Gut Microbes and Metabolites Shift in Real Time During Rectal Cancer Chemoradiotherapy

When a patient with rectal cancer begins weeks of concurrent chemoradiotherapy, the treatment is aimed at a tumor deep in the pelvis. Yet a new study suggests the blast of radiation and daily capecitabine tablets also reshapes an ecosystem that was never on the treatment plan: the trillions of microbes living in the gut. In one of the most finely sampled looks at this phenomenon to date, researchers in Beijing tracked the fecal microbiome and metabolome of rectal cancer patients at nine time points spanning the entire course of therapy, capturing in real time which bacterial species rose, which fell, and which chemical signatures of the gut environment changed alongside them.

The study, published in Frontiers in Microbiology, enrolled 19 patients with histologically confirmed rectal adenocarcinoma at the First Medical Center of the Chinese PLA General Hospital between August 2022 and August 2023. Seventeen patients ultimately entered the longitudinal analysis after two were excluded because a fecal sample could not be collected within a prespecified window spanning three radiotherapy fractions. The researchers collected stool specimens before the first dose of capecitabine and before the start of radiotherapy, then at intervals after every three fractions through the full course of treatment, generating a dense temporal map of the gut ecosystem under therapeutic stress.

Technically, the team combined full-length 16S rRNA gene sequencing on a PacBio Sequel IIe system with untargeted liquid chromatography-mass spectrometry metabolomics. The long-read approach allowed species-level taxonomic resolution that short-amplicon sequencing often cannot achieve, while the metabolomic platform captured nearly 12,000 features that were filtered down to 2,873 for statistical modeling. To identify genuine temporal trends rather than noise, the researchers fitted negative-binomial generalized linear mixed models for microbial counts and linear mixed-effects models for metabolite intensities, with each patient included as a random intercept to account for repeated measures within individuals. All P values were corrected using the Benjamini-Hochberg false discovery rate, a conservative approach that guards against mistaking statistical flukes for real biological signals.

The results were striking in their breadth. Nineteen genera and 27 of the 176 analyzed species showed significant linear time trends during treatment. Among the decliners were three species with well-documented links to colorectal cancer: Parvimonas micra, Peptostreptococcus stomatis, and Gemella morbillorum. All three showed steep negative trajectories, with P. stomatis falling at an estimated rate of -0.641 per interval and G. morbillorum at -0.589, both with false discovery rates well below 0.001. On the rising side were Lachnoclostridium pacaense, Blautia hydrogenotrophica, [Eubacterium] sulci, and Coprococcus eutactus, species previously associated in the literature with fermentative and short-chain fatty acid-related metabolic capacity.

At the metabolite level, the picture was equally dynamic. Of the 2,873 features analyzed, 393 changed significantly over the course of treatment, with 246 decreasing and 147 increasing. Among the most prominent decliners were features annotated as glucosyringic acid, 4-hydroxyestrone-2-S-glutathione, and several phospholipids including PS (18:2/14:0) and PC (LTE4/15:0). Rising features included those annotated as lumirubin, a bilirubin degradation product, and chenodeoxycholic acid, a primary bile acid. The researchers were careful to note that these are putative annotations pending confirmation with authentic standards, but the patterns suggest that the fecal chemical environment undergoes substantial remodeling during pelvic chemoradiotherapy.

One of the most intriguing findings emerged from the integration of the two data types. In a two-stage analysis, the team first used Spearman correlation to identify 2,527 taxon-metabolite pairs, then selected the 50 with the strongest correlations and tested them in mixed-effects models that included time as a covariate. Seventeen of those 50 pairs survived false discovery rate correction. Notably, Coprococcus eutactus was positively associated with a feature annotated as stercobilin, another bilirubin-derived compound, while the genus Escherichia showed the opposite relationship with the same feature. Ruminococcus gnavus was negatively associated with a feature annotated as prostaglandin F3alpha. These associations generate experimentally testable hypotheses about which gut organisms might be linked to which metabolic products during cancer therapy, though the authors emphasize that statistical association does not demonstrate that a given microbe produces or regulates the corresponding metabolite.

The study also included a descriptive comparison with 20 healthy volunteers who provided a single fecal sample. Pretreatment patient samples had lower Shannon diversity at both the genus and species levels, and Bray-Curtis community profiles differed significantly between the two groups by analysis of similarities. Several taxa previously associated with colorectal cancer, including P. micra, P. stomatis, and G. morbillorum, were enriched in patients relative to controls. However, the researchers were transparent about the limitations of this comparison: the healthy volunteers were substantially younger on average, at 35 versus 57 years, and the groups were not matched for age or sex. The findings are therefore descriptive rather than evidence of disease-specific microbial signatures.

To strengthen the case-control findings, the team cross-referenced their results against GMrepo, a curated database of human gut microbiome data. Five species could be matched between the cohort and the database, and all five showed the same directional pattern: A. equolifaciens, P. stomatis, and G. morbillorum were higher in colorectal cancer, while H. parainfluenzae and P. plebeius were lower. When the researchers then examined the longitudinal trajectories of these five species during treatment, only P. stomatis and G. morbillorum showed significant declines. The other three showed no significant time trends, demonstrating that being associated with a disease and being responsive to its treatment are not the same property.

The authors are careful about what their data can and cannot support. Because capecitabine and radiotherapy were administered together, the separate contributions of each modality cannot be disentangled. Time was modeled as a linear variable, which could miss early peaks or late plateaus. The 16S rRNA approach yields relative abundances rather than absolute bacterial loads, so a declining relative count does not necessarily mean fewer bacteria in absolute terms. The metabolite annotations remain putative without targeted validation, and the pathway-level analysis, which flagged primary bile acid biosynthesis and glycerophospholipid metabolism at nominal thresholds, did not survive multiple-testing correction. No clinical endpoints such as tumor regression or treatment toxicity were linked to the microbial or metabolic trajectories.

Despite these caveats, the dense sampling design represents a methodological advance. Most prior studies of the gut microbiome in rectal cancer treatment have relied on one or two time points, making it impossible to distinguish transient fluctuations from sustained trends. By collecting samples at nine prespecified intervals anchored to radiotherapy fractions, this study provides a template for how future investigations might capture the temporal dynamics of the gut ecosystem during cancer therapy. The researchers call for multicenter prospective studies with absolute microbial quantification, shotgun metagenomics, targeted metabolite measurement, and prespecified clinical endpoints to determine whether any of the observed signals can eventually serve as biomarkers of treatment response or toxicity. For now, the study stands as a detailed exploratory map of what happens to the gut’s microbial and chemical landscape when the full force of chemoradiotherapy is directed at a rectal tumor.

Subject of Research: Longitudinal gut microbiome and metabolome changes during chemoradiotherapy for rectal cancer

Article Title: Longitudinal integrated analysis of gut microbiota and metabolites in rectal cancer patients undergoing chemoradiotherapy

Article References: Longitudinal integrated analysis of gut microbiota and metabolites in rectal cancer patients undergoing chemoradiotherapy. (n.d.). https://doi.org/10.3389/fmicb.2026.1934703

Image Credits: AI Generated

DOI: 10.3389/fmicb.2026.1934703

Keywords: gut microbiome, rectal cancer, chemoradiotherapy, metabolomics, 16S rRNA sequencing, capecitabine, colorectal cancer, microbiota, bile acids, stercobilin, longitudinal study, Frontiers in Microbiology