2010P Non-invasive radiomics signature predicts response to combination chemoimmunotherapy in extensive-stage small cell lung cancer (ES SCLC)

ES SCLC is a highly aggressive malignancy with poor outcomes. While immunotherapy (IO) has shown promise in improving response rates in some patients (pt) with ES SCLC, there are no reliable biomarkers to predict which pt are most likely to benefit. In this study, we aimed to explore the potential of radiomic texture features from inside (intratumoral) and immediately outside (peritumoral) the tumor on pre-treatment CT scans to predict overall survival (OS) and response to chemoIO in ES SCLC pt. The study included 142 ES SCLC pt from three institutions, with measurable lung lesions on baseline chest CT scans. A maximum of two lung lesions were annotated. Pt with best objective response per RECIST 1.1 of complete or partial response were categorized as responders, while those with stable or progressive disease were considered non-responders. A total of 860 intratumoral and peritumoral texture features were extracted, and a radiomic risk score (RRS) was generated using the least absolute shrinkage and selection operator. Cox regression model was used to predict OS. The model was trained on 97 ES-SCLC pt (St) treated with chemotherapy at University Hospital Cleveland Medical Center and validated on 45 ES-SCLC pt (Sv) treated with chemoIO at Roswell Park Cancer Institute (Sv1; n=34) and Emory University (Sv2; n=11). The discriminative ability of the features was assessed using Kaplan-Meier and log-rank tests, and the probability of treatment response was evaluated using a previously trained linear discriminant classifier (LDA). A combination of peritumoral and intratumoral radiomic features on pretreatment CT showed an AUC of 0.73 in St and 0.72 in independent external Sv to predict response to treatment. The RRS was significantly associated with OS in both St (HR: 2.27; 95% CI, 1.7–2.98; P<0.001; C-index=0.70) and Sv (HR: 2.26; 95% CI, 1.32–3.74; P=0.016; C-index=0.65). Pre-treatment CT radiomic features are potentially prognostic and predictive of response to chemotherapy or chemoIO in ES SCLC pt. These findings may have implications for guiding treatment decisions and improving outcomes by better pt stratification. Additional independent, multi-site validation is warranted.