Background: Postoperative complications are prone to occur in patients after radical pancreaticoduodenectomy (PD). This study aimed to construct and validate a model for predicting postoperative major complications in patients after PD.
Methods: The clinical data of 360 patients who underwent PD were retrospectively collected from two centers between January 2019 and December 2023. Visceral adipose volume (VAV) and subcutaneous adipose volume (SAV) were measured using three-dimensional (3D) computed tomography (CT) reconstruction. According to the Clavien-Dindo classification system, the postoperative complications were graded. Subsequently, a predictive model was constructed based on the results of least absolute shrinkage and selection operator (LASSO) multivariate logistic regression analysis and stepwise (stepAIC) selection. The nomogram was internally validated by the training and test cohort. The discriminatory ability and clinical utility of the nomogram were evaluated by area under the receiver operating characteristic (ROC) curve (AUC), calibration curve, and decision curve analysis (DCA).
Results: The major complications occurred in 13.3% (n = 48) of patients after PD. The nomogram revealed that high VAV/SAV, high system inflammation response index (SIRI), high triglyceride glucose-body mass index (TyG-BMI), low prognostic nutritional index (PNI) and CA199 ≥ 37 were independent risk factors for major complications. The C-index of this model was 0.854 (95%CI [0.800-0.907]), showing excellent discrimination. The calibration curve demonstrated satisfactory concordance between nomogram predictions and actual observations. The DCA curve indicated the substantial clinical utility of the nomogram.
Conclusion: The model based on clinical and CT indices demonstrates good predictive performance and clinical benefit for major complications in patients undergoing PD.
Keywords: Pancreaticoduodenectomy; Postoperative complications; Predictive model; Three-dimensional CT reconstruction.
© 2024 Wang et al.