Faecal microbiota of schoolchildren is associated with nutritional status and markers of inflammation: a double-blinded cluster-randomized controlled trial using multi-micronutrient fortified rice

The study addresses how gut microbiota composition in school-aged children relates to nutritional status and whether micronutrient-fortified rice can modulate the microbiota. While infant and adult microbiomes are well studied, data on older children, particularly in non-Western settings, are limited. Malnutrition in low- and middle-income countries remains prevalent, including stunting and micronutrient deficiencies (iron, zinc, vitamin A). Prior literature links diet and micronutrient status to gut microbiota, and microbiota immaturity has been associated with severe malnutrition. The FORISCA project previously showed that fortified rice improved several micronutrient biomarkers and cognition but increased hookworm infection, with evidence of inflammation in a subset of children. This study aims to characterize baseline faecal microbiota in Cambodian schoolchildren, examine associations with age, sex, nutritional and micronutrient status, inflammation, and parasitic infection, and evaluate how two formulations of fortified rice alter microbiota composition and predicted function over six months.

- Gut microbiota influences nutrient processing and non-communicable diseases; composition stabilizes around age 2–3 but can vary across lifespan. - Diet is a key driver of microbiota variation, yet large interindividual diversity exists; data from non-Western cohorts, especially older children, are scarce. - Severe acute malnutrition is associated with microbiota immaturity and can be causally linked to disease progression in gnotobiotic models; microbiota-directed complementary foods improved growth in Bangladeshi children. - Micronutrient status interacts with microbiota: gut microbes can contribute to B-vitamin availability; zinc needs are reduced in germ-free animals, whereas iron needs are not; iron supplementation can decrease Bifidobacterium in some studies. Evidence is mixed across models and iron forms. - In Cambodia, stunting and anaemia have been prevalent; zinc deficiency is common. Prior FORISCA results reported improved micronutrient status and cognition with fortified rice but increased hookworm infections and both systemic and gut inflammation in subsets of children. - There is a gap in randomized trials assessing dietary micronutrient interventions on microbiota in large cohorts of school-aged children in low-resource settings.

Design and participants: Substudy within FORISCA, a double-blind, cluster-randomized, placebo-controlled trial in six schools in Kampong Speu, Cambodia (Nov 2012–Jul 2013). A random subset of 380 children (aged 6–14; exclusion: <6 or >14 years, disabilities, severe anaemia Hb <70 g/L; no antibiotics in past 3 months) with complete stool and biomarker data at both time points was analyzed. Interventions: Daily school meals for six months with one of three rices (cluster randomized by school): - Placebo: normal rice. - UR-original: fortified with iron, zinc, vitamins B1 and B9; provided slightly more Fe and Zn than UR-improved. - UR-improved: UR-original plus vitamins A, B3, and B12 (with slightly less Fe and Zn than UR-original). Ethics: Approved by Cambodian NECHR, Ministry of Education, and PATH; ClinicalTrials.gov NCT01706419; written parental consent and child assent. All children dewormed (single 500 mg mebendazole) after baseline collection. Data collection: Anthropometry (weight, height; WHO 2006 standards for HAZ and BAZ; stunting HAZ<-2; underweight BAZ<-2). Blood (non-fasting morning serum) for ferritin, soluble transferrin receptor, retinol-binding protein, CRP, AGP (ELISA, VitMin Lab), and serum zinc (AAS, Hanoi NIN). Faecal calprotectin for gut inflammation. Parasitology by Kato-Katz (eggs/g) at CNM. Microbiota profiling: Faecal DNA (QIAamp DNA Stool Mini kit). 16S rRNA V3–V5 amplicon sequencing (Illumina MiSeq, 2×300 bp; primers 357F/926R). Four randomized sequencing batches; no-template PCR control included. DADA2 used for denoising/chimera removal; due to low overlap, only forward reads analyzed. Taxonomy via RDP Naïve Bayes classifier against SILVA v138.1. Chloroplast/mitochondria removed. Rarefaction at 3,500 reads/sample for diversity. Core microbiota assessed at genus level (present in ≥95% of samples). Sequence data: PRJNA882252. Statistical analyses: - Alpha diversity: Shannon, Pielou’s evenness; Kruskal–Wallis tests (P<0.05). Linear mixed models with age group and sex covariates used to validate significant features; multiple testing via Benjamini–Hochberg where applicable. - Beta diversity: PERMANOVA (999 permutations) on Bray-Curtis, Jaccard, weighted and unweighted UniFrac; baseline vs endline Bray-Curtis with subject stratification (adonis2). - Differential taxa: ASVs with <5% prevalence removed; LEfSe (Wilcoxon P<0.05; LDA>2) to identify biomarkers across metadata groups, followed by linear mixed models adjusting for age and sex with BH correction (q<0.1) for robustness. - Longitudinal intervention effects: QIIME2 q2-longitudinal for pairwise differences (alpha) and distances (beta) between baseline and endline within groups; Mann–Whitney U with BH FDR. - Multivariable associations: MaAsLin2 linear mixed models at genus level (genera present in ≥10% of samples; TSS transformation; fixed effects: group and time; random effect: subject; q<0.25). - Functional prediction: PICRUSt2 to infer MetaCyc/KEGG functions (placement via EPA-NG/gappa; hidden state prediction via castor; pathways via MinPath). Group comparisons via STAMP v2.1.3 (Welch’s t-test; BH FDR<0.05).

- Cohort and baseline status (n=380; mean age 9.7 years, 53% boys): anaemia 20%, stunting 45%, iron deficiency 51%, zinc deficiency 89%, vitamin A deficiency <8%, systemic inflammation 37%, gut inflammation 3%, parasite infection 27%. - Microbiota composition: Dominated by Firmicutes, Bacteroidota, Proteobacteria, and Actinobacteriota; notably high Lactobacillaceae (Bacilli, Lactobacillales, Lactobacillus). Core genera (≥95% prevalence) limited to five genera. Total reads post-filter: 29,910,230 across 760 samples. - Diversity associations at baseline: • Alpha diversity: Older children (10–14y) had higher Shannon and Pielou’s indices than younger (6–9y). Lower diversity/evenness in anaemic, iron deficiency anaemia, and vitamin A-deficient children. Zinc deficiency, stunting, underweight, haemoglobinopathy, systemic/gut inflammation, and parasites not associated with alpha diversity. • Beta diversity: Significant differences by age (Bray-Curtis, Jaccard, Unweighted UniFrac), iron deficiency (Bray-Curtis, Jaccard), vitamin A deficiency (Bray-Curtis, weighted UniFrac). Sex differences observed on Jaccard. - Differential taxa (LEfSe; robust after age/sex adjustment): • Anaemia: enriched Ruminococcus torques group, Anaerostipes, Prevotella; non-anaemia: Limosilactobacillus, Erysipelatoclostridiaceae, Klebsiella. • Iron deficiency anaemia: enriched Prevotella; non-IDA: Monoglobales/Oscillospirales; both non-anaemic and non-IDA associated with Clostridia, Lachnospiraceae, Faecalibacterium, Eubacterium hallii. • Vitamin A deficiency: Bacilli (Lactobacillaceae), Gordonibacter, Romboutsia; sufficient vitamin A: Desulfovibrionia, Oscillospirales, Lachnospirales. • Stunting: Prevotella 7, Holdemanella; non-stunted: Clostridia, Marinifilaceae, Succinivibrionaceae, Odoribacter, Alistipes, Raoultibacter. • Systemic inflammation: Firmicutes enriched; no specific gut inflammation associations (low prevalence). • Parasites: Desulfovibrionia, Hungateiclostridiaceae, Erysipelotrichaceae, Weissella, Sarcina; parasite-free: Kurthia. • Age: Younger associated with Bacilli/Lactobacillus, Butyricoccaceae, Anaerococcus, Agathobacter; older with Fusobacteriota, Negativicutes, Oscillospirales. Sex differences also noted (girls: Clostridiales, Staphylococcales; boys: Actinobacteria, Desulfobacterota, Oscillospiraceae, Eubacterium, Coprococcus, Anaerovoraceae, Ligilactobacillus). - Intervention effects on diversity (q2-longitudinal, baseline vs endline): Significant within- and between-group changes. Multiple significant pairwise distances across Bray-Curtis, Jaccard, unweighted and weighted UniFrac; alpha diversity pairwise differences significant for some group comparisons (e.g., Pielou’s and Faith PD between Placebo and UR-original; UR-improved vs UR-original). - Intervention effects on taxa (MaAsLin2 vs Placebo): • UR-original (n=114): 42 genus associations (22 negative, 20 positive). • UR-improved (n=122): 94 genus associations (17 negative, 77 positive). • In total, 136 genera differentially affected; Lachnospiraceae members were notably sensitive and also associated with non-deficient vitamin A/iron status at baseline. - Predicted functions (PICRUSt2/STAMP): Differences by status were small in magnitude (mean proportion differences ≤0.18%). Non-anaemic and non-IDA children had more amino acid metabolism functions; vitamin A deficiency associated with reduced abundance of three pathways; zinc deficiency linked to increased formaldehyde oxidation pathway. Over time, stronger functional shifts: Placebo showed increases in fatty acid biosynthesis; endline fatty acid metabolism functions more frequent in Placebo and UR-original; stearate biosynthesis higher in UR-original than UR-improved.

This large, longitudinal randomized trial in Cambodian schoolchildren shows that gut microbiota composition is characterized by an unexpectedly high abundance of Lactobacillaceae and is associated with specific aspects of nutritional and inflammatory status. Diversity and composition differed by age and sex, underscoring their roles as confounders. Anaemia and vitamin A deficiency were linked to lower alpha diversity and distinctive taxa, whereas zinc deficiency showed no compositional associations, possibly due to the very high prevalence of deficiency or limited discriminatory power. Stunting and parasitic infections were associated with specific bacterial groups, aligning with literature on helminth–microbiota interactions. Both fortified rice interventions altered microbiota structure over six months, with UR-improved (broader micronutrient premix including vitamin A) and UR-original (higher Fe/Zn) producing distinct compositional patterns. Lachnospiraceae emerged as sensitive to both nutritional status and intervention, suggesting potential as biomarkers of micronutrient adequacy. Predicted functional analyses indicated modest baseline differences by status but more pronounced longitudinal shifts dominated by lipid metabolism pathways, particularly fatty acid biosynthesis, rather than clear signatures of iron, zinc, or vitamin A handling. Overall, findings support the hypothesis that micronutrient status and multi-micronutrient fortification are linked to microbiota composition in school-aged children, providing potential avenues for microbiota-informed nutritional strategies in low-resource settings.

The study provides a detailed characterization of faecal microbiota in Cambodian schoolchildren and demonstrates associations with micronutrient deficiencies (iron, vitamin A), stunting, inflammation, and parasitic infections. Six months of school-based consumption of fortified rice significantly altered microbiota diversity and composition, with distinct effects between UR-original and UR-improved formulations, and functional shifts mainly in lipid metabolism pathways. The prominent presence of Lactobacillaceae and sensitivity of Lachnospiraceae to nutritional status and intervention suggest potential microbial biomarkers of micronutrient adequacy. Future directions include validating these findings with shotgun metagenomics and metabolomics to directly assess functional capacity, disentangling micronutrient-specific effects using single-nutrient or factorial designs, evaluating clinical correlations (e.g., diarrhoeal morbidity), and investigating helminth–microbiota interactions, particularly hookworms, in relation to nutritional outcomes.

- Attribution of effects to specific micronutrients is difficult because the two fortified rice formulations differed in both composition and amounts (e.g., UR-original higher Fe/Zn; UR-improved added vitamin A, B3, B12 but less Fe). - Functional inference relied on 16S rRNA predictive tools (PICRUSt2) rather than shotgun metagenomics, limiting resolution and accuracy of functional conclusions. - No positive control was included in sequencing runs; only forward reads used due to low overlap, potentially reducing taxonomic resolution. - Low prevalence of gut inflammation (~3%) likely limited power to detect associations with gastrointestinal inflammation. - Morbidity data (e.g., diarrhoeal episodes) were not collected alongside interventions, limiting interpretation of potential clinical effects of micronutrient-related microbiota changes. - Study conducted in one province/ethno-geographic region, which may limit generalizability, although it reduces confounding by geography and culture.