Abstract 4393: Tumor volumetric analysis to correlate disease burden with response to dual immune checkpoint blockade in metastatic NSCLC

Abstract Background: Compared to other well-studied biomarkers, the relationship between overall tumor burden and immune-checkpoint inhibitor (ICI) efficacy is poorly understood. Methods: We identified patients with stage IV NSCLC without EGFR, ALK or ROS1 alterations treated on the LONESTAR protocol with nivolumab + ipilimumab with formal 3-month radiographic response evaluation. Response was annotated using RECIST 1.1. Manual segmentation of primary and measurable metastatic lesions on pre-treatment CT scans was performed, and tumor volume was calculated from these measurements. Tumor burden was defined as the involvement status, volume, or lesion counts within specific organs (lungs, pleura, lymph nodes, liver, adrenals, bone, and soft tissue) and across whole body. We explored the impact of baseline tumor burden on ICI response and its association with other clinical features. Results:152 patients were included; patients with progressive disease (PD; n=50) at 3 months were compared to all others (n=102). Overall disease volume correlated to tumor volume at the primary lesion (spearman rho=0.69), lymph node (rho=0.45), soft tissue (rho=0.23), lung metastases (rho=0.22), and pleura (rho=0.20); overall lesion counts correlated to lesion counts in lymph nodes (rho=0.61), lung metastases (rho=0.45), and bone (rho=0.16). On univariate analysis, overall tumor volume (P=2e-53) and lesion counts (P=2e-51) had the strongest association with PD, outperforming PD-L1 (P=0.02), ECOG performance status (PS, P=7e-16), and clinical variables including prior treatment and smoking status. PD correlated with higher disease burden in all individual organs except bone; adrenal metastases inversely correlated with PD. Volumetric features complemented PD-L1 and other clinical risk factors to predict PD; in a composite model (ClinVol) AUC=0.82, sensitivity=66%, specificity=93%, while PD-L1 alone yielded an AUC=0.62, sensitivity=94%, specificity=28%. Similar trends were observed for subgroup analyses stratified by brain metastasis status or smoking history, in which volumetric measures increased the power of PD-L1 to predict PD by a consistent AUC increase of ~0.2. Upon subgroup analysis stratified by PD-L1, volumetric models showed robust stratification of PD, with AUC of 0.97, 0.81, 0.81 respectively for PD-L1 <1%, 1-49%, and ≥50%. Conclusion: Higher baseline tumor burden was associated with increased likelihood of disease progression on combination ICI therapy. ClinVol modeling improves predictive accuracy of 3-month progression. Citation Format: Muhammad Aminu, Natalie I. Vokes, Maliazurina B. Saad, Hui Li, Lingzhi Hong, Mohamed S. Mohamed, John Boom, Pingjun Chen, Mehmet Altan, Saumil Gandhi, Stephen Swisher, Mara B. Antonoff, Jenny V. Pozadzides, George Blumenschein Jr, Don L. Gibbons, Tina Cascone, Yasir Y. Elamin, Xiuning Le, Marcelo V. Negrao, Ferdinandos Skoulidis, Anne S. Tsao, Janet Tu, J.Jack Lee, Jianjun Zhang, John V. Heymach, Jia Wu. Tumor volumetric analysis to correlate disease burden with response to dual immune checkpoint blockade in metastatic NSCLC. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 4393.