MAPPING THE GUT MICROBIOME SIGNATURE ALONG THE COLORECTAL ADENOMA-CARCINOMA CONTINUUM

Introduction: Adenomas are major precursors of colorectal cancers (CRC). Individuals after adenoma resection remain at a higher CRC risk than those with no adenomas. The gut microbiome is associated with CRC, but its dynamics over time following adenoma resection remain unknown. We aimed to characterize the patterns of species-level genome bins (SGBs) in patients with adenoma removal or CRC compared to healthy subjects.

Methods: We analyzed gut metagenomes of 354 patients after adenoma resection (mean years between resection and stool collection=12) and 354 matched polyp-free individuals from the Micro-N, a microbiome cohort within Nurses’ Health Study II (adenoma dataset). We also included publicly available metagenomes from 882 CRC cases and 929 healthy individuals from 11 external datasets. We calculated the standardized mean difference (SMD) between cases and controls for each SGB within each dataset and aggregated the effect sizes from the CRC datasets using a random-effects meta-analysis. To identify microbes with similar trends in CRC and adenoma cases, we employed a genetic algorithm to solve an optimization problem designed to reduce the SGB number while increasing the accuracy for distinguishing adenoma cases from controls based on the CRC case-control comparison. We assessed the associations between diet/lifestyle and the identified SGBs in the adenoma dataset using Spearman’s correlations.

Results: The microbiome profile of adenoma and CRC cases compared to their corresponding controls was correlated (rho=0.29, p<0.0001). We identified 41 SGBs with similar SMD trends in CRC and adenoma, including Blautia spp. and R. torques. A classifier based on the abundances and SMD values of these 41 SGBs in the CRC datasets showed moderate internal discrimination using leave-one-out cross-validation (area under the curve (AUC)=0.64) and similar discrimination in the adenoma dataset (AUC=0.67). Similar accuracy was noted when adenoma dataset-derived classifier was applied to distinguish CRC (AUC=0.61) and adenoma (AUC=0.61) from controls. For adenomas, the discriminatory accuracy was similar regardless of the time interval between resection and stool collection. However, CRC biomarkers specific to late-stage CRC and enriched in mucosa (e.g., F. nucleatum and P. micra) were not detectable in adenoma cases. SGBs’ enrichment in adenoma cases was inversely correlated with healthy diet/lifestyle factors and positively with unhealthy factors, with stronger correlations in adenoma cases than controls.

Conclusions: Individuals after adenoma resection showed similar microbial changes as observed in CRC, which persisted years after resection, potentially reflecting their increased CRC risk. These microbial changes might reflect the influence of persistent unhealthy dietary and lifestyle behaviors, underscoring the importance of lifestyle modification after adenoma resection for CRC prevention.