Abstract
BACKGROUND AND AIMS: Inflammatory bowel disease (IBD) is associated with a higher risk of developing colorectal cancer, according to the inflammation-dysplasia-cancer (IDC) sequence from inflammation to colitis-associated colorectal cancer (CAC). The objective of this study was to identify and generate a transcriptomic signature and score, related to the IDC sequence, that could ultimately classify dysplasia and cancer in IBD. METHODS: Demographics, clinical parameters, histological characteristics, and RNA-sequencing data were evaluated on 134 formalin-fixed paraffin-embedded lesions from 2 independent cohorts of IBD patients with low- or high-grade dysplasia (LGD, HGD) and/or CAC. An ordinal logistic regression screened for significant IDC sequence-associated genes that were computed in a transcriptomic signature score. RESULTS: Principal component analysis and unsupervised clustering on 1% of the most variable genes showed a good clustering between the 4 lesion groups (Normal Mucosa, Inflamed Mucosa, LGD/HGD, and CAC). A gene signature was identified on 27 genes that correlated with the lesion groups in the exploratory cohort. The most weighted gene in this transcriptomic signature was the long non-coding regulatory RNA KCNQ1OT1, a gatekeeper against genomic instability and transposon activation. Based on the expression of these 27 genes, we built and validated a transcriptomic signature score to classify dysplasia and CAC. The overall accuracy of the transcriptomic signature score was 85.71% in the exploratory cohort and 90.91% in the validation cohort. CONCLUSION: We identified a tissue-based transcriptomic score to classify IDC lesions in IBD patients and uncovered some of the pivotal genes in carcinogenesis related to inflammation in IBD.