Development and validation of LCMM prediction algorithms to estimate recovery pattern of postoperative AKI in type A aortic dissection: a retrospective study.

Abstract

Postoperative acute kidney injury (PO-AKI) is a prevalent complication among patients with acute type A aortic dissection (aTAAD) for which unrecognized trajectories of renal function recovery, and their heterogeneity, may underpin poor success in identifying effective therapies. This was a retrospective, single-center cohort study in a regional Great Vessel Center including patients undergoing aortic dissection surgery. Estimated glomerular filtration rate (eGFR) recovery trajectories of PO-AKI were defined through the unsupervised latent class mixture modeling (LCMM), with an assessment of patient and procedural characteristics, complications, and early-term survival. Internal validation was performed by resampling. A total of 1,295 aTAAD patients underwent surgery and 645 (49.8%) developed PO-AKI. Among the PO-AKI cohort, the LCMM identified two distinct eGFR trajectories: early recovery (ER-AKI, 51.8% of patients) and late or no recovery (LNR-AKI, 48.2% of patients). Binary logistic regression identified five critical determinants regarding poor renal recovery, including chronic kidney disease (CKD) history, renal hypoperfusion, circulation arrest time, intraoperative urine, and myoglobin. LNR-AKI was associated with increased mortality, continuous renal replacement therapies, mechanical ventilation, ICU stay, and hospital stay. The assessment of the predictive model was good, with an area under the curve (AUC) of 0.73 (95% CI: 0.69-0.76), sensitivity of 61.74%, and specificity of 75.15%. The internal validation derived a consistent average AUC of 0.73. The nomogram was constructed for clinicians' convenience. Our study explored the PO-AKI recovery patterns among surgical aTAAD patients and identified critical determinants that help to predict individuals at risk of poor recovery of renal function.