+ Rectal cancer is the 8th most common neoplasia worldwide and accounts for 3.5% of cancer-related fatalities. The standard treatment for locally advanced rectal cancer (LARC) includes neoadjuvant therapy (radiotherapy with or without chemotherapy) followed by resection surgery. However, patients exhibit highly variable responses to neoadjuvant treatment, ranging from complete response to no tumour regression. Currently, there are no clinically validated approaches to predict response, and neoadjuvant treatment is administered to most patients without knowing whether it will provide meaningful benefit. Accurate predictive biomarkers would thus be very useful to stratify LARC patients, identifying those likely to achieve a complete response and those unlikely to respond, who may be better suited for immediate surgery, being spared from unnecessary toxicity and the risk of disease progression during ineffective treatment. Given this unmet need, we have analysed publicly available Affymetrix microarray data comprising pre-treatment biopsy gene expression profiles of 13 LARC patients, as well as matched clinical data, including response to neoadjuvant therapy. These biopsies were previously processed by laser capture microdissection to separate stroma and epithelium for tissue-specific analyses. Using Lasso regression, we were able to derive a response score based on the weighted expression of candidate genes in the tumour stroma. This score successfully separated LARC patients according to their response to neoadjuvant therapy, and a permutation analysis supported its robustness. These findings, however, are based solely on computational analyses of a third-party dataset. Experimental validation in an independent patient cohort is being planned to confirm their predictive value.
0 commit comments