Background:
Colorectal cancer (CRC) develops over a long time, often arising from precursor lesions that can progress to malignancy and these steps are characterized by molecular alterations. Using small RNA-Sequencing (sRNASeq) we provide an overview of small non-coding RNAs (sncRNAs) landscape including miRNAs and less explored piRNAs, tRNAs, and snoRNAs in relation to cancer progression from precursor lesions, non-advanced adenomas (nAA) and advanced adenomas (AA) to late stages of CRC.
Objectives:
We aim to highlight the potential of stool-based sncRNA biomarkers as a non-invasive tool for early detection and CRC prevention, complementing standard screening programmes.
Methods:
sRNA-seq was performed on stool samples from subjects with nAA (n = 60), AA (n = 95) and CRC (n = 79), as well as colonoscopy-negative controls (n = 226). As the data were generated from two independent cohorts, a negative binomial regression was used to integrate the datasets and correct for cohort-specific batch effects. After following our pipeline (Pardini et al, Gastroneterology 2023), reads unmapped to the human miRNome were subsequently aligned to an in-house sncRNA reference, and differential expression analyses were conducted using DESeq2, followed by integrative analyses with mixOmics.
Results:
A total of 18,139 sncRNAs were identified, with piRNAs (n = 13,274), miRNAs (n = 2,630), and tRNAs (n = 1,095) representing the most abundant biotypes in stool samples. Prior to data integration, cohort-related batch effects explained on average 5% of the variance in miRNA expression and 6% in other sncRNAs. Following batch-correction, this contribution was reduced to 1% and 0.05%, respectively. Globally, 1,499 sncRNAs were found to be differentially expressed across lesion types, with a progressive increase in the number of dysregulated features as the disease progressed. These included 479 miRNAs, 376 piRNAs, 232 tRNAs, and 76 snoRNAs. Notably, 95 miRNAs displayed lesion-specific alterations, together with 267 other sncRNAs (29 tRNAs, 44 snoRNAs, and 142 piRNAs). Interestingly, snoRNAs were the only biotype exclusively altered in stool samples from CRC patients, with no significant changes observed in nAA or AA. Sparse PCA performed on the complete set of DE sncRNAs revealed a separation of patients with pre-cancerous lesions and those with CRC, primarily driven by changes in snoRNA and miRNAs expression levels. This clustering pattern was consistently observed using alternative clustering approaches.
Conclusions:
We showed that stool-derived sncRNA profiles could reflect the progressive molecular changes underlying CRC progression, from precursor lesions to the later stages. In particular, distinct sncRNA signatures could help in the discrimination between individuals with nAA or AA and those with cancer. Overall, our findings further highlight the potential of stool-based sncRNAs signatures as non-invasive complementary tool for CRC prevention and early detection.