WCLC 2025 - Posters & ePosters-EP.06.68 Clonal Hematopoiesis (CH)-Derived Mutations Detected in Non-Small Cell Lung Cancer (NSCLC) Liquid Biopsies (LBx)

EP.06.68 Clonal Hematopoiesis (CH)-Derived Mutations Detected in Non-Small Cell Lung Cancer (NSCLC) Liquid Biopsies (LBx)

Back to course

Pdf Summary

This study investigates clonal hematopoiesis (CH) and its impact on liquid biopsy (LBx) interpretation in non-small cell lung cancer (NSCLC). CH involves mutations in hematopoietic stem cells and can present variants overlapping with tumor mutations, complicating interpretation. Using data from 21,456 NSCLC patient plasma LBx samples analyzed with Foundation Medicine’s FoundationOneLiquid CDx assay, researchers applied a machine learning (ML) algorithm to classify variant origins as tumor-somatic, germline, or CH-derived. This algorithm was trained and validated on paired plasma and white blood cell (WBC) samples, achieving high sensitivity and specificity without dependence on WBC sequencing.<br /><br />Key findings show that established NSCLC target mutations, such as classic EGFR drivers, KRAS G12C, BRAF V600E, and MET exon 14 skipping, are rarely affected by CH origin variants, with CH-derived mutations being virtually absent among these critical targets. However, certain emerging targets and other genes frequently mutated in NSCLC show significant CH confounding—for example, 70% of TP53 Y220C mutations, 25% of non-V600E BRAF mutations, and 5% of SMARCA4 mutations detected in LBx are CH-derived. The study highlights that ruling out CH origin by additional testing (e.g., tissue biopsy, WBC sequencing, or predictive algorithms) is advisable for these variants to prevent misinterpretation.<br /><br />Overall, the research demonstrates that CH does not substantially interfere with detection of established NSCLC therapeutic targets in liquid biopsies, supporting their reliability for clinical decisions. Yet, for some emerging or less common mutations, consideration of CH origin is critical. The study provides comprehensive prevalence data on CH-derived mutations in NSCLC-relevant genes and emphasizes the utility of ML-based variant origin prediction to enhance accuracy in cfDNA-based precision oncology.

Asset Subtitle

David Kozono

Meta Tag

Speaker David Kozono

Topic Pathology and Biomarkers

Keywords

clonal hematopoiesis

liquid biopsy

non-small cell lung cancer

FoundationOneLiquid CDx

machine learning algorithm

tumor-somatic mutations

CH-derived mutations

EGFR mutations

TP53 Y220C mutation

precision oncology