Gut microbiota profiling for risk stratification of surgical intervention in preterm infants with necrotizing enterocolitis: a retrospective cohort study.

内容摘要

To investigate whether gut microbiota characteristics are associated with surgical intervention status in preterm infants with necrotizing enterocolitis (NEC) and to evaluate their potential discriminatory value for surgical risk stratification. This retrospective study included 56 preterm infants with NEC admitted to Northwest Women's and Children's Hospital between May 2020 and May 2023, including 33 managed non-surgically and 23 who underwent surgery, as well as 30 preterm infants without NEC as controls. Blood samples were collected to measure prostaglandin E2 (PGE2), interleukin-6 (IL-6), interleukin-10 (IL-10), tumor necrosis factor-&#x3b1; (TNF-&#x3b1;), and C-reactive protein (CRP). Fecal samples were subjected to 16S rRNA gene sequencing to assess microbial diversity and taxonomic composition. Sequencing quality was evaluated using rarefaction analysis, &#x3b2;-diversity differences were tested by PERMANOVA, and differential taxa were further reanalyzed using CLR transformation with FDR correction. Associations between gut microbiota indicators and surgical intervention were analyzed using multivariable logistic regression adjusted for clinically relevant covariates, including gestational age, birth weight, postnatal age at NEC diagnosis, Bell stage, antibiotic exposure, feeding type, length of hospital stay, and probiotic use. Receiver operating characteristic (ROC) curves were used to evaluate discriminatory performance for distinguishing surgical NEC from non-surgical NEC within this retrospective cohort. No significant differences were observed among the three groups in sex, mode of delivery, 1-min Apgar score, or maternal complications (all P > 0.05). In contrast, gestational age was lower in the NEC groups than in controls, and infants in the surgical group had a younger postnatal age at NEC diagnosis, a higher proportion of Bell stage III disease, and more frequent exposure to prolonged antibiotic treatment and formula-predominant feeding. The proportion of meconium-stained amniotic fluid was also significantly higher in the surgical group (47.8% vs. 18.2% and 6.7%, P = 0.001). Compared with the non-surgical and control groups, the surgical group showed significantly higher levels of IL-6, IL-10, CRP, and TNF-&#x3b1; (all P < 0.001), along with lower microbial diversity as indicated by reduced Chao and Shannon indices (both P < 0.001). At the taxonomic level, the surgical group exhibited lower relative abundances of Firmicutes, &#x3b3;-Proteobacteria, Bifidobacterium, and Lactobacillus, but higher relative abundances of Proteobacteria, Salmonella, and Clostridium (all P < 0.001). In multivariable analysis adjusted for gestational age, birth weight, postnatal age at diagnosis, Bell stage, antibiotic exposure, feeding type, length of hospital stay, and probiotic use, Bifidobacterium (adjusted OR 0.63, 95% CI 0.44-0.89), Chao index (adjusted OR 0.72, 95% CI 0.56-0.93), and Shannon index (adjusted OR 0.68, 95% CI 0.49-0.94) remained inversely associated with surgical intervention, whereas Proteobacteria (adjusted OR 1.34, 95% CI 1.08-1.67) and Clostridium (adjusted OR 1.51, 95% CI 1.12-2.04) remained positively associated with surgery (all P < 0.05). The combined model incorporating Firmicutes, Proteobacteria, &#x3b3;-Proteobacteria, and Clostridium achieved an AUC of 0.904 (95% CI 0.823-0.985), with 78.62% sensitivity and 88.54% specificity. Gut microbiota dysbiosis in preterm infants with NEC was associated with surgical intervention status, and part of these associations remained significant after adjustment for key clinical covariates. A combined microbiota-based model showed potential discriminatory value for surgical risk stratification within this retrospective cohort; however, because groups were defined according to final treatment outcome, these findings should not be interpreted as evidence of true prospective prediction or guidance of surgical timing. These findings require confirmation in larger prospective multicenter studies.