ESTRO 2025 - Abstract Book

S419

Clinical - Biomarkers

ESTRO 2025

tumor stage (staging recoded to TNM8 for all patients), CRP, albumin, and LDH. EDTA-plasma concentrations of GM CSF, IFN-α2a, IFN-β, IFN-y, IL-1α, IL-1β, IL-2, IL-6, IL-8, IL-10, IL-12p70, TNF-α, IP-10, MCP1 and PD-L1 were analyzed using the U-plex assay (MesoScaleDiscovery). 1-year overall survival (OS)-based cut-offs of each analyte was based on max Long-Rank statistics. OS was calculated from start of RT. Significance level was set at 0.05, using Holm Bonferroni correction when appropriate. Results: The development cohort included 331 patients (median OS: 22.0 months, 95% CI 17.6-26.4) and the independent validation cohort comprised 40 patients (median OS: 22.5 months). From the 15 proteins analyzed, only IL-6 (OR: 2.957, 95% CI: 1.837-4.758, p<0.001), TNF-α (OR: 2.249, 95% CI 1.363-3.711, p=0.002), IP-10 (OR: 2.292, 95% CI: 1.419 3.702) and PD-L1 (OR: 3.967, 95% CI 2.423-6.496, p<0.001) were significantly associated with OS. In the multivariable logistic regression analysis, PS ≥2 (OR=3.76, 95% CI 1.67-8.46, p=0.001), stage IIIC (OR=2.76, 95% CI 1.25-6.09, p=0.012), IL-6 (OR=2.06, 95% CI 1.22-3.48), IP-10 (OR=1.75, 95% CI 1.03-2.98), and PD-L1 (OR=3.34, 95% CI 1.97-5.67, p<0.001) were identified as independent risk factors of early mortality. A nomogram was developed to calculate the risk of receiving futile treatment for each patient (Figure 1). The AUC of the development and validation cohort was 0.762 (95% CI 0.708-0.816) and 0.739 (95% CI 0.575-0.903), respectively. Patients classified as intermediate (median OS: 12.0 months, 95% CI 1.8-22.2) or high (median OS: 6.0 months, 95% CI 2.8-9.3) risk to receive futile treatment presented 21.5% of the total cohort.

Conclusion: A prognostic model, based on clinical and blood biomarkers, was developed that can identify the patients who are at high risk of early mortality during and after CRT. These patients may benefit from alternative treatment strategies, such as upfront immunotherapy.

Keywords: NSCLC, chemoradiation, prediction model