Impact of Endoscopic Ultrasound on Diagnosis and Management of Pancreatic Cystic Lesions: Results from the Bicentric Retrospective GAME-CYST Study

BACKGROUND AND AIMS. Pancreatic cystic lesions (PCLs) are increasingly identified on cross-sectional imaging, creating a clinical dilemma between preventing malignant progression and avoiding unnecessary surgical or surveillance burdens. Although endoscopic ultrasound (EUS) is widely utilised as a second-line investigation, its real-world decision-making impact across unselected cohorts remains poorly quantified. We aimed to assess the incremental clinical value of EUS by measuring the "Game Changer" rate—defined as a categorical modification in presumed diagnosis, management recommendation, or both—and identifying its clinical and radiological predictors. METHODS Bicentric retrospective cohort study (Padua and Santiago de Compostela, 2015–2025) of consecutive adults with presumed PCLs undergoing index EUS (± FNA/FNB). Pre-EUS and post-EUS diagnosis and management were reconstructed from medical records. The primary endpoint was the Game Changer rate; predictors were explored by logistic regression and concordance with histology in resected patients. RESULTS. Index EUS was a Game Changer in 21.1% of patients (n = 68/322, 95% CI: 16.8%–26.0%), with consistent rates between centres (Padua 23.3% vs Santiago 18.5%, p = 0.293). Diagnostic reclassification occurred in 16.8% (n = 54), predominantly through increased identification of serous cystic neoplasms (SCNs; rising from 7.1% to 10.9%, p = 0.012) and complete resolution of all 10 baseline indeterminate cysts. Additionally, four mucinous lesions (1.2%) were reclassified as pancreatic adenocarcinoma. Management pathways changed significantly post EUS (Bowker’s test, p < 0.001) in 12.7% (n = 41): 15 patients (4.7%) escalated from surveillance to surgery, whereas surveillance was discontinued in 23 patients (7.1%), expanding the discharge group from 4.7% to 11.5% (p < 0.001). An additional 14 patients (4.3%) had their surveillance interval shortened due to intermediate-risk endosonographic features. Morphology alone accounted for 47.1% of modifications, while EUS-guided sampling (performed in 36.0%) contributed to 52.9%. On multivariable analysis, baseline worrisome features (OR 3.13, 95% CI: 1.51–6.50; p = 0.002) and FNA/FNB execution (OR 3.49, 95% CI: 1.81–6.75; p < 0.001) independently predicted a Game Changer, whereas comorbidity burden was inversely associated (ACCI: OR 0.84, 95% CI: 0.73–0.97; p = 0.021). In 46 resected patients, post-EUS presumed diagnosis matched overall neoplastic lineage in 95.7%, yielding advanced neoplasia (high-grade dysplasia or invasive carcinoma) in 60.9%. Procedure-related adverse events occurred in 2.6% of sampled patients (0.93% overall), with zero mortality. DISCUSSION AND CONCLUSIONS. EUS changed the diagnosis or management in about one in five patients with PCLs, and its impact was bidirectional and not limited to tissue acquisition: almost half of the changes derived from morphology alone, whereas sampling was decisive for detecting malignancy and withdrawing surgical indications. None of the 23 discontinuations concerned a final IPMN, being due to serous cystic neoplasms, pseudocysts or no cyst at EUS. The benefit was concentrated in patients with worrisome features and in those undergoing sampling. Main limitations are the retrospective design, no control group or systematic post-discharge follow-up, and histology only in resected patients (14.3%). Prospective studies with long-term follow-up are needed to confirm that these changes improve patient outcomes.