Response Evaluation and Survival Prediction After PD-1 Immunotherapy in Patients with Non-Small Cell Lung Cancer: Comparison of Assessment Methods.

Abstract

Immunotherapy using programmed death-1 blockers is a promising modality for non-small cell lung cancer (NSCLC). Therefore, defining the most accurate response criteria for immunotherapy monitoring is of great importance in patient management. This study aimed to compare the correlation between survival outcome and response assessment by PERCIST, version 1.0; immunotherapy-modified PERCIST (imPERCIST); RECIST, version 1.1; and immunotherapy-modified RECIST (iRECIST) in NSCLC patients. Methods: Seventy-two patients with NSCLC who were treated with nivolumab or pembrolizumab and had baseline and follow-up 18F-FDG PET/CT data were analyzed. The patients were categorized into responders (complete or partial response) and nonresponders (stable or progressive disease) according to PERCIST1 and PERCIST5 (analyzing the peak SUV normalized by lean body mass [SULpeak] of 1 or up to 5 lesions), imPERCIST1, imPERCIST5, RECIST, and iRECIST. The correlation between achieved response and overall survival (OS) was compared. Results: The overall response rate and the overall disease control rate of the study population were 29% and 74%, respectively. The OS and progression-free survival (PFS) of patients with complete and partial response were statistically comparable. The OS and PFS were significantly different between responders and nonresponders (20.3 vs. 10.6 mo, P = 0.001, for OS and 15.5 vs. 2.2 mo, P < 0.001, for PFS, respectively). Twenty-three (32%) patients with progressive disease according to PERCIST5 had controlled disease according to imPERCIST5; follow-up of patients showed that 22% of these patients had pseudoprogression. The overall incidence of pseudoprogression was 7%. The response rate was 25% and 24% according to PERCIST1 and PERCIST5 (P = 0.2) and 32% and 29% according to imPERCIST1 and imPERCIST5 (P = 0.5), respectively, indicating no significant difference between analyzing the SULpeak of only the most 18F-FDG-avid lesion and analyzing up to the 5 most 18F-FDG-avid lesions. Conclusion: The achieved response by all conventional and immunotherapy-modified methods correlated strongly with patients' survival outcome, with significantly longer OS and PFS in responders than in nonresponders according to all assessed definitions. The most 18F-FDG-avid lesion according to PERCIST and imPERCIST accurately reflects the overall metabolic response.