Multi-omics profiling of tumor DNA, mRNA, and plasma ctDNA in advanced lung cancer.

Abstract

287 Background: The extensive landscape of therapeutic biomarkers in lung cancer has catalyzed the evolution of comprehensive genomic profiling (CGP). In this study, we evaluated the technical performance and clinical utility of a multi-omic approach, integrating DNA and mRNA sequencing for tumor profiling, combined with ctDNA analysis for real-time monitoring. Methods: FFPE samples were collected from 180 patients with advanced lung cancer. Genomic DNA was sequenced using a 504-gene panel with high-density probes (K-4Care, Gene Solutions) to identify actionable biomarkers. Whole mRNA sequencing was performed to detect fusions and predict tissue of origin using the OriCUP machine learning model. For longitudinal ctDNA monitoring, serial plasma samples from 39 patients were analyzed using a combined tumor-informed and tumor-agnostic approach. ctDNA-based molecular response results were subsequently correlated with progression-free survival (PFS). Results: Targeted FFPE DNA sequencing identified actionable mutations in 62.7% (113/180) of the samples. The high-density probe configuration significantly enhanced the sensitivity to detect and quantify copy number variations (CNVs), specifically for MET , EGFR , and MYCN amplifications. Notably, this optimized design enabled the robust detection of MTAP-CDKN2A homozygous deletions, an emerging therapeutic target, in 13.3% (24/180) of cases. For fusion detection, mRNA sequencing showed 20% higher sensitivity compared to DNA sequencing in reference samples. However, with 35% of clinical FFPE samples showing poor RNA quality, combined DNA and mRNA sequencing was critical to maximize fusion detection. For tissue of origin, the OriCUP model predicted lung origin with 88.2% accuracy. In the longitudinal cohort of 39 patients, tumor-agnostic plasma profiling detected resistance mutations in 15.4% (6/39) of cases. Molecular responders, defined by a > 50% reduction in ctDNA level from baseline, demonstrated significantly longer PFS compared to molecular non-responders (HR = 9.53, p < 0.0001; 12-month PFS: 95.5% vs. 33.6%). Conclusions: Integrating genomic with transcriptomic profiling provided a more comprehensive molecular landscape than DNA-only CGP alone. The incorporation of ctDNA analysis further helped identify resistance mutations and evaluate treatment response.