Risk prediction of continuous renal replacement therapy in patients with acute kidney injury after lung transplantation.

Background: This study used least absolute shrinkage and selection operator (LASSO) regression analysis to identify the influencing factors of continuous renal replacement therapy (CRRT) implementation in patients with acute kidney injury (AKI) following lung transplantation (LTx), and subsequently constructed a nomogram model for predicting CRRT risk based on these determinants. The model aims to provide decision-making support for early clinical intervention in high-risk populations. Methods: This retrospective study collected clinical data and laboratory parameters from patients who underwent LTx at the Second Affiliated Hospital of Zhejiang University School of Medicine between June 2018 and January 2024. Postoperative AKI was diagnosed and staged according to the serum creatinine (sCr)-based criteria defined by the Kidney Disease: Improving Global Outcomes guidelines. Patients were grouped by CRRT receipt. Variables were initially screened using LASSO regression to minimize overfitting, followed by multivariate logistic regression analysis to construct a CRRT risk prediction model. The predictive efficacy and clinical practicability of the model were evaluated using receiver operating characteristic curve analysis, calibration curves, and decision curve analysis. Results: A total of 448 LTx patients were included in this study, of whom 340 (75.9%) developed creatinine-elevated AKI. Among these AKI patients, 79 (23.2%) received CRRT. Multivariate logistic regression analysis identified the following independent risk factors for CRRT in post-LTx AKI patients: advanced age, increased intraoperative blood loss, higher intraoperative fluid input, bilateral LTx, prolonged postoperative extracorporeal membrane oxygenation (ECMO) support duration, delayed time to peak sCr, rapid sCr elevation rate, and greater magnitude of sCr rise. Conversely, preoperative mechanical ventilation (MV) emerged as a protective factor. Validation analyses demonstrated that the CRRT risk prediction nomogram model, incorporating nine variables (age, preoperative MV, LTx type, intraoperative blood loss, intraoperative net fluid balance, postoperative ECMO duration, time to sCr peak, sCr elevation rate, and rise magnitude), exhibited robust predictive performance and clinical applicability, significantly enhancing patient clinical benefit. Conclusions: The risk prediction nomogram model developed in this study demonstrates high accuracy and clinical utility. This model facilitates early identification of high-risk patients by clinicians and provides critical guidance for optimizing early AKI intervention strategies and formulating personalized CRRT protocols.