Gut Virus Signatures Predict Necrotizing Enterocolitis Risk in Preterm Infants
A study published September 16, 2026, in Gut found that patterns in gut viruses and antibiotic-resistance genes predicted necrotizing enterocolitis (NEC) in preterm infants before symptoms appeared, correctly flagging early-onset cases with 81% sensitivity and late-onset cases with 83% accuracy.
Researchers analyzed 1825 stool samples collected over time from 43 preterm infants who later developed NEC and 86 gestational-age- and birthweight-matched preterm infants who did not, across three U.S. hospitals. Using metagenomic and metatranscriptomic sequencing plus machine learning, the team tracked gut phages (viruses that infect bacteria) and the resistome (the collection of antibiotic-resistance genes bacteria carry) in the weeks before disease onset. Cases were split into early-onset NEC (diagnosed within 40 days of birth) and late-onset NEC (diagnosed after 40 days) because the two forms showed different warning patterns. Early-onset NEC was predicted by phage-bacteria interaction signatures with 75% accuracy and 81% sensitivity. Late-onset NEC was predicted by antibiotic-resistance gene profiles with 83% accuracy.
NEC Prediction Accuracy: Early-Onset (75%) vs Late-Onset (83%)
The authors report that phages in NEC cases showed high DNA abundance but low gene activity, a pattern called a lysogenic lifestyle, meaning the viruses insert into bacterial DNA and stay dormant rather than multiplying and destroying their bacterial hosts. In this dormant state, the authors say, these phages carry metabolic genes that may help pathobionts, bacteria that can turn harmful under certain gut conditions, survive and persist. The findings suggest early and late NEC follow separate microbial pathways rather than one shared signature, one shaped by virus-bacteria interactions and the other by antibiotic-resistant bacteria.
The cohort included only 43 NEC cases, a small number for a disease with variable presentation, and the design is observational, so it identifies patterns linked to NEC without proving what causes the disease. The abstract does not list funding or conflict-of-interest information.
Data Panel
- Who
- 43 preterm infants who developed necrotizing enterocolitis (NEC) and 86 gestational-age- and birthweight-matched preterm infants who did not, across three U.S. hospitals
- Design
- Observational study using metagenomic and metatranscriptomic sequencing of 1825 longitudinal stool samples, analyzed with machine learning; cases split into early-onset (within 40 days of birth) and late-onset (after 40 days) NEC
- Dose
- Not applicable — no intervention was tested; the study analyzed naturally occurring stool samples collected over time
- Primary result
- Early-onset NEC predicted by phage-bacteria interaction signatures with 75% accuracy and 81% sensitivity; late-onset NEC predicted by antibiotic-resistance gene (resistome) profiles with 83% accuracy
- Secondary
- Phages in NEC cases showed high DNA abundance but low gene expression, a lysogenic pattern, and these phages carried metabolic genes that may enhance survival of pathobionts, bacteria that can turn harmful under certain gut conditions
- Funding / conflicts
- Not reported in the abstract
Dr. Axe's Take
This tells me the earliest weeks of a preemie's gut microbiome are more consequential than most parents realize, and that viruses living inside gut bacteria are not passive passengers. For a baby in the NICU, I would focus on what already has evidence behind it for lowering NEC risk: exclusive breast milk feeding, skin-to-skin kangaroo care, and minimizing unnecessary antibiotic courses when medically appropriate, always in partnership with the NICU team. I would never use a study like this to second-guess a doctor's antibiotic decisions for a preterm infant. What I want to see next is whether stool-based phage testing becomes a bedside tool in the first weeks of life.
— Dr. Axe