Introduction

Endoscopic submucosal dissection (ESD) is widely implemented in Asia. BEST J-score is a clinically useful prediction model validated in 2021, based on this population, to stratify the risk of delayed bleeding (DB) after gastric ESD. Experience from the Western world is still scarce.

Aims & Methods

The aim of our study is to validate BEST-J score in a Western sample.
Retrospective analysis with a prospectively maintained database of all patients undergoing ESD at a Western tertiary center from June 2016 to December 2022.
Gastric ESD performed by a single operator were included in this study. BEST-J score was applied for all procedures, which comprised 10 variables (intake of warfarin, direct oral anticoagulant (DOAC), chronic kidney disease with haemodialysis, intake of P2Y12 receptor antagonist, aspirin, cilostazol, tumour size >30 mm, lower-third in tumour location, presence of multiple tumours and interruption of each kind of antithrombotic agents). The primary endpoint was to validate BEST-J score accuracy in predicting DB (up to 28 days after ESD).

Results

Of 477 ESD performed, a total of 252 ESD met the inclusion criteria; 56.5% were male, with a mean age of 68 years (SD 8 years). Of this, 30 patients (15.3%) were under anticoagulant therapy, 19 of them (63.3%) with DOAC. Antiplatelet agents were taken by 48 patients (19%). From these, 5 (2.0%) presented DB following ESD, with a median time to bleeding of 11 days.
According to BEST-J score, bleeding risk was low (0-1 points) in 205 procedures (81,3%), intermediate (2 points) in 18 (7.1%), high (3-4 points) in 26 (10.3%) and very high (≥5 points) in 3 (1.2%).
BEST-J score presented an acceptable accuracy predicting DB in our sample, with an AUC=0.777 (p=0.034, CI 95%, 0.58-0.975). The optimal cut-off value to predict DB was a BEST-J score ≥ 3, which matches the cut-off value for high-risk of bleeding in the original investigation. This value had a sensitivity of 60.0%, specificity of 88.7% and a negative predictive value of 99.1%.

Conclusion

The BEST-J score presents an acceptable accuracy for post-ESD bleeding and it is particularly useful in identifying low risk DB patients. Therefore, this model is a good clinical decision-making support tool in the Western population.