Multi-omic ctDNA-based MRD for predicting clinical outcomes in advanced NSCLC receiving chemoimmunotherapy.

Abstract

8576 Background: Immune checkpoint inhibitors (ICIs) have revolutionized treatment for advanced non–small cell lung cancer (aNSCLC); however, many patients achieving radiographic partial or complete response (PR/CR) relapse early. This underscores the need for more precise risk stratification after an initial response. We investigated whether circulating tumor DNA -based minimal residual disease (ctDNA-MRD) assessment could stratify patients with durable benefit versus those at risk of early progression after chemoimmunotherapy (Chemo-IO). Methods: We analyzed 152 retrospective and 60 prospective aNSCLC patients who achieved PR/CR after first-line (1L) PD-1-based Chemo-IO. Plasma was collected at the first imaging-confirmed PR/CR (retrospective) or at baseline and Cycle 5 Day 1 (C5D1; prospective). ctDNA-MRD was assessed using a tumor-naïve, mutation-based panel and a fragmentomics model derived from low-pass whole-genome sequencing (LP-WGS). In the retrospective cohort, 89 patients were used for LP-WGS model training and 63 for validation. Risk stratification was evaluated using progression-free (PFS) and overall survival (OS). Results: In the retrospective validation cohort (n = 63), both the mutation- and fragmentomic-based assays stratified PFS and OS. For the mutation-based assay, median PFS (mPFS) was 27.0 vs. 10.1 months (negative vs. positive; Hazard ratios [HR] 0.414, 95% confidence intervals [CI] 0.211‒0.814, p = 0.008) and median OS (mOS) was not reached (NR) vs. 17.5 months (HR 0.341, 95% CI 0.151‒0.768, p = 0.007). For the LP-WGS-based model, mPFS was 27.0 vs. 13.8 months (HR 0.412, 95% CI 0.216‒0.785, p = 0.005) and mOS was NR vs. 25.1 months (HR 0.355, 95% CI 0.158‒0.795, p = 0.009). Integration using an either-assay-positive definition further improved prognostic discrimination, with mPFS NR vs. 13.8 months (HR 0.379, 95% CI 0.201‒0.714, p = 0.002) and mOS NR vs. 25.1 months (HR 0.324, 95% CI 0.143‒0.730, p = 0.004). In the prospective cohort (n = 60), both assays at C5D1 identified patients at higher risk of progression: mutation-based mPFS 18.3 vs. 6.1 months (HR 0.184, 95% CI 0.075‒0.457, p < 0.0001); LP-WGS-based mPFS NR vs. 5.3 months (HR 0.178, 95% CI 0.070‒0.456, p < 0.0001). Integration-based mPFS was NR vs. 6.5 months (HR 0.130, 95% CI 0.039‒0.440, p = 0.0001). Multivariable analysis confirmed the independent and complementary prognostic value of both approaches across cohorts. Conclusions: Both mutation- and fragmentation-based ctDNA-MRD assays effectively stratified risk among aNSCLC patients with radiographic response to 1L Chemo-IO. Their integration further improved prognostic performance, refining risk beyond imaging-defined PR/CR and informing post-response management decisions.