Non-Invasive Genomic and Transcriptomic Characterization of Small Cell Lung Cancer Using Circulating Tumor DNA

Chirayu Mohindroo; Brett Schroeder; Rajesh Kumar; Melissa Abel; Parth Desai; Nobuyuki Takahashi; Gavriel Fialkoff; Nir Friedman; Vladimir B. Teif; Anish Thomas

doi:10.1016/j.jtho.2025.07.054

Background Small Cell Lung Cancer (SCLC) is driven by the expression of distinct genomic and epigenomic regulators, which create specific therapeutic vulnerabilities. However, the clinical application of these findings is often constrained by the difficulty in obtaining tumor biopsies, especially during relapse. In this context, we introduce three non-invasive techniques for monitoring SCLC with baseline and sequential plasma samples and matched tumor transcriptomes. Methods We conducted an observational study using plasma collected from with relapsed/refractory SCLC patients enrolled in therapeutic clinical trials with either chemotherapy and/or immunotherapy-based regimens. (NCT02484404; NCT02487095; NCT02769962; NCT03554473 NCT03896503). We applied three techniques to infer SCLC gene expression programs from plasma: cell free DNA(cfDNA) low pass whole genome (0.1X coverage) sequencing, chromatin immunoprecipitation of cell-free nucleosomes carrying active chromatin modifications followed by sequencing (cfChIP-seq) (n=299) and nucleosomal patterns reflected in cell-free DNA fragmentation (n=109) . Results of these analysis were integrated with clinical data and matched tumor transcriptomes. Results cfDNA showed strong correlation with the disease trajectory. Low baseline tumor burden derived from cfDNA (p = 0.010), fragmentomes (p < 0.001), and cfChIP-seq (p < 0.001) were predictive of longer overall survival (OS). We observed mutations in TP53 (90.2%) and RB1 (83.6%), along with alterations in Myc family genes, Notch signaling genes, cell cycle regulators, and chromatin modifiers, which were highly concordant with the tumor samples. Phylogenetic analysis of cfDNA demonstrated a linear evolutionary pattern, indicating a stable genetic landscape dominated by truncal clones with TP53 and RB1 alterations that did not change significantly over the course of treatment. All three approaches were additionally able to infer gene expression patterns of tumors. Specifically, cfChIP-seq showed a significant positive correlation with gene expression in over 25% of the genes (623 genes with q < 0.05; Pearson correlation 0.33 - 0.93). Strong correlations were observed between cfChIP-seq and tumor RNA-seq data for three key transcription factors—ASCL1, NEUROD1, and POU2F3—(Pearson r = 0.93, 0.85, and 0.97, p < 1×10-5, <7×10-5, and <1×10-5, respectively), which were absent from healthy control cfDNA. Using a multi-step classifier, cfChIP-seq was able to discriminate between SCLC lineage-defining. Conclusion transcription factor subtypes. Fragmentomics analysis successfully identified ASCL1 transcription factor binding sites (TFBS), which were used to stratify SCLC into high and low neuroendocrine (NE) groups. The ASCL1 TFBS was concordant with tumor transcriptome data, as evidenced by correlations with the NE50 score (r = 0.3, p = 0.05) and the SCLC-nonNE score (r = -0.31, p = 0.04). CfDNA can inform classification of SCLC into subtypes based on lineage-specific transcription factors. It is also predictive and prognostic of disease trajectory in SCLC patients. Longitudinal tracking of cfDNA could offer valuable insights into resistance mechanisms to chemotherapy and immunotherapy.

Non-Invasive Genomic and Transcriptomic Characterization of Small Cell Lung Cancer Using Circulating Tumor DNA

View Online

Abstract

Details

Metrics