The Impact of Minimal Residual Disease (MRD) Testing on the Decision-Making Process in Non-Small-Cell Lung Cancer...

Background: Minimal residual disease (MRD) assessment is an emerging tool for refining the risk of relapse following definitive therapy in non-small-cell lung cancer (NSCLC). However, data regarding its clinical impact on the decision-making process remain limited. We evaluated MRD feasibility and its impact in the real-world setting. Methods: A pooled retrospective analysis of longitudinal MRD data in NSCLC patients (n = 34: Signatera™ (Exome), n = 25, Guardant Reveal™, n = 9) was implemented. Co-primary endpoints: MRD feasibility and clinical impact on management (changes in surveillance intensity or therapy escalation/de-escalation). Secondary endpoints: sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), accuracy for recurrence detection, and MRD lead time. Results: MRD was feasible in 32/34 patients (94.1%); longitudinal testing included two samples in 15 patients (44.1%) and three samples in 2 patients (5.9%). Signatera™ (Exome) failed in 2/25 (8.0%) due to insufficient tissue. MRD influenced management in 20/34 (58.8%) patients, most commonly supporting therapy de-escalation (15/34, 44.1%), followed by imaging surveillance modification (3/34, 8.8%) and therapy escalation (2/34, 5.9%). In univariable analysis, tumor grade and STAS were associated with MRD-driven management impact, but neither remained significant in multivariable analysis. With a median follow-up of 18.9 months (IQR 8.5-30.7), MRD was positive in 6/32 (18.8%), while recurrence/progression occurred in 10/32 (31.3%) patients. MRD yielded 21 true negatives, five true positives, five false negatives (including two isolated brain recurrences), and one false positive, corresponding to a sensitivity of 50.0%, specificity of 95.5%, PPV of 83.3%, NPV of 80.8%, and an accuracy of 81.3%. The median MRD lead time (n = 5) was 1.31 months (range, 0.46-5.52). Conclusions: In this real-world cohort, MRD testing was feasible and frequently guided clinical decisions, mainly supporting treatment de-escalation. MRD was highly specific but less sensitive. Prospective studies are needed to define optimal testing intervals and validate MRD-guided strategies.