OA02.5. Combined Intraoperative Fluorescence Angiography With Indocyanine Green and Postoperative Endoscopic Ischemia Algorithm Predicts Cervical Anastomotic Leak After Esophagectomy

Abstract

Esophageal Cancer: Surgical Treatment of Esophageal Cancer – early outcomes and complications Anastomotic leakage remains a serious complication following esophagectomy with gastric conduit reconstruction. We developed a postoperative endoscopic mucosal ischemic index (Kobayashi et al. Langenbecks Arch Surg. 2023). We evaluated whether a predefined algorithm combining intraoperative indocyanine green (ICG) perfusion assessment with postoperative endoscopic findings improved leak prediction. Patients who underwent esophagectomy with gastric conduit reconstruction via the retrosternal route were prospectively analyzed (June 2020–Dec 2025). Intraoperative ICG fluorescence assessment after conduit creation classified perfusion as sufficient (SP) or insufficient (IP) using a predefined threshold. All patients underwent postoperative upper endoscopy on postoperative days 4–15 to grade conduit mucosal ischemia using a standardized approach. The primary outcome was clinically relevant cervical anastomotic leakage. The concordance between ICG and endoscopic ischemia was assessed. The predictive performances of (1) ICG alone, (2) endoscopy alone, and (3) the predefined combined algorithm were evaluated using multiple performance metrics, including sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), accuracy, F1 score, and area under the curve (AUC). Ninety-one patients were included, and anastomotic leakage occurred in 5. Leakage was observed in 0/73 SP versus 5/18 IP on ICG (P<0.001). Endoscopic mucosal ischemia was also associated with leakage: 0/79 without ischemia versus 5/12 with ischemia (P<0.001). ICG and endoscopic ischemia were in agreement in 89.0% of cases. PPV improved from 27.8% (ICG, 5/18) and 41.7% (endoscopy, 5/12) to 50.0% (algorithm), accompanied by higher accuracy (85.7%, 92.3%, and 94.5%) and AUC (0.924, 0.959, and 0.971), respectively. Decision analysis: Although endoscopy was performed in all patients, a modeled strategy reserving postoperative endoscopy for ICG-IP patients would avoid endoscopy in 73/91 (80.2%) patients with no missed leaks in this cohort. A predefined algorithm integrating intraoperative ICG perfusion assessment with postoperative endoscopic mucosal ischemia grading improved precision and AUC while maintaining perfect sensitivity in this cohort. While endoscopy was performed in all patients, a modeled strategy limiting endoscopy to ICG-IP patients suggests that approximately 80% of endoscopies could be avoided without missing leaks. Larger cohorts and external validation are warranted. Video Description