Impact of Reconstruction Route on Postoperative Morbidity After Esophagectomy: Analysis of Esophagectomies in the Japanese National Clinical Database

Abstract Background Esophagectomy followed by gastric conduit reconstruction is a standard surgical procedure for esophageal cancer. However, there is no evidence of the superiority or inferiority of the posterior mediastinal (PM) versus the retrosternal (RS) reconstruction route with regard to short‐term outcomes after esophagectomy. We aimed to elucidate whether the reconstruction route can affect the short‐term outcomes after esophagectomy followed by gastric conduit reconstruction. Methods We reviewed the clinical data of patients who underwent esophagectomy between 2016 and 2018 from the Japanese National Clinical Database. This study included 9786 patients who underwent gastric conduit reconstruction through the PM or RS route with cervical anastomosis. Results Of the 9786 patients analyzed, 3478 and 6308 underwent gastric conduit reconstruction thorough the PM and RS routes, respectively. The incidence of anastomotic leak and surgical site infection (SSI) was significantly lower in the PM group than in the RS group (11.7% vs 13.8%, P = .005 and 8.4% vs 14.9%, P < .001, respectively), while the incidence of pneumonia was higher in the PM group (13.7% vs 12.2%, P = .040). Generalized estimating equation logistic regression analysis revealed a higher risk of anastomotic leak and SSI (odds ratio [OR], 1.32; 95% confidence interval [CI], 1.15–1.51; P < .001 and OR, 2.06; 95% CI, 1.78–2.38; P < .001, respectively) and a lower risk of pneumonia (OR, 0.86; 95% CI, 0.75–0.98; P = .028) in the RS group than in the PM group. Conclusion The findings of this study will help surgeons to design the reconstruction route following esophagectomy.


| INTRODUC TI ON
Esophageal cancer is one of the leading causes of cancer-related deaths worldwide. 1 Esophagectomy plays an important role in the multidisciplinary treatment strategies for esophageal cancer. [2][3][4] In the Asia-Pacific region, including Japan, the vast majority of esophageal cancers are squamous cell carcinomas (SCC) located at the upper to lower thoracic esophagus. [4][5][6] Surgical treatment of thoracic esophageal SCC generally consists of subtotal esophagectomy, twoor three-field lymphadenectomy, and reconstruction using organs such as the stomach. [6][7][8] Recent advances in surgical techniques and perioperative management have enabled us to safely perform onestage esophagectomy with radical lymphadenectomy followed by reconstruction; however, esophagectomy remains a highly invasive procedure that can lead to severe morbidities such as anastomotic leak, respiratory complications, and cardiac events. [9][10][11] The stomach is the most commonly used organ for reconstruction following esophagectomy. 3,6,8 Reconstruction using a gastric conduit can be performed via the posterior mediastinal (PM), retrosternal (RS), or subcutaneous (SC) route, among which PM and RS comprise the vast majority of reconstruction routes due to advantages over the SC route, such as shorter reconstruction route and fewer cosmetic changes after esophagectomy. 6,8 Esophagogastric anastomosis is generally made at the cervical incision in cases of RS reconstruction. Cervical or intrathoracic anastomosis can be used in cases of PM reconstruction, and cervical anastomosis is commonly performed in several Japanese institutes. 6 There are some controversies regarding the superiority or inferiority of PM reconstruction to RS reconstruction; however, there is no evidence on the impact of the reconstruction route on short-term outcomes after esophagectomy based on large-scale clinical data.
In this study we reviewed the clinical data of 17,478 patients who underwent esophagectomy followed by one-stage reconstruction using the National Clinical Database (NCD), a nationwide, web-based, data entry system in Japan. 10 -14 We sought to elucidate whether the reconstruction route can affect the shortterm outcomes after esophagectomy followed by gastric conduit reconstruction.

| NCD data registration
Details of the data registration system of the Japanese NCD are available elsewhere. 10 -14 The NCD started data registration in January 2011, and ~1,500,000 cases are registered annually from over 5000 institutions, which corresponds to >95% of surgeries in Japan. 14 For the eight major gastroenterological surgery procedures, including esophagectomy, that were determined to represent the performance of surgery in each specialty, the input of detailed items, such as lab data and operative morbidities, are requested. Since January 2016, more detailed information, including the reconstruction organ, reconstruction route, and the location of anastomosis have been included in the requested items for esophagectomy cases. The Union for International Cancer Control TNM staging version 7 was adopted to classify the pretreatment tumor stages. All variables, definitions, and inclusion criteria for the NCD were accessible online by the participating institutions (http://www.ncd.or.jp/). The NCD supports an E-learning system that can be used by participants to enter consistent data. The patient variables in the NCD were almost identical to those used by the American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP). 15,16

| Study design
This was a retrospective cohort study. We reviewed the clinical data of patients who underwent esophagectomy between 2016 and 2018 from the NCD Japan. According to the inclusion criteria, only thoracic esophageal cancer patients who underwent esophagectomy followed by gastric conduit reconstruction through the RS or PM route with cervical anastomosis were included in this study. Patients with metastatic or recurrent diseases, those with clinical T4, TX, NX, or M1 tumors, those who underwent emergency operation, those <18 y, and those with missing data were excluded from this study. interval [CI], 1.15-1.51; P < .001 and OR, 2.06; 95% CI, 1.78-2.38; P < .001, respectively) and a lower risk of pneumonia (OR, 0.86; 95% CI, 0.75-0.98; P = .028) in the RS group than in the PM group.

Conclusion:
The findings of this study will help surgeons to design the reconstruction route following esophagectomy.

K E Y W O R D S
anastomotic leak, national clinical database, pneumonia, posterior mediastinal route, retrosternal route, surgical site infection

| Endpoints
The primary outcome measure was the incidence of major postoperative morbidities, including anastomotic leak; surgical site infection (SSI), including superficial incisional SSI, deep incisional SSI, and organ space SSI with or without leak; and pneumonia. The secondary endpoints included the 30-d and operative mortality rates, operation time, and bleeding. Operative mortality included all patients who died within the index hospitalization period, regardless of the length of hospital stay (up to 90 d), and any patient who died after hospital discharge (up to 30 d after surgery).

| Statistical analysis
For categorical variables, the proportion of patients experiencing the abovementioned outcomes was compared between the PM and RS groups using Pearson's chi-squared test, and continuous variables were compared using the Mann-Whitney U-test.  (1-8, 9-18, 19-41, ≥42 cases/year) were included in the multivariable analysis based on our previous study. 17 All statistical tests were two-sided, and P < .05 was considered significant.
All statistical analyses were performed using R v. 3.6.3 (2020; R Foundation for Statistical Computing, Vienna, Austria).

| Study population
Between January 2016 and December 2018, the NCD registered 17,478 patients who underwent esophagectomy followed by onestage reconstruction at Japanese institutes. According to the inclusion and exclusion criteria, 9786 thoracic esophageal cancer patients who underwent esophagectomy followed by gastric conduit reconstruction through the RS or PM route with cervical anastomosis were included for analysis. Of the total analysis population, 3478 (35.5%) and 6308 (64.5%) patients underwent gastric conduit reconstruction thorough the PM route and RS route with cervical anastomosis, respectively ( Figure 1).

| Risk profile
Of the 9786 thoracic esophageal cancer patients who underwent esophagectomy followed by gastric conduit reconstruction thorough either the PM or RS route with cervical anastomosis, 80.2% were male and 41.8% were elderly patients ≥70 y old. Among these patients, 36

| Background characteristics of the PM and RS groups
Most of the preoperative variables, such as age, BMI, ASA-PS, and previous cerebrovascular disease were equivalent between the PM and RS groups; however, the rates of DM with insulin use, congestive F I G U R E 1 Selection process for the study population heart failure, and past cardiac surgery were significantly higher in the PM group than the RS group (3.5% vs 2.4%, P = .002; 0.3% vs 0.1%, P = .024; and 1.2% vs 0.2%, P < .001, respectively) ( Table 1). Regarding preoperative lab data, the number of patients with serum creatinine ≥1.2 mg/dl was higher in the PM group than the RS group (7.4% vs 5.8%, P = .001) ( Table 1). In contrast, clinical T stage and N stage were higher in the RS group than the PM group (P = .002 and .019, respectively) ( Table 1). There was no significant difference in the histological type of esophageal cancer between the two groups (P = .695) ( Table 1). MIE was performed more frequently in the PM group than the RS group (P < .001) ( Table 1). The RS route was more commonly selected for reconstruction after esophagectomy in hospitals with a higher esophagectomy case volume per year (P < .001) ( Table 1).

| Effect of reconstruction route on operation time and bleeding
The median operation time and bleeding in the 9786 patients who underwent esophagectomy followed by gastric conduit reconstruc-

| Effect of reconstruction route on postoperative morbidities
Major postoperative morbidities, including anastomotic leak, SSI, and pneumonia were observed in 13.0%, 12.6%, and 12.7%, respectively, of the 9786 patients who underwent esophagectomy followed by gastric conduit reconstruction with cervical anastomosis ( Table 2). The rates of anastomotic leak and SSI were significantly lower in the PM group than the RS group (11.7% vs 13.8%, P = .005 and 8.4% vs 14.9%, P < .001, respectively). In contrast, the rate of pneumonia was higher in the PM group than the RS group (13.7% vs 12.2%, P = .040) ( Table 2).

| Mortality
The mortality rates of esophagectomy followed by gastric conduit reconstruction are presented in Table 2. The 30-d mortality was 0.6%, and the overall operative mortality was 1.0%. There was no significant difference in either 30-d or operative mortality rates between the PM and RS groups (0.5% vs 0.7%, P = .445 and 0.9% vs 1.0%, P = .835, respectively).

| Risk comparison of postoperative morbidities between the PM and RS groups
As shown in Figure 2, GEE logistic regression analysis revealed that

| D ISCUSS I ON
In this study we analyzed 9786 thoracic esophageal cancer patients in the NCD who underwent esophagectomy followed by gastric conduit reconstruction with cervical anastomosis. Our results demonstrated the RS route as a risk factor for anastomotic leak and SSI and the PM route as a risk factor for pneumonia. To the best of our knowledge, this is the first and largest comparison report based on a nationwide database, to elucidate the impact of reconstruction route on the short-term outcomes after esophagectomy.
In daily medical practice, the route used as the first choice for reconstruction after esophagectomy is dependent on the institutional policy or surgeons' preference. However, the final reconstruction route is generally decided based on the tumor stage and the patient's comorbidities, such as DM, which is a known risk factor for anastomotic leak and SSI, 11,18 and cardiac dysfunction, which may be affected by the stomach in the RS route. 19 the front of the heart. In cases of local recurrence following PM reconstruction, tumors may invade the gastric conduit, leading to secondary complications, such as bleeding and stenosis, and difficulty in radiation therapy. 21 These general backgrounds in deciding the reconstruction route were reflected in the present study as the difference between the PM and RS groups is shown in Table 1.
In previous reports using NCD data, anastomotic leak and SSI were the most frequent surgical complications (approximately 13% and 14%, respectively), and pneumonia was the most frequent nonsurgical complication (approximately 15%) among major morbidities after esophagectomy. 10,11,22 In the present study, the rates and risks of anastomotic leak, SSI, and pneumonia were investigated as major morbidities after esophagectomy, as well as candidates of reconstruction route-related morbidities clinically.
Previous studies, including one randomized control trial (RCT) and five retrospective studies that enrolled a relatively small number of patients, have compared postoperative morbidities between the PM and RS routes in patients undergoing esophagectomy followed by gastric conduit reconstruction with cervical anastomosis. 21,[23][24][25][26][27] Whereas two retrospective studies showed a higher incidence of anastomotic leak in the RS group, 24,25 the others showed no significant differences between the PM and RS routes, 21,23,26,27 and the impact of reconstruction route on anastomotic leak remained controversial. In the present study, anastomotic leak and SSI were more frequently observed in the RS group than the PM group, and multivariable analyses identified the RS route as an independent risk factor for anastomotic leak and SSI. The higher rate and risks of SSI appear to be affected by the higher incidence of anastomotic leak in the RS group. However, the mechanisms by which the RS route in- present study suggest an impact of the reconstruction route on the oncological outcomes after esophagectomy. Indeed, previous studies on esophageal SCC and adenocarcinoma showed an association of postoperative complications such as pneumonia with poor oncological outcomes. [41][42][43] Therefore, the reconstruction route following esophagectomy may impact survival by affecting the risks of postoperative morbidities.
The limitations of the current study include a lack of anastomotic procedures, such as hand-suture and stapling anastomosis, and a lack of long-term outcomes over 3 mo and quality of life after surgery. In addition, our study did not capture known shortand long-term complications, such as anastomotic stenosis, gastroesophageal reflux, and malnutrition because the NCD did not collect these data. The raw data outcomes of anastomotic leak and SSI could be underestimated, and those of pneumonia could be overestimated by hospital volume, which may be associated with postoperative complications, as speculated by its impact on risk-adjusted mortality following esophagectomy 44 and the preference of the RS route for reconstruction in hospitals with a higher esophagectomy case volume per year (Table 1). Training status and compliance, and the certification status of the institute and surgeon could also influence the outcomes. 45 However, these factors were included in the multivariable analysis as a hospital esophagectomy case volume in this study, and the outcomes shown in Table 2 were further supported by the multivariable analysis.
A major strength of this study is that this is a large-scale comparison report based on a nationwide database. In contrast to RCTs, which generally enroll patients below a certain age limit, without serious comorbidities, in limited institutions the NCD registered almost all surgical cases in Japan. The present study analyzed a large number of patients who underwent esophagectomy in the whole of Japan using the NCD, and unveiled the current status of reconstruction route choice after esophagectomy. Moreover, we examined the impact of the reconstruction route on postoperative morbidities after esophagectomy, without excluding elderly patients or those with serious comorbidities.
In conclusion, the present study first analyzed the impact of the reconstruction route on postoperative morbidities using the nationwide clinical database of patients undergoing esophagectomy for esophageal cancer. The RS route was identified as a risk factor for anastomotic leak and SSI, and the PM route was identified as a risk factor for pneumonia after esophagectomy. Although there may be some bias in the indication of reconstruction route based on the patients' condition and tumor stage, risk-adjusted models in the present study provided important information about the risks of major morbidities, which should be considered for the indication of reconstruction route preoperatively.

ACK N OWLED G M ENTS
We thank all data managers and hospital staff who participated in this National Clinical Database project for their efforts in data entry. We also thank Drs. Nao Ichihara, MD, MPH, PhD, and Yoshihiro Hiramatsu MD, PhD, and Kinji Kamiya MD, PhD for their advice.