), stratified by time (calendar month) of enrolment. The geometric mean ratios (GMRs) comparing early BCG vaccinated with non-vaccinated infants were analysed collectively for the non-specific responses to LPS, Pam3CSK4, PMA/ionomycin, CL075 and medium alone using Tobit regression. The study included infants randomised from March to December 2011. Season defined by the month of randomisation as November–December, n=101 versus March–October, n=366. *p<0.05; ***p<0.001 for test of BCG; ΔΔp<0.01 for interaction between season and BCG.