Pretreatment radiomic biomarker for immunotherapy responder prediction in IB-IV stage NSCLC: A multi-center retrospective study.

Abstract

e21060 Background: The current biomarker of immune checkpoint inhibitors (ICIs) efficacy is not sufficient. This study investigated the correlation between radiomic biomarker and immunotherapy responder in patients with IB-IV stage non-small-cell lung cancer (NSCLC), including its biological explanation for ICIs treatment response prediction. Methods: CT images from 319 pretreatment NSCLC patients receiving immunotherapy between January, 2015 and November, 2021 were retrospectively collected and were distributed into a discovery (n = 214), independent validation (n = 54), and external test cohort (n = 51). 851 features were extracted from both tumoral and each peri-tumoral VOIs. The reference standard is the durable clinical benefit (DCB) and the crucial exclusion criteria are missing pretreatment CT scans. The predictive value of combined radiomic signature (CRS) was assessed in a cohort of 98 resectable NSCLC patients receiving ICIs. The association of radiomic features and tumor immune landscape on the online dataset (n = 60) was also evaluated. A clinical model was constructed and incorporated with radiomic signature to improve performance further. Results: CRS discriminated between DCB and non-DCB patients in the training cohort (area under the curve (AUC): 0.82, 95% CI: 0.75 to 0.88)) and independent cohort (AUC: 0.75, 95%CI: 0.62 to 0.88). In this study, the predictive value of CRS was better than PD-L1 expression (AUC of PD-L1 subset: 0.59, 95%CI: 0.50 to 0.69) or clinical model (AUC of validation cohort: 0.66, 95%CI: 0.51 to 0.81). When predicting pathological response, CRS divided patients into a major pathological response (MPR) and non-MPR group (AUC: 0.76, 95%CI: 0.67 to 0.81). Moreover, CRS showed a promising stratified ability on overall survival (OS; HR: 0.48, 95%CI: 0.27 to 0.89; p = 0.020) and progression-free survival (PFS; HR: 0.43, 95%CI: 0.26 to 0.74; p = 0.002). Biological pathway analysis on online dataset revealed that dysregulation of Hedgehog and Hippo signaling pathways got involved in the prediction of radiomic features. Conclusions: By analyzing both tumoral and peri-tumoral regions of CT images in a radiomic strategy, we developed a non-invasive biomarker distinguishing responder from ICIs therapy, which might inform clinical decisions on the use of ICIs in advanced as well as resectable NSCLC patients.