Trimodality Therapy for Esophageal Cancer: Radiation to the Gastric Conduit is Not Associated with Post-operative Anastomotic Complication

Since publication of the CROSS study, the majority of patients with locally advanced esophageal cancer are treated with neoadjuvant chemoradiotherapy prior to esophagectomy (trimodality therapy, TMT). The stomach, specifically the fundus and greater curvature, is used to re-create neo-esophagus and is often located in the pre-operative radiation (RT) field. This study aims to determine if RT to the stomach and its substructures correlate with anastomotic complication (AC). From 2007 to 2016, patients with esophageal cancer treated with TMT at a tertiary academic institution with restored digital RT plans were included in this IRB approved study. Clinical and pathologic data were obtained retrospectively. The stomach and its substructures were contoured on the preoperative RT planning scans and dosimetric parameters were extracted. AC was defined as a leak and/or a stricture. Fisher-exact and Wilcoxon rank-sum tests were used. 96 patients were included in the analysis. The median age was 63 (55-70); 82% of patients were male. Majority of the tumors were located in the distal esophagus (52%) and gastroesophageal junction (GEJ, 30%) versus 15% and 3% in the middle and proximal esophagus, respectively. Adenocarcinoma was the predominant histology (80%). Carboplatin and paclitaxel was used in majority of the patients (36%). Median RT dose was 45 Gy (range, 41.4-54 Gy). Transhiatal esophagectomy was the common surgery performed (50%), followed by Ivor-Lewis (33%), 3-incision (14%) and other (3%). AC developed in 57% of the patients (47 strictures,49%; 16 leaks, 17%; 8 both, 8%). Diabetes (DM) was more common in patients with AC (n=16, 29% vs. n=7, 17%, p=0.172). Esophagectomy type was associated with AC (p<0.000). The predominant type of surgery in patients with AC was transhiatal (62%) followed by 3-incision (18%), Ivor-Lewis (16%) and others (4%). The mean dose to the stomach, fundus, and greater curvature was 27 Gy (18-35 Gy), 33 Gy (23-45 Gy), and 23 Gy (13-33 Gy), respectively. RT dose to the stomach and its substructures were not associated with AC (p>0.05). Of the 47 patients with distal/GEJ tumor location and esophagectomies with cervical anastomoses, 72% of them developed AC (31 strictures, 66%; 11 leaks, 23%; 8= both, 17%). DM was more common in patients with AC (n=9, 27% vs. 0%, p=0.047). RT was not associated with AC (p>0.05). Our analysis does not demonstrate an association between RT to gastric substructures that form the neo-esophagus and AC. However, it does show an increased risk of AC in patients who undergo esophagectomies with cervical anastomosis. In this population, the majority with a leak had anastomosis in the neck and distal esophageal or GEJ tumor undergoing cervical anastomosis. Patients with distal tumors may be better served with anastomosis in the thorax rather than neck. Further analysis in a larger study population is needed to clarify associations.