Patients with walled-off pancreatic necrosis (WOPN) are burdened by a high risk of mortality. Primary aim was to identify early clinical and procedural risk factors for mortality after endoscopic ultrasound (EUS)-guided drainage of WOPN. Secondary aim was to compare clinical and procedural characteristics before and after EUS-guided WOPN drainage between survivors and patients experiencing WOPN-related death.

In a retrospective study, all patients with WOPN treated with EUS-guided drainage with lumen apposing metal stent (LAMS) were enrolled. WOPN characteristics were assessed by the QNI system. Risk factors were searched among pre-procedural, procedural and early (occurring <14 days from EUS-guided drainage) post-procedural items. Inclusion criteria: 1) age >18 years; 2) WOPN treated with LAMS. Exclusion criteria: 1) pancreatic pseudocystis; 2) post-surgical pancreatic collections; 3) pancreatic neoplasms. Data expressed as median [range], differences assessed with the X² test, the Student t-test or the Mann-Whitney u-test, risk factors searched by using uni- and multivariate logistic regression.

61 patients with WOPN treated by EUS-guided drainage with LAMS were enrolled (age: 63 [33-86] years). Overall, 8 (13.1%) patients died during hospitalization due to WOPN-related complications. Endoscopic necrosectomy was performed in a comparable proportion of patients between survivors and patients experiencing mortality (35 [66.04%] vs 6 [75%], p=0.92). In patients experiencing mortality, higher median BMI (24.2 [14-39] vs 27.7 [24.5-40.5], p=0.027), pre-procedural procalcitonin levels (0.13 [0-16] vs 2.5 [0.2- 20.8], p=0.0003), frequency of N2 (≥30% necrosis) (12 [22.6%] vs 6 [75.0%], p=0.009) and frequency of post-procedural percutaneous drainage placement (11 [18.9%] vs 5 [62.5%], p=0.038) were observed. At multivariate analysis, intensive care unit recovery (4.4 [1.9-7.5], p=0.0009), higher pre-procedural procalcitonine values (1.9 [1.05-3.4], p=0.04), and duodenal double-gating (2.05 [1.08-3.7], p=0.03) were identified as risk factors for mortality, while a BMI <25 was protective (0.24 [0.15-0.37], p=0.005).

In the present study only non-modifiable predictive factors for mortality have been identified. Nonetheless, recocgnition of early predictive factors for WOPN-related mortality may suggest the need for treatment optimization or accelerated step-up strategy.