Fecal microbiome signatures for early detection of colorectal cancer and precursor lesions

Background: Colorectal cancer (CRC) remains a major global health burden, and the development of non-invasive biomarkers for early detection is a critical unmet need. Growing evidence indicates that intestinal microbiome dysbiosis contributes to the adenoma-carcinoma sequence and may support CRC risk stratification. Aim: To investigate whether intestinal microbiome profiling from residual material of Fecal Immunochemical Test (FIT) samples can serve as a biomarker for CRC screening. Methods: Residual stool samples from Brazilian FIT-positive individuals (n=133) were classified according to colonoscopy as no lesions (N=30), non-advanced adenoma (NAA=46), advanced adenoma (AA=13), or colorectal cancer (CA=44). In three individuals, the microbiome profile was compared on three collection tubes (two FIT tubes and Omnigut). Bacterial DNA from FIT samples underwent 16S rRNA sequencing on Oxford Nanopore MinION (R10.4.1 flow cells, for 48 hours). Reads were processed using Dorado, aligned with Emu to a curated full-length 16S database, and analyzed in QIIME2 using Python. Alpha diversity (25 different metrics), beta diversity (Jaccard, Bray-Curtis), and differential abundance (ANCOMBC) were assessed. Results: Alpha diversity was significantly higher in CA samples compared with N and NAA patients (Chao 1, p < 0.001 and p=0.006; Observed features, p < 0.001 and p=0.007 ), with no difference compared to AA. Beta diversity (PERMANOVA, p = 0.001 for both metrics) showed significant compositional separation among CA and AA (Jaccard, p=0.001 , Bray Curtis, p=0.001 ), NAA (Jaccard, p=0.001 , Bray Curtis, p=0.001 ), and N (Jaccard, p=0.001 , Bray Curtis, p=0.003 ). Differential abundance indicated increasing dysbiosis as lesions progressed. Compared with N samples, NAA exhibited 42 differentially abundant species (9 decreased, 33 increased), AA showed 65 species (15 decreased, 50 increased), and CRC showed pronounced dysbiosis with 179 species (28 decreased, 151 increased). Preliminary inspection highlights enrichment of Fusobacterium nucleatum in CA samples. Analysis of the taxonomic profiles revealed no differences in species composition among the three collection tubes analyzed. Conclusion: Microbiome profiling using routine residual FIT samples is feasible and reveals a marked diversity shift along the adenoma-carcinoma spectrum. These findings support the potential of FIT-derived microbiome signatures as non-invasive biomarkers for CRC early detection in Brazilian screening populations.