Do Higher Radiation Doses with Concurrent Chemotherapy in the Definitive Treatment of Esophageal Cancer Improve Outcomes? A Meta-Analysis and Systematic Review

Background: To investigate the effects and safety profile of radiation dose escalation utilizing computerized tomography (CT) based radiotherapy techniques (including 3-Dimensional conformal radiotherapy, intensity-modulated radiotherapy and proton therapy) in the definitive treatment of patients with esophageal carcinoma (EC) with definitive concurrent chemoradiotherapy (dCCRT). Methods: All relevant studies utilizing CT-based radiation planning, comparing high-dose (≥ 60 Gy) versus standard-dose (50.4 Gy) radiation for patients with EC were analyzed for this meta-analysis. Results: Eleven studies including 4946 patients met the inclusion criteria, with 96.5% of patients diagnosed with esophageal squamous cell carcinoma (ESCC). The high-dose group demonstrated a significant improvement in local-regional failure (LRF) (OR 2.199, 95% CI 1.487-3.253; P<0.001), two-year local-regional control (LRC) (OR 0.478, 95% CI 0.309-0.740; P=0.001), two-year overall survival (OS) (HR 0.744, 95% CI 0.657-0.843; P<0.001) and five-year OS (HR 0.683, 95% CI 0.561-0.831; P<0.001) rates relative to the standard-dose group. In addition, there was no difference in grade ≥ 3 radiation-related toxicities and treatment-related deaths between the groups. Conclusion: Under the premise of controlling the rate of toxicities, doses of ≥ 60 Gy in CT-based dCCRT of ESCC patients might improve locoregional control and ultimate survival compared to the standard-dose dCCRT. While our review supports a dose-escalation approach in these patients, multiple ongoing randomized trial initial and final reports are awaited to evaluate the effectiveness of this strategy.


Introduction
Globally, esophageal cancer (EC) is one of the most common causes of cancer-related death [1,2] . Collectively, the 5-year overall survival (OS) rate for all patients is 19% [2] . For non-operable, localized EC, definitive concurrent chemoradiotherapy (dCCRT) is usually the standard treatment approach. Currently, the dCCRT radiation dose recommended by National Comprehensive Cancer Network (NCCN) is 50.4 Gy Ivyspring International Publisher [3] , which is primarily based on results of the Radiation Therapy Oncology Group (RTOG) 85-01 and 94-05 randomized trials [4][5][6] . However, the optimal radiation dose in this scenario remains controversial. Many studies have demonstrated that local-regional failure (LRF) remains a common failure pattern for EC patients following dCCRT, most likely to occur within the original gross tumor volume (GTV), even in patients achieving clinical complete response (cCR) following treatment [7][8][9][10][11] . In addition, LRF rates of esophageal squamous cell carcinoma (ESCC) appear to be higher than that of esophageal adenocarcinoma (EAC) [12,13] . Because patients with EAC generally have lower cCR rates with chemoradiation and are frequently considered for surgical resection following neoadjuvant treatment (with resultant high rates of local control), these patients have not been the primary focus of dose escalation studies. However, patients with ESCC often have additional comorbidities and generally achieve higher cCR rates with chemoradiation, leading to consideration of non-operative approaches. Therefore, the role of dose escalation is more relevant in patients with ESCC, particularly given that LRF rates are high following standard chemoradiation approaches.

Methods
Studies published prior to February 2020 comparing radiation dose and disease-related outcomes in nonoperable EC patients were included. Search keywords included "esophageal or oesophageal" and "carcinoma or cancer or neoplasm" and "chemoradiotherapy or chemoradiation or radiochemotherapy or irradiation or chemo-irradiation" and "dose or dose escalation or high dose". After retrieval, we manually filtered the articles by abstract and/or full text review. Inclusion criteria included: 1) CT-based radiotherapy techniques, such as 3-D, IMRT, or proton beam therapy, were utilized (articles that applied 2D radiotherapy techniques were excluded); 2) dCCRT was used (articles that reported sequential treatments, neoadjuvant or adjuvant CRT combined with surgery, palliative CRT or radiotherapy alone were excluded); 3) subject patients were stratified into high-dose (≥ 60 Gy) and standard-dose (approximately 50 Gy) groups, with comparative data provided; 4) accurate statistical methods, valid data, and clear conclusions were given; 5) hazard ratio (HR) and 95% confidence intervals (CI) were provided or could be calculated.
Outcome data included cCR rate, LRF rate, two-year LRC rate, two-and five-year OS, grade ≥ 3 radiation-related toxicities and treatment-related deaths.
Case-control study evaluation guidelines were applied in order to evaluate the quality of each manuscript for the following criteria: 1) whether gender, age, and tumor location were clearly stated; 2) whether the comparability of the two groups was analyzed; 3) whether the statistical method was appropriate; 4) whether biases were discussed in the study. A point was assigned for each of these four items, with a total score of ≥ 3 indicative of reliable quality. Two researchers independently reviewed the literature according to the unified quality standard, with results crosschecked. If there were some different opinions, a third researcher would be invited to solve the disagreement.
Data were analyzed using Stata version 11.0. Hazard ratio (HR), odds ratio (OR) and 95% CI were used to measure effect size. A Q test was applied to test for result heterogeneity. If P > 0.05, the fixed effect model was used for statistical consolidation. If P ≤ 0.05, the random effect model was used. The combined effect size was tested utilizing the z test. Funnel plots were created to evaluate the risk of publication bias.
Three articles analyzed the LRF rate of the two groups [14,17,25] . High-dose group had a significant advantage over the standard-dose group in this respect (OR 2.199, 95% CI 1.487-3.253; P<0.001, Figure  3).

Sensitivity analysis
Sensitivity analysis showed that the new combined HR of two-year OS rate were different from the original HR while other items were similar to the original HR/OR ( Table 3).

Discussion
This study demonstrated that a radiation dose of ≥ 60 Gy utilizing CT-based radiation techniques for the dCCRT of esophageal cancer might decrease LRF and improve two-year LRC, and five-year OS rates of patients without increasing toxicity rates compared to standard RT doses. As there were approximately 97% of patients with ESCC in this study, whether this conclusion was applicable to patients with EAC requires further verification.
RTOG 85-01 and RTOG 94-05/INT 0123 trials established a dose of 50.4 Gy in the dCCRT for patients with inoperable, localized EC [4][5][6] . The RTOG 94-05 trial represents the only large randomized controlled trial (RCT) evaluating RT dose in this setting, comparing the effect of high-dose (64.8 Gy) versus standard-dose (50.4 Gy) treatment. Study results showed that the high-dose arm failed to demonstrate any survival benefit while showing a higher treatment-related mortality rate [6] . However, there were caveats, including that seven of the eleven treatment-related deaths in the high-dose group occurred before 50.4 Gy, i.e. dose-escalation was not the cause of death in these patients. Additionally, patients in the high-dose group underwent a treatment break for side effects recovery, resulting in a significantly prolonged treatment time. Finally, patients in the high-dose group received significant lower doses of 5-fluorouracil (5-FU) compared to those in the standard-dose group. These and other factors have the potential to influence ultimate outcomes and potentially negate any benefits of dose-escalation. Additionally, radiation technique in this study was based on what are largely historical 2D approaches, as well as utilized fields larger than that used in contemporary practice, which may increase toxicity rates and not be well suited for dose-escalation.
More recently, improvements in radiotherapy techniques (including CT-based planning), have led to reevaluation of dose-escalation approaches in the dCCRT of esophageal cancer. In a study by Suh et al [21] , the high-dose group showed significant improvement in 2-year LRC rate (69% vs. 32%, P<0.01). Ren and colleagues found that the 10-year LRC rate increased by more than 20% (52% vs. 29.8%, P=0.028) with higher radiation doses [19] . Similarly, Kim et al demonstrated that their high-dose group experienced significantly lower LRF-alone rates (25.9% vs. 39.2%, P=0.029) and a significantly higher 2-year LRC rate (69.1% vs. 50.3%, p=0.002) relative to the standard-dose group [17] . He et al also showed that their high-dose group had a significantly lower LRF rate (17.9% vs. 34.3%, P=0.024) [14] . Welsh et al. use simultaneous integrated boost (SIB)-IMRT technique to dose-escalate gross disease and demonstrated that this approach could reduce LRF rates for patients with unresectable locally advanced EC (29% vs. 54%, P=0.01) [25] . Along these lines, the current meta-analysis also demonstrated that the high-dose group had significantly lower LRF rates (OR 2.199, 95% CI 1.487-3.253; P<0.001) and higher two-year LRC rates (OR 0.478, 95% CI 0.309-0.740; P=0.001) compared to the standard-dose group.
Given the above, the role of increased LRC on survival benefits has also been explored. Some studies have indicated that higher radiation dose may lead improved survival for ESCC patients undergoing dCCRT [15-17, 19-21, 23-24] , while others have not [14,18,[26][27][28][29][30] . A recent registry study reported by Japanese investigators demonstrated 5-year OS rates for stage groups I, II/III and IV were 64.2%/57.2%, 35.0%/27.0% and 18.0%/15.3% in the 50.4 Gy and 60 Gy groups, respectively, with no superiority of the high-dose group in any stage group [28] . A meta-analyses by Song et al in 2015 included 55 articles and concluded that high-dose radiation increased treatment response and 5-year OS rates while decreasing LRF and distant failure rates with without additional toxicity, notably for ESCC patients [31] . Another study by Chen et al analyzed 18 articles and also found that a dose of ≥ 60 Gy appeared to improve OS and LRC, notably in Asian patients [32] . Luo et al. reached a similar conclusion [33] . In spite of this, the studies analyzed in these three meta-analyses included many series implementing 2D radiation techniques, which may not reflect the real impact of dose-escalation in modern practice and planning. In the current study, we defined inclusion criteria to include only studies utilizing CT-based radiation planning techniques. Our results demonstrated that a dose of ≥ 60 Gy appeared to increase 5-year OS compared to standard doses, without increasing treatment related toxicities, supporting this strategy in the modern planning approaches in these patients. Based on the current data, the chemotherapy regimens used were not significantly different between the two groups, respectively, in most enrolled studies. While as the limitation of the data, we cannot analyze the effects of the different chemotherapy regimens. As previous researches reported [34,35] , different chemotherapy regimens in the definitive chemoradiotherapy may not affect the survival of patients.
In view of the above results, it is important to review the underlying principles of dose escalation. As reported by Fletcher over four decades ago [36] , biologically, doses of 45-50 Gy are generally adequate to control microscopic disease, ≥ 60 Gy required to control gross disease, and nearly 100 Gy to cure solid tumors. In other diseases, including non-small cell lung cancer, it has been estimated that there is an approximate 1% improvement in long-term LRC and 3% decrease in the hazard from death with each additional Gy of radiation delivered [37] . To our knowledge, the failure of RTOG 94-05 was related to high rates of toxicities in 2D era. A latest study in Japan used proton beam therapy for esophageal cancer and reported lower rates of toxicities in comparison to photon radiotherapy [38] . The biological effective dose (BED) in this study was up to 87.2 Gy (67.2 -96.1 Gy). The 3-year, 5-year OS rate and five-year LC rate was 66.7%, 56.3% and 64.4%, respectively. The 5-year OS rate based on stage IV were 28.3%. 3D radiotherapy techniques could control the toxicities and side effects, so higher doses may bring survival benefits to patients. In most Asian countries, where ESCC is the predominant histological type, doses of ≥ 60 Gy are much common. Exemplifying this, in the current meta-analysis, ESCC was the predominant histology (approximately 96.5% of patients) and eight of eleven articles included were from Asian countries/only two from the west, making our results particularly relevant to patients with ESCC from Asian countries.
There are limits to this meta-analysis. Firstly, articles were searched among published articles written in English and statistical analysis was limited to the published data. So, publication and language bias maybe exist. Secondly, ten of eleven included studies were retrospective studies (with the inherit limitations), with the one prospective randomized study only included 28 cases. It was a pity that no randomized trial and prospective study with large cases have been published yet (trials ongoing). But two studies with large cases (Chen et al [15] and Ren et al [19] ) constructed a propensity score matched cohort to balance observable potential confounders. Additional RCTs are needed to verify our conclusion, with many ongoing ( Table 5). Thirdly, our data include that the combined HR/OR of two-year OS rate were different from the original results following sensitivity analysis. Additionally, we did not perform subgroup analyses with regard to tumor site, stage and pathological type, with locally advanced, thoracic ESCC tumors comprising most of our population.
Finally, although cCR is defined as not visible tumor following dCCRT [39] , in the clinical practice, specific evaluation methods and criterion of cCR may not be uniform, creating uncertainty as to whether cCR following dCCRT is able to guide subsequent treatment recommendations and requires further study. These factors might have influenced our findings and conclusion.
In conclusion, our study demonstrated that, compared to the standard radiation doses, dose-escalated (≥60 Gy), CT-based radiotherapy techniques might improve ultimate disease-related outcomes in patients with inoperable ESCC under the premise of controlling toxicity rates, and represents the first meta-analysis comparing high-versus standard-radiation doses utilizing CT-based/modern radiotherapy techniques. While our review supports a dose-escalation approach in these patients, multiple ongoing randomized trial initial and final reports are awaited to evaluate the effectiveness of this strategy. In addition, as there were approximately 97% of patients with ESCC in this study, whether this conclusion was applicable to patients with EAC similarly requires further verification.