WCLC 2025 - Posters & ePosters-P2.06.77 A Perioperative Circulating Tumor DNA Model for Landmark Detection in Non Small Cell Lung Cancer

Pdf Summary

This study developed and validated a perioperative circulating tumor DNA (ctDNA) model to improve detection of molecular residual disease (MRD) in patients with non-small cell lung cancer (NSCLC) after surgery. While ctDNA-based MRD has been recognized as a strong prognostic biomarker, its clinical utility in guiding adjuvant therapy is limited by the low sensitivity of single time-point MRD assessments.<br /><br />The researchers included 163 NSCLC patients with undetectable MRD within 90 days post-surgery as a training set and selected features from perioperative and baseline ctDNA data using LASSO regression. They constructed a Cox-XGBoost survival model incorporating five key features: baseline cell-free DNA (cfDNA) concentration, baseline ctDNA status and level, landmark cfDNA concentration, and tumor stage (TNM). The model's predictive performance was assessed by concordance index and feature importance measures.<br /><br />The final model stratified patients with undetectable MRD into two distinct risk groups with significant differences in disease-free survival (DFS). In the training cohort, the low-risk group had a 2-year DFS of 98%, whereas the high-risk group had a 72% rate. Validation in an independent TRACERx cohort of 92 patients confirmed strong prognostic discrimination, with 2-year DFS rates of 85.5% for low-risk versus 52.8% for high-risk patients.<br /><br />This perioperative ctDNA model offers improved sensitivity over single landmark MRD detection and may better identify patients with truly favorable prognosis after lung cancer surgery. The authors suggest further validation to establish the model’s utility in guiding more personalized adjuvant therapy decisions, potentially enabling safe de-escalation in low-risk individuals. Overall, this approach enhances prognostic stratification beyond conventional MRD assays by integrating multiple perioperative ctDNA features.

Asset Subtitle

Jia-Tao Zhang

Meta Tag

Speaker Jia-Tao Zhang

Topic Pathology and Biomarkers

Keywords

perioperative circulating tumor DNA

ctDNA

molecular residual disease

MRD detection

non-small cell lung cancer

NSCLC

LASSO regression

Cox-XGBoost survival model

disease-free survival

adjuvant therapy guidance