A Phase III Multicenter Randomized Clinical Trial of 60 Gy versus 50 Gy Radiation Dose in Concurrent Chemoradiotherapy for Inoperable Esophageal Squamous Cell Carcinoma

Esophageal carcinoma is associated with high mortality, with a 5-year overall survival (OS) rate of approximately 20% (1). The landmark RTOG 85-01 trial established concurrent chemoradiotherapy (CCRT) with 50 Gy conventional fractionated radiotherapy as the standard dose for treating inoperable locally advanced esophageal carcinoma (2). However, locoregional failure rates approaching 50% led to the consideration of radiotherapy dose escalation as a possible solution (3‒5). The RTOG 94-05 study concluded that dose escalation benefited neither locoregional control nor survival by comparing a high dose (64.8 Gy) with the standard dose (50.4 Gy) in CCRT for esophageal carcinoma (6). To date, the dose of 50–50.4 Gy has remained the standard for CCRT in most Western countries.

Unlike Western countries, esophageal squamous cell carcinoma (ESCC) accounts for 95% of all esophageal carcinoma patients in China (7). Studies comparing 50 Gy versus 70 Gy doses for esophageal carcinoma in China over the last few decades found that 70 Gy did not improve local control or survival and was associated with a higher incidence of treatment discontinuation than 50 Gy (8‒10). All of the above conclusions were drawn using two-dimensional radiotherapy technology, which has apparent dosage deficiency and cold spots compared with three-dimensional conformal radiotherapy (11); these may be the main reasons for a high local failure rate.

Currently, clinicians in China have commonly adopted 60 Gy dose using modern radiotherapy technologies based on the belief that 50 Gy is inadequate for ESCC, given the different biological characteristics between ESCC and adenocarcinoma (12). However, no prospective evidence was available that 60 Gy improves disease control compared with 50 Gy, although several retrospective studies have shown that the outcomes promoted by increasing the radiotherapy dose (12‒14). Here, we conducted a prospective, randomized, multicenter, phase 3 clinical trial comparing the clinical outcomes of ESCC following the delivery of 60 Gy in 30 fractions or 50 Gy in 25 fractions via modern radiotherapy technologies combined with CCRT.

This randomized, phase 3 clinical trial was approved by the institutional review boards of each institution and conducted according to the tenets of the Declaration of Helsinki. All patients provided written informed consent for participation in the study. The study protocol was registered at ClinicalTrials.gov (NCT01937208) and was included in the Supplementary Materials.

Eligible patients were histologically or cytologically confirmed to have ESCC and satisfied the condition that their disease was medically inoperable. All cases were stage IIA‒IVA, according to the 6th edition of the American Joint Committee on Cancer staging manual for esophageal carcinoma. Further inclusion criteria included ages 18‒70 years, Karnofsky Performance Status (KPS) ≥ 70, no prior anticancer therapy, measurable target lesions according to RECIST 1.1, life expectancy ≥3 months, and adequate bone marrow, hematologic, hepatic, and renal function. Patients with invasion of the tracheobronchial tree or tracheoesophageal fistula, multiple esophageal carcinomas, or extension of the carcinoma to within 2 cm of the gastroesophageal junction were excluded.

Patients were randomly assigned in a 1:1 ratio to the 60 Gy group or the 50 Gy group using a stratified permuted block method.

The whole course of treatment was recorded in our previous study (15). The radiotherapy was delivered using intensity-modulated radiotherapy (IMRT) or image-guided radiotherapy (IGRT; IMRT verified by cone-beam CT). Middle or lower thoracic ESCC were recommended to use IGRT technology for better motion management; however, it was not mandatory. All patients underwent CT simulation in the supine position, with CT images obtained at 5 mm thickness throughout the entire neck, thorax, and upper abdomen. The gross target volume (GTV) included the primary tumor (GTV-T) and lymph node metastasis (GTV-N). GTV-T included all the esophageal tumors identified through CT scan, esophageal barium swallow radiography, endoscopy, endoscopic ultrasonography, and PET/CT (PET/CT was not mandatory in this trial). GTV-N was diagnosed if at least one of the following criteria were met: Nodes >1.0 cm at the shortest axis in the intrathoracic and intra-abdominal region; nodes >0.5 cm beside the paraesophageal, tracheoesophageal sulcus, and diaphragmatic crura on CT scans; or with >3.0 maximum standardized uptake of ¹⁸F-deoxyglucose on PET/CT images (16‒18). The clinical target volume for the primary tumor (CTV-T) was defined as a 3.0-cm superior and inferior margin and a 0.6-cm lateral margin from the GTV-T according to the treatment guideline of radiotherapy for Chinese esophageal carcinoma (19). The recommended clinical target volume for lymph node metastasis (CTV-N) included the cervical and upper mediastinal nodes for lesions in the cervical and upper thoracic esophagus, the upper and middle/lower mediastinal regions in the middle thoracic esophageal carcinoma, and the middle/lower mediastinal and abdominal lymph node regions for lesions in the lower thoracic esophagus. The planning target volume (PTV) was generated by adding a uniform 0.5-cm margin around the CTV (CTV-T+CTV-N).

The plans were optimized on the basis of a dose-volume histogram. The prescribed isodose curve covered 95% of the PTV, and the 95% isodose curve corresponded to 99% of the PTV. The maximum allowable dose within the PTV was 110% of the prescribed dose. Inhomogeneity corrections were also performed. The following dose constraints to the organs-at-risk (OAR) were specified: (i) The mean lung dose (MLD) was ≤15 Gy, the V20 (percentage of the total lung volume receiving ≥ 20 Gy) was ≤30%, V30 ≤ 20%, and V5 ≤ 60%. (ii) The mean heart dose (MHD) was ≤25 Gy; V30 < 60%, and V40 < 50%. (iii) The maximum spinal cord dose was ≤ 45 Gy. (iv) For the liver, V50 < 30% and V30 < 60%. (v) For the kidney, V20 < 40%. If these criteria could not be met, MLD was set to <17 Gy, lung V20 < 34%, and MHD to <28 Gy. The radiotherapy treatment plans were confirmed by researchers at each center and were centrally reviewed if the radiotherapy plan could not meet dose distribution or sparing of OAR in the protocol. Using the eligibility criteria for each beam and other criteria, including 3% dose difference and a 3-mm distance to agreement, which were recommended by the American Association of Physicists in Medicine Task Group 119, the planar dose distributions assessed were more than 95% of the data points (20). Pre-treatment quality assurance verification for each IMRT plan was performed for central review using fluency maps measured with an electronic portal imaging device. If the patients received IGRT, cone-beam CT online verification was performed for 5 consecutive days in the first week of radiotherapy and once a week thereafter.

On day 1 of radiotherapy, docetaxel was administered to all patients (25 mg/m² intravenously, over 60 minutes) followed by cisplatin (25 mg/m² intravenously, over 30 minutes). Dexamethasone (10-mg, intravenously) and diphenhydramine (50 mg, intramuscularly) were administered 30 minutes before treatment. This was repeated weekly alongside concurrent chemotherapy for 5 weeks. Chemotherapy was discontinued if grade 3/4 toxicity developed and restarted with a 10% dose reduction if the severity decreased to ≤ grade 2. After completing CCRT, followed by an interval of 3‒4 weeks, patients received two cycles of consolidation chemotherapy with docetaxel 70 mg/m² combined with cisplatin 25 mg/m² on days 1‒3.

All enrolled patients were examined and graded weekly throughout the treatment. The highest score was recorded as the patient's toxicity grade. Routine examinations, including physical examination, serum chemistry profile, barium swallow, and chest and upper abdomen CT scans, were performed to evaluate the response at 4 weeks after CCRT and consolidation chemotherapy completion. Then, it was performed every 3 months for the first year, every 4 months for 2 years, and every 6 months subsequently. Additional diagnostic procedures were performed if clinically indicated.

The primary endpoint was locoregional progression-free survival (LRPFS) and the analysis was based on time from randomization to the diagnosis of a locoregional recurrence as the first event. Locoregional failure was defined as the presence of primary tumor or regional lymph node progression on CT confirmed by pathology or identified by the investigator and the senior radiologist through integration with other imaging examinations, including PET/CT. Censoring occurred for patients experiencing other events before locoregional recurrence (distant recurrence alone or survival at last follow-up). The secondary endpoints included OS, PFS, patterns of failure, and toxicity. The OS and PFS were calculated from the randomization date to the date of death, disease progression, or last follow-up time. Treatment-related toxicity was evaluated and graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events version 4.0 (21).

It was estimated that the 3-year LRPFS rate would be 50% in the 50 Gy group and 65% in the 60 Gy group. Thus, 140 events would be required to detect an absolute 15% improvement in 3-year survival (corresponding HR, 0.62) by 80% power at a two-sided 5% significance level. On the basis of 2 years of accrual and 2 more years of follow-up, 320 patients were needed for accrual, allowing for a 24% drop-out rate. Categorical variables between the 60 Gy and 50 Gy groups were compared using Pearson's χ² test or Fisher's exact test. Independent t tests were used to compare the dosimetric parameters of the OAR between the two groups. LRPFS, PFS, and OS rates were calculated using the Kaplan‒Meier method with log-rank tests. Cox proportional hazard models were used to calculate the hazard ratio (HR) for recurrence or death. All analyses were performed using Statistical Package for Social Sciences (SPSS) 23.0 (SPSS Inc.) and GraphPad Prism 8.3.0 (GraphPad Software).

The data generated in this study are not publicly available due to possible compromise patient privacy but are available upon reasonable request from the corresponding author.

Between May 10, 2013 and May 16, 2017, a total of 324 patients from 22 sites were randomized to receive 60 Gy (experimental group, n = 161) or 50 Gy (control group, n=163; Fig. 1). Five were lost to the first follow-up and efficacy evaluation (1 in the 60 Gy group, 4 in the 50 Gy group); therefore, 319 patients were defined as the Full-Analysis-Set population. Baseline characteristics were similar between the two groups (Table 1). The actual percentage of patients who received the PET/CT staging was 15.4% (49 of 319). Five cases (3.1%) in the 60 Gy group and two (1.3%) in the 50 Gy group applied for central review. Following discussion by the expert group, 3 cases received a modified target volume, whereas the other 4 received an optimized plan. Finally, all patients met the protocol requirements. There were no statistically significant differences between the 60 Gy and 50 Gy groups in terms of dosimetric comparisons of the OAR, except for lung V30 (11.67 ± 3.77 vs. 10.64 ± 3.95, P = 0.03), heart V50 (4.93 ± 5.35 vs. 2.41 ± 3.03, P < 0.001), and the maximum dose to the spinal cord (41.60 ± 4.68 vs. 39.77 ± 6.63, P = 0.01).

The vast majority (87.8%) of patients in this study used IMRT technology, whereas the remainder (12.2%) received IGRT. In addition, the total radiotherapy completion rates were 88.2% and 96.9% in the 60 Gy and 50 Gy groups (P < 0.01), respectively (Supplementary Table S1). Of the remaining patients who did not complete RT, 7 received 55–58 Gy, 8 received 50–54 Gy, and 4 received 30–48 Gy in the 60 Gy group, whereas 5 received 20–48 Gy in the 50 Gy group. The reasons for suspending radiotherapy in the 60 Gy group mainly included severe esophagitis, poor KPS score, pulmonary infection, and fistula (n = 6, 6, 6, and 1 patients, respectively). Radiotherapy was discontinued in the 50 Gy group due to poor tolerance and fistula (n = 4 and 1 patients, respectively). The full five courses of weekly concurrent chemotherapy were completed for 58.1% of patients, 3–4 courses were completed for 30.0% of patients, and 0–2 courses were completed for 11.9% of patients in the 60 Gy group, compared with 63.5%, 27.0%, and 9.4%, respectively, in the 50 Gy group. Owing to esophagitis and poor physical condition, the full two courses of consolidation chemotherapy were completed by 46.3% of patients in the 60 Gy group, compared with 55.3% in the 50 Gy group. The objective response rates evaluated at 4 weeks after completion of all treatments were 90.6% [25.3% complete response (CR) and 65.3% partial response (PR)] in the 60 Gy group and 93.4% (26.1% CR and 67.3% PR) in the 50 Gy group.

With a median follow-up of 34.0 (range, 1.4–89.3) months for all 319 patients, and 55.2 (1.5‒89.3) months for the 153 surviving patients, there were no statistically significant differences in LRPFS, OS, or PFS rates. In the 60 Gy and 50 Gy groups, the 1-, 2-, and 3-year LRPFS rates were 75.6%, 56.9%, and 49.5% versus 72.1%, 57.2%, and 48.4%, respectively, with a median LRPFS of 36.0 and 34.5 months, respectively [HR, 1.00; 95% confidence interval (CI), 0.75‒1.35; P = 0.98; Fig. 2A]. The OS rates were 83.7%, 65.1%, and 53.1% versus 84.8%, 62.8%, and 52.7%, with a median survival of 45.3 and 41.2 months, respectively (HR, 0.99; 95% CI, 0.73‒1.35; P = 0.96; Fig. 2B). The PFS rates were 71.2%, 54.4%, and 46.4% compared with 65.2%, 52.1%, and 46.1%, respectively, with a median PFS of 27.7 and 25.5 months, respectively (HR, 0.97; 95% CI, 0.73‒1.30; P = 0.86; Fig. 2C). Analyses for population who had completed the prescribed radiotherapy dose in both arms demonstrated that the 3-year LRPFS and OS rates were 49.6% versus 50.1% (P = 0.94) and 53.7% versus 54.5% (P = 0.85), respectively. Subgroup analysis also demonstrated that the LRPFS and OS did not depend on patient baseline clinical features, response to the first 40 Gy irradiation, or the number of concurrent chemotherapy courses (Supplementary Figs. S1 and S2, respectively).

By the last follow-up, the rates for any failure were 46.3% and 45.3% in the 60 Gy and 50 Gy groups, respectively, including locoregional failure in 26.9% versus 26.4%, distant metastasis in 16.3% versus 16.4%, and both locoregional and distant failure in 3.1% versus 2.5%, respectively; these values were not significantly different (Table 2). Most of the locoregional failures occurred in the GTV (60 Gy group, 81.3%; 50 Gy group, 84.8%).

All 324 intention-to-treat populations were analyzed for toxicity. The treatment-related grade 3 or higher adverse events are presented in Table 3. All adverse events of grade 3 or higher were recorded in 54.7% of patients in the 60 Gy group versus 50.9% in the 50 Gy group (P = 0.50). The incidence of grade 3 or higher radiotherapy–related pneumonitis was higher in the 60 Gy than the 50 Gy group (7.5% vs. 3.1%, nominal P = 0.03). Seven patients in the 60 Gy group experienced grade 5 side effects possibly attributed to CCRT: 4 cases of esophageal fistula (2 of them only received 30 Gy and 50 Gy), 1 case of esophageal bleeding, and 2 pulmonary-related deaths. Eight patients experienced grade 5 side effects in the 50 Gy group, including 4 with esophageal fistula (1 of them only received 20 Gy), 2 cases of esophageal bleeding, and 2 pulmonary-related deaths.

This multicenter, randomized, prospective study compared a higher radiotherapy dose (60 Gy) with the standard dose (50 Gy) to assess definitive CCRT using modern IMRT or IGRT technology for inoperable ESCC. Our trial demonstrated no significant differences in the primary endpoint of LRPFS between the 60 Gy and 50 Gy groups.

Using a dose of 50‒50.4 Gy, established as the standard for esophageal carcinoma, almost 50% of patients developed locoregional recurrence after CCRT (3, 22), which indicates a need for local therapy improvement. Compared with the RTOG 94-05 trial, our study demonstrated an advantage in LRPFS and lower treatment-related mortality in the high-dose group (6), which may be attributed to modern techniques and advances in radiotherapy. However, the dose escalation of 10 Gy does not offer any locoregional control or survival benefits, and the 3-year LRPFS in the 60 Gy group was not even close to the hypothesized 65%. These findings are consistent with the results of three similar prospective studies reported previously in different periods (6, 23, 24). Therefore, it can be seen that radiotherapy dose escalation does not necessarily translate into improvement of efficacy in clinical practice due to radiosensitivity, OAR tolerance, and other reasons.

Some retrospective studies in recent years have reported that dose escalation in the definitive treatment of locally advanced esophageal carcinoma may improve local control but not OS. He and colleagues (25) reported that a high dose (>50.4 Gy) had significantly lower local failure rate but no benefit in regional failure or 5-year OS. Similar data reported by Brower and colleagues (26) showed that the OS for patients with esophageal carcinoma remained unchanged when the radiotherapy dose was escalated to >50.4 Gy. Ke and colleagues (27) declared that escalated radiotherapy doses for inoperable ESCC affected neither OS nor disease-free survival and did not reduce locoregional or distant failure. However, other retrospective analyses showed different results. Chen and colleagues (28) found that a higher (≥60 Gy) than standard (50‒50.4 Gy) radiotherapy dose may improve survival. One systematic review of the clinical benefit of different radiotherapy doses in CCRT for esophageal carcinoma using modern radiotherapy techniques concluded that high-dose radiotherapy (particularly ≥ 60 Gy) induced a favorable benefit-risk profile by improving the OS without increasing severe toxicities (29).

The recently published ARTDECO study indicated that dose escalation from 50.4 Gy to 61.6 Gy at the primary tumor did not result in a significant increase in local control or OS, and no dosage effect was observed in either adenocarcinoma or squamous cell carcinoma (24). However, it should be noted that 61.6 Gy only covered the primary tumor site, and 38% of patients were not diagnosed with ESCC in the ARTDECO study. In comparison, the prescribed dose was delivered consistently to the planning target in our trial, and all patients enrolled had ESCC. In this study, there was a lower 3-year LRPFS (50% vs. 59% for the high-dose group; 48% vs. 53% for the standard-dose group) but a higher OS at 3 years (53% vs. 39% and 53% vs. 42%, respectively), compared with the ARTDECO trial, which might be attributed to the higher proportion of patients with T4 disease (23% vs. 7%) in our study. However, all our patients were pathologically confirmed as squamous cell carcinoma (100% vs. 62%), which shows different risk factors and biological characteristics from adenocarcinoma (30). Referring to the long-term results of NEOCRTEC5010 and CROSS randomized trials, the survival of patients with ESCC was demonstrated to be superior to that of patients with adenocarcinoma (31‒33). Furthermore, the younger median age of our patients (63 vs. 71 years) may be an alternative explanation for better OS.

Our results raise the question: Why did dose escalation not improve efficacy? In our study, locoregional failure was still the most common type of progression, particularly in the GTV, although the incidence rate was significantly lower than that in previous trials (6, 34). We speculated that a dose of 50 Gy combined with concurrent chemotherapy might be sufficient to kill radiosensitive ESCC, whereas the radioresistant tumor cells would not be killed even with ≥60 Gy doses. Future molecular/genetic studies on the heterogeneity of the esophageal carcinoma are necessary for the development of individualized treatment. The implementation of PET/CT also has the potential capability to select esophageal carcinoma patients who may benefit from escalated irradiation doses according to metabolic response levels (35‒37).

Our study also found that radiotherapy completion rates in the 60 Gy group were lower than those in the 50 Gy group (88.2% vs. 96.9%, P < 0.01) with a higher incidence of grade 3+ radiotherapy–related pneumonitis. Although the treatment compliance was statistically acceptable, the therapeutic advantage of the 60 Gy group may be diluted. We then conducted additional analyses for the population who had completed the prescribed radiotherapy dose in both arms, and no significant difference was seen in these values as well.

Our study has some limitations. First, most of the radiotherapy plans were reviewed and confirmed at each sub-center, and lack of central review, to some extent, had been shown to impact overall results of studies in the past (38, 39). Second, some occult metastatic disease that was not diagnosed or properly included in the radiotherapy fields might be missed for the low implementation of PET/CT scan. Third, more closely monitoring is required in future investigator initiated trials.

In conclusion, this study revealed that LRPFS was similar for patients in the 60 Gy and 50 Gy groups, but severe radiotherapy pneumonitis was significantly more common in patients in the 60 Gy group. Fifty Gy should be the recommended radiotherapy dose of CCRT in the treatment of ESCC.

No disclosures were reported.

Y. Xu: Conceptualization, resources, data curation, supervision, funding acquisition, writing–original draft, writing–review and editing. B. Dong: Data curation, software, formal analysis, writing–original draft, writing–review and editing. W. Zhu: Resources, data curation. J. Li: Resources, data curation. R. Huang: Resources, data curation. Z. Sun: Resources, data curation. X. Yang: Resources, data curation. L. Liu: Resources, data curation. H. He: Resources, data curation. Z. Liao: Investigation, visualization, methodology. N. Guan: Data curation, software. Y. Kong: Resources, data curation, validation, investigation. W. Wang: Resources, data curation. J. Chen: Resources. H. He: Resources. G. Qiu: Resources, data curation. M. Zeng: Resources, data curation. J. Pu: Resources, data curation. W. Hu: Resources, data curation. Y. Bao: Resources, data curation. Z. Liu: Resources, data curation. J. Ma: Resources, data curation. H. Jiang: Resources, data curation. X. Du: Resources, data curation. J. Hu: Resources, data curation. T. Zhuang: Resources, data curation. J. Cai: Resources, data curation. J. Huang: Resources, data curation. H. Tao: Resources, data curation. Y. Liu: Resources, data curation. X. Liang: Resources, data curation. J. Zhou: Resources, data curation. G. Tao: Resources, data curation. X. Zheng: Resources, data curation. M. Chen: Conceptualization, supervision, funding acquisition, validation, visualization, methodology, project administration, writing–review and editing.

This research was supported by the Basic Public Welfare Research Program of Zhejiang Province under award number LGF21H160005 (to Y. Xu). We express our gratitude to all those who helped us during the project process. The number of participants in the project was not listed owing to limitations on space. Special acknowledgments should be given to Yan Xu for secretarial work.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.