Skip to main content
News

Multimodal cfDNA MRD Testing Detects Molecular Recurrence Ahead of Imaging in Resected Non-Small Cell Lung Cancer

Edited by 

Clinical Summary: 

  • Design/Population: This prospective cohort study evaluated longitudinal tumor-naïve multimodal circulating cell-free DNA (cfDNA) minimal residual disease (MRD) monitoring in patients with resected stage I–III non-small cell lung cancer (NSCLC), using postoperative plasma samples collected during surveillance.

  • Key Outcomes: Longitudinal MRD monitoring demonstrated 77.8% sensitivity, 89.2% specificity, and 96.5% negative predictive value for recurrence detection. MRD positivity was strongly associated with recurrence risk, with molecular signals preceding radiographic recurrence in 21 of 27 patients and a median lead time of 264 days.

  • Clinical Relevance: These findings support the potential of tumor-naïve multimodal cfDNA MRD monitoring to improve postoperative recurrence-risk stratification and identify molecular relapse before radiographic detection, while highlighting the complementary value of mutation-independent cfDNA signals.

Results from a prospective study demonstrated that longitudinal tumor-naïve multimodal circulating cell-free DNA (cfDNA) minimal residual disease (MRD) monitoring identified patients at increased risk of recurrence following curative-intent resection of non-small cell lung cancer (NSCLC), with molecular signals detected ahead of radiographic recurrence.

These results were presented by Di Lu, MD, PhD, Nanfang Hospital, Guangzhou, Guangdong, China, at the Association for Molecular Pathology (AMP) Annual Meeting and Expo. 

In this prospective cohort study, researchers evaluated 212 patients with resectable stage I–III NSCLC who underwent curative-intent surgery, analyzing 1207 postoperative plasma samples collected at an early landmark time point and every 3 to 6 months during surveillance.

The tumor-naïve MRD approach integrated mutation-based, normalized-depth, and fragmentomic cfDNA signals using ultra-deep molecular-barcoded sequencing with matched white blood cell filtering.

Recurrence-free survival was assessed using Kaplan-Meier analysis and Cox regression. Longitudinal MRD status was evaluated through patient-level positivity and time-dependent modeling, with additional analyses examining individual cfDNA signal contributions, dynamic patterns of MRD positivity, and recurrence characteristics.

At the early postoperative landmark assessment, integrated MRD positivity was strongly associated with recurrence risk (hazard ratio [HR], 9.72; 95% confidence interval [CI], 4.39 to 21.52; P = 2.06 × 10⁻⁸).

Longitudinal MRD monitoring demonstrated an even stronger association with recurrence risk (HR, 16.64; 95% CI, 6.67 to 41.53; P = 2.66 × 10⁻¹⁶). Across longitudinal surveillance, the integrated MRD approach achieved a sensitivity of 77.8%, specificity of 89.2%, and negative predictive value of 96.5%.

When MRD conversion timing was incorporated into the analysis, longitudinal MRD positivity remained associated with recurrence in both unadjusted and stage-adjusted models. In the unadjusted analysis, the HR for recurrence was 19.74 (95% CI, 7.92 to 49.23; P < .001). After adjustment for disease stage, the association remained substantial (HR, 10.57; 95% CI, 4.01 to 27.87; P < .001).

Integrated MRD monitoring detected molecular evidence of recurrence before radiographic confirmation in 21 of 27 patients who experienced recurrence, with a median lead time of 264 days.

Among 41 patients with longitudinal integrated MRD positivity, 8 were mutation-negative, including 5 patients who experienced recurrence and were identified through normalized-depth or fragmentomic signals. In these cases, nonmutational cfDNA signals preceded recurrence by 9 to 1105 days, highlighting the complementary contribution of mutation-independent features to MRD detection.

Dynamic-pattern analyses further distinguished patients with sustained molecular negativity from those with single or late MRD positivity and those with repeated or persistent positivity. The greatest recurrence risk was observed among patients with repeated or persistent MRD positivity, suggesting that longitudinal patterns may provide additional prognostic information beyond a single MRD assessment.

Exploratory analyses of recurrence characteristics suggested that repeated or persistent MRD positivity was more common among patients experiencing distant or systemic recurrence. In contrast, MRD-negative recurrence events included locoregional, central nervous system, selected distant, and unknown-site recurrences.

These findings suggest that combining mutation-based and nonmutational cfDNA signals with repeated longitudinal assessments may improve the identification of patients at elevated risk of recurrence following NSCLC resection.


Source: 

Lu D, Cai K, Du Y, et al. Longitudinal tumor-naïve multimodal cfDNA MRD monitoring for postoperative recurrence-risk stratification in NSCLC. Presented at the AMP Annual Meeting and Expo. November 10-14, 2026; Seattle, Washington. Abstract ST064. 

© 2026 HMP Global. All Rights Reserved.
Any views and opinions expressed are those of the author(s) and/or participants and do not necessarily reflect the views, policy, or position of Oncology Learning Network or HMP Global, their employees, and affiliates.