Supraclavicular and/or celiac lymph node metastases from thoracic esophageal squamous cell carcinoma did not compromise survival following neoadjuvant chemoradiotherapy and surgery

This study is to evaluate the prognostic significance of supraclavicular and/or celiac lymph node (LN) metastases in locally advanced thoracic esophageal squamous cell carcinoma (ESCC) patients treated with neoadjuvant chemoradiotherapy (NACRT) and surgery. Among the total 199 patients, 75 (37.7%) had supraclavicular and/or celiac LN metastasis. Surgery was performed following NACRT in 168 patients (84.4%). After the median 18.7 (1.0-147.2) months’ follow-up, 2-year rates of progression-free survival (PFS) and overall survival (OS) in all patients were 48.1% and 65.7%, respectively. In multivariate analyses, negative surgical margin (p < 0.001), ypT0 stage (p = 0.004), and ypN0 stage (p = 0.020) were significantly favorable factors for PFS, and negative surgical margin (p < 0.001) was the only significantly favorable factor for OS. Metastasis to the supraclavicular and/or celiac LNs was significant factor neither for PFS (p = 0.311) nor OS (p = 0.515). Supraclavicular and/or celiac LN metastasis did not compromise the clinical outcomes following NACRT and surgery.


INTRODUCTION
The prognosis of resectable esophageal cancer remains still unsatisfactory with 5-year overall survival (OS) rates of 15%~34% [1,2]. A multicenter randomized trial demonstrated that the addition of neoadjuvant chemoradiotherapy (NACRT) and surgery improved OS at 5 years when compared to surgical resection alone in patients with resectable stage (T2-3N0-1M0) (47% vs. 34%, p = 0.003) [3]. Several meta-analyses also concluded that NACRT and surgery showed survival benefit when compared to surgery with or without neoadjuvant chemotherapy in locally advanced esophageal cancer [4,5]. Thus, NACRT and surgery has been now recommended as the standard treatment in resectable locally advanced esophageal cancer.
Esophageal cancer can spread bidirectionally through the lymphatics to reach remote locations ranging from the cervical to abdominal lymph nodes (LNs). The 6 th edition of American Joint Committee on Cancer (AJCC) staging system [6] defined the regional and non-regional LNs based on the primary tumor location: for thoracic esophageal cancer, upper periesophageal, subcarinal, and lower periesophageal LNs were regional LNs; while cervical and celiac LNs were non-regional LNs. Cervical LN metastasis from upper thoracic esophageal cancer and celiac LN metastasis from lower thoracic esophageal cancer were classified as M1a stage considering the distance between the primary tumor and metastatic LNs, however, metastases to more distantly located LNs were Clinical Research Paper classified as M1b together with distant organ metastasis. The 7 th edition of AJCC staging system re-defined the regional LNs as any periesophageal LN from cervical to celiac LN and removed M1a and M1b classifications, but incorporated the numbers of metastatic LN instead [7].
The previous clinical trials, which established NACRT and surgery as the current standard for locally advanced esophageal cancer, mostly included the patients with cM0, but not cM1a/b, stage according to the 6 th edition of AJCC staging system. The current study aimed to evaluate the prognostic influence of cM1a/b by the 6 th AJCC system in patients with locally advanced thoracic esophageal squamous cell carcinoma following NACRT and surgery.

Patient characteristics
The patients' characteristics are summarized in

Treatment compliance
Compliance to NACRT was very high: 196 patients (98.5%) could complete the planned radiotherapy (RT)

Significance of supraclavicular and/or celiac LN metastasis
Among all, the patients having metastasis to the supraclavicular and/or celiac LNs (cM1a/b group) showed numerically lower, but not significant, PFS (40.1% vs. 53.5%, p = 0.204) and OS (62.5% vs. 68.4%, p = 0.362) at 2 years, when compared to those having regional LN metastasis (cM0 group). Among 168 patients who underwent surgery, pCR rate was numerically higher in

DISCUSSION
Supraclavicular and celiac LNs were defined as non-regional LNs of thoracic esophageal cancer in the 6 th AJCC staging system. However, a number of evidence had suggested that supraclavicular and/or celiac LN metastasis does not compromise prognosis compared to other regional LN metastasis [8][9][10][11]. Thus, in the 7 th AJCC staging system, discrimination of cervical and celiac LNs as non-regional LNs was abandoned. However, the prognostic impact of supraclavicular and/or celiac LN metastasis in the setting of NACRT and surgery has not been properly addressed yet, since most previous trials, including the CROSS trial, did not include the patients with cM1a/b stage by the 6 th AJCC staging system [3,12,13]. The current study revealed that the supraclavicular and/or celiac LN metastasis did not compromise survival outcomes in the patients who tolerated tri-modality treatment of NACRT and surgery.
Several prognostic factors have been suggested in esophageal cancer following NACRT and surgery such as tumor location, cT stage, surgical margin status, and pCR rate [14][15][16][17]. The present study also confirmed that negative resection margin was a significantly favorable prognostic factor for both PFS and OS in multivariate analysis (p < 0.001). pCR rate following NACRT is known as 25%~30% [3,15,16] though its prognostic significance is still controversial [18,19]. In the current study, pCR was achieved in 26.2% and it was not a significant factor on OS at 2 year (79.8% vs. 64.3%, p = 0.118). This might be partly explained by the fact that OS at 2 years of patients with pT0N+ and pT+N0 (76.3% and 66.3%) were not different from that of patients achieving pCR (79.8%) ( Figure 2).
Recently, the prognostic implication of FDG-PET parameters following NACRT was raised. Our institution previously reported the impact of FDG-PET parameter following NACRT in esophageal cancer patients [20]. In the present study, we could not incorporate the FDG-PET parameters into the prognostic factor analysis because the raw data of FDG-PET parameters were not available in 20% of the patients, whose FDG-PET CT was done at other hospitals. However, based on the available data in the previous study, SUVmax value reduction rates of 72.1% in main tumor and 50.7% in LNs estimated pCR with the best accuracy while total lesion glycolysis did not.
The clinical outcomes by the current study were favorably comparable with those by previous randomized studies, even though 37.7% of patients in the current study had M1a/b disease. The previous randomized studies reported OS rates of 60%~67 % at 2 years and 39%~47% at 5 years following NACRT and surgery [3,12,13]. In this study, the 2-and 5-year OS rates were 69.3% and 53.9%, respectively. Furthermore, the rates of R0 resection and postoperative mortality in hospital or within 30 days were 91.1% and 6.5%, respectively, which were also comparable with the previous studies (92.0% and 5.9% in the CROSS trial and 93.8% and 11.1% in the FFCD 9901 trial, respectively) [3,13].
This study has a few limitations. First, this study included rather a limited number of patients to reach concrete conclusions. However, considering that we included only the patients with squamous cell carcinoma of the thoracic esophagus, who were treated in a highly homogenous manner, 199 patients is not a small number. Squamous cell carcinoma is the most prevalent histologic type in Asian countries such as China, Japan, and Korea. In fact, adenocarcinoma composed of only 2.5% of the patients who received NACRT for esophageal cancer in our institution. Second, a selection bias might have existed as NACRT and surgery was mainly recommended to those who had favorable performance status and were expected to tolerate large RT volume. In fact, chemotherapy alone before surgery was mainly recommended to the patients who had marginal performance status. Thus, our results might not be applicable to all patients having supraclavicular and/or celiac LN metastasis.
In summary, celiac and/or supraclavicular LN metastasis did not compromise the survival in patients with locally advanced esophageal squamous cell carcinoma who received NACRT and surgery. Further larger prospective studies should be warranted to establish the optimal treatment of thoracic esophageal cancer with celiac and/or cervical LN.

Patients
NACRT and surgery was recommended as the definitive therapy to 201 patients for having locally advanced thoracic esophageal squamous cell carcinoma at the authors' institute from January 2003 till July 2014. Following approval by our institutional review board, we retrospectively reviewed the medical records. After excluding two patients, who abandoned their treatment during NACRT course and were lost to our follow-up, analyses were done on 199. In all patients, endoscopy with biopsy, chest computed tomography (CT), and whole body 18F-fluorodeoxyglucose positron emission tomography-computed tomography ( 18 F-FDG PET-CT) were performed for diagnosis and staging purpose. Endoscopic ultrasonography (EUS) was performed in 116 patients (58.3%). Primary tumor location was allocated www.impactjournals.com/oncotarget according to the uppermost involvement of the primary tumor according to the AJCC 7 th edition.

Treatment scheme
RT was to deliver 45 Gy in 25 fractions by 1.8 Gy per fraction over 5 weeks until March of 2009 in 54 patients (27.1%) and 44 Gy in 22 fractions by 2.0 Gy per fraction thereafter in 145 (72.9%). All patients received 3-dimensional conformal RT (3D-CRT), typically through 3 or 4 coplanar fields using 4-, 6-, or 10-MV photons from linear accelerator. Primary tumor and metastatic LNs were delineated as the gross tumor volume (GTV) based on CT, PET-CT, and endoscopic findings. The clinical target volume (CTV) of primary tumor included primary GTV plus 2~3 cm margins in the craniocaudal directions and 0.5 cm margin in the circumferential direction. The nodal CTV was delineated by putting 1 cm margin in all directions from the nodal GTV. Elective irradiation to the supraclavicular and/or celiac LNs was not employed in the current study. The planning target volume (PTV) was defined as 0.5~0.7 cm margin in all directions from the CTV to account for the respiratory motion and daily setup errors.
Two cycles of intravenous chemotherapy (5-flurouracil (5-FU) 1,000 mg/m 2 /day for 4 consecutive days plus cisplatin 60 mg/m 2 /day on the 1 st day) were administered at 3 weeks' interval, and the 1 st cycle was to planned on the 1 st day of RT. Tumor response and resectability was re-evaluated before surgical resection by chest CT and FDG PET-CT in 3~4 weeks of NACRT completion.
Esophageal resection through thoracotomy was combined with either two-field or three-field lymphadenectomy. Two-field (thoracic and abdominal) LN dissection was performed mainly in the patients with lower thoracic esophageal cancer while three-field (cervical, thoracic, and abdominal) LN dissection was done mainly in those with upper or mid-thoracic esophageal cancer. Following NACRT, adjuvant chemotherapy or thoracic RT were optionally recommended to the patients with good performance status and risk factors based on surgical and pathological findings: 3 cycles of 5-FU and cisplatin chemotherapy was delivered in 36 patients (21.4%); 20 Gy in 10 fractions thoracic RT was in 13 (7.7%), respectively. Usual indications for adjuvant therapy included multiple pathologic lymph node metastases, close or positive resection margin, and/or difficulty of dissection experienced at time of surgery due to bulky primary tumor and/or metastatic LN.

Statistical analyses
The durations of PFS and OS were calculated from the first date of NACRT till the date of any event of recurrence and death or the last follow-up, respectively. The distribution of categorical variables was analyzed by the chi-square test. The survival rates were calculated by the Kaplan-Meier method and were compared by logrank test for univariate analysis. Multivariate analysis was described with HR and their 95% CI, derived from the cox proportional hazards model. Two-sided p-values of 0.05 or less were considered statistically significant. SPSS Statistics version 20 (SPSS Inc., an IBM Company, Chicago, IL) was used for the analysis.