Concurrent Chemoradiotherapy Versus Radiation Monotherapy for Patients with Locally Advanced Esophageal Squamous Cell Carcinoma in the Era of Intensity Modulated Radiotherapy: A Propensity Score-matched Analysis

Background: To investigate the survival benet of concurrent chemoradiotherapy for patients with locally advanced esophageal squamous cell carcinoma during the years of intensity-modulated radiotherapy. Methods: Medical records of 1273 patients with esophageal squamous cell carcinoma who received intensity-modulated radiotherapy from January 2005 to December 2017 in the CAMS were retrospectively reviewed. 683 patients received concurrent chemoradiotherapy, 590 patients received radiotherapy alone. Propensity score matching (PSM) method was used to eliminate baseline differences between the two groups. Survival and toxicity prole were evaluated afterwards. Results: After a median follow-up time of 50.4 months (3.2-157.4 months), both overall survival and progression-free survival of the concurrent chemoradiotherapy group were better than those of the radiotherapy group, either before or after PSM. After PSM, the 1-, 3-, 5-year OS of radiotherapy and concurrent chemoradiotherapy groups were 63.3% vs 72.2%, 31.6% vs 42.2% and 28.5% vs 38.1%, respectively (p=0.003). The 1-year, 3-year and 5-year PFS rates of radiotherapy and concurrent chemoradiotherapy group were 44.3% vs 48.6%, 23.4% vs 31.2% and 15.8% vs 25.2%, respectively (p=0.002). The rates of ≥ grade 3 leukopenia and radiation esophagitis in the concurrent chemoradiotherapy were higher than those in the radiotherapy alone group (p<0.05). There was no signicant difference in the probability of radiation pneumonia between the two groups (p=0.359). Multivariate logistic regression analysis showed ≥ 70 years old, female, KPS ≤ 70, stage I-II, and patients diagnosed at earlier years (2005-2010) had lower probability of receiving concurrent chemoradiation (p<0.05). Multivariate Cox analysis indicated that female, stage I-II, EQD2 ≥ 60Gy and concurrent chemotherapy were favorable prognostic factors for both OS and PFS.

Results: After a median follow-up time of 50.4 months (3.2-157.4 months), both overall survival and progression-free survival of the concurrent chemoradiotherapy group were better than those of the radiotherapy group, either before or after PSM. After PSM, the 1-, 3-, 5-year OS of radiotherapy and concurrent chemoradiotherapy groups were 63.3% vs 72.2%, 31.6% vs 42.2% and 28.5% vs 38.1%, respectively (p=0.003). The 1-year, 3-year and 5-year PFS rates of radiotherapy and concurrent chemoradiotherapy group were 44.3% vs 48.6%, 23.4% vs 31.2% and 15.8% vs 25.2%, respectively (p=0.002). The rates of ≥ grade 3 leukopenia and radiation esophagitis in the concurrent chemoradiotherapy were higher than those in the radiotherapy alone group (p<0.05). There was no signi cant difference in the probability of radiation pneumonia between the two groups (p=0.359). Multivariate logistic regression analysis showed ≥ 70 years old, female, KPS ≤ 70, stage I-II, and patients diagnosed at earlier years (2005-2010) had lower probability of receiving concurrent chemoradiation (p<0.05). Multivariate Cox analysis indicated that female, stage I-II, EQD2≥60Gy and concurrent chemotherapy were favorable prognostic factors for both OS and PFS.
Conclusions: Concurrent chemotherapy can bring survival bene ts to patients with locally advanced esophageal squamous cell carcinoma receiving intensity-modulated radiotherapy. For patients who cannot tolerate concurrent chemotherapy, radiation monotherapy is an effective alternative with promising results.

Background
Esophageal cancer (EC) is a common malignancy with poor prognosis, and China is the country with the largest number of new cases and deaths every year. According to a national epidemiological research conducted in 2015 [1], the numbers of new diagnosed cases and deaths of esophageal cancer in China have reached 478,000 and 375,000, ranking third and fourth respectively among all malignant tumors. Unlike most Western countries, the pathological type of esophageal cancer in China is still dominated by squamous cell carcinoma (SCC), and patients with esophageal squamous cell carcinoma account for more than 90% of all EC patients [2,3].
Due to insu cient promotion of annual endoscopy screening, a certain portion of patients in China present with advanced disease at diagnosis and are not suitable for esophagectomy. For these patients, de nitive concurrent chemoradiotherapy (CCRT) has become the standard of care as recommended by most guidelines since the RTOG 85 − 01 study [4][5][6]. However, in clinical practice, a considerable number of patients could not tolerate concurrent chemoradiation for various reasons (i.e., advanced age, poor general condition, poor nutritional status, obvious weight loss, patient refusion, etc.). For this group of patients, the most commonly used treatment is radiation monotherapy.
In the RTOG85-01 study, the 5-year survival rate of patients in the radiotherapy-only group was just 0, which indicates that radiation monotherapy is very ineffective in treating locally advanced esophageal cancer. According to the clinical experience and some contemporary large retrospective analysis results [7][8][9], however, the survival outcome of patients receiving radiotherapy alone is much better than that of RTOG 85 − 01. Besides, as a research started in the 1980s, RTOG 85 − 01 employed the conventional two-dimensional radiotherapy (2DRT), whereas radiation techniques have been evolving rapidly in last few decades. The emergence of three-dimensional conformal radiotherapy (3DCRT) and intensity-modulated radiotherapy (IMRT) has greatly improved the conformality of dose distribution, making it possible to increase the dose of the planning target volume without increasing the irradiation dose to adjacent normal tissues [10]. Many studies have shown that advanced radiation techniques have reduced the incidence of side effects and improved the local control of esophageal cancer patients [11]. Unfortunately, there are very limited reports on the survival and prognosis of patients receiving radiation monotherapy using the aforementioned advanced radiation technique (i.e., IMRT), let alone studies comparing the survival results of patients receiving radiotherapy alone with that of patients receiving CCRT.
Therefore, we conducted this retrospective study to collect data of patients receiving IMRT alone and compare with the survival of patients receiving CCRT in the same period, in the hope of providing a baseline reference of treatment standard and prognosis for patients with advanced esophageal cancer.

Treatment
All patients received computed tomography (CT) simulation and IMRT. The gross tumor volume (GTV-T) was de ned as the primary tumor. The GTV-T would be determined using all available resources (physical examination, upper gastrointestinal contrast, endoscopy, EUS, contrast CT-thorax/abdomen, contrast MRIthorax/abdomen, PET-CT, etc.). The gross lymph nodes volume (GTV-N) were de ned as any lymph node diagnosed as or highly-suspected as metastatic. The clinical target volume (CTV) comprised of a 0.5-0.8 cm lateral margin and a 3.0 cm longitudinal margin around GTV, a 0.5 cm uniform margin around GTVnd, and relevant lymphatic drainage areas. Planning target volume (PTV) was derived from CTV plus a uniform 0.5 cm margin. The boost volume (PGTV) was created by expanding GTV-T by 1.0 cm craniocaudally and 0.5 cm radially, and the GTV-N by a uniform 0.5 cm margin.
The median dose of CCRT and RT alone group was 59.92 Gy (range: 40.0-70.0 Gy) and 60.0 Gy (range: 40.0-70.0 Gy), respectively. The prescription to PTV varies from 1.8 Gy/fraction to 2.2 Gy/fraction whereas the prescription to PGTV varies from 2.0 Gy/fraction to 2.4 Gy/fraction (1 fraction per day, 5 fractions per week). As the fractionation schemes have slight variations between patients, we converted the radiation dose of each patient into equivalent dose in 2.0 Gy/fration (EQD2) in the nal analysis.
All patients received CT simulation, and their image data were registered in the treatment planning system (Pinnacle; Philips Medical Systems, Hanover, MA, USA).
It is required that at least 95% of the PTV received the prescribed dose. Dose volume histograms (DVH) was adopted to assess the quality of the treatment plan and the exposure of organs at risk (OAR). The volume of lung tissue receiving 20 Gy or more should not exceed 28% of the total lung volume (V20 lung < 28%). The mean dose of lung tissue should be lower than 16 Gy (Dmean lung < 16Gy). Other dose constraints to OARs include: V40 heart < 30%, V30 heart < 40%, V40 stomach < 40%, Dmax stomach < 55-60Gy, V40 small intestine < 40%, Dmax small intestine < 55Gy, V30 liver < 30%, V20 kidney < 30% and Dmax spinal cord PRV < 45 Gy. Imageguided radiotherapy (IGRT) was applied to all patients either by electronic portal imaging device (EPID) or cone beam computed tomography (CBCT).
The most common reasons for not receiving concurrent chemotherapy include: 1) obvious weight loss before treatment, weak general condition, poor performance status; 2) elderly patients; 3) serious cardiopulmonary complications or insu cient hepatic or renal functions; 4) patients' or relatives' refusal.

Outcome measures
Acute and late toxicities were scored according to the Common Toxicity Criteria for Adverse Events, version 4.0 [12]. In brief, patients were assessed every 3 months for the rst 2 years after RT, every 6 months for the next 3 years, and then once annually. Assessments included barium esophagram; CT of neck, chest and upper abdomen with contrast; ultrasonography of the neck and upper abdomen; and conventional blood and biochemistry studies. Positron emission tomography-CT (PET-CT) and ne-needle aspiration cytology were performed if needed. Bone scan was performed in case of bone pain or abnormally elevated serum alkaline phosphatase (ALP), and cranial MRI was performed if clinically indicated.
Overall Survival (OS) was measured as the interval between the beginning of RT to the date of death from any cause or nal follow-up. Progression-free survival (PFS) was de ned as the interval between the end of RT and the date of rst recurrence or death from any cause. Patients who had not experienced progression or death by the last follow-up were administratively censored.

Statistical Analysis
Descriptive statistics, including frequencies and percentages for categorical variables, and mean and standard deviation for quantitative variables, were computed to summarize patient characteristics for the entire cohort and for each treatment group. Between-group comparisons to evaluate imbalances in covariates were conducted using t-tests and Chi-square test for quantitative and categorical variables respectively. The incidence of toxicities was compared by Chi-square test and Fisher's exact test.
OS time and PFS time were estimated for RT alone and CCRT group using Kaplan-Meier (KM) plots both before and after PSM. Log-rank test was carried out to compare event time distributions between two groups. A Cox regression model was used to perform multivariate analyses of the effect of covariates on OS and PFS after PSM. The results of the Cox models were expressed as hazard ratios (HRs) along with the 95% CI estimates. Multivariable logistic regression models were used to identify factors associated with undergoing concurrent chemotherapy. The results of these logistic regression analyses were expressed as adjusted odds ratios (ORs) along with the corresponding 95% con dence intervals (95% CIs).
To adjust unbalanced covariates, PSM [13] method was used. The propensity score for each patient was estimated with a logit model that included the following variables: age, sex, KPS, TNM stage, tumor location, radiation dose and the year of diagnosis. Exact matching method was performed on TNM stage and sex whereas nearest matching method was adopted for the remaining variables. Setting caliper = 0.10, matching ratio = 1:1, two comparable groups of patients were created with 294 patients in each group. The signi cance level was set as p-value less than 0.05. All computations were conducted in R 2.13.0.

Patient and treatment characteristics
A total of 1273 patients were involved in this study, including 590 patients in the CCRT group and 683 patients in the RT alone group. The patient, tumor, and treatment characteristics were summarized in Table 1. The propensity score-matched cohort included 294 patients in the CCRT group and 294 patients in the RT alone group, and all the selected covariates were well-balanced between the two matched groups (see Table 1).
The addition of concurrent chemotherapy has further improved both OS and PFS. Figure 2 presents the survival curves comparing RT alone with CCRT both before and after PSM. Compared with the RT alone group, the CCRT group had better OS and PFS rates whether before or after PSM (before-match OS: log-rank p < 0.001 [ Fig. 2A]; after-match OS: log-rank p = 0.003 [ Fig. 2B]; before-match PFS: log-rank p = 0.003 [ Fig. 2C]; aftermatch PFS: log-rank p = 0.02 [ Fig. 2D

Subgroup Analysis
To further explore the potential bene ciaries of concurrent chemotherapy, we conducted several subgroup analyses of OS in the PSM sample. Results showed that male (p = 0.004), age < 70 years (p = 0.001), stage III-IV (p = 0.004), EQD2 ≥ 60 Gy (p = 0.004), and patients diagnosed between 2011-2017 (p = 0.006) could bene t from concurrent chemoradiotherapy. The subgroup analysis of age and radiation dose is shown in Fig. 3. Among the 588 paired patients, the vast majority were non-elderly patients (n = 486), and most patients received high-dose radiotherapy (n = 433). Among 102 elderly patients (aged over 70 years) in the matched sample, CCRT did not show survival bene t over RT alone (p = 0.8). In non-elderly patients, the OS rate of the CCRT group was better than that of the RT alone group (p < 0.001). Among 155 patients receiving low-dose radiotherapy (EQD2 < 60 Gy) in the matched sample, CCRT did not show survival bene t over RT alone (p = 0.12). In high-dose group, the OS rate of the CCRT group was better than that of the RT alone group (p = 0.004).

Multivariate Analyses in after-PSM Cohort
The multivariate Cox regression analysis for OS and PFS after PSM is summarized in

Toxicities
We further compared the pro le of treatment-related toxicities between RT alone and CCRT group in the matched cohort. As shown in Table 3, the incidences of ≥ grade 2 anemia, thrombocytopenia, ≥ grade 2/3 leukopenia and radiation esophagitis in the CCRT group were signi cantly higher than those in the RT alone group (all of which p < 0.05). The most common Grade 3-4 toxicities were leucopenia (13.7%) and radiation esophagitis (11.3%) in the CCRT group, while the most common Grade 3-4 toxicity in the RT alone group was radiation esophagitis (5.0%). There were six and three cases of treatment-related deaths in the CCRT and RT alone group respectively, all of which were attributed to Grade 5 radiation pneumonitis. No signi cant difference was seen in the incidences of radiation pneumonitis between the two groups (p = 0.36). Factors found to be signi cantly associated with undergoing concurrent chemotherapy were assessed using a multivariable logistic regression analysis that included all previously mentioned covariates as shown in

Discussion
Our research con rms the e cacy of concurrent chemotherapy in treating patients with locally advanced ESCC by combining with IMRT. According to the matched results, the 5-year OS of the CCRT group was as high as 38.1%, and the concurrent chemotherapy increased the 1-year, 3-year, and 5-year OS rates of RT alone group by 8.9%, 10.6%, and 9.6%, and the 1-year, 3-year, and 5-year PFS rates by 4.3%, 7.8%, and 9.4%, respectively. Although the incidence of toxicities also increased accordingly, the general incidence of Grade 3-4 toxicities in the CCRT group was still within the acceptable range (the most common Grade 3-4 side effects is leukopenia, the incidence of which is 13.7%). Besides, CCRT did not increase the incidence of radiation-related pneumonitis or treatment-related mortality, indicating that for patients with locally advanced ESCC, de nitive IMRT concurrent with chemotherapy are safe and tolerable.
In the subgroup analysis, for elderly patients (aged 70 years or older), CCRT did not show a signi cant survival advantage over RT alone. This result is consistent with some recently published retrospective analyses of elderly patients [14]. This can be related to the poor tolerance of elderly patients for CCRT. One should note that there is still lack of consensus on the optimal treatment option for elderly patients with esophageal cancer, and corresponding prospective randomized studies are needed to further explore the role of concurrent chemotherapy in elderly patients, as well as the best concurrent chemotherapy regimen and other issues.
Although since the publication of RTOG 94 − 05 [15] study, 50.4 Gy became the recommended radiation dose for EC patients receiving non-surgical treatment, physicians of our center are still more accustomed to applying higher dose prescriptions. Besides, in our multivariate Cox analysis after PSM, higher radiation dose (EQD2 ≥ 60Gy) was found to be an independent protective prognostic factor for both OS and PFS. RTOG 94 − 05 is a research conducted in the era of 2DRT. In recent years, the emergence of more advanced radiation techniques (3DCRT, IMRT, etc.) has made it possible to increase the dose of target volumes while reducing the exposure of adjacent normal tissues [10,11]. Numerous recently-published retrospective analysis studies [16,17] and metaanalysis [18] also showed that patients with ESCC may bene t from high-dose radiotherapy. At present, a series of prospective clinical trials [19][20][21][22] applying simultaneous integrated boost-IMRT are in progress. Their results con rmed the safety and feasibility of high-dose IMRT concurrent with chemotherapy in patients with ECE, and satisfactory local control and OS has been achieved.
It is true that the proportion of patients receiving concurrent chemotherapy in this study is relatively low when compared with most contemporaneous large-scale real-world reports in developed countries [23]. Multivariate logistic regression analysis showed that the factors affecting the rate of concurrent chemotherapy were age, gender, TNM stage and year of diagnosis. In addition, we believe that this phenomenon is also related to the advanced TNM stage, large tumor burden, poor general condition, and poor economic status of patients treated in our center. However, our research also con rmed that relatively satisfactory survival rates could be achieved even with IMRT monotherapy for patients with locally advanced ESCC. In the pre-matched original sample, a total of 590 patients received IMRT monotherapy as the radical treatment, of which over 80% had stage III-IV disease. For patients with such advanced disease at initial diagnosis, RT alone still resulted in 1-year, 3-year and 5-year OS rates of 65.3%, 32.6%, and 24.8%, indicating its e cacy in treating patients with advanced ESCC. Our results are consistent with other recently published researches [24,25], and show a much better survival than that with patients treated by 2DRT [7]. All these results suggest that for patients with locally advanced ESCC who cannot receive CCRT, radiation monotherapy can be an effective and safe alternative with promising survival results.