Abstract
OBJECTIVES: To compare the value of tumor stroma ratio (TSR) and radiomic signature from baseline MRI for stratifying the risk of distant metastases (DM) in patients with locally advanced rectal cancer (LARC). MATERIALS AND METHODS: Data from 302 patients with LARC who underwent neoadjuvant chemoradiotherapy and total mesorectal excision in our hospital between 2015 and 2018 were retrospectively reviewed, and the patients were randomly allocated into the training and validation cohorts in a ratio of 7:3. Patients were followed-up for more than 3 years postoperatively with metachronous DM as the endpoint. Independent risk factors for DM-free survival (DMFS) were analyzed using Cox regression. The TSR of endoscopic biopsy specimens was scored automatically. Totally 1229 radiomic features of each tumor were extracted from baseline MRI, and the Radscore was calculated. RESULTS: The median follow-up time was 54.3 (51.6-57.1) months, and the 3-year DMFS was 83.8%. The best cutoff value of the TSR to distinguish a patient's DM risk was 0.477 (Sen=70.8%, Sep=78%, P <0.001). Increased TSR (HR=3.072, P =0.006) and Radscore (HR=719.231, P =0.023), advanced MR-evaluated T stage (HR=2.660, P =0.023) and ypN (HR=2.362, P =0.028) stage were independent risk factors for DMFS. The area under the curve of the combined model was significantly higher than that of the radiomic model ( P =0.013) but without a significant advantage over the TSR model ( P =0.086). CONCLUSION: TSR of colonoscopic biopsies can independently stratify DM risk in patients with LARC. The TSR model is the most convenient and efficient method for DM risk stratification in LARC.