Original Abstract
BACKGROUND: Circulating tumor DNA (ctDNA)-based assays enable the detection of minimal residual disease (MRD) after surgery and are predictive of recurrence in lung cancer patients. However, the optimal technical strategy remains a subject of active debate. MATERIALS AND METHODS: We conducted a head-to-head comparison of tumor-naïve fixed panel, tumor-informed fixed panel, and tumor-informed personalized panel assays to analyze ctDNA-MRD in 417 plasma samples from 84 patients with stage I-III lung cancer. RESULTS: At the landmark time point, the personalized assay achieved superior performance compared with the tumor-naïve and tumor-informed assays, with a sensitivity of 40.9% (P = 0.003), a specificity of 100.0% (P < 0.001), a negative predictive value of 81.7% (P < 0.001), and a positive predictive value of 100.0% (P < 0.001). MRD positivity by the personalized assay predicted significantly shorter disease-free survival (P < 0.001; HR = 13.98; 95% CI: 1.95-100.37). Longitudinal MRD monitoring using personalized assay improved relapse prediction, providing median lead times of 7.0 months before radiologic recurrence. CONCLUSION: The personalized assays may optimize MRD detection and inform individualized postoperative surveillance in lung cancer.