3D modeling to predict vascular involvement in resectable pancreatic adenocarcinoma

Background: Current management of patients with borderline resectable pancreatic adenocarcinoma (BR-PDAC) depends on the degree of involvement of the major arterial and venous structures. The aim of this study was to evaluate 3D segmentation and printing to predict tumor size and vascular involvement of BR-PDAC to improve pre-operative planning of vascular resection and better select patients for neoadjuvant therapy.

Methods: We retrospectively evaluated 16 patients with BR-PDAC near vascular structures who underwent pancreatoduodenectomy (PD) with or without vascular resection between 2015 and 2021. The pre-operative computed tomography (CT) images were processed by segmentation with 3D reconstruction and printed as 3D models. Two radiologists specialized in pancreatic imaging and two pancreatic surgeons blindly and independently analyzed the pre-operative CT scans and 3D models using a defined checklist. Their evaluations were compared to the pre-operative 2D-CT reports utilized for patient management. A positive delta was defined by the 3D analysis resulting in greater accuracy in predicting vascular involvement as proven intraoperatively or histopathologically.

Results: Fourteen PD, one total pancreatectomy, and one exploratory laparotomy were performed. Ten patients had a positive delta concerning vascular involvement of the superior mesenteric or portal vein. Tumor extension was also more accurately evaluated by 3D modeling than by 2D-CT (p < 0.05).

Conclusions: Our pilot study demonstrates that 3D segmentation can provide additional information for choosing the best treatment strategy and surgical plain in patients with BR-PDAC. Especially for upcoming mini-invasive techniques like laparoscopic and robotic resections, better pre-operative planning is essential to allow safety and prevent vascular injury.