Predicting the need for step-up after EUS-guided drainage of peripancreatic fluid collections, including Quadrant-Necrosis-Infection score validation: a prospective cohort study

Abstract

Background and aims: Factors predicting the need for step-up procedures after endoscopic ultrasound (EUS)-guided fluid collection drainage (EUS-FCD) of peripancreatic fluid collections (PFCs) were explored in retrospective studies restricted to walled-off necrosis (WON) and lumen-apposing metal stents (LAMSs).

Methods: All consecutive candidates for EUS-FCD from 2020 to 2024 were included in a Prospective Registry of Therapeutic EUS (PROTECT, NCT04813055), with prospective monthly follow-up evaluating clinical success, adverse events, and recurrences. Prospectively assessed baseline clinical and morphologic factors, including the Quadrant-Necrosis-Infection (QNI) classification, were included in a stepwise logistic regression model to predict the need for step-up. The agreement between EUS and radiology in assessing the extent of necrosis was compared with the use of Cohen's kappa.

Results: Seventy patients (29 postsurgical collections, 21 pseudocysts, and 20 WONs) were treated with double-pigtail plastic stents (DPPSs) in 59% of cases and LAMSs in 41%. Clinical success was 92.9%, with a need for step-up (mostly endoscopic necrosectomy) in 35.7% of cases. Necrosis ≥60% (odds ratio [OR], 7.7; 95% confidence interval [CI], 1.4-43) and being in the high-risk QNI group (OR, 4.6; 95% CI, 1.4-15) were the only independent predictors of any step-up. The same factors predicted the endoscopist's decision to allocate PFCs to LAMSs vs DPPSs. The high-risk QNI group was associated with a significantly longer hospital stay (12 days vs 4 days; P = .004). EUS tended to upscale the necrotic content compared with preprocedural radiology (κ = 0.31).

Conclusions: The extent of necrosis and the QNI classification strongly correlated with the need for step-up and allocation to LAMS versus DPPS drainage, suggesting a central role in treatment personalization.