This study aimed to evaluate the long-term effect of esophagectomy in patients with esophageal squamous cell carcinoma (ESCC) by comparing the chemoradiotherapy (CRT)-only group and the trimodality treatment (TMT) group who received concurrent CRT followed by surgery.
We included 412 operable ESCC patients treated with TMT or CRT between January 2005 and December 2015. The oncological outcomes of the two groups were compared using a weighted Cox proportional-hazards model with inverse probability of treatment weighting (IPTW).
The median survival time was 64 and 32 months in the TMT (n=270) and CRT (n=142) groups, respectively (p < 0.001). After IPTW, the median overall survival (OS) remained significantly higher in the TMT group than in the CRT group (61 months vs. 32 months, p=0.016). Moreover, the TMT group showed a better local recurrence–free rate (LRFR; p < 0.001) and distant metastasis–free rate (p=0.007). In the subgroup of patients with clinical complete response (cCR), the OS was not significantly different between the two groups, both before and after IPTW adjustment (p=0.35 and p=0.93). However, among non-cCR patients, the OS was significantly higher in the TMT group (64% vs. 45%, p < 0.001).
In patients with locally advanced ESCC, TMT was superior to CRT in terms of OS and LRFR. Such difference was more prominent in the non-cCR subgroup. In patients who achieved cCR, esophagectomy was effective in improving LRFR but not OS, suggesting that esophagectomy may be omitted in complete responders.
Esophageal cancer is the seventh most common malignancy and the sixth most common cause of cancer-related mortality globally [
However, the term “response” includes a wide range of disease status in the real-world setting, from partial responses of a slight decrease in the disease status to complete disappearance of the tumor. Moreover, it may be difficult to evaluate the clinical response due to treatment-induced edema and esophagitis, which do not subside at the time of post-treatment reevaluation. Therefore, it is necessary to accurately define the criteria for “response”, which is complicated due to the individual variability among clinicians in interpreting treatment responses. To minimize these variations and possible confusion, the use of the criteria of “complete response” may be more reasonable and practical despite the controversy on the evaluation methods.
We assume that surgical resection may be omitted only in responders to CRT, because patients may miss the optimal timing for surgical resection and cure. However, there is a limited amount of available data on the survival benefit of esophagectomy in patients who show a clinically good response after CRT for locally advanced esophageal squamous cell carcinoma (ESCC), and the available results are not consistent [
We identified 730 patients who were treated with TMT or CRT at our center between January 2005 and December 2015 for locally advanced esophageal cancer with squamous cell carcinoma. The inclusion criteria were as follows: (1) histologically confirmed resectable but advanced ESCC (cT2–4/anyN/M0 or anyT/N+/M0 stage), (2) medically operable status, (3) Eastern Cooperative Oncology Group (ECOG) performance status of ≤ 2, and (4) no history of thoracic surgery. The exclusion criteria were as follows: (1) double primary cancer, (2) suboptimal radiotherapy (RT) dose (< 38 Gy), and (3) insufficient follow-up duration (< 3 months) without oncologic event. Accordingly, 412 patients were included in this analysis (
The initial diagnostic evaluation included detailed medical history, physical examination, laboratory blood analysis, esophagogastroduodenoscopy (EGD), endoscopic ultrasound, computed tomography (CT) scans of the chest and abdomen, esophagography, and positron emission tomography–CT (PET-CT) scan. Four weeks after the completion of CRT, treatment response was evaluated with EGD and biopsy, chest CT scan, and PET-CT.
Clinical complete response (cCR) after CRT was defined as the absence of residual tumor on endoscopy with biopsy and metabolic complete remission (CR) on PET scan. Metabolic CR was defined as the complete resolution of fluorodeoxyglucose uptake in the primary tumor and metastatic lymph nodes or indistinguishable initial tumor site from the surrounding tissue in cases of diffuse esophagitis with increased uptake within a radiation field.
After treatment, regular follow-up examinations were performed every 3 months during the first 2 years, and every 6 months thereafter until 5 years. Toxicities during and after treatment were evaluated using the Common Terminology Criteria for Adverse Events (ver. 4.03). Surgical complications were assessed by the Clavien-Dindo classification.
Treatment strategies for locally advanced esophageal cancer were primarily determined by a multidisciplinary team. Several patients who could not be assessed by the multidisciplinary team were assessed by individual members of the multidisciplinary team, including thoracic surgeons, medical oncologists, gastroenterologists, and radiation oncologists.
The median prescribed radiation dose was 46 Gy for neoadjuvant treatment and 54 Gy for definitive CRT. LightSpeed RT (GE Medical Systems, Palo Alto, CA) was used for CT simulation with intravenous contrast enhancement. The gross tumor volume (GTV) was delineated on each slice of the acquired CT images and was assisted by the information from PET-CT, chest CT, and EGD. During the period of 3D treatment, primary tumor and mediastinal lymph node were treated with margins of 5 cm in the cranio-caudal direction and 2 cm in the lateral direction. The supraclavicular lymph nodes were included when the GTV existed in the upper thoracic esophagus, and the celiac trunk was included when the GTV was in the mid or distal thoracic esophagus. After the introduction of intensity-modulated radiotherapy (IMRT), the clinical target volume (CTV) margin was reduced to 3 cm in the S-I direction and 1 cm in the radial direction. The planning target volume margin was 7 mm in the radial direction and 10 mm in the cranio-caudal expansion of the CTV.
All patients were treated with capecitabine-cisplatin (XP) or 5-fluorouracil–cisplatin (FP) chemotherapy for concurrent CRT. For XP chemotherapy, patients received capecitabine 1,600 mg/m2/day for 5 days plus cisplatin 30 mg/m2/day on the first day, weekly. For FP chemotherapy, patients received cisplatin 60 mg/m2/day on the first day plus 5-fluorouracil 1,000 mg/m2/day on the second day for 4 days, every 3 weeks.
All patients were routinely evaluated for operability by an experienced thoracic surgeon. Surgery was performed 6 to 8 weeks after the completion of CRT using either the Ivor-Lewis or McKeown approach.
Continuous variables were compared using the t test, and categorical variables were compared using the Fisher exact test or chi-squared test. The rates of overall survival (OS), local recurrence–free rate (LRFR), and distant metastasis–free rate (DMFR) were estimated using the Kaplan-Meier method and compared using the log-rank test. Inverse probability treatment weighting (IPTW) analysis based on the propensity score was used to reduce the impact of selection bias and potential confounding factors. Propensity scores were calculated using a logistic regression model using the following variables: sex, age, Charlson-Deyo score, alcohol, smoking, ECOG performance status, tumor location, and tumor stage. The absolute standardized differences (STDs) were used to check the balance after IPTW, and weighted Cox regression models with robust standard errors were used for the comparison of survival after IPTW adjustment. Also, tests for interaction were performed to assess the heterogeneity of treatment effect among the cCR subgroups. p-values less than 0.05 were considered statistically significant. All statistical analyses were performed with SAS ver. 9.4 (SAS Institute Inc., Cary, NC) and R ver. 3.6.1 (R Foundation for Statistical Computing, Vienna, Austria).
Among the 412 study patients, 270 patients (65.5%) received TMT and 142 (34.5%) received CRT. The baseline characteristics of the patients are summarized in
After the completion of CRT, all patients were evaluated by EGD with/without biopsy, CT scan, and PET-CT. One patient did not undergo a PET-CT scan but was confirmed with a residual tumor on EGD. The median time from the last day of CRT to PET-CT was 0.9 months (range, 0.3 to 2.8). Metabolic CR was achieved in 241 patients (58.6%). Of the 412 patients, 119 (28.8%) showed a complete response on EGD. Among the 353 patients (85.7%) who underwent endoscopic biopsy, 324 (91.8%) had negative biopsy results.
A total of 92 (22.3%) patients showed cCR. In the TMT group, 119 (44.1%) patients had a pathologic complete remission (pCR), which was more common in the cCR group than in the non-cCR group (36/58 [62.1%] vs. 83/212 [39.2%], p=0.003).
At the time of analysis (May 2020), 150 patients were alive (TMT, 109 [40.3%]; CRT, 41 [28.9%]). The median follow-up duration was 39.2 months (range, 5.3 to 164.0) in the entire cohort and 67.5 months (range, 13.6 to 164.0) in the living patients. The median OS duration was 64 months (95% confidence interval [CI], 44 to 74) and 32 months (95% CI, 18 to 38) for the TMT group and the CRT group, respectively (p < 0.001) (
The patterns of failure are summarized in
In the CRT group, the major patterns of failure were local recurrence in 10 (29.4%), distant metastasis in three (8.8%), and both local recurrence and distant metastasis in five patients (14.7%). The TMT group had a significantly better LRFR and DMFR than did the CRT group (hazard ratio [HR], 0.362; 95% CI, 0.252 to 0.519; p < 0.001 and HR, 0.585; 95% CI, 0.388 to 0.883; p=0.011, respectively). These results were also observed in multivariate and IPTW-adjusted analyses (
In the cCR group, OS and DMFR were not significantly different between the TMT group and the CRT group (HR, 1.027; 95% CI, 0.561 to 1.880; p=0.93 and HR, 0.905; 95% CI, 0.324 to 2.524; p=0.85, respectively) (
In the non-cCR subgroup, the TMT group showed significantly superior rates of OS, DMFR, and LRFR compared with the CRT group (p < 0.001, p=0.001, and p < 0.001, respectively) (
Acute complications after TMT and CRT are summarized in
Currently, TMT is regarded as the standard treatment for locally advanced esophageal cancer, with its role having been established by the results of randomized controlled trials (RCTs) on adenocarcinoma [
In order to focus on testing the possibility of safely omitting esophagectomy, remnant tumors in the mediastinum should be considered for performing a successful study. Accordingly, the study by Piessen et al. [
Other investigators also tried to identify the benefit of esophagectomy in cCR patients with SCC, but none has demonstrated significant advantages in terms of survival. Castoro et al. [
We have published a similar study with more advanced evaluation methods such as PET scan, and reported that while esophagectomy resulted in significantly improved disease-free survival, it did not result in a significant improvement in OS [
Although the exact reason why we could not observe a significant difference in OS in the cCR patients is unclear, the following factors may have been involved. First, surgical mortality might have been involved in the nullification of the survival benefits of esophagectomy. For example, robot-assisted surgery was introduced to our center for esophageal cancer in 2010, and there may have had been a period of higher rates of complication and mortalities during the learning curves of the surgeons. Yet, the transient increase in mortality was not enough to offset the difference in LRFR. Second, salvage esophagectomy might reduce the mortality from local recurrence in CRT patients. As salvage esophagectomy was performed in 15.8% of cases with locally recurrent tumors and was not enough to explain the negative result, it is worth noting that 80% of those cases had R0 resection without any in-hospital mortality or death within 90 days after salvage surgery. Markar et al. [
In terms of radiation dose, our patients received a neoadjuvant dose of around 45 Gy and some patients received 50–54 Gy for definitive aim. Some investigators insisted on the use of 60 Gy or higher for definitive CRT considering that local recurrence occurs in about half of patients as in our current study. Yet, there is not enough empirical evidence to change the current standard dose, which was established based on RCTs; in the near future, however, it may be feasible to administer higher doses through more advanced techniques such as IMRT and Proton. In order to determine the role of esophagectomy after CRT, we chose a rather narrow range of conventional radiation doses because we sought to perform response evaluation at one month after treatment. In order to use a higher radiation dose, a treatment break of 1 month should be considered prior to the additional dose, whose effect is not established at present.
The present study has several limitations. First, it is a retrospective analysis and the result might have been influenced by potential selection bias. Although IPTW-adjusted analysis was performed to adjust for the differences in patient characteristics, unobserved confounding factors may have still been present. Second, the rate of pCR was higher than that of cCR, which suggests that our criteria for cCR may have been overly strict and not representative of the real-world setting. Third, we focused on high-grade treatment-related toxicities and may have underestimated lower grade toxicities and their possible effects. Forth, histological confirmation of recurrent lesions could not be performed in every patient.
Despite these limitations, our study provides clinically meaningful results because the treatments were performed according to a prospectively established study protocol by an experienced multidisciplinary esophageal cancer team. Moreover, we applied the IPTW adjustment to perform a reliable analysis. Although some of our results did not support our initial hypothesis, this study showed that esophagectomy after CRT was associated with improved survival compared with CRT and we hope that our study may be used as a reliable reference for future studies.
Esophagectomy after CRT was associated with significantly better survival results and lower rates of local recurrence and distant metastasis rates compared with CRT. As such effects were more prominent in patients who did not achieve cCR, esophagectomy may be considered in such patients. In complete responders, however, the treatment decision should be made by considering the pros and cons of esophagectomy, which was effective for improving LRFR but not OS.
Supplementary materials are available at Cancer Research and Treatment website (
This study was approved by the institutional review board of Asan Medical Center (Seoul, Korea; IRB number: 2020-1562) and written informed consent was waived because of the retrospective study.
Conceived and designed the analysis: Yu J, Kim JH.
Collected the data: Yu J, Song KJ, Jang JY, Jo YY, Yoo YJ.
Contributed data or analysis tools: Kim JH, Kim SB, Park SR, Kim YH, Kim HR, Lee HJ, Song HJ.
Performed the analysis: Yu J.
Wrote the paper: Yu J, Kim JH.
Review the manuscript: Kim JH.
Conflict of interest relevant to this article was not reported.
We thank Dr. Joon Seo Lim from the Scientific Publications Team at Asan Medical Center for his editorial assistance in preparing this manuscript.
Flow diagram of patient selection. CR, complete response; CRT, chemoradiotherapy; dCRT, definitive chemoradiotherapy; ESCC, esophageal squamous cell carcinoma; nCRT, neoadjuvant chemoradiotherapy; RT, radiotherapy.
Kaplan-Meier survival analysis for overall survival (OS) after inverse probability of treatment weighting adjustment. The OS of the trimodality treatment (TMT) group was significantly better than that of the chemoradiotherapy (CRT) group (p=0.016).
Kaplan-Meier survival analysis for overall survival (A), local recurrence (B), and distant metastasis (C) after inverse probability of treatment weighting adjustment stratified by clinical complete response (cCR) and treatment. (A) In the cCR group, the overall survival of the trimodality treatment (TMT) group was comparable to that of the chemoradiotherapy (CRT) group (p=0.93); in the non-cCR group, the overall survival of the TMT group was significantly better than that of the CRT group (p < 0.001). (B) Local recurrence–free rate of the TMT group was significantly higher than that of the CRT group in both the cCR group (p=0.003) and the non-cCR group (p < 0.001).
Patient characteristics
Trimodality (n=270) | CRT (n=142) | p-value | |
---|---|---|---|
Male | 256 (94.8) | 135 (95.1) | 0.91 |
Female | 14 (5.2) | 7 (4.9) | |
≤ 60 | 127 (47.0) | 39 (27.5) | < 0.001 |
> 60 | 143 (53.0) | 103 (72.5) | |
Mean±SD | 61.36±7.09 | 66.14±8.8 | < 0.001 |
0–1 | 267 (98.9) | 140 (98.6) > | 0.99 |
2 | 3 (1.1) | 2 (1.4) | |
0 | 199 (73.7) | 88 (62.0) | 0.029 |
1 | 56 (20.8) | 42 (29.6) | |
2 | 12 (4.4) | 6 (4.2) | |
3 | 3 (1.1) | 6 (4.2) | |
No | 33 (12.2) | 18 (12.7) | 0.89 |
Yes | 237 (87.8) | 124 (87.3) | |
No | 53 (19.6) | 32 (22.5) | 0.49 |
Yes | 217 (80.4) | 110 (77.5) | |
≤ 2 | 112 (41.5) | 63 (44.7) | 0.53 |
> 2 | 158 (58.5) | 78 (55.3) | |
Negative | 74 (27.4) | 33 (23.2) | 0.36 |
Positive | 196 (72.6) | 109 (76.8) | |
≤ II | 100 (37.0) | 56 (39.7) | 0.60 |
> II | 170 (63.0) | 85 (60.3) | |
Well | 33 (12.2) | 15 (10.6) | 0.021 |
Moderate | 205 (75.9) | 94 (66.2) | |
Poor | 24 (8.9) | 22 (15.5) | |
Upper | 35 (13.0) | 25 (17.6) | 0.21 |
Mid | 127 (47.0) | 55 (38.7) | |
Lower | 108 (40.0) | 62 (43.7) | |
CR | 58 (21.5) | 34 (23.9) | 0.57 |
Non-CR | 212 (78.5) | 108 (76.1) |
Continuous variables were compared using the t test, and categorical variables were compared using the Fisher’s exact test or chi-squared test. CR, complete response; CRT, chemoradiotherapy; ECOG, Eastern Cooperative Oncology Group; SD, standard deviation.
Patterns of failure
TMT | CRT | Total | |
---|---|---|---|
58 | 34 | 92 | |
Local recurrence | 2 (3.4) | 10 (29.4) | 12 (13.0) |
Distant metastasis | 5 (8.6) | 3 (8.8) | 8 (8.7) |
Both | 4 (6.9) | 5 (14.7) | 9 (9.8) |
212 | 108 | 320 | |
Local recurrence | 24 (11.3) | 30 (27.8) | 54 (16.9) |
Distant metastasis | 18 (8.5) | 14 (13.0) | 32 (10.0) |
Both | 26 (12.3) | 18 (16.7) | 44 (13.8) |
Values are presented as number (%). cCR, clinical complete res-ponse; CRT, chemoradiotherapy; TMT, trimodality treatment.
Hazard ratios for oncological outcomes in the entire cohort
Oncologic outcomes | Method | HR |
95% CI | p-value |
---|---|---|---|---|
Overall survival | Univariate | 0.627 | 0.488–0.805 | < 0.001 |
Multivariable-adjusted |
0.651 | 0.487–0.870 | 0.004 | |
IPTW-adjusted | 0.693 | 0.514–0.933 | 0.016 | |
Local recurrence–free rate | Univariate | 0.362 | 0.252–0.519 | < 0.001 |
Multivariable-adjusted |
0.310 | 0.209–0.460 | < 0.001 | |
IPTW-adjusted | 0.352 | 0.235–0.528 | < 0.001 | |
Distant metastasis–free rate | Univariate | 0.585 | 0.388–0.883 | 0.011 |
Multivariable-adjusted |
0.474 | 0.303–0.740 | 0.001 | |
IPTW-adjusted | 0.529 | 0.332–0.843 | 0.007 |
CI, confidence interval; HR, hazard ratio; IPTW, inverse probability of treatment weighting.
Chemoradiotherapy compared with trimodality treatment,
A multivariate analysis was performed using the variables used to calculate the propensity score.
IPTW-adjusted hazard ratios for oncological outcomes in the subgroups stratified by clinical complete response
Clinical response | Oncologic outcomes | HR |
95% CI | p-value | p-value for interaction |
---|---|---|---|---|---|
cCR | Overall survival | 1.027 | 0.561–1.880 | 0.93 | 0.13 |
Local recurrence–free rate | 0.247 | 0.097–0.624 | 0.003 | 0.44 | |
Distant metastasis–free rate | 0.905 | 0.324–2.524 | 0.85 | 0.24 | |
Non-cCR | Overall survival | 0.610 | 0.463–0.805 | < 0.001 | |
Local recurrence–free rate | 0.367 | 0.250–0.539 | < 0.001 | ||
Distant metastasis–free rate | 0.465 | 0.299–0.722 | 0.001 |
cCR, clinical complete response; CI, confidence interval; HR, hazard ratio; IPTW, inverse probability of treatment weighting.
Adjusted hazard ratio, chemoradiotherapy compared with trimodality treatment.
Treatment-related acute toxicity (grade ≥ 3)
Grade | TMT group (n=270) | No. | |
---|---|---|---|
From CRT | 3 | Dysphagia | 2 |
Odynophagia | 2 | ||
Hemorrhage | 1 | ||
From surgery | 3 | Vocal cord palsy | 19 |
Pneumonia | 9 | ||
Infection | 8 | ||
Anastomosis site leakage | 7 | ||
Chylothorax | 2 | ||
Fistula | 1 | ||
Diaphragmatic hernia | 1 | ||
Cardiac toxicity | 1 | ||
4 | Pneumonia | 9 | |
Chylothorax | 3 | ||
Anastomosis site leakage | 3 | ||
Cardiac toxicity | 2 | ||
Infection | 1 |
One case of grade 3 dysphagia was detected in the CRT group.
CRT, chemoradiotherapy; TMT, trimodality treatment.