Objective: Environmental enteric dysfunction (EED) can be assessed by the lactulose:mannitol (L:M) test. Our objective was to determine if selected host fecal transcripts were correlated with EED, and whether transcripts and clinical characteristics could be used to predict EED in rural African children.
Methods: Demographic and sanitation characteristics, along with L:M testing and host fecal transcript analyses from 798 asymptomatic Malawian children aged 12 to 61 months were compared with linear growth over the subsequent 3 months. Fecal host mRNA analysis included quantification of expression of 18 transcripts associated with L:M. Permeability was categorized as normal (L:M ≤ 0.15), moderate (0.15<L:M<0.45) and severe (L:M ≥ 0.45), and random forest predictive models were created.
Results: L:M was inversely correlated with linear growth over the subsequent 3 months (r = -0.32, P < 0.001) and severe EED was associated with stunting (P < 0.0001). Age younger than 24 months, weight-for-height z score <0, domesticated animals in the child's sleep environment, lack of a pit latrine combined with a potentially contaminated water source, and a recent history of diarrhea were associated with severe EED. A random forest model using CD53, HLA-DRA, MUC12, and TNF was 84% sensitive for severe EED and 83% sensitive for no EED.
Conclusions: Selected host fecal transcripts can be used in a random forest model as a noninvasive biomarker for categories of EED in rural African children.