Key points are not available for this paper at this time.
Abstract Background: For resectable gastric cancer, postoperative adjuvant chemotherapy remains the standard modality for preventing recurrence and improving survival. Although a previous study proposed a model for stratifying risk and determining the likelihood of chemotherapeutic benefit following surgery, this model requires molecular genomic analysis of tumour tissues and only uses unidimensional data. However, a recent study revealed that a multiomics approach reflects more compressive cancer characteristics, providing more precise prediction of treatment response. Recent advancements indicate that radiomics features are predictive biomarkers of treatment response. The present study investigated whether computed tomography (CT) image-based quantitative radiomics combined with genomics data can provide clinically actionable information for a more detailed stratification of cancer patients. Methods: This retrospective multicohort study included patients with advanced gastric cancer treated with surgery plus adjuvant chemotherapy or surgery alone. Two filters were used to extract 1,106 radiomics features from preoperative venous-phase CT images. Dimensional reduction was performed via Cox regression with the LASSO method, following which the coefficients were used to determine the Rad-score. The optimal cutoff point was determined via the Kaplan-Meier method and used to divide the cohort into high- and low-risk groups. The primary endpoints were the predictive and prognostic performance of the Rad-score and nomogram in terms of overall survival and the value of adding the Rad-score to genetic information. Findings: The optimal Rad-score (11·978379207) successfully stratified patients into low- and high risk groups. Improved survival was noted among patients defined to have a ‘poor prognosis’ with GZMB–, WARS–, and SFRP+ genetic test results and among high-risk patients based on the Rad-score when no adjuvant chemotherapy was provided. Interpretation: Our findings suggest that the Rad-score can be used to further stratify patients in the response group, preventing potentially ineffective treatment and unnecessary adverse events in those with advanced gastric cancer
Kim et al. (Fri,) studied this question.