Association of single-click radiomic classifier with response and prognosis in non-small cell lung cancers (NSCLC) treated with immune checkpoint inhibitors.

• DOI: 10.1200/jco.2023.41.16_suppl.8574
• Publication Date: 2023-06-04T16:12:44Z
• Authors: Amogh Hiremath; Haojia Li; Allison Clement; Amit Gupta; Vamsidhar Velcheti; Anant Mababhushi; Nathaniel Braman
• Keywords: Atezolizumab

8574 Background: Radiomics has shown promise to non-invasively phenotype disease and address the limitations of extant biomarkers (e.g. PD for immune checkpoint inhibitors (ICI) in cancers, such as NSCLC. However, considerable barriers clinical adoption these tools remain, their dependence on precise annotation tumor extent by experienced users. Here, we demonstrate a radiomic solution that requires only single user mouse click within one or more target lesions baseline CT scan, contour tumors 3D generate patient-level prediction response outcome ICI treated NSCLC patients. Methods: 1778 scans from 1261 patients were used develop validate an interactive, semi-automated tool predicting outcomes prior therapy. A based deep learning contouring model was trained validated 1146 patients, then create annotations analysis. least absolute shrinkage selection operator (LASSO) Cox proportional hazards utilized select features associated with post-ICI overall survival (OS) derive risk score training cohort (n=74) can separate into high low groups. The tested held out pre-treatment CTs 41 recipients 2 institutions association OS, progression-free (PFS), objective (OR). Results: total 77 identified segmented testing set. Average volume per lesion 54.10 mL patient 101.60 mL. OR observed 48.70% threshold -0.31 defining groups chosen optimal separation set (HR=2.59 [95% 1.48~4.50], p=0.0009). Radiomic significantly stratified OS (C-index=0.64, HR=3.03 1.15~8.02], p=0.03) PFS (C-index=0.59, HR=3.20 1.13~9.10], p=0.03). IO group independently prognostic variables (Table 1) further predicted AUC=0.74 0.71-0.78]. Conclusions: From lesions, our able predict prognosis radiology scan. Additional multi-site validation prospective evaluation will assess value radiomics classifier decision support clinic. [Table: see text]