ctDNA-derived copy-number signatures associated with immune checkpoint inhibitor resistance beyond PD-L1 and TMB in advanced NSCLC.

Background: Immune checkpoint inhibitors (ICIs) have improved outcomes in advanced non-small cell lung cancer (NSCLC), but clinical benefit remains heterogeneous and is not fully explained by programmed death-ligand 1 (PD-L1) expression or tumour mutational burden (TMB). We explored whether whole-exome sequencing (WES) of circulating tumour DNA (ctDNA) could identify mutational and copy-number (CN) signatures associated with outcome in advanced NSCLC treated with first-line ICI. Methods: Baseline plasma ctDNA from 37 patients with advanced NSCLC treated with first-line ICI or chemo-ICI was analysed by WES. Single-base substitution (SBS), indel (InD) and CN signatures were inferred using COSMIC-based signature frameworks. SBS signatures reflect point mutation patterns, including clock-like processes that accumulate with age; indel signatures reflect small insertions and deletions; and CN signatures summarise structural changes such as chromosomal gains and loss of heterozygosity. Signatures were integrated with PD-L1 status, TMB, progression-free survival (PFS) and overall survival (OS). Patients were classified into four clinicogenomic groups according to PD-L1 expression and PFS. Results: Median PFS and OS were 9 and 20 months, respectively. Single-nucleotide-variant-derived TMB (TMB-SNV) was similar in patients with adverse and favourable outcomes, and patients with TMB-SNV >10 mutations/Mb were evenly distributed between both groups. The SBS profile was mainly represented by the clock-like signatures SBS1, SBS5 and by the tobacco-like signature SBS92, whereas indel signatures were mostly represented by InD4a and InD10. CN signatures showed the most apparent separation between outcome groups. CN9 exposure, a marker of structurally unstable genomes, was higher in adverse-outcome patients and most prominent in Group 1 (PD-L1 ≥50% and adverse outcome), whilst CN21 exposure was highest in Group 4 (PD-L1 ≥50% and favourable outcome). No signature retained statistical significance after correction for multiple testing. Conclusions: In this exploratory cohort, ctDNA-derived CN signatures may capture structural genomic phenotypes associated with ICI outcome beyond PD-L1 and TMB. CN9 emerged as a candidate resistance-associated signature, whereas CN21 was associated with more favourable disease control. These findings are hypothesis-generating and require validation in larger prospective cohorts.