Developing a predictive model for healing of clinically relevant pancreatic fistulas post-pancreatoduodenectomy based on clinical and imaging nutritional status.

Background: Clinically relevant pancreatic fistula (CR-PF) is a common complication following pancreaticoduodenectomy, and delayed healing (DH) of the fistula can lead to poor prognosis. We aimed to predict DH of CR-PF based on postoperative clinical and imaging-related inflammation and nutritional status in patients.

Methods: We retrospectively collected data from 121 patients who developed grade B CR-PF following pancreaticoduodenectomy (PD) at two centers between 2018 and 2023. Patients were categorized into early and delayed healing groups based on the median healing time. Postoperative contrast-enhanced CT was used to measure pancreatic thickness (PT), main pancreatic duct diameter (DPD), and psoas muscle index (PMI). Logistic regression analysis was used to identify risk factors associated with delayed healing and to construct a predictive model. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC), calibration curve, and decision curve analysis (DCA).

Results: Postoperative neutrophil-to-lymphocyte ratio, prognostic nutritional index, PMI, PT, and pancreatic texture (soft) were identified as independent risk factors for DH in patients with Grade B CR-PF. A comprehensive model was constructed based on these independent predictors. In the validation group, the model demonstrated an AUC of 0.840 (95% CI 0.711-0.969), and DCA and calibration curve indicated good predictive consistency and clinical utility.

Conclusion: The model constructed based on dual-center data can accurately predict DH of Grade B CR-PF, providing reference value for clinicians in nutritional support and surgical interventions for CR-PF.