227. DETERMINING SEVERITY OF ESOPHAGEAL ANASTOMOTIC LEAK IN PATIENTS AFTER ESOPHAGECTOMY: DEVELOPMENT OF THE SEAL SCORE

Abstract

Abstract Anastomotic leak (AL) is a common and severe complication after esophagectomy, but it is largely unknown how to assess leak severity when diagnosed. We aimed to develop the Severity of Esophageal Anastomotic Leak (SEAL) score, a tool to determine leak severity at diagnosis of AL. The TENTACLE—Esophagus study is an international, retrospective cohort study in 71 centres worldwide and included patients with AL after esophagectomy between 2011 and 2019. The primary endpoint was 90-day mortality. To develop the SEAL score, multivariable logistic regression was performed to determining 90-day mortality risk. Leak-related parameters at diagnosis identified as prognostic factors were included as predictors for the SEAL score. Four classes of leak severity were defined (mild, moderate, severe and critical) and the score was internally validated. The study protocol is accessible at www.tentaclestudy.com. 1509 patients with AL were included and the 90-day mortality rate was 11.7%. Twelve leak-related predictors factors were included in the SEAL score, including post-operative day of diagnosis, leucocyte count, organ failure, defect circumference and presence of intrathoracic fluid collections. The score showed good calibration and discrimination with a c-index of 0.77 (95%CI 0.73–0.81). More severe leaks were associated with significant increase in length of (intensive care unit (ICU)) stay, healing time, Comprehensive Complication Index (CCI) and Esophagectomy Complications Consensus Group (ECCG) classification (Table 1). The SEAL score was incorporated in an online tool for use in individual cases: https://www.tentaclestudy.com/seal-score. The SEAL score grades leak severity into four classes based on individual predictions of 90-day mortality by combining 12 leak-related predictors. The SEAL score also reflected morbidity and can be used to assess AL severity, may guide treatment decisions and may be used in research on AL.