A microbiota-directed complementary food intervention in 12-18-month-old Bangladeshi children improves linear growth

Abstract

Background: Globally, stunting affects ∼150 million children under five, while wasting affects nearly 50 million. Current interventions have had limited effectiveness in ameliorating long-term sequelae of undernutrition including stunting, cognitive deficits and immune dysfunction. Disrupted development of the gut microbiota has been linked to the pathogenesis of undernutrition, providing potentially new treatment approaches.

Methods: 124 Bangladeshi children with moderate acute malnutrition (MAM) enrolled (at 12-18 months) in a previously reported 3-month RCT of a microbiota-directed complementary food (MDCF-2) were followed for two years. Weight and length were monitored by anthropometry, the abundances of bacterial strains were assessed by quantifying metagenome-assembled genomes (MAGs) in serially collected fecal samples and levels of growth-associated proteins were measured in plasma.

Findings: Children who had received MDCF-2 were significantly less stunted during follow-up than those who received a standard ready-to-use supplementary food (RUSF) [linear mixed-effects model, β_{treatment group x study week} (95% CI) = 0.002 (0.001, 0.003); P = 0.004]. They also had elevated fecal abundances of Agathobacter faecis, Blautia massiliensis, Lachnospira and Dialister, plus increased levels of a group of 37 plasma proteins (linear model; FDR-adjusted P < 0.1), including IGF-1, neurotrophin receptor NTRK2 and multiple proteins linked to musculoskeletal and CNS development, that persisted for 6-months post-intervention.

Interpretation: MDCF-2 treatment of Bangladeshi children with MAM, which produced significant improvements in wasting during intervention, also reduced stunting during follow-up. These results suggest that the effectiveness of supplementary foods for undernutrition may be improved by including ingredients that sponsor healthy microbiota-host co-development.

Funding: This work was supported by the BMGF (Grants OPP1134649/INV-000247).

Keywords: Childhood undernutrition; Human gut microbiome development and repair; Microbiome-directed therapeutic foods; Plasma protein mediators of growth/postnatal development; Post-treatment follow-up; Stunting and wasting.

Fig. 1

Anthropometric trajectories during treatment with MDCF-2 or RUSF and during the two year post-intervention follow-up period. (a) Trial design, including timing of biospecimen collection for analysis of the microbiome and plasma proteome. (b and c) Weight-for-length Z-score (WLZ) trajectories during the three months of MDCF-2 or RUSF treatment (panel b) and during the two years of post-intervention follow-up (panel c). (d and e) Length-for-age Z-score (LAZ) trajectories during the three months of MDCF-2 versus RUSF treatment phase of the trial (panel d) or the two years of post-intervention follow-up (panel e). Each plot indicates a simple linear fit of the data for the MDCF-2 (blue) and RUSF (red) treatment groups along with the 95% confidence interval. The linear, mixed effects model for each anthropometric assessment (Eq. 2; Methods) was used to determine the coefficient (β) and statistical significance (P) for the specified effect. For the treatment phase analysis, the beginning of nutritional intervention was used at the baseline measurement, whereas for the follow-up analysis, the baseline used was the end of 3-month nutritional intervention.

Fig. 2

Differences in bacterial MAG abundances between fecal samples collected from MDCF-2- and RUSF-treated trial participants during the post-intervention follow-up. (a) The enrichment of groups of positively (red) or negatively (blue) WLZ-associated MAGs after ranking by differential abundance between the MDCF-2 or RUSF groups at end of treatment and at the 1-, 6-, or 12- month post-intervention time points. Normalized Enrichment Scores (NES) were determined by GSEA. Points representing significant NES significant (q < 0.05) are outlined in black. (b) Taxonomic assignments of MAGs driving the enrichment of positively WLZ-associated MAGs in the analysis depicted in (a). Species assignments are shown for taxa containing >1 MAG.

Fig. 3

Effects of MDCF-2 on the plasma proteome of children with MAM during nutritional intervention and in the follow-up period. (a) Volcano plot of WLZ-associated plasma proteins that are also associated with LAZ responses during the post-treatment follow-up period. The linear model described under the plot was applied to relate the rate of change for each participant in the levels of proteins during the treatment-phase to the rate of change of LAZ during the follow-up. Significant LAZ-predictive proteins (q < 0.1) are colored in orange. (b) Annotated functions of the 37 plasma proteins associated with the effect of MDCF-2 on LAZ during the follow-up period. (c) Association of the 37 proteins with treatment arm during the 3-month treatment phase and at the 6-month follow-up time point. One-way t-tests were applied to the β3 coefficient in Eq. 9 and the β1 coefficient in Eq. 10 (see Methods, q < 0.05) at each time point to test the statistical significance of their enrichment in either the MDCF-2 or RUSF treatment arm.