Predictors for the Development of Neurological Immune-Related Adverse Events of Immune Checkpoint Inhibitors and Impact on Mortality

Abstract

Objective To report the incidence, predictors for development, impact on mortality, and impact on pre-existing neurological conditions of neurological immune-related adverse events (irAEs) in a large clinical cohort. Background Immune checkpoint inhibitors (ICI) are associated with irAEs. Although neurological complications have been described, little is known about risk factors for their development and their impact on mortality. The impact of ICIs on pre-existing neurological conditions is also not well understood. Design/Methods Patients who received ICI between January 2013 and December 2018 were identified using a tertiary cancer center registry. Patient demographics, cancer characteristics, treatment type, and concurrent oncologic therapy were summarized using descriptive statistics. Patients with neuro-irAE were compared to those without neuro-irAE during the study timeframe. Odds ratios from univariable and penalized multivariable logistic regression models were calculated to identify potential predictors for developing a neuro-irAE. The impact of a neuro-irAE on overall survival was estimated by Kaplan-Meier and multivariable Cox proportional-hazard models. Results Overall frequency of neurological irAEs was 2.3% (28/1228). Peripheral nervous system complications such as myasthenia gravis, myositis, and neuropathies were the most frequent (53.6%). Melanoma, younger age, prior chemotherapy, prior resection, CTLA-4 ICI exposure, and combination ICI exposure had significantly higher odds for developing a neuro-irAE (p <0.05), but these findings were not statistically significant in the multivariable models. Those with a neuro-irAE had greater survival at 3 years compared to those without a neuro-irAE (69% vs 55%, p = 0.004), but after adjusting for patient and cancer characteristics, this effect was no longer statistically significant. Relapse rate of pre-existing neurological condition after exposure to ICI was 15.4% (2/13). Conclusions Neuro-irAEs are rare and are not associated with an increase in mortality. Potential predictors for the development of neuro irAEs are younger age, melanoma, prior chemotherapy and resection, CTLA-4, or combination ICI exposure. Relapse of a pre-existing neurological condition was uncommon.