Cohort study on immune checkpoint inhibitor-associated acute kidney injury: Incidence, risk factors, and management strategies.

INTRODUCTION: Case studies and retrospective chart reviews of health system data have demonstrated an increased risk of nephrotoxicity in patients receiving immune checkpoint inhibitors compared to clinical trials. This study investigated the frequency, causes, and risk factors for acute kidney injury in a real-world, rural setting. METHODS: This was a retrospective cohort study of patients who received at least one dose of a checkpoint inhibitor at a rural health system from May 2013 to February 2020 and who received at least one dose of a checkpoint inhibitor. Electronic and manual chart review helped to determine the incidence of, risk factors for, and renal outcomes and management strategies of checkpoint inhibitor-related acute kidney injury. Multivariable Fine and Gray subdistribution hazard models were used to assess the impact of patient characteristics on the incidence of sustained acute kidney injury and checkpoint inhibitor-induced acute kidney injury. RESULTS: After exclusion criteria, 906 patients who received at least one dose of a checkpoint inhibitor at Marshfield Clinic Health System during the study period were included. The incidence of acute kidney injury of any duration and due to any cause was 36.1%, while sustained acute kidney injury occurred in 28.7% of patients. Checkpoint inhibitor-related acute kidney injury was thought to have occurred in 2.7% of patients. Baseline estimated glomerular filtration rate < 60 was the sole predictor of checkpoint inhibitors-related acute kidney injury. Most patients with suspected checkpoint inhibitor-related acute kidney injury were managed with corticosteroids, and 62.5% experienced complete renal recovery. CONCLUSIONS: Ours is the first retrospective cohort study to test whether baseline Eastern Cooperative Oncology Group score and checkpoint inhibitor place in therapy were associated with checkpoint inhibitor-related acute kidney injury, and neither of these data points were found to be predictive. Even after expanding the parameters and methodologies of our study as compared to other retrospective cohort studies, we found only three baseline characteristics to be predictive of sustained acute kidney injury: Baseline eGFR, loop diuretic, and spironolactone use. For checkpoint inhibitor-related baseline, eGFR alone was predictive.