Tianyi Zhang, Hang Zhou, Zhaohui Ni, Qin Wang, Jiajia Wu, Qian Chen, Ming Qiu, Yue Wang, Tingting Fu, Mingyu Ye, Jin Zhang, Wei Xue and Shan Mou
Background: Acute kidney injury (AKI) during the perioperative period mainly causes the loss of renal function after nephrectomy, while accurate prognostic models in which AKI as an independent predicting variable are still absent.
Methods: A retrospective study of 528 patients with renal cell carcinoma who underwent nephrectomy between January 2013 and December 2016 was carried out. The endpoint was the time to the incident stage 3 or higher chronic kidney disease (CKD), or any claim for initiation of dialysis. Cox proportional hazards regression analysis was conducted to construct the final model. Internal validation was performed using 10-fold cross-validation. The model was evaluated in discrimination by c statistics and area under curve (AUC) values, and calibration by calibration plots.
Results: Among 528 admitted patients, 232 (43.9%) developed AKI, and stage 3 or higher CKD occurred in 8.9% of people during the follow-up time. AKI was significantly correlated with poor prognosis (HR=3.079, P<0.001), and after the adjustment of conventional predictors, AKI was still independently related to kidney function deterioration, and the correlation was influenced by the severity of AKI. Five variables were selected to establish the prognostic model, including age, surgery type, preoperative estimated glomerular filtration rate, preoperative blood urea nitrogen and AKI. The model had good discrimination, with a C-harrell statistic of 0.92 (95% CI 0.89 to 0.95), AUC values varied from 87.7 to 95.7 at different time points.
Conclusion: AKI during the perioperative period is an independent predicting factor of stage 3 or higher CKD after nephrectomy.
