BACKGROUND: Lymph node metastasis is important for the management and surgical procedures of patients with colorectal cancer. Preoperative identification of D3-metastasis could help evaluate the necessity of D3 lymphadenectomy. No adequate noninvasive method has been developed to detect the status of lymph nodes. OBJECTIVE: To establish a ctDNA methylation-based method to preoperatively identify lymph node metastasis and D3-metastasis in colorectal cancer. DESIGN: This is a prospective and paired diagnostic study. After an analysis of the genome-wide DNA methylation landscape, differential biomarkers were selected. A ctDNA methylation-based model for identifying lymph node metastasis (cMCL) was constructed with a machine learning algorithm. Its performance was compared with that of traditional methods. A nomogram was constructed by incorporating the ctDNA methylation-based model for identifying lymph node metastasis and clinicopathological predictors to identify D3-metastasis. In addition, a traditional clinicopathological method was constructed with the same method but without ctDNA methylation-based model for identifying lymph node metastasis. Comparison between the performance of the two models was performed. SETTINGS: A single cancer center in China. PATIENTS: A total of 206 patients with stage I-III colorectal cancer were recruited between June 2022 and September 2023. All patients underwent radical surgical resection and D3 lymphadenectomy without neoadjuvant treatment. MAIN OUTCOME MEASURES: The performance of identifying lymph node metastasis and D3-metastasis. RESULTS: After selecting 27 differential biomarkers, the ctDNA methylation-based model for identifying lymph node metastasis was constructed and yielded a sensitivity of 82.6% and a specificity of 73.3%. The accuracy was superior to that of CT (77.2% vs. 66.5%, p = 0.019). Afterward ctDNA methylation-based model for identifying lymph node metastasis-nomogram was constructed to predict D3-metastasis. The accuracy was also superior to that of traditional method (82.6% vs. 67.4%, p < 0.001). LIMITATION: This study included a small number of patients. CONCLUSIONS: The proposed novel approach based on ctDNA methylation could accurately identify lymph node metastasis and D3-metastasis in colorectal cancer preoperatively and inform tailored treatment. See Video Abstract . Clinical trial registration number: NCT05558436.
Ji et al. (Thu,) studied this question.