Variations in survival and perioperative complications between hospitals based on data from two phase III clinical trials for oesophageal cancer

Transcription

1 Original article Variations in survival and perioperative complications between hospitals based on data from two phase III clinical trials for oesophageal cancer K. Kataoka 1, K. Nakamura 1, J. Mizusawa 1, H. Fukuda 1, H. Igaki 2,S.Ozawa 6, K. Hayashi 4,K.Kato 3, Y. Kitagawa 5 and N. Ando 7 1 Japan Clinical Oncology Group (JCOG) Data Centre/Operations Office, Centre for Research Administration and Support, National Cancer Centre, 2 Oesophageal Surgery Division and 3 Gastrointestinal Medical Oncology Division, National Cancer Centre Hospital, 4 Department of Chemotherapy and Palliative Care, Tokyo Women s Medical University, and 5 Department of Surgery, Keio University School of Medicine, Tokyo, 6 Department of Gastrointestinal Surgery, Tokai University School of Medicine, Isehara, and 7 International Goodwill Hospital, Yokohama, Japan Correspondence to: Dr K. Kataoka, JCOG Operations Office, Centre for Research Administration and Support, National Cancer Centre, Tsukiji, Chuo-ku, Tokyo, 14 45, Japan ( kozokataoka@hotmail.co.jp) Background: Variations in institutional practice may contribute to different outcomes of cancer treatment. The impact of interinstitutional heterogeneity on outcomes between hospitals after oesophagectomy has not been examined previously using data from surgical clinical trials. Methods: The data from two phase III trials for oesophageal cancer were used. Japan Clinical Oncology Group (JCOG) 924 involved oesophagectomy (92-OP) versus oesophagectomy plus postoperative chemotherapy (92-POST), with accrual from 1992 to JCOG997 involved postoperative chemotherapy (99-POST) versus preoperative chemotherapy (99-PRE), with accrual from 2 to 26. Hospitals contributing fewer than three patients were excluded. The influence of time and preoperative chemotherapy on interinstitutional heterogeneity related to and 5-year overall survival were evaluated by comparisons within and between these trial groups. Heterogeneity was estimated by a mixed-effects model after adjusting for age, sex, performance status, location of the primary tumour and clinical stage. Results: Twelve hospitals in 92-OP (114 patients), 13 in 92-POST (114), 19 in 99-POST (158) and 18 in 99-PRE (154) were eligible. There was considerable heterogeneity in predicted postoperative complications in both groups in JCOG924 (median 31 3 (range ) per cent), and in 99-PRE (35 2 ( ) per cent) but not in 99-POST (27 7 ( ) per cent) from JCOG997. A similar pattern was seen for predicted overall survival (92-POST: 66 4 (range ) per cent; 99-PRE: 55 9 ( ) per cent; 99-POST: 44 4 ( ) per cent). Conclusion: Interinstitutional heterogeneity regarding complications and survival after oesophagectomy is a problem that merits wider consideration. Presented to the 68th Annual Meeting of the Japan Oesophageal Society, Tokyo, Japan, July 214 Paper accepted 27 March 215 Published online 11 June 215 in Wiley Online Library ( DOI: 1.12/bjs.9839 Introduction Disparity in the quality of cancer treatments between hospitals can lead to differences in outcome. Such interinstitutional heterogeneity tends to be greater with more complex treatments such as oesophageal cancer 1 4,and some studies have suggested that morbidity and survival following surgical treatments for oesophageal cancer show wider variation between institutions than seen in other gastrointestinal cancers 5 8. Most reports, however, are observational studies based on nationwide cancer registries 1 8 that may not contain information about patient selection, specific treatments or postoperative complications. To eliminate some of the biases inherent in registry data, the present study relied on data collected from two prospective studies involving patients with homogeneous backgrounds and treatment modalities. Hospital volume has a key influence on mortality after oesophagectomy 5 9. In this context, oesophagectomy has come to be performed in high-volume hospitals in Japan, as well as Western countries Although this should 215 BJS Society Ltd BJS 215; 12:

2 Outcomes between hospitals after oesophagectomy 189 All registered patients in JCOG924 n = 242 All registered patients in JCOG997 n = 33 Surgery alone n = 122 (17 hospitals) Surgery + postoperative chemotherapy n = 12 (17 hospitals) Surgery + postoperative chemotherapy n = 166 (24 hospitals) Preoperative chemotherapy + surgery n = 164 (24 hospitals) Excluded: 8 patients from 5 hospitals with 2 patients Excluded: 8 patients from 5 hospitals with 2 patients Excluded: 6 patients from 4 hospitals with 2 patients Excluded: 1 patients from 6 hospitals with 2 patients 92-OP n = 114 (12 hospitals) 92-POST n = 114 (13 hospitals) A 99-POST n = 158 (19 hospitals) B 99-PRE n = 154 (18 hospitals) Analysis of interinstitutional heterogeneity in overall survival Excluded: 4 patients did not undergo surgery Excluded: 1 patients did not undergo surgery 228 patients in 92-OP + 92-POST groups underwent surgery (13 hospitals) C 154 patients in 99-POST group underwent surgery (19 hospitals) D 144 patients in 99-PRE group underwent surgery (18 hospitals) Analysis of interinstitutional heterogeneity in Fig. 1 Flow chart for the Japan Clinical Oncology Group (JCOG) 924 and JCOG997 trials. A Evaluation of the effect of time on the interinstitutional heterogeneity in overall survival (92-POST versus 99-POST); B evaluation of the influence of preoperative chemotherapy on interinstitutional heterogeneity in overall survival (99-POST versus 99-PRE); C evaluation of the effect of time on the interinstitutional heterogeneity in (92-OP + 92-POST versus 99-POST); and D evaluation of the influence of preoperative chemotherapy on the interinstitutional heterogeneity in (99-POST versus 99-PRE). 92-OP, oesophagectomy alone group in JCOG924; 92-POST, oesophagectomy + postoperative chemotherapy group in JCOG924; 99-POST, oesophagectomy + postoperative chemotherapy group in JCOG997; 99-PRE, preoperative chemotherapy + oesophagectomy group in JCOG997 improve outcomes, it is unclear whether interinstitutional heterogeneity still remains among high-volume hospitals. The first aim of this study was to evaluate the effect of time on heterogeneity between hospitals regarding and overall survival (OS) after oesophagectomy using data from two trials (Japan Clinical Oncology Group (JCOG) and JCOG ) conducted in different periods. The second was to evaluate the influence of introducing preoperative chemotherapy on heterogeneity for these outcomes. 215 BJS Society Ltd BJS 215; 12:

3 19 K. Kataoka, K. Nakamura, J. Mizusawa, H. Fukuda, H. Igaki, S. Ozawa et al. Methods Patients registered in two clinical trials (JCOG924 and JCOG997) conducted by JCOG were included after excluding those from institutions that contributed fewer than three patients. JCOG924 was undertaken to confirm the superiority of oesophagectomy followed by postoperative chemotherapy (fluorouracil plus cisplatin) over oesophagectomy alone for resectable oesophageal squamous cell carcinoma. Details of eligibility and main outcomes have been reported previously 13. Based on the results of JCOG924, JCOG997 was conducted to confirm the superiority of preoperative chemotherapy (fluorouracil plus cisplatin) followed by oesophagectomy over oesophagectomy followed by chemotherapy (fluorouracil plus cisplatin) for resectable oesophagaeal squamous cell carcinoma. Again, eligibility criteria and main outcomes from JCOG997 have been reported elsewhere 14. JCOG997 involved the same operative procedure and postoperative chemotherapy as JCOG924. The influence of time when treatments were undertaken on interinstitutional heterogeneity was evaluated using a comparison between 92-OP + 92-POST and 99-POST for, and a comparison between 92-POST and 99-POST for 5-year OS. The relationship between preoperative chemotherapy and interinstitutional heterogeneity was evaluated by comparing postoperative complication rates and 5-year OS of 99-POST and 99-PRE. These comparisons were evaluated using the predicted data estimated by a mixed-effects model and the unadjusted actual observed data. If interinstitutional heterogeneity was observed in 99-PRE, associations between the outcomes (postoperative complication rate, 5-year OS and blood loss) and institutional factors (total number of patients with oesophageal cancer of any clinical stage who underwent surgery during the accrual period and number of patients with clinical stage II III oesophageal cancer who underwent surgery during the accrual period) were examined. As no data regarding blood loss during surgery were collected in JCOG924, operative blood loss was assessed only in the analysis aiming to determine the cause of heterogeneity in 99-PRE. The protocol of this ancillary study (JCOG1312-A) was approved by the JCOG Protocol Review Committee and al Review Board of National Cancer Centre, Tokyo. Endpoints OS was estimated from the date of registration until death from any cause. Surviving patients and patients who were lost to follow-up were censored at the latest contact date. Table 1 Patient characteristics 92-OP 92-POST 99-POST 99-PRE (n = 114) (n = 114) (n = 158) (n = 154) Age (years)* 6 (4 75) 6 (4 74) 61 (39 75) 61 (38 75) Sex ratio (M : F) 14 : 1 11 : : : 19 ECOG performance status 62 (54 4) 67 (58 8) 114 (72 2) 117 (76 ) 1or2 52(45 6) 47 (41 2) 44 (27 8) 37 (24 ) 3or4 () () () () Tumour location Upper third 1 (8 8) 19 (16 7) 16 (1 1) 1 (6 5) Middle or lower third 14 (91 2) 95 (83 3) 142 (89 9) 144 (93 5) Clinical tumour category ct1 22 (19 3) 28 (24 6) 5 (3 2) 4 (2 6) ct2 16 (14 ) 28 (24 6) 36 (22 8) 34 (22 1) ct3 72 (63 2) 55 (48 2) 117 (74 1) 116 (75 3) ct4 4 (3 5) 3 (2 6) () () Clinical node status cn 31 (27 2) 31 (27 2) 52 (32 9) 56 (36 4) cn+ 83 (72 8) 83 (72 8) 16 (67 1) 98 (63 6) Histology of resected specimen Squamous cell carcinoma 112 (98 2) 113 (99 1) 147 (93 ) 137 (89 ) Adenosquamous carcinoma () 1 ( 9) () () Other 2 (1 8) () 11 (7 ) 17 (11 ) Values in parentheses are percentages unless indicated otherwise; *values are median (range). 92-OP, oesophagectomy alone group in Japan Clinical Oncology Group (JCOG) 924; 92-POST, oesophagectomy + postoperative chemotherapy group in JCOG924; 99-POST, oesophagectomy + postoperative chemotherapy group in JCOG997; 99-PRE, preoperative chemotherapy + oesophagectomy group in JCOG997; ECOG, Eastern Cooperative Oncology Group. 215 BJS Society Ltd BJS 215; 12:

4 Outcomes between hospitals after oesophagectomy 191 Rate of was defined as the frequency of patients who experienced any of the following : respiratory-related complications (including pneumonia, acute respiratory distress syndrome, pneumothorax and pulmonary embolism), acute circulatory failure, myocardial infarction, recurrent nerve palsy, anastomotic leakage, haemothorax, pleural empyema, chylothorax, bowel obstruction, sepsis, cholangitis, anastomotic stricture and wound infection. Postoperative complications from the end of an operation to the first postoperative discharge were counted. They were evaluated before the administration of postoperative chemotherapy in all patients. Statistical analysis A mixed-effects model was used for the main analysis. Variance in the rate of among hospitals was estimated using a mixed logistic regression model in which institution was taken as a random effect and the other clinical background factors (age, sex, Eastern Cooperative Oncology Group (ECOG) performance status, location of the primary tumour, clinical stage) were taken as fixed effects. The variance in OS was estimated using a mixed proportional hazards model in which institution was taken as a random effect and other clinical background factors (age, sex, ECOG performance status, primary tumour location, tumour (T) category and node (N) status) were taken as fixed effects. The exponential survival distribution was used as a baseline hazard function of mixed proportional hazards because it matched with the data adequately. The SAS (SAS Institute, Cary, North Carolina, USA) procedure PROC GLIMMIXED was used for the analysis of complications and NLMIXED for analysis of survival. The predicted OS of each hospital was translated into a 5-year OS to facilitate the interpretation Observed probability of a 92-OP + 92-POST Observed probability of b 99-POST Observed probability of c 99-PRE Fig. 2 Observed interinstitutional heterogeneity in in a 92-OP (oesophagectomy) + 92-POST (oesophagectomy + postoperative chemotherapy), b 99-POST (oesophagectomy + postoperative chemotherapy) and c 99-PRE (preoperative chemotherapy + oesophagectomy). Error bars represent 95 per cent credible intervals 215 BJS Society Ltd BJS 215; 12:

5 192 K. Kataoka, K. Nakamura, J. Mizusawa, H. Fukuda, H. Igaki, S. Ozawa et al. of results. To establish the reason for the observed interinstitutional heterogeneity, Spearman s correlation coefficients were calculated for the relationships between predicted outcomes and institutional factors 15,16. Differences in 5-year OS of more than 5 per cent and variations in the frequency of of more than 1 per cent were considered to represent a moderate degree of interinstitutional heterogeneity. Differences in 5-year OS of more than 1 per cent and variations in the frequency of of more than 2 per cent were considered to indicate a large degree of heterogeneity. The observed postoperative complication rate and the observed 5-year OS rate in each institution were calculated by the Kaplan Meier method. All statistical analyses were performed with SAS version 9.2. Results In the main trials, there were 17 participating hospitals in JCOG924 and 24 in JCOG997. The 17 hospitals in JCOG924 performed 243 oesophagectomies in total and the 24 hospitals in JCOG997 undertook 392 oesophagectomies during the accrual period. In JCOG997, 75 per cent of patients completed two courses of postoperative chemotherapy and 85 4 per cent completed two courses of preoperative chemotherapy. Twelve hospitals in 92-OP (114 patients), 13 in 92-POST (114), 19 in 99-POST (158) and 18 in 99-PRE (154) were included in the presentanalysis (Fig. 1). Four patients from 99-POST and ten from 99-PRE who did not undergo surgery were excluded from the analysis of heterogeneity in. Patient characteristics are shown in Table 1. The observed interinstitutional heterogeneity in the rate of in 92-OP + 92-POST, 99-POST and 99-PRE is shown in Fig. 2. The observed interinstitutional heterogeneity in 5-year OS in 92-POST, 99-POST and 99-PRE is illustrated in Fig. 3. Observed 5-year overall survival a 92-POST 1 Observed 5-year overall survival Observed 5-year overall survival c 99-PRE b 99-POST Fig. 3 Observed interinstitutional heterogeneity in overall survival in a 92-POST (oesophagectomy + postoperative chemotherapy), b 99-POST (oesophagectomy + postoperative chemotherapy) and c 99-PRE (preoperative chemotherapy + oesophagectomy) 215 BJS Society Ltd BJS 215; 12:

6 Outcomes between hospitals after oesophagectomy a 92-OP + 92-POST Predicted probability of Predicted probability of b 99-POST c 99-PRE Predicted probability of Fig. 4 Predicted interinstitutional heterogeneity in in a 92-OP (oesophagectomy) + 92-POST (oesophagectomy + postoperative chemotherapy), b 99-POST (oesophagectomy + postoperative chemotherapy) and c 99-PRE (preoperative chemotherapy + oesophagectomy). Error bars represent 95 per cent credible intervals Predicted interinstitutional heterogeneity in the rate of is shown in Fig. 4. A large degree of heterogeneity was observed in 92-OP + 92-POST (median 31 3 (range ) per cent) (Fig. 4a), but none in 99-POST (27 7( ) per cent) (Fig. 4b). Predicted interinstitutional heterogeneity in 5-year OS is shown in Fig. 5. This was considered moderate in 92-POST (66 4 ( ) per cent) (Fig. 5a), but not in 99-POST (44 4 ( ) per cent) (Fig. 5b). A large degree of interinstitutional heterogeneity in (35 2 ( ) per cent) (Fig. 4c) and a moderate degree of interinstitutional heterogeneity in 5-year OS (55 9 ( ) per cent) (Fig. 5c) were detected in 99-PRE. Predicted interinstitutional heterogeneities at the trial level are shown in Fig. S1 (supporting information). Correlations identified between predicted outcomes and institutional factors in 99-PRE are shown in Table 2. The total number of patients with oesophageal cancer of any stage who underwent surgery did not correlate with blood loss (ρ = 26, P = 919), (ρ= 2, P = 938) or 5-year OS (ρ= 356, P = 147). The number of patients with clinical stage II III oesophageal cancer who underwent surgery did not correlate with blood loss (ρ= 7, P = 782) or (ρ= 231, P = 356), but did correlate with 5-year OS (ρ= 499, P = 35). Discussion This study indicated that the disparities in complication rates and survival after surgery for oesophageal cancer among hospitals reduced over time. This study examined data from randomized phase III trials, which had the advantages of dealing with homogeneous patient groups, the ability to evaluate the impact of new treatment 215 BJS Society Ltd BJS 215; 12:

7 194 K. Kataoka, K. Nakamura, J. Mizusawa, H. Fukuda, H. Igaki, S. Ozawa et al. Predicted 5-year overall survival a 92-POST Predicted 5-year overall survival b 99-POST Predicted 5-year overall survival c 99-PRE Fig. 5 Predicted interinstitutional heterogeneity in overall survival in a 92-POST (oesophagectomy + postoperative chemotherapy), b 99-POST (oesophagectomy + postoperative chemotherapy) and c 99-PRE (preoperative chemotherapy + oesophagectomy) Table 2 Exploratory analysis of the cause of interinstitutional heterogeneity in 99-PRE Total no. of patients with oesophageal cancer who underwent surgery at each hospital (2 26) Any stage Clinical stage II III ρ P ρ P Postop. complication rate year overall survival Blood loss PRE, preoperative chemotherapy + oesophagectomy group in Japan Clinical Oncology Group (JCOG) 997. modalities and to assess interinstitutional heterogeneity among high-volume centres. Multi-institutional cancer clinical trials often include many participating hospitals and the number of accrued patients in each hospital is relatively small. As shown in Figs 2 and 3, it is difficult to evaluate interinstitutional heterogeneities of institutions with a limited number of patients using observed values. With a fixed model, it is likely to be unsuccessful because the predicted regression coefficient may be unstable and loss of efficiency may be high. Some studies have indicated that use of a mixed-effects model with institution taken as a random effect is one of the promising methods when interinstitutional heterogeneity is evaluated in a multi-institutional prospective study 15,17.Itwasonthis basis that the mixed-effects model was used to evaluate the interinstitutional heterogeneity in the present study. There has been an improvement in survival among patients undergoing oesophagectomy in Japan and elsewhere in recent years 6,18. Improvements in surgical performance, the use of postoperative chemotherapy and perioperative supportive care could all contribute to better outcomes, and play a part in minimizing differences between centres. Discussion among the participating 215 BJS Society Ltd BJS 215; 12:

8 Outcomes between hospitals after oesophagectomy 195 investigators might also have contributed to the reduction in interinstitutional heterogeneity. Although preoperative chemotherapy is generally considered to be better tolerated than postoperative regimens, a large degree of interinstitutional heterogeneity was seen in terms of and a moderate degree in efficacy in the analysis of 99-PRE. This suggests that there is the potential to increase interinstitutional heterogeneity in trials as new treatment modalities are introduced. The fact that this applied to complication rates after oesophagectomy in JCOG924, when many Japanese surgeons had already been performing a three-field operation for more than 1 years 19 21, implies that this effect can persist for a long time, at least for a highly technical treatment. Associations between hospital volume or volume and morbidity or survival have been described extensively 1 4. Although some have suggested that surgeon volume has a greater effect on survival than hospital volume 2, a study 9 in Japan found that hospital volume was associated with mortality rate after oesophagectomy. In the present study, an analysis of 99-PRE demonstrated that the number of patients with clinical stage II III oesophageal cancer who underwent surgery at each hospital correlated with 5-year OS. This effect was not due to intraoperative blood loss or, suggesting that hospital volume mainly influences long-term outcomes. The present study has three main limitations. Patients were registered after they had recovered from oesophagectomy in JCOG924, so some patients who experienced serious were not registered as they were not considered fit for chemotherapy. The extent to which this occurred at each institution was not assessed. Postoperative complications were not defined in accordance with objective criteria in either trial, as there were no established criteria or agreed definitions for many surgical complications at the time of these studies. This means that the total number of complications may have been underestimated and no attempt could be made to grade the severity of complications. Predictors of random effect could have varied depending on the procedure used in the mixed-effects model because explicit integral computation could not be performed in this study. Small amounts of variance might not have been detected, which could explain the lack of interinstitutional heterogeneity in 99-POST. These studies suggest that heterogeneity in surgical performance decreased with time, but the effect does persist for many years after the introduction of a highly technical procedure like oesophagectomy. When a new treatment modality is introduced, such as preoperative chemotherapy, interinstitutional heterogeneity in outcomes once again increases. These effects should be taken into account in the interpretation of outcomes. Acknowledgements The authors thank all participating patients, investigators and members of the JCOG Data Centre/Operations Office. This study was supported in part by the National Cancer Centre Research and Development Funds (23-A-16, 26-A-4) from the Ministry of Health, Labour and Welfare of Japan. Disclosure: The authors declare no conflict of interest. References 1 Bachmann MO, Alderson D, Edwards D, Wotton S, Bedford C, Peters TJ et al. Cohort study in South and West England of the influence of specialization on the management and outcome of patients with oesophageal and gastric cancers. Br J Surg 22; 89: Derogar M, Sadr-Azodi O, Johar A, Lagergren P, Lagergren J. Hospital and surgeon volume in relation to survival after esophageal cancer surgery in a population-based study. J Clin Oncol 213; 31: McCulloch P, Ward J, Tekkis PP. Mortality and morbidity in gastro-oesophageal cancer surgery: initial results of ASCOT multicentre prospective cohort study. BMJ 23; 327: Wouters MW, Karim-Kos HE, le Cessie S, Wijnhoven BP, Stassen LP, Steup WH et al. Centralization of esophageal cancer surgery: does it improve clinical outcome? Ann Surg Oncol 29; 16: Birkmeyer JD, Stukel TA, Siewers AE, Goodney PP, Wennberg DE, Lucas FL. Surgeon volume and operative mortality in the United States. N Engl J Med 23; 349: Finks JF, Osborne NH, Birkmeyer JD. Trends in hospital volume and operative mortality for high-risk surgery. NEngl JMed211; 364: Gruen RL, Pitt V, Green S, Parkhill A, Campbell D, Jolley D. The effect of provider case volume on cancer mortality: systematic review and meta-analysis. CA Cancer J Clin 29; 59: Hollenbeck BK, Dunn RL, Miller DC, Daignault S, Taub DA, Wei JT. Volume-based referral for cancer surgery: informing the debate. J Clin Oncol 27; 25: Fujita H, Ozawa S, Kuwano H, Ueda Y, Hattori S, Yanagawa T. Esophagectomy for cancer: clinical concerns support centralizing operations within the larger hospitals. Dis Esophagus 21; 23: Forshaw MJ, Gossage JA, Stephens J, Strauss D, Botha AJ, Atkinson S et al. Centralisation of oesophagogastric cancer 215 BJS Society Ltd BJS 215; 12:

9 196 K. Kataoka, K. Nakamura, J. Mizusawa, H. Fukuda, H. Igaki, S. Ozawa et al. services: can specialist units deliver? Ann R Coll Surg Engl 26; 88: Al-Sarira AA, David G, Willmott S, Slavin JP, Deakin M, Corless DJ. Oesophagectomy practice and outcomes in England. Br J Surg 27; 94: Birkmeyer JD, Finlayson EV, Birkmeyer CM. Volume standards for high-risk surgical procedures: potential benefits of the Leapfrog initiative. Surgery 21; 13: Ando N, Iizuka T, Ide H, Ishida K, Shinoda M, Nishimaki T et al.; Japan Clinical Oncology Group. Surgery plus chemotherapy compared with surgery alone for localized squamous cell carcinoma of the thoracic esophagus: a Japan Clinical Oncology Group Study JCOG924. J Clin Oncol 23; 21: Ando N, Kato H, Igaki H, Shinoda M, Ozawa S, Shimizu H et al. A randomized trial comparing postoperative adjuvant chemotherapy with cisplatin and 5-fluorouracil versus preoperative chemotherapy for localized advanced squamous cell carcinoma of the thoracic esophagus (JCOG997). Ann Surg Oncol 212; 19: Yamaguchi T, Ohashi Y. Investigating centre effects in a multi-centre clinical trial of superficial bladder cancer. Stat Med 1999; 18: Yamaguchi T, Ohashi Y, Matsuyama Y. Proportional hazards models with random effects to examine centre effects in multicentre cancer clinical trials. Stat Methods Med Res 22; 11: Legrand C, Duchateau L, Sylvester R, Janssen P, van der Hage JA, van de Velde CJ et al. Heterogeneity in disease free survival between centers: lessons learned from an EORTC breast cancer trial. Clin Trials 26; 3: Ando N, Ozawa S, Kitagawa Y, Shinozawa Y, Kitajima M. Improvement in the results of surgical treatment of advanced squamous esophageal carcinoma during 15 consecutive years. Ann Surg 2; 232: Akiyama H, Tsurumaru M, Udagawa H, Kajiyama Y. Radical lymph node dissection for cancer of the thoracic esophagus. Ann Surg 1994; 22: Baba M, Aikou T, Yoshinaka H, Natsugoe S, Fukumoto T, Shimazu H et al. Long-term results of subtotal esophagectomy with three-field lymphadenectomy for carcinoma of the thoracic esophagus. Ann Surg 1994; 219: Matsubara T, Ueda M, Yanagida O, Nakajima T, Nishi M. How extensive should lymph node dissection be for cancer of the thoracic esophagus? J Thorac Cardiovasc Surg 1994; 17: Supporting information Additional supporting information may be found in the online version of this article: Fig. S1 Predicted interinstitutional heterogeneity in a in Japan Clinical Oncology Group (JCOG) 997, b overall survival in JCOG924, and c overall survival in JCOG997 (Word document) 215 BJS Society Ltd BJS 215; 12: