Definitive radiotherapy dose escalation with chemotherapy for treating non-metastatic oesophageal cancer

The locoregional failure rate remains high after concurrent chemoradiotherapy with standard-dose radiotherapy (RT, 50–50.4 Gy) for oesophageal cancer (EC). This retrospective study evaluated whether RT dose escalation was effective among 115 consecutive patients with non-metastatic EC (July 2003 to November 2016). Forty-four patients received an RT dose of <66 Gy and 71 patients received ≥66 Gy, with most patients receiving concurrent cisplatin plus fluorouracil. The median follow-up was 12 months for all patients (52 months for 18 surviving patients). The ≥66 Gy group had significantly higher 3-year rates of overall survival (17.9% vs. 32.1%, p = 0.026) and local progression-free survival (46.1% vs. 72.1%, p = 0.005), but not disease progression-free survival (11.4% vs. 21.9%, p = 0.059) and distant metastasis-free survival (49% vs. 52.6%, p = 0.852). The ≥66 Gy group also had significantly better 5-year overall survival compared with 41.4–65.9 Gy. The only significant difference in treatment-related toxicities involved acute dermatitis (7% vs. 28%, p = 0.009). Inferior overall survival was associated with poor performance status, clinical N2–3 stage and not receiving maintenance chemotherapy. In conclusion, patients with inoperable EC experienced better survival outcomes and acceptable toxicities if they received higher dose RT (≥66 Gy) rather than lower dose RT (<66 Gy).

The median ages were 62 years (range: 32-85 years) for the <66 Gy group and 60 years (range: 39-86 years) for the ≥66 Gy group (p = 0.523). The ≥66 Gy group had a significantly higher proportion of patients with a good performance status (Eastern Cooperative Oncology Group performance status [ECOG PS] of 0) (p = 0.027). There were no significant differences between the <66 Gy and ≥66 Gy groups in terms of their sex, tumour histology, grade, location, length, clinical T stage, clinical N stage, reasons for no surgery, chemotherapy regimens, maintenance chemotherapy status, and RT technique.
The median follow-up durations were 12 months (range: 2-103 months) for all patients and 52 months (range: 12-100 months) for the 18 surviving patients. The 3-year and 5-year overall survival (OS) rates for all patients were 26.6% and 18.4%, respectively. The 3-year rates of disease progression-free survival (DPFS), local progression-free survival (LPFS), and distant metastasis-free survival (DMPFS) were 17.8%, 62.8%, and 51.5%, respectively. Table 2 shows the results of the univariate analyses for OS, DPFS, LPFS, and DMPFS. The comparisons of the <66 Gy and ≥66 Gy groups revealed significant differences in the 3-year rates of OS (117.9% vs. 32.1%, p = 0.026) and LPFS (46.1% vs. 72.1%, p = 0.005), but not in DPFS (11.4% vs. 21.9%, p = 0.059) and DMPFS (49% vs. 52.6%, p = 0.852). The median OS values in the ≥66 Gy and <66 Gy groups were 13 months and 10.2 months, respectively, which corresponded to 5-year OS rates of 24.1% and 10.2%, respectively. The survival curves of each group are plotted in Fig. 1. In the univariate analyses, poor OS was associated with older age (≥60 years), ECOG PS of 1-2, tumour length of >5 cm, clinical N2-3 stage, and not receiving maintenance chemotherapy. The subgroup analysis excluded 3 patients who received an RT dose of <41.4 Gy, but still revealed a significantly higher 3-year OS rate in the ≥66 Gy group than in the 41.4-65.9 Gy group (32.1% vs. 17.1%, p = 0.027) ( Fig. 2A). We also calculated the OS from the latest date of RT to death or the last follow-up, which still revealed a significantly better 3-year OS rate in the ≥66 Gy group than in the 41.4-65.9 Gy group (32.2% vs. 14.6%, p = 0.037) (Fig. 2B). We further excluded 13 patients with RT dose <50 Gy. The 3-year OS rate was significantly higher in the ≥66 Gy group relative to the 50-65.9 Gy group, with rates of 32.1% and 16.1%, respectively (p = 0.035) (Fig. 3).

Discussion
The present study investigated the relationship between RT dose and survival among patients with non-metastatic EC who underwent definitive RT with chemotherapy. The results indicate that a total RT dose of ≥66 Gy at the primary tumour was associated with significantly better OS and LPFS than a dose of <66 Gy, and this difference persisted after excluding patients who received an inadequate RT dose (<41.4 Gy or <50 Gy). It is possible that the results were related to the immortal time bias 12 , as we initially evaluated survival from the start date of RT for both groups. However, we also evaluated the survival outcomes from the last date of RT, which eliminated the immortal time bias and revealed that the significant difference in OS persisted between the two groups (Fig. 2B).
Although various studies have evaluated RT dose escalation for treating EC, the potential clinical benefits remain controversial 5,6,13-17 . Brower et al. used the National Cancer Data Base in the US to evaluate RT dose escalation for patients with stage I-III EC who underwent CCRT 13 , and reported that higher RT doses (51-54, 55-60, or >60 Gy) did not improve OS relative to a dose of 50-50.4 Gy. However, their cohort only included a limited number of patients who underwent IMRT or 3DCRT (39%), and the predominance of cases with an unknown RT modality (61%) might have influenced the apparent effect of RT dose on survival in the modern era of RT. Recently, Xu et al. reported an abstract of a randomized study to evaluate the clinical benefit of high-dose RT using modern techniques 14 . They found no significant differences in local/regional progression-free survival, DPFS, OS and toxicity between the high-dose (60 Gy) and low-dose (50 Gy) group. Chen et al. performed a population-based propensity score-matched study using Taiwanese registry data from patients with EC who underwent IMRT or 3DCRT 15 , and reported that an RT dose of ≥60 Gy may provide better 5-year OS than 50-50.4 Gy (22% vs. 14%, p < 0.05). He et al. retrospectively reviewed 193 patients with EC who underwent CCRT 16 , and reported that the high-dose group (>50.4 Gy) had a significantly lower local failure rate (17.9% vs. 34.3%, p = 0.024) than the low-dose group (≤50.4 Gy), but there was no significant difference in the 5-year OS rates (41.7% vs. 33.0%, p = 0.617). Cao et al. evaluated outcomes among 115 patients with squamous cell carcinoma of the cervical oesophagus who underwent definitive RT with or without chemotherapy 17   The present study revealed that many radiation oncologists in our department did not follow the NCCN guidelines' recommendations regarding the RT dose for treating EC, as most patients (96/ 18 , and reported that an RT dose of >60 Gy was associated with better pathological complete remission than lower doses (59.1% vs. 36.4%). Moreover, Welsh et al. reported that unresectable EC with gross tumour volume (GTV) failure after CCRT was associated with shorter survival than cases without GTV failure, which suggested that local control helped improve survival 3 . Therefore, we assume that higher RT doses in the present study provided improved local control that translated into improvements in OS.
Regarding treatment-related toxicity, one systematic review has indicated that there were no differences in grade ≥3 acute or late esophagitis between high-dose RT (≥60 Gy) and standard-dose RT 5 . Furthermore, other toxicities are rare and moderately tolerable. Two retrospective studies have also evaluated treatment-related toxicities, with Kim et al. reporting no significant differences between the <60 Gy and ≥60 Gy groups 6 . However, relative to a low-dose group (≤50.4 Gy), He et al. reported that higher doses (>50.4 Gy) were associated with higher rates of grade 3 skin reactions (12.5% vs. 2.2%, p < 0.001) and oesophageal stricture (32.1% vs. 18.2%, p = 0.037) 16 . In the present study, the only dose-related difference in treatment-related toxicity was observed for acute dermatitis (Table 4), although only 1 patient in the ≥66 Gy group experienced grade 3 acute dermatitis.
The present study has several limitations. The first is the retrospective design and the variable chemotherapy regimens. The second is the different patient characteristics for each group, with the high-dose group having more favourable performance status, which might have influenced the results, although a multivariate Cox proportional hazard model was used to adjust for potential confounding factors. Some may argue that patients in the low-dose group were treated with more palliative intent. However, we analysed the chemotherapy regimens and maintenance chemotherapy status between the <66 Gy and ≥66 Gy groups. No significant differences were found between the 2 groups in terms of chemotherapy regimens and maintenance chemotherapy status, which means that the patients in the low-dose group were not treated with more palliative intent. The third is the regional differences in the histological type of EC, with our cases predominantly involving squamous cell carcinoma (95%).  Similar results were reported in the 2015 report by the Taiwan Department of Health 19 , although squamous cell carcinoma is declining in North America and Europe, with concurrent increases in adenocarcinoma of the distal oesophagus and gastroesophageal junction (approximately 70% of oesophageal carcinomas in the US) 20 . Thus, the results of the present study might not be generalizable to regions where the dominant tumour histology is adenocarcinoma. The fourth is the hidden bias which is a common issue for observational study 21 . For example, body weight loss, cachexia, or nutrition status was not controlled in the current study 22 .
In conclusion, the present study's results suggest that higher RT doses (≥66 Gy), when administered using modern RT technique (3DCRT, IMRT, or VMAT), are feasible for patients with inoperable and non-metastatic EC. In addition, the higher RT doses were associated with significantly better LPFS and OS than for RT doses of <66 Gy. The two groups had similar toxicity incidences and severities, with the only significant differences observed for acute dermatitis, which is a manageable event. Furthermore, inferior OS was independently predicted by poor performance status, clinical N2-3 stage, and not receiving maintenance chemotherapy. Nevertheless, these findings require validation in prospective randomized trials using modern RT techniques.

Methods
Data source. We retrospectively evaluated 115 patients with histologically-confirmed EC who underwent definitive RT with chemotherapy at our department between July 2003 and November 2016. All patients had initially undergone a physical examination, chest radiography, barium swallow, chest computed tomography (CT), abdominal sonography, and upper gastrointestinal panendoscopy and/or fluorodeoxyglucose positron emission tomography (PET-CT). Patients were excluded if they had undergone primary surgery or had distant metastasis. Data were collected regarding demographic characteristics, performance status, pathology, imaging results, disease stage, reason for not undergoing surgery, RT, chemotherapy, and follow-up results. Disease staging was based on the 7 th version of the American Joint Committee on Cancer guidelines (7 th AJCC, 2009). The patients who were diagnosed within 2003-2009 were retrospectively staged in accordance with the 7 th AJCC guidelines by reviewing their medical records and images. The retrospective study protocol was approved by the institutional review board of the Tri-Service General Hospital in Taiwan (1-101-05-041). All methods were performed in accordance with the relevant guidelines and regulations.  GTV contained the primary tumour and metastatic lymph nodes, and was determined using CT, barium swallow, and endoscopy with or without PET-CT. The clinical target volume (CTV) was obtained by expanding the GTV by 3-5 cm at the cranial and caudal margins and by 0.3-2 cm at the transversal margins. The planning target volume included the CTV in the initial phase or the GTV in the boost phase, with additional margins of 0.3-1 cm in all directions to account for organ movement. All patients had undergone VMAT, IMRT, or 3DCRT, with sequential boost or simultaneous integrated boost approaches being used in our department to treat EC. During the sequential boost approach, the irradiated field encompassed the CTV and the regional lymph node drainage area, with a dose of 41.4-50.4 Gy in 1.8-2-Gy fractions, which was followed by a total RT dose of 50.4-70 Gy to the GTV. During the simultaneous integrated boost approach, VMAT or IMRT was used to simultaneously deliver a high dose (50.4-70 Gy) to the GTV and a lower dose (45-60 Gy) to the CTV and the regional lymph node drainage area. The irradiated tumour length was identified using the CT, barium swallow scans, endoscopy and/or PET-CT. The total RT dose included the doses to the primary tumour from the external beam RT, and patients were grouped according to GTV RT doses of <66 Gy (median: 54 Gy, range: 14.4-65 Gy) or ≥66 Gy (median: 66 Gy, range: 66-70 Gy).
Chemotherapy. The chemotherapy regimens were selected based on the patient's age, general physical condition, and the oncologist's preference. Most patients (n = 95) received cisplatin 60-100 mg/m 2 on day 1 plus 5-fluorouracil (5-FU) 500-1000 mg/m 2 daily on days 1-4 through continuous intravenous infusion every 3-4 weeks. Two cycles of cisplatin plus 5-FU were administered concurrently during RT. The other regimens during RT were cisplatin monotherapy (n = 14) with 30-40 mg/m 2 weekly or 60-100 mg/m 2 every 3-4 weeks, and 5-FU monotherapy (n = 6) with 500-1000 mg/m 2 daily on days 1-4 every 3-4 weeks. Patients received additional maintenance chemotherapy if a medical oncologist determined that their medical condition and disease status would allow this.
Follow-up and toxicity assessment. Toxicity assessments were performed weekly during the RT. The first follow-up evaluation was performed at 1 month after RT and follow-ups were then continued at 3-6-month intervals. During the follow-up, the patients underwent physical examinations, blood and biochemical testing, chest CT, barium scans, and abdominal sonography. Upper gastrointestinal panendoscopy with biopsy or PET-CT were performed if clinically indicated. Toxicities were scored using version 4.0 of the Common Terminology Criteria for Adverse Events, and toxicities were considered acute if they developed within 1 month after completing RT.
Statistical analysis. The OS interval was calculated from the start of RT to death or the last follow-up. The DPFS interval was calculated from the start of RT to the first instance of recurrence, death, or the last follow-up. The LPFS and DMPFS intervals were calculated from the start of RT to the last follow-up or the first instance of local or distant recurrence, respectively. Categorical variables were analysed using the chi-square test, and survival analyses were performed using the Kaplan-Meier method. Univariate and multivariate analyses were performed using the log-rank test and a Cox proportional hazard regression model, respectively. A Cox proportional hazards model was used to determine the HR and 95% CI values. Differences were considered statistically significant at a p-value of <0.05. All analyses were performed using SPSS software (version 18.0; SPSS, Chicago, IL).
Data availability. The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.