Real-world landscape transition of death causes in the immunotherapy era for metastatic non-small cell lung cancer.

Background: With approval of anti-PD-1/PD-L1, metastatic non-small cell lung cancer (NSCLC) has entered the era of immunotherapy. Since immune-related adverse events (irAEs) occur commonly in patients receiving anti-PD-1/PD-L1, the landscape of death causes may have changed in metastatic NSCLC. We aim to compare patterns of death causes in metastatic NSCLC between the pre-immunotherapy and immunotherapy era to identify the consequent landscape transition of death causes. Methods: In this cohort study, 298,48patients with metastatic NSCLC diagnosed between 2000 and 2018 were identified from the Surveillance, Epidemiology, and End Results Program. Unsupervised clustering with Bayesian inference method was performed for all patients' death causes, which separated them into two death patterns: the pre-immunotherapy era group and the immunotherapy era group. Relative risk (RR) of each death cause between two groups was estimated using Poisson regression. Reduced death risk as survival time was calculated with locally weighted scatterplot smooth (Lowess) regression. Results: Two patterns of death causes were identified by unsupervised clustering for all patients. Thus, we separated them into two groups, the immunotherapy era (2015-2017, N=40,172) and the pre-immunotherapy era (2000-2011, N=166,321), in consideration of obscure availability to immunotherapy for patients diagnosed in 2012-2014, when the follow-up cutoff was set as three years. Although all-cause death risk had reduced (29.2%, 13.7% and 27.8% for death risks of lung cancer, non-cancer and other cancers), non-cancer deaths in the immunotherapy era (N=2,100, 5.2%; RR=1.155, 95%CI: 1.101-1.211, P<0.001) significantly increased than that in the pre-immunotherapy era (N=7,249, 5.0%), which included causes of chronic obstructive pulmonary disease, cerebrovascular disease, pneumonia and influenza, septicemia, infectious diseases, accidents and adverse effects, hypertension, and chronic liver disease and cirrhosis. However, cancer-caused deaths (excluding lung cancer) had no significant changes. Conclusions: The real-world landscape of death causes has changed in metastatic NSCLC when entering the immunotherapy era, and the increased non-cancer diseases may contribute to the changes that may be associated with commonly occurring irAEs.