Endoscopic versus percutaneous drainage placement after left-sided pancreatectomy: multi-state model for defining objectifiable variables on interventional treatment for pancreatic fistula.

Background: Postoperative pancreatic fistula (POPF) remains a prevalent complication after left-sided pancreatectomy. Interventional treatment is regularly required and relies on interdisciplinary concepts, including interventional radiology and endoscopy. Evidence on the correct indication and clinical algorithms for available treatment modalities are needed.

Methods: During a 5-year observational period, we followed the clinical courses of 200 patients after left-sided pancreatectomy. Patients who fulfilled clinical and radiographic criteria for interventionally-relevant(IR-) POPF were identified. Interventional treatment groups were stratified via established step-up concepts for descriptive analyses and via a decision-based algorithm for comparative analyses. We used a Multi-State Time-to-Event Model (MSM) to assess clinical resolution efficacy.

Results: Following surgery, patients with IR-POPF needed multiple (58.2 %) and multimodal (49.4 %) interventions to achieve final resolution. Technical success rates were higher in ED (57.7 %) versus PD (32.7 %), and resulted in earlier resolution (8.4d, 95%C [5.5-11.2d]) when compared to the PD group (23.4d, 95%CI[2.8-17.9d]). At any given time, probability of reaching clinical resolution were higher for ED in multifaceted MSM analysis (HR 1.39, 95%CI[0.854, 2.25]).

Conclusion: The present study highlights the importance of endoscopy-guided drainage placement after left-sided pancreatic resection. CRP value appeared to be a single objectifiable parameter for optimized treatment allocation.