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

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.