CT characteristics of postoperative fluid collections are associated with treatment intensity after pancreatoduodenectomy
Emil Pettersson, Inkeri Lehto, Tiina Lehtimäki, Katarina Johansson, Caj Haglund, Kethe Hermunen, Hanna SeppänenAbstract
Introduction
Postoperative complications after pancreatoduodenectomy frequently manifest as intra-abdominal fluid collections visualized on computed tomography (CT), with features such as wall enhancement, gas, hemorrhage, inflammatory changes, and loculated morphology. Many collections are benign, reflecting normal postoperative changes, making it difficult to distinguish those requiring treatment. This study aimed to evaluate whether CT morphologic features associate with treatment intensity.
Methods
This retrospective single-centre study included 117 consecutive patients undergoing pancreatoduodenectomy with postoperative CT performed. Intra-abdominal fluid collections were identified and characterized by radiologists using predefined variables, including loculated morphology, demarcation, wall enhancement, perilesional fat stranding, gas, hematoma, and size. A total of 99 patients had fluid collections; 76 had a single collection, 18 had two, and 5 had three. For patient-level analysis, a single index collection was selected and defined as the largest collection for that patient. Associations between CT features and treatment intensity groups (no treatment, antibiotics, percutaneous drainage, reoperation) were analysed using SPSS 30.
Results
Collection size increased with treatment intensity (P = 0.002). Loculated morphology was associated with higher treatment intensity (P = 0.001) and independently predicted reoperation (OR 7.1, P = 0.023). Surrounding edema showed a non-significant trend (P = 0.067). Other CT features were not associated with outcomes.
Discussion
Loculated morphology was associated with higher treatment intensity (P = 0.001) and independently predicted reoperation after adjustment for size (OR 7.1, P = 0.023). Collection size was associated with treatment intensity (P = 0.002). These findings suggest that CT morphology can aid in risk assessment after pancreatoduodenectomy.