Construction and evaluation of risk prediction model for kidney injury in patients treated with PD-1 inhibitors.

OBJECTIVE: Construction of a nomogram model for predicting the risk of kidney injury associated with programmed cell death protein-1 (PD-1) inhibitors.

METHODS: Patients using PD-1 inhibitors from January 2018 to January 2021 in our center were retrospectively analyzed and followed up until January 2022. Kidney injury was defined as acute kidney disease and disorders (AKD). Univariate and multivariate logistic regression models were used to identify independent risk factors for PD-1 inhibitor associated kidney injury, and nomogram prediction models were further established.

RESULTS: A total of 447 patients were included, of whom 71 patients developed AKD. Multivariate logistic regression analysis identified comorbid extrarenal immune related adverse events (irAEs), lower baseline hemoglobin (Hb) level, higher baseline serum creatinine (SCr) and direct bilirubin (DBIL) level as independent risk factors for the development of PD-1 inhibitor associated kidney injury (P < 0.05). A nomogram prediction model was established based on the above independent risk factors. The area under the receiver operating characteristic (ROC) curve (AUC) of the training set and validation set of the model was 0.703 (95% CI 0.628~0.777), 0.791 (95% CI 0.671~0.911), respectively, which suggested that the predictive ability of the model was acceptable. The calibration curves for both the training and validation sets hovered around the ideal line of 45 degrees, suggesting good calibration of the model. Clinical decision curve analysis (DCA) showed that the constructed nomogram curve was far from the two polar end lines, suggesting better clinical benefit of the model.

CONCLUSION: There are many independent risk factors of PD-1 inhibitor related kidney injury. The predictive ability and calibration of the nomogram model established in this study were acceptable and had clinical utility.