Performance and incremental value of preoperative multimodal imaging and perioperative clinical characteristics in predicting postoperative major adverse events in type A aortic dissection.

BACKGROUND: Postoperative major adverse events (MAEs) can occur after surgical repair of acute Stanford type A aortic dissection (ATAAD), and conventional clinical assessment may not fully capture the prognostic information provided by preoperative multimodal imaging. This study aimed to develop and externally validate a staged risk prediction model for postoperative MAEs and to evaluate the incremental value of preoperative multimodal imaging relative to conventional clinical and laboratory assessment in the prediction of MAEs.

METHODS: In this retrospective, two-center cohort study, patients with surgically treated ATAAD were screened between April 2016 and April 2026. After application of predefined exclusion criteria, 1,960 patients were included, comprising a development cohort from Center I (n=1,498) and an external validation cohort from Center II (n=462). The development cohort was further randomly divided into a training set (n=1,048) and an internal validation set (n=450). Preoperative multimodal imaging and perioperative clinical data were collected, including computed tomography angiography (CTA), transthoracic/transesophageal echocardiography (TTE/TEE), admission laboratory indices, disease-severity indicators, and surgery-related variables. Candidate predictors were screened via least absolute shrinkage and selection operator (LASSO) regression and multivariable logistic regression. Five progressively expanded models were constructed to assess the incremental value of Penn classification, echocardiography, CTA, and intraoperative variables. Model performance was evaluated via receiver operating characteristic analysis, calibration curves, decision curve analysis (DCA), net reclassification improvement (NRI), and integrated discrimination improvement (IDI).

RESULTS: In the development cohort, postoperative MAEs occurred in 375 patients and did not occur in 1,123 patients. LASSO-based multivariable analysis identified age, lactate level, neutrophil-to-lymphocyte ratio (NLR), Penn classification, severe true lumen compression, mesenteric artery involvement, and cardiopulmonary bypass (CPB) time as the core predictors for incorporation into the final nomogram. Model discrimination improved progressively with staged expansion. In the training set, the areas under the curve of models 1-5 were 0.725, 0.768, 0.782, 0.852, and 0.868, respectively; in the internal validation set, they were 0.718, 0.755, 0.776, 0.841, and 0.858; and in the external validation cohort, they were 0.705, 0.742, 0.765, 0.832, and 0.848, respectively. The largest incremental gain was observed after addition of multimodal CTA variables. The final nomogram showed good calibration and favorable clinical utility in both the training and external validation cohorts. Risk stratification in the development cohort demonstrated a marked stepwise increase in observed MAE incidence from 3.7% in the low-risk group to 27.9% in the intermediate-risk group and 74.2% in the high-risk group (P for trend <0.001).

CONCLUSIONS: A staged prediction model integrating preoperative multimodal imaging and perioperative clinical characteristics provided effective risk stratification for postoperative MAEs among patients with ATAAD. Preoperative CTA contributed the greatest incremental improvement beyond baseline clinical assessment, and the final nomogram demonstrated good discrimination, calibration, and clinical utility. This multimodal framework may support more precise perioperative risk assessment and individualized management for patients with surgically treated ATAAD.