A Novel Radiomic Signature for Prediction of Lymph Node Metastasis in T1 Colorectal Cancer: A Multicenter Study

Background: To develop and validate a radiomic signature for the prediction of lymph node metastasis (LNM) in T1 colorectal cancer (CRC) patients by utilizing common, non-invasive, clinical imaging data. With this signature, patients with T1 CRC at a lower-risk of LNM may avoid extensive radical surgical treatments required in high-risk cases. Methods: This retrospective analysis included 559 T1 CRC patients enrolled at three hospitals. Key radiomic features were extracted from CT images and T2w-MR images using a coarse-to-fine feature selection strategy. These features were used to generate three radiomic signatures based on CT, MR and combined (CT&MR) imaging data. The performance of these radiomic signatures were compared within the primary and external validation cohorts. Furthermore, their prediction performance and clinical usefulness were measured and compared with those of a clinical information-based model. Findings: The radiomic signature combining CT and MR images achieved AUCs of 0.872 (95% CI, 0.770-0.974) in the primary cohort and 0.812 (95% CI, 0.709-0.915) in validation cohort 1, which were significantly better than those of other prediction models. More importantly, the signature achieved high negative predictive values (95.5% in the primary cohort and 97.9% in the validation cohort 1). The decision curve analysis demonstrated the clinical usefulness of this signature. Interpretation: The study suggests that the radiomic signature can accurately predict LNM in cases of T1 CRC. Furthermore, combining CT and MR images can achieve better performance. Specifically, the radiomic signature can potentially reduce overtreatment in patients who were misdiagnosed as LNM+. Funding: This work was supported by grants from the National Natural Science Foundation of China (Nos. 81922040, 81920108026, 81871964, 81930053, 81227901, and 81527805), the National Ten Thousand Plan Young Top Talents (for Dr. Yanlei Ma), the Shanghai Young Top Talents (for Dr. Yanlei Ma. No. QNBJ1701), the Shanghai Science and Technology Development Fund (No.19410713300), CSCO-Roche Tumor Research Fund (No. Y-2019Roche-079), the Fudan Outstanding Young Talent Training Plan (No.YJYQ201601), and the Shanghai Pujiang Program (No.17PJD007), the Beijing Natural Science Foundation under Grant No. 7182109, the National Key Research and Development Plan of China under Grant Nos. 2017YFA0205200, the Youth Innovation Promotion Association CAS (grant number 2019136). Declaration of Interest: The authors indicated no potential conflicts of interest. Ethical Approval: The study protocol complied with the Declaration of Helsinki and was approved by the ethics committee of each participating hospital.