Predictive biomarkers in the gut microbiome and metabolome for severe acute radiation enteritis in cervical cancer radiotherapy.

Introduction: Radiation enteritis (RE), a common side effect, is a growing health concern, particularly the severe acute form of RE (SARE), among cervical cancer patients undergoing radiation therapy. Currently, there is no noninvasive diagnostic method for SARE. This study aimed to identify gut microbiomics- and metabolomics-based signatures, and assess their predictive value for SARE.

Methods: Samples from 50 cervical cancer patients receiving volumetric modulated arc therapy (VMAT) were collected for gut microbiota and metabolomic profiling. 16 S rDNA amplicon sequencing analyzed gut microbiota, and nontargeted liquid chromatography-mass spectrometry determined metabolomic profiles. Multivariate and pathway analyses identify independent metabolites associated with SARE. A predictive nomogram for SARE, combining multi-omics-based signatures and clinical characteristics, was constructed and evaluated using the area under the receiver operating characteristic curve (AUC) and calibration curve.

Results: Fecal microbiome analysis showed characteristic alterations in SARE, mainly including Faecalibacterium enterotype-3, Escherichia, and Shigella enterotype-2. Metabolomic analyses identified a panel of molecules significantly associated with SARE. Furthermore, an intuitive nomogram consisting of these multi-omics signatures (serum COX-2 and fecal phenylethylamine), combined with clinical characteristics with predictive value, was constructed to predict SARE. Notably, the evaluation of model performance suggested an excellent predictive discrimination for SARE [AUC, 0.975; 95% confidence interval (CI), 0.953-0.998]. Calibration curve analysis showed an adequate calibration for the model and good consistency between the predicted SARE cases with this newly developed model and the actual SARE cases.

Conclusion: This study identified noninvasive signatures, including COX-2 and phenylethylamine, as promising predictive biomarkers for SARE and developed an intuitive nomogram with good predictive accuracy for SARE in cervical cancer patients.