Predictive factors for major postoperative complications related to gastric conduit reconstruction in thoracoscopic esophagectomy for esophageal cancer: a case control study

From 2006 to 2015, 105 patients with esophageal cancer underwent esophagectomy and lymph node dissection at the Department of Surgery at Nagasaki University Hospital. Treatment plans for each patient were provided according to the clinical guidelines edited by the Japan Esophageal Society [13]. We chose open esophagectomy for the patients with severe adhesions in the chest or invasive neoplasia with lymph node involvement. Thirty patients were excluded because they required open esophagectomy with lymph node dissection. Seventy-five consecutive patients with esophageal cancer who underwent TE were retrospectively studied to evaluate the predictive factors for major complications related to gastric conduit after TE. The rules for classification and staging corresponded to the 7th edition of the International Union Against Cancer (UICC)/American Joint Committee on Cancer (AJCC) Tumor Node Metastasis (TNM) staging system [14].

TE was performed from the right side in the left lateral position. Esophagectomy with lymphadenectomy in the mediastinum and around both recurrent nerves were performed. In the abdominal section, hand-assisted laparoscopic gastrectomy was performed to remove the mobilized esophagus with lymphadenectomy around the left gastric artery and aorta. After mobilization of the full stomach and esophagus, a gastric conduit was created by dividing the lesser curve of the stomach. The right gastric and right gastroepiploic artery provided the vascular supply to the created gastric conduit. In 73 patients, the gastric conduit was pulled up in the post-sternal route; in 2 patients, it was pulled up in the post-mediastinal route. The esophagogastrostomy was performed in the neck by end-to-side anastomosis. A 21-mm or 25-mm intraluminal stapler was used as the stapling device (CDH21, CDH25, Ethicon Ltd., Edinburgh, United Kingdom). The inserted part of the gastric conduit was crossed by linear stapling. All staple lines were oversewn.

Three-field lymph node dissection (3FLND) was performed in patients who had upper thoracic esophageal cancer or middle or lower thoracic esophageal cancer with lymph node metastasis in the neck region or around the right recurrent nerve [15].

This study was approved by the Ethics Committee of Nagasaki University Hospital (16082215). The written informed consent from the patients was waved from the Ethics Committee because the information on the opportunity to opt out was presented on the web site (http://​www.​mh.​nagasaki-u.​ac.​jp/​research/​rinsho/​patients/​open_​surgery2.​html). This study was registered in the UMIN Clinical Trials Registry as UMIN000024436.

The nasogastric tube was removed before anastomosis. On the first postoperative day (1POD), transintestinal nutrition was started from a jejunostomy feeding tube to prevent postoperative complications [16]. In the first three postoperative days, the patients without hoarseness and aspiration pneumonia started to drink fluids, followed by a soft diet.

Major complications related to gastric conduit after TE were defined as anastomotic leakage, refractory anastomotic strictures, and gastric conduit necrosis. Anastomotic leakage was defined as fistula formation that required any invasive treatment (Clavien-Dindo classification of grade III or more). Anastomotic strictures were defined as the presence of a lumen requiring endoscopic balloon dilatation for the passage of a normal endoscope (9.2 mm diameter) with symptomatic dysphagia. Refractory esophageal strictures were defined as more than 5 sessions of balloon dilation 6 months after the operation [17, 18]. Gastric conduit necrosis was defined as a severe ischemic condition that required resection of the gastric graft.

The data are expressed as the means ± standard deviation (SD) or medians and interquartile ranges (IQR). The relationships among major complications related to gastric conduit and age, body mass index (BMI), total operation time, operation time of thoracic surgery, and C-reactive protein (CRP) at 1POD were evaluated using Student’s t-tests. The relationships between major postoperative complications related to gastric conduit reconstruction and other values were evaluated using Wilcoxon’s tests. The relationships among anastomotic leakage and age, body mass index (BMI), total operation time, operation time of thoracic surgery, and C-reactive protein (CRP) at 1POD were evaluated using Student’s t-tests. The relationships between anastomotic leakage and other values were evaluated using Wilcoxon’s tests. The relationships among refractory anastomotic strictures and age, body mass index (BMI), total operation time, operation time of thoracic surgery, and C-reactive protein (CRP) at 1POD were evaluated using Student’s t-tests. The relationships between refractory anastomotic strictures and other values were evaluated using Wilcoxon’s tests. Receiver operating characteristic (ROC) curves and the area under the ROC curve (AUC) were used to assess the feasibility of using CRP and creatine phosphokinase (CPK) at 1POD as diagnostic tools for major complications related to gastric conduit [19]. The 95% CI values greater than 0.5 for AUC indicated that prediction was better than chance [20]. The patients were divided into two groups according to the cut-off values of CRP and CPK at 1POD. The relationships of categorical clinical factors between the groups were analyzed using chi-square tests or Fisher’s exact tests. A Fisher’s exact test was applied if the theoretical frequency was less than five. Probability values (P) less than 0.05 were considered statistically significant. Multiple logistic regression (stepwise) models were developed, and odds ratios (OR) were used to evaluate predictive factors associated with major complications related to gastric conduit. The Cochrane-Armitage trend test was used to test for a linear trend in the proportion of patients who developed major postoperative complications related to gastric conduit reconstruction according to numbers of predictive factors. All statistical analyses were performed using SAS-JMP programs for Windows (SAS Institute Inc., Cary, NC).

The clinical characteristics of the 75 patients, which included 18 females and 57 males, are summarized in Table 1. The average age of all patients was 61.3 ± 8.1 years. The average BMI of all patients was 21.3 ± 2.7. Preoperative chemotherapy was performed in 51 patients (68.0%). Three patients (4.0%) were diagnosed with adenocarcinoma, and 72 patients (96.0%) were diagnosed with squamous cell carcinoma. According to the TNM classification, 47 patients (62.7%) had tumors more advanced than stage I. 3FLND was performed in 23 patients (30.7%). The average operating time was 605 ± 114 min. The median estimated blood loss was 370 g (IQR 270–600). Blood transfusion was performed in 7 patients (9.3%). The median length of the postoperative hospital stay was 27 days (IQR 20–39).

The major complications related to gastric conduit after TE are summarized in Fig. 1. Twenty-three patients (30.7%) developed major complications related to gastric conduit reconstruction after TE. Anastomotic leakage occurred in 17 patients who required drainage to manage infectious conditions. No patients died within 30 days after the operation due to anastomotic leakage. A stricture occurred in 33 patients who required endoscopic balloon dilation. Twenty patients developed simple esophageal strictures without other gastric conduit-related complications. Seven patients developed anastomotic leakage followed by simple esophageal strictures. Six patients developed refractory esophageal strictures. All patient with refractory strictures developed symptomatic strictures within 2 months after TE (28.0 ± 7.0 days). Two patients developed anastomotic leakage followed by refractory esophageal strictures. Two patients had gastric conduit necrosis, and one of these two patients died due to non-occlusive mesenteric ischemia after resection of the necrotic gastric conduit. The length of postoperative hospital stay after TE in the patients with major complications related to gastric conduit (39 days, IQR 28–47) was significantly longer than in those without these complications (22 days, IQR 19–28) (P <  0.01).

The predictive factors for developing major complications related to gastric conduit are shown in Table 2. In a univariate analysis, the predictive factors for developing major complications related to gastric conduit included age, 3FLND and levels of CRP and CPK at 1POD. The AUC for CRP and CPK at 1POD was 0.684 (95%CI; 0.546–0.796) and 0.670 (95%CI; 0.514–0.796). ROC curve analysis also identified the following cut-off values for CRP and CPK at 1POD: 9.6 × 10^4 μg/L and 1164 IU/L, respectively (Fig. 2). At a threshold of 9.6 × 10^4 μg/L for CRP at 1POD, the optimal sensitivity and specificity were 73.9% and 65.4%, respectively, in patients developing major complications related to gastric conduit. At a threshold of 1164 IU/L for CPK at 1POD, the sensitivity and the specificity were 69.6% and 75.0%, respectively.

The predictive factors for developing major complications related to gastric conduit are shown in Tables 3 and 4. In a univariate analysis, the predictive factors for developing anastomotic leakage related to gastric conduit included levels of CRP and CPK at 1POD.

When a multiple logistic regression analysis was performed to evaluate confounding factors, 3FLND and the levels of CPK and CRP at 1POD were found to be significantly associated with developing major complications related to gastric conduit (Table 5). The incidence rates of these complications for 0, 1, 2, and 3 predictive factors were 5.3% (1/19), 18.8% (6/32), 58.8% (10/17), and 85.7% (6/7), respectively (Fig. 3). There was a strong trend toward increasing the prevalence of major complications related to gastric conduit based on the number of predictive factors (P <  0.01). The accuracy of 2 or more factors for major complications related to gastric conduit after TE was 0.800.