Oesophagectomy in the East versus the West: comparison of two national audit databases

Oesophageal cancer is an aggressive disease. It metastasizes, even at an early stage1,2. Consequently, it is the sixth leading cause of cancer death3. To improve clinical outcomes, multimodal treatment, including surgery, chemotherapy, and radiotherapy has all been improved4. Although surgery is the mainstay, treatment strategies vary among countries. For surgically resectable oesophageal cancer, neoadjuvant chemotherapy (NAC) followed by surgery is standard in Japan and some other countries in East Asia5,6. Neoadjuvant chemoradiotherapy (NACRT) plus surgery has been adopted as the standard of care in Western countries and, based on the CheckMate 577 study, in which the efficacy of nivolumab (as adjuvant immunotherapy) after NACRT plus surgery was demonstrated, nivolumab is now a treatment option7. More recently, neoadjuvant triplet chemotherapy was shown to be superior to NACRT in oesophageal adenocarcinoma in the ESOPEC trial8. Treatment strategies and guidelines have been established based on several pivotal clinical trials. However, the generalizability of many studies according to geography, ethnic group, and country is largely unknown. Although chemotherapeutic drugs are expected to be effective regardless of tumour location, careful consideration should be given to new surgical strategies because of diversity regarding surgery type and quality. In addition to surgery, the healthcare system, treatment costs, hospital volume, and patient background vary significantly among countries. To interpret surgical outcomes for each study, it is necessary to identify the underlying similarities and differences in the background of patients who underwent oesophagectomy, as well as risk factors for postoperative complications. To date, several studies have used databases consisting of patients from several countries. Aiming to benchmark complications associated with oesophagectomy, the Esophageal Complications Consensus Group was developed, involving 24 high-volume oesophageal surgical centres in 14 countries, which successfully reported the definitions of postoperative complications9. Subsequently, complication rates after oesophagectomy among several international databases were compared with those of an unselected international cohort10. However, as the established international consensus was only based on data from high-volume centres, the demographics of each country were not fully represented. There have been few international collaborative studies that have comprehensively reviewed patient characteristics and surgical outcomes using national databases that have covered almost the total population of each country. The aim of this study was to identify differences in characteristics and short-term outcomes of patients who underwent oesophagectomy in Japan and the Netherlands using independent patient data from both national audits. This was a retrospective cohort study of patients who underwent oesophagectomy between 2015 and 2019 and who were registered in the Japanese National Clinical Database (NCD) and the Dutch Upper Gastrointestinal Cancer Audit (DUCA). The Japanese NCD is a nationwide surgical registry that was launched in 2011 and, currently, approximately 1 500 000 surgeries are registered annually from over 5000 institutions11,12. For oesophagectomy procedures, the Japanese NCD developed risk calculation models that have been used in clinical practice13. In addition, various clinical issues, including the clinical impact of minimally invasive oesophagectomy, type of reconstruction, and patient variables, have been evaluated14–16. Approximately 6000 oesophagectomies are registered annually12. The DUCA was also established in 2011 and is a mandatory database for centres performing oesophageal resections in the Netherlands17,18. Data were retrieved from the Japanese NCD and the DUCA. This study was approved by the Ethical Review Board of Keio University School of Medicine (date of approval 28 July 2020; approval number 20200109). The scientific committees of the DUCA and the Dutch Upper Gastrointestinal Cancer Group approved the study protocol. Dutch legislation did not require informed consent or ethical approval. Because of the difference in the annual number of patients included in the Japanese NCD and the DUCA, data from the last 2 years, 2018 and 2019, were used from the Japanese NCD and data from all years were used from the DUCA. The incidences of major surgical outcomes were similar between 2015–2017 and 2018–2019 in the Japanese NCD (data not shown). Extracted variables included age, sex, BMI, ASA grade, histology, location, pT, pN, pM, co-morbidities, weight loss, surgical variables, and postoperative complications. Primary outcomes were the incidences of pneumonia and anastomotic leakage. A risk factor analysis was included for these two complications. Three analyses were performed. First, patient demographics for all oesophageal malignancies were compared. Patients who underwent surgery without resection of the primary tumour, such as a bypass or drainage procedure, were excluded. The second analysis focused on patients undergoing transthoracic oesophagectomy. Patients undergoing a transhiatal approach, total or proximal gastrectomy, and oesophagectomy without reconstruction were excluded. Incidences of postoperative death and complications, such as any complication, wound-related complications, anastomotic leakage, pancreatic fistula, pneumonia, renal failure, complications of the central nervous system, acute myocardial infarction, pulmonary embolism, venous thromboembolism, and sepsis, were assessed. Blood transfusion rates were compared. A subgroup analysis was performed for patients undergoing minimally invasive oesophagectomy. Third, two multivariable logistic regression models were constructed to predict the incidences of postoperative pneumonia and anastomotic leakage, separately for the Japanese NCD and the DUCA. Preoperative predictors commonly available in the Japanese NCD and the DUCA, including sex, age, BMI, ASA grade, histology, TNM classifications, co-morbidities (chronic lung disease, prior myocardial infarction, prior cerebellar vascular disease, presence of weight loss, diabetes mellitus, congestive heart failure, peripheral vascular disease, and kidney dysfunction), tumour location, organ reconstruction, cervical anastomosis, and the use of minimally invasive techniques, were used. All analyses were conducted using SAS software (SAS institute, Cary, NC, USA; version 9.4). Two-sided P values <0.050 were considered statistically significant. The profiles of patients who underwent surgery for oesophageal or oesophagogastric junction cancer with resection of the primary tumour are shown in Table S1. The mean annual number of resections for oesophageal cancer was 6673 in the Japanese NCD (5.3 cases per 100 000 population) and 551 in the DUCA (3.1 cases per 100 000 population). Patients in the Japanese NCD were older, had a lower BMI, and had a lower ASA grade. Squamous cell carcinoma was predominant in the Japanese NCD (84.5%) and adenocarcinoma was predominant in the DUCA (79.3%), corresponding to the locations of the primary tumours. Data of patients undergoing transthoracic oesophagectomy are shown in Table S2. Similar trends in differences between the Japanese NCD and the DUCA regarding patient background, including sex, histology, and the location of the primary tumour, were observed. Regarding co-morbidities, the rates of diabetes and prior cerebellar vascular disease were not different between the audits. Prior myocardial infarction and peripheral vascular disease both occurred more frequently in the Dutch cohort. The ASA grade was higher for patients in the DUCA. The distribution of surgical procedures is shown in Table 1 and Fig. 1. In the DUCA, 85.0% of oesophagectomies were totally minimally invasive. In contrast, in the Japanese NCD, a minimally invasive thoracic approach was performed in 73.0% of patients. Of the patients, 21.5% underwent both an open abdominal dissection and an open thoracic dissection. Reflecting the difference in tumour location, even within transthoracic oesophagectomy with reconstruction, anastomosis location was different between the Japanese NCD and the DUCA. A cervical anastomosis was constructed in 83.5% of the patients registered in the Japanese NCD versus 33.8% of the patients registered in the DUCA. The percentage of R2 resection was higher in the Japanese NCD, at 2.7%. Patient characteristics, surgical variables, and surgical outcomes from the Japanese National Clinical Database and the Duch Upper Gastrointestinal Cancer Audit NCD, National Clinical Database; DUCA, Duch Upper Gastrointestinal Cancer Audit. Surgical variables Values are n (%) unless otherwise indicated. NCD, National Clinical Database; DUCA, Dutch Upper Gastrointestinal Cancer Audit; NA, not applicable. Surgical variables Values are n (%) unless otherwise indicated. NCD, National Clinical Database; DUCA, Dutch Upper Gastrointestinal Cancer Audit; NA, not applicable. Table 2 and Fig. 1 show the short-term outcomes after transthoracic oesophagectomy with reconstruction. The 30-day mortality rate in the Japanese NCD and the DUCA was 0.8% and 2.7% respectively. The incidences of anastomotic leakage and pneumonia were 13.9% and 14.8% respectively in the Japanese NCD and 19.2% and 17.8% respectively in the DUCA. Postoperative complications Values are n (%) unless otherwise indicated. NCD, National Clinical Database; DUCA, Dutch Upper Gastrointestinal Cancer Audit. Postoperative complications Values are n (%) unless otherwise indicated. NCD, National Clinical Database; DUCA, Dutch Upper Gastrointestinal Cancer Audit. Data of open and minimally invasive oesophagectomy groups are shown in Table S3. Overall, patient variables, such as sex, ASA grade, co-morbidities, and TNM classifications, were almost identical between open and minimally invasive thoracic oesophagectomy groups in the Japanese NCD and the DUCA. In the Japanese NCD, there was a difference between open and minimally invasive oesophagectomy with respect to histology and the location of the primary tumour. Because the open approach was preferred for an intrathoracic anastomosis in Japan, distal oesophageal adenocarcinoma was most frequently treated using open thoracotomy. In Table 3, short-term outcomes are shown for open and minimally invasive thoracic oesophagectomy. Surgical outcomes were nearly comparable between open and minimally invasive thoracic oesophagectomy in the Japanese NCD and the DUCA. Comparison of postoperative complications between open and minimally invasive oesophagectomy Values are n (%) unless otherwise indicated. NCD, National Clinical Database; DUCA, Dutch Upper Gastrointestinal Cancer Audit. Comparison of postoperative complications between open and minimally invasive oesophagectomy Values are n (%) unless otherwise indicated. NCD, National Clinical Database; DUCA, Dutch Upper Gastrointestinal Cancer Audit. Surgical procedures for the four groups are shown in Table S4. A non-radical resection occurred more often in the open group of the Japanese NCD. Minimally invasive oesophagectomy was not an alleviating factor for postoperative pneumonia in the Japanese NCD. In the DUCA, a minimally invasive thoracic approach reduced the risk of pneumonia (OR 0.64 (95% c.i. 0.45 to 0.92)). Based on risk factor analysis for postoperative anastomotic leakage, a cervical anastomosis was considered a significant risk factor for leakage in the Japanese NCD and the DUCA. The incidence of postoperative complications after oesophagectomy was nearly comparable between the Japanese and Dutch cohorts. Some remarkable differences were observed between cohorts regarding baseline characteristics, such as age, BMI, and histology, as well as the approach used for an intrathoracic anastomosis. These differences have to be taken into account when interpreting clinical studies from either Eastern or Western countries. Patient characteristics such as BMI and ASA grade might be linked to differences in short-term outcomes, including mortality. Both a lower BMI and a higher BMI have been reported to have a negative impact on postoperative complications using a Japanese nationwide database19. The optimal BMI differs between Japan and the Netherlands due to differences in ethnicity20,21. It is therefore difficult to estimate how the differences in BMI and age between the Japanese NCD and the DUCA would affect surgical outcomes. Distal oesophageal adenocarcinoma was predominant in the Netherlands. As a result, an intrathoracic anastomosis with a gastric conduit was used in most patients. In Japan, many tumours were located in the mid-thoracic oesophagus and a cervical anastomosis was used. This influenced the approach, which was usually open for an intrathoracic anastomosis. Experience with minimally invasive intrathoracic anastomoses was limited in Japan. Overall, most differences between cohorts were caused by differences in patient characteristics and traditional surgeon-related factors. In the risk factor analysis for postoperative pneumonia, a minimally invasive thoracic approach did not influence the incidence of pneumonia after surgery in the Japanese NCD. This result is discordant with outcomes of a previous randomized trial conducted in the Netherlands, which demonstrated that a minimally invasive thoracic approach reduced the risk of postoperative pulmonary complications22. This suggests that improvements in perioperative management may reduce the overall incidence of postoperative complications and minimize the difference in outcomes between open and minimally invasive oesophagectomy. Preoperative treatment of oesophageal cancer differs between Japan and the Netherlands. In Japan, based on JCOG9907 and JCOG1109, NAC has been adopted as the standard treatment4,5,23. In the Netherlands, based on the CROSS trial, NACRT became standard24. This might affect surgical outcomes. However, in the JCOG1109 study, in which NAC and NACRT were compared, no significant differences in postoperative complications were observed. The present results, which show that overall postoperative complications were nearly comparable between the Japanese NCD and the DUCA, are therefore consistent with the findings in the aforementioned trial. For other gastrointestinal cancer surgeries, few studies have compared national databases. In 2015, Anazawa et al.25 performed a collaborative study with the Japanese NCD and the American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP), with the goal of creating a standardized surgery database to improve quality. They also included patients undergoing colorectal and pancreatic resections. Significant differences between patient populations were observed, including age, BMI class, and chronic kidney disease. The results of the present study are consistent with these findings. Limitations of this study include the retrospective design and use of audit data. More detailed data may be lacking. Of note, both national databases were carefully validated in a previous study and were shown to reflect current patient status and surgical outcomes after oesophagectomy11,18,26,27. The authors have no funding to declare. S.M. and P.V.D.S. authors have contributed equally to the work. and B.P.L.W. and Y.K. shared senior authorship. Satoru Matsuda (Project administration, Conceptualization, Data curation, Investigation, Methodology, Resources, Writing—original draft, Writing—review & editing), Pieter van der Sluis (Conceptualization, Data curation, Investigation, Writing—review & editing), Hiraku Kumamaru (Data curation, Methodology, Formal analysis, Writing—original draft, Writing—review & editing), Feike Kingma (Data curation, Investigation), Hirofumi Kawakubo (Writing—review & editing), Jelle Ruurda (Writing—review & editing), Masayuki Watanabe (Writing—review & editing), Hiroaki Miyata (Writing—review & editing), Sjoerd Lagarde (Writing—review & editing), Hiroya Takeuchi (Writing—review & editing), Richard van Hillegersberg (Writing—review & editing), Ken Shirabe (Writing—review & editing), Bas P. L. Wijnhoven (Conceptualization, Project administration, Writing—review & editing), and Yuko Kitagawa (Conceptualization, Project administration, Writing—review & editing) The authors declare no conflict of interest. Supplementary material is available at BJS online. The data sets generated during the present study are available from the corresponding author upon reasonable request.