The Gut Microbiome Profile Predicts the Severe Acute Toxicities in Patients With Locally Advanced Rectal Cancer Undergoing Neoadjuvant Chemoradiation.

Purpose/Objective(s) Occurrence of severe acute toxicities imposes a major impediment to the completion of neoadjuvant chemoradiation (nCRT) in patients with locally advanced rectal cancer (LARC). Predicting the acute toxicities before treatment initiation is an unmet need. Given the role of the gut microbiome in inflammation and immunomodulation, this study aims to develop a microbiome-based classifier to predict the severe acute toxicities in patients with LARC undergoing nCRT. Materials/Methods 167 stool specimens were collected before and after nCRT from 84 patients with LARC, followed by 16S rRNA gene sequencing. Gastrointestinal toxicities (GIT) and myelosuppression (MS) within 90 days of treatment initiation were recorded and graded according to the Common Terminology Criteria for Adverse Events (CTCAE v5). Based on comparison analysis, two microbiome-based classifiers were established in a training cohort of 37 baseline specimens and tested in a validation cohort of 47 baseline specimens, for predicting severe GIT and severe MS respectively. Results In the baseline specimens, an inverse correlation was observed between the microbial diversity and the severity of MS (Chao1 index, P = 0.043). Comparison analysis in baseline specimens revealed a significant overrepresentation of Akkermansia muciniphila and Coprococcus eutactus, the latter of which is a butyrate producer, in patients with Grade 0-1 GIT and in patients with Grade 0-1 MS; and an overrepresentation of several  β-glucuronidase-producing microbes including Escherichia fergusonii, which is also a pathogen causing diarrhea and bacteremia, in patients with Grade 2+ GIT and in patients with Grade 2+ MS. Beneficial commensal Bifidobacterium adolescentis was also found to be significantly enriched in patients with Grade 0-1 MS. Following nCRT, a significant decrease in the relative abundance of the butyrate-producing microbe, Faecalibacterium prausnitzii; and a significant increase in the relative abundance of the β-glucuronidase-producing microbe, Bacteroides fragilis were observed in patients with Grade 2+ GIT and in patients with Grade 2+ MS. Moreover, a significant decrease in the relative abundance of Fusobacterium was also noticed in patients with Grade 0-1 GIT and in patients with Grade 0-1 MS. With random forest analysis, a classifier for predicting severe GIT (Grade 3 or higher) was constructed, reaching an area under the curve (AUC) value of 85.71% [95% confidence interval (CI): 73.19% - 98.24%] in the training cohort and 75.58% (95% CI: 58.15 – 93.02%) in the validation cohort. The random forest classifier for predicting severe MS (Grade 2 or higher) showed a more robust performance, with the AUC value of 88.57% (95% CI: 74.67% - 100.0%) in training cohort and 80.62% (95% CI: 67.72% - 93.53%) in validation cohort. Conclusion This study highlighted the potential value of the gut microbiome profile in patients with LARC to predict the severe GIT and severe MS of nCRT. Further validation with a larger sample size is needed.