2026-10-14T05:34:33Z https://api.figshare.com/v2/oai

oai:figshare.com:article/32640738 2026-10-01T16:28:36Z portal_1172 item_type_8 month_year_10_2026

Molecular risk stratification in early cancers detected during bowel cancer screening Ryan Byrne (11513863) PUREID: 629006973 Colorectal Cancer Lymph Node Metastasis Biomarker Gene Expression Profiling Gene Expression Signatures Colorectal Neoplasms Computational Biology Bioinformatics KRAS Early Detection of Cancer Precision Medicine Bowel cancer screening has led to more colorectal cancers (CRCs) being detected earlier, including at the T1 stage, when they have only invaded as far as the submucosal layer of the bowel wall. This limited invasion makes T1 CRCs amenable to endoscopic resection, which carries less morbidity and mortality than formal surgical resection. In the absence of lymph node metastasis (LNM), endoscopic resection alone is sufficient treatment. Conversely, the presence of LNM warrants formal surgical resection to remove the affected portion of bowel and lymph nodes. However, there is no reliable way to determine LNM status preoperatively. Current approaches use histological features to stratify cases into high-risk and low-risk groups, with high-risk patients treated with surgical resection. However, many patients are stratified as high-risk, only to be found to have no LNM when surgery is performed, meaning they are unnecessarily exposed to surgical risks. To address this, this project aimed to identify molecular features associated with LNM status in T1 CRC, offering a potential avenue for more accurate risk stratification. As this project aimed to employ gene signatures, first, some challenges associated with gene signature application were explored, with commonly used gene signatures found to be confounded by tumour composition (Chapter 3) and RAS-related gene signatures found to have limited association with KRAS mutational status in established human CRC (Chapter 4). Then, using a cohort of T1 CRCs (n=230), the existing histopathology-based approach to T1 risk stratification was shown to lack specificity (Chapter 5). Finally, transcriptional profiling and in silico analysis of the same T1 cohort found no transcriptional features or signatures that could reliably discriminate between LNM-positive and LNM-negative T1 CRC (Chapter 6). In summary, this work highlights the potential for misinterpretation of gene signatures and suggests that transcriptional features cannot reliably discriminate between LNM-positive and LNM-negative T1 CRC. Thesis is embargoed until 31 July 2030. 2026-10-01T16:28:36Z Text Thesis 10.17034/32640738.v1 https://figshare.com/articles/thesis/Molecular_risk_stratification_in_early_cancers_detected_during_bowel_cancer_screening/32640738 All Rights Reserved Open Access after 2030-07-31