Abstract 3770: DNA methylation-based model predicts occult lymph nodal metastasis for resectable NSCLC patients

Abstract Background: Patients with non-small cell lung cancer (NSCLC) harboring pathologic lymph nodal metastases (LNDM) (pN1-2) have substantially higher recurrence risks. However, 10∼15% of patients staged as cN0 via preoperative CT have lymph node involvement confirmed by pathologic evaluation. Endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA) improves preoperative nodal staging, but extensive sampling of multiple stations increases procedural risk and complexity. Thus, exploring LNDM biomarkers are needed. This was aimed to develop a tumor DNA methylation-based model to predict intrathoracic nodal metastases indicating more precised patient population receiving EBUS-TBNA evaluation. Methods: Primary tumor tissue and paired preoperative plasma from 68 patients with resectable NSCLC was retrospectively collected (pN0=24, pN1=24, pN2=20, all pN2 were mediastinal nodal metastases). Tumor DNA was extracted and subjected for enzyme conversion and methylation sequencing (EM-Seq). Differentially methylated CpG sites associated with nodal involvement were identified by comparing pN0 vs pN1/2 (Wald test). A lymph-node-metastasis methylation score was constructed using LASSO regression and its performance assessed by ROC analysis. Promoter-localized differentially methylated sites were subjected to Gene Ontology (GO) and pathway enrichment analyses. Further validation of this model on preoperative plasma cell-free DNA is undergoing. Results: Differentially methylated 6,745 CpG sites between pN0 and pN1/2 tumors (p<0.05) were disclosed. An 8-CpG methylation score discriminated patients with and without nodal metastases with an AUC of 0.962 (cut-off 2.0; sensitivity 1.0; specificity 0.841). The methylation score increased with pathologic nodal stage (median pN0=0.76, pN1=2.74, pN2=3.30, pN1 vs pN0 p<0.01, pN2 vs pN0 p<0.01, pN2 vs pN1 p=0.26). No correlation to primary tumor size (r=-0.08, p=0.505) was found, making it only LNDM relevant. Among differentially methylated loci, 28% (1,872 sites) were located in promoter regions. In node-positive tumors, enriched GO terms included epithelial cell polarity and cell-cell adhesion, whereas node-negative tumors showed enrichment of phenylalanine, tyrosine, and glycine/serine metabolism. Conclusions: A DNA methylation model strongly associated with the presence and extent of intrathoracic nodal involvement was developed for resectable, cN0 NSCLC. This model has the potential to preoperatively identify patients who most warrant EBUS-TBNA. Its predictive performance in pre-treatment plasma is being validated, and plasma-based results will be presented at the AACR. Citation Format: Long Jiang, Xing Li, Yanhua Chen, Changbin Zhu, Ziming Li. DNA methylation-based model predicts occult lymph nodal metastasis for resectable NSCLC patients [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 3770.