Tolerance and dose-volume relationship of intrathoracic stomach irradiation after esophagectomy for patients with thoracic esophageal squamous cell carcinoma

Purpose To identify the tolerance of radiation with a high prescribed dose and predictors for the development of intrathoracic stomach toxicity in patients with thoracic esophageal squamous cell carcinoma (SCC) after esophagectomy followed by gastric conduit reconstruction. Methods and Materials From 2011 to 2013, 105 patients after esophagectomy were treated with postoperative radiotherapy. The intrathoracic stomach was outlined with the calculation of a dose-volume histogram (DVH) for the initial intended treatment of 6020 cGy or 6300 cGy. The volume of the intrathoracic stomach receiving each dose was recorded at 10-Gy intervals between 10 and 40 Gy and at 5-Gy intervals between 40 and 60 Gy. The grade of toxicities was defined by the National Cancer Institute Common Toxicity Criteria version 4.0. Results The mean and maximum doses of the intrathoracic stomach were 2449 ± 986 cGy and 6519 ± 406 cGy, respectively. Sixteen (15.2%) and three (2.9%) experienced Common Toxicity Criteria Grade 2 and Grade 3 acute gastric toxicity. There were no Grade 4 toxicities. Fourteen patients (13.3%) exhibited late gastric complications possibly related to radiation. The volume percent of the intrathoracic stomach receiving at least 50 Gy (V50) was strongly associated with the degree of toxicity (p = 0.024, respectively). Multivariate analysis of patient and treatment-related factors revealed no other significant predictors of severe toxicities. Conclusions The intrathoracic stomach is well tolerated with a high-dose irradiation for patients with esophageal SCC receiving radiotherapy after esophagectomy. A strong dose-volume relationship exists for the development of Grade 2 acute intrathoracic stomach toxicity in our study.


INTRODUCTION
Existing evidences have indicated that overall survival (OS) could be improved using neoadjuvant chemoradiotherapy followed by surgery for esophageal cancer [1]. However, neoadjuvant therapy might increase the risk of postoperative morbidity or perioperative mortality [2], so a considerable number of patients with local advanced thoracic esophageal squamous cell carcinoma (SCC) in our country perform surgery as their initial treatment. However, sometimes it is difficult to achieve the purpose of complete resection because of surgeons' skills. Moreover, according to previous studies the recurrence rate of SCC is as high as 40%-50% after radical surgery, and locoregional recurrence accounts for more than half of treatment failures [3][4], even among patients with a pathologically complete response to neoadjuvant chemoradiotherapy [5]. Recurrences in supraclavicular and superior mediastinal areas were the most common failures [6][7]. It further necessitates the need for adjuvant therapy to decrease the likehood of local recurrence, especially for patients with positive lymph nodes [8][9]. Radiotherapy(RT) is also a crucial treatment for locoregional failures. Usually, the radiation dose is -as high as possible for patients with limited lesions.
Stomach is the first choice for esophageal replace ment following esophagectomy. Recently, gastric tube has gained wide acceptance for esophageal reconstruction, which significantly improve the quality of postoperative life. In our center most of patients received esophagectomy with gastric tube reconstruction and intrathoracic anastomosis via the retrosternal route. As a result, the intrathoracic stomach is often incidentally irradiated in postoperative thoracic radiotherapy. Therefore, when formulating radiation plans for patients who have undergone surgery, the intrathoracic stomach need to be protected to avoid severe complications such as marginal ulcers, bleeding, perforation and anastomotic fistula, as RT-induced injury could occur hours to weeks after the first treatment [10]. Gastroduodenal (GD) tolerance to RT has been investigated in abdominal malignancies [11][12], but the application of these results to esophageal SCC patients requires caution. The reason is there are very few published reports of acute and late RT effects on the intrathoracic stomach, and whether the effects would influence the functions of the gastric substitute is not well known.
Due to technical limitations of 2dimensional RT, the actual dose distribution of organs at risk could not be analyzed. Currently, developed three-dimensional conformal RT (3DCRT) has been succeed in providing the possibility of analyzing dose-toxicity relationship and decreasing certain toxicities in esophageal SCC patients.
In the present study, various clinical and dose-volume histogram (DVH) parameters were analyzed based on a widely used radiation system in order to identify a potentially safe dose tolerance of the intrathoracic stomach and risk factors for gastric toxicity by observing the radiation-induced adverse side effects in a group of patients from a randomized phase II clinical trial (http:// ClinicalTrials.gov website, number NCT01391572) who received thoracic RT after esophagectomy.

Patient population and treatment
Between May 2011 and December 2013, patients from 6 centers with esophageal SCC undergoing surgical resection followed by gastric tube reconstruction enrolled our clinical trial. The patients were included if their pathological stage was T 3-4 N 0-3 M 0 according to the AJCC/UICC TNM staging system (Version 7.0, 2009) and if they did not receive neoadjuvant therapy. Patients without postoperative RT or available DVH data were ineligible ( Figure 1).
All patients underwent esophagectomy through right thorax and abdominal incisions and done in esophageal reconstruction. The intrathoracic stomach was formed from the distal aspect of the lesser curvature of the stomach with application of linear staplers. It was created by resection of the lesser curvature and formation of the gastric conduit (4-6 cm in diameter). And then, the tubular stomach was pulled upward to the cervical or aortic arch part through the posterior mediastinal route and performed two-layer anastomosis. All included patients were never diagnosed with gastric ulcers, reflux esophagitis and other serious gastrointestinal diseases that would preclude safe administration of treatment. Components of the pre-radiation process included the following: a complete history and physical examination; complete blood cell counts; serum biochemical assays; barium esophagram to exclude evidence of gastric perforation, anastomotic fistula, or deep ulceration to the mediastinum; chest CT scans; and ultrasonographic examination to rule out distant metastases in the neck, liver, kidney, spleen, and retroperitoneal lymph nodes.
All patients underwent CT-based treatment simu lation while supine, and 5-mm-thick images were obtained throughout the entire neck, thorax, and upper abdomen. The clinical tumor volumes (CTVs) encompassed the tumor bed and/or the bilateral supraclavicular and upper mediastinal lymphatic drainage areas. Planning target volumes (PTVs) were defined as the CTV plus a uniform 1cm margin. According to the protocol for estimating the optimal radiation volume of postoperative radiation, all patients were assigned into either the largefield group (including tumor bed, bilateral supraclavicular and upper mediastinal lymphatic drainage areas) or the smallfield group (only the tumor bed area) by random number table. A simultaneous integrated-boost intensity-modulated radiotherapy (SIB-IMRT) technique was used, and treatment plans were generated by the Pinnacle treatment planning system (Philips Medical Systems). Radiation was delivered with 6MV photons by a linear accelerator. The prescribed doses of the tumor bed area were 60.2 Gy (in 28 fractions of 2.15 Gy/fraction) using a 6MV Xray for patients with T3stage disease or 63 Gy (in 28 fractions of 2.25 Gy/fraction) for patients with T4stage disease. The lymphatic drainage area in the largefield group was prescribed a dose of 50.4 Gy (Figure 2). The goals were to deliver the prescription dose to at least 95% of the PTV and 95% of the prescribed dose to at least 99% of the PTV. The normal tissue constraints met the following criteria: (1) maximum spinal cord dose ≤ 45 Gy; (2) lung V20 ≤ 25% and mean lung dose (MLD) ≤ 15 Gy; and (3) mean heart dose ≤ 30 Gy. The intrathoracic stomach was not designated as a constrained structure for the original treatment plans. All patients were required to fast during simulation and irradiation.

Follow-up and diagnosis of gastric toxicity
Followup occurred 4-6 weeks after treatment completion and every 3 months thereafter. Complete history and physical examinations, including an evaluation of digestive symptoms, were performed at each visit. Re-examinations included endoscopy, cervical ultrasounds, chest-enhanced CT scans, abdominal ultrasound screening, and, when necessary, bone emission computed tomography (ECT) and positron emission tomography (PET)/CT. Gastric toxicity was defined as the new development of or the aggravation of endoscopic abnormalities, such as erosive gastritis or a gastric ulcer in close proximity to the RT field following RT. Assessment of digestive symptoms was adapted from the Common Toxicity Criteria for Adverse Events, version 4.0.

Dosimetric evaluation of the intrathoracic stomach
Intrathoracic stomach contour (PTV sto ) was defined as a three-dimensional expansion of the intrathoracic tubular stomach. DVH were recorded for the lung, heart, spinal cord, intrathoracic stomach and PTV in all patients. The dosimetric parameters from DVHs were as follows: (1) D max : maximum dose, (2) D mean : mean dose (3) V dose : percentage volume receiving more than the irradiated dose and (4) aV dose : absolute volume receiving more than the irradiated dose. The range of the intrathoracic stomach volume in 10 Gy (V 10 ) to 40 Gy (V 40 ), at 10 Gy intervals and in 40Gy (V 40 ) to 60Gy (V 60 ), at 5 Gy intervals was acquired. Patients with Grade 2 toxicity and above were analyzed further to determine their risk of developing toxicity according to the dose-volume parameters.

Statistical analysis
The primary endpoint was the occurrence and severity of toxicity. Continuous variables were summarized by descriptive statistics, and categorical variables were tabulated as frequencies and percentages. Dosi metric comparisons between smallfield and largefield group were performed with independent sample t-tests in all patients. The association between toxicity grade (0-1 vs. ≥ 2) and irradiated stomach volume and other clinical and treatment-associated factors were analyzed by multiple logistic regression. Factors analyzed included the following: age, gender, irradiation volume, anastomotic location, surgery procedure and location of the intrathoracic stomach. The area under the curve (AUC) was calculated to determine the cut-off value of the best predictors. Statistical analyses were performed with Statistical Package for the Social Sciences software (Version 13.0, Chicago, IL, USA). All tests were two sided, and P-values less than 0.05 were considered significant.

RESULTS
A total of 105 patients with esophageal SCC were identified for analysis. Table 1 lists the patient and tumor characteristics. The median age of patients included in the study was 58 years (range, 40-71). The median followup time was 16.0 months (range, 3.0-37.3 months). The cases included 79 (75.2%) cervical anastomoses and 26 (24.8%) aortic anastomoses. The median volume of the intrathoracic stomach was 229.06 ml (73.24-440.50 ml), the D max of PTV sto was 6519 ± 406 cGy, and the D mean was 2449 ± 986 cGy. Among the 51 patients in the smallfield group, the D mean and D max of PTV sto were 2062 cGy and 6374cGy, respectively, compared with 2813 cGy and 6657 cGy, respectively among the 54 patients in the largefield group (both p < 0.001, respectively). Independent sample ttests found all mean aV dose and V dose parameters but aV 60 and V 60 significantly higher in the largefield group than the smallfield group (p < 0.05, showed in Table 2).
Treatment toxicities were classified according to the clinical symptoms. Overall, the maximum acute gastric toxicities encountered during RT were Grade 0 in 82 patients (78.1%), Grade 1 in 4 patients (3.8%), Grade 2 in 16 patients (15.2%) and Grade 3 in 3 patients (2.9%). There were no Grade 4 toxicities and no treatmentrelated deaths. None of patients required a treatment break because of gastric toxicity. The cases exhibited Grade 2 and above acute digestive reactions, including 2 with abdominal pain (1.9%), 9 with anorexia (8.6%), 8 with sour regurgitation (7.6%), and 4 with gastrectasia (3.8%). All 3 cases with Grade 3 toxicity was anorexia. The median time to the onset of these symptoms was 16 days (range, 1-40 days).
By multiple logistic regression analysis, V 50 was the only predictive factor for Grade 2 and above gastric toxicity (p = 0.024, Table 3). ROC curve analysis showed that the cutoff value of V 50 was 14.05% (0.815, 95% CI:0.685-0.946; the sensitivity and specificity were 82.4%, and 61.3%, respectively) and that the rates of Grade 2 and above acute and late toxicities were 19.1% for V50 < 14.05% and 34.5% for V50 > 14.05%.

DISCUSSION
Our results show that ≥ grade 2 acute and late gastric toxicity occurred in 18.1% and 14.3% of patients, respectively, and that an RT dose-volume effect for stomach toxicity was shown. V 50 was the most predictive factor for ≥ grade 2 toxicity for the stomach.
In the era of 3DCRT, upper digestive tract tolerance to RT has been investigated in abdominal malignancies. The overall incidence of GD toxicity after RT was reported to be between 5.7% and 23.1% in hepatocellular carcinoma [13][14][15]. The risk of grade 2 and greater GD toxicities was reported to be between 33% and 80% when combined with chemotherapy in pancreatic cancer [11][12]16]. In two retrospective studies for esophageal cancer, the risk of acute toxicity in the upper aerodigestive tract www.impactjournals.com/oncotarget and stomach was 2.3%-11.9% for the small T portal group and 12%-18.6% for the large T portal group [8,17], which was similar to our results. Cosset et al. [18] reported severe late gastric complications included ulcers (n = 25) and severe gastritis (n = 2) among 516 patients with Hodgkin's disease treated by RT close to 40 Gy. Chen et al. also reported the risk of late complications in their study, 1 of 355 patients experienced grade 2-3 gastric bleeding and another 4 patients experienced grade 5 gastric bleeding after postoperative radiation. In a phase II trial of postoperative concurrent chemoradiotherapy, 8% patients experienced grade 3-4 upper digestive tract toxicity, and 6% patients required an unplanned hospitalization [19]. The use of concurrent chemotherapy, differences in RT volume and prescribed dose, and selection bias for the study population may explain the different rates of modest complications in these studies. The irradiated dose in our study was higher when being converted into the biologically effective dose (BED), while partial patients' irradiated volumes were smaller than in previous studies.
In early reports, the gastric ulceration and perforation rates were 4% and 2% vs. 16% and 14% after doses <50 Gy vs. ≥ 50 Gy [10]. Emami et al. [20] demonstrated the tolerance dose for late gastric ulceration was 50, 55, and 60 Gy for the whole stomach, 2/3 of the stomach, and 1/3 of the stomach, as a suggestion for TD5/5 (the probability of 5% complication within 5 years). However, they did not offer estimates to predict acute toxicities. Currently, for stomach tolerance, the current Quantitative Analysis of Normal Tissue Effects in the Clinic (QUANTEC) report recommends dose constraints of the stomach for patients with abdominal tumors as follows: whole stomach 50 Gy (range of maximum 45-54 Gy) and partial volume dose limits of 2% V 50 -10% V 45 . [21]. However, these data may not be applicable for esophageal SCC patients because (1) anatomical structures change after surgery, generally causing some scar tissue in the thoracic and peritoneal cavity; (2) intrathoracic stomach tube for upper digestive tract reconstruction contributes to the development of digestive diseases; and (3) the prescribed dose of thoracic radiation always exceeds 50 Gy, even more than 60 Gy. A dose of 50 Gy has endured as a broad dose limit guideline when irradiated fields encompass a large portion of the stomach, albeit with rather limited support from actual published data [10]. However, nearly all of the patients in our study received intrathoracic stomach irradiation   with a D max more than 60 Gy. The mean V 60 for the group was 12.1%, which was much higher than the QUANTEC standards. According to the study of Emami et al., 60 Gy of 1/3 stomach was tolerant for TD5/5; therefore, we believed the risk of severe toxicities due to RT in our study would not be higher than the existing data. Meanwhile, we think it is still important to ensure that hotspots are minimized when using most conformal techniques. Some of the previously summarized studies reported that RT toxicity was also related to irradiated volume and preformed dose-volume analysis for the stomach. Nakamura et al. [12] reported that V 50 of the stomach ≥ 16 cm 3 may be the best predictor for ≥ grade 2 acute gastrointestinal toxicity. Kim et al. showed a dosevolume analysis of GD toxicity in cirrhotic patients with HCC and suggested that V 35 ≥ 5% could predict ≥ grade 3 GD toxicity [22]. However, there is a lack of data on the evaluation of toxicity based on dose-volume analysis for the intrathoracic stomach using DVH parameters. In the current study, we confirmed the dosevolume effect for digestive toxicity. We found that V 50 for the intrathoracic stomach was the most predictive factor for ≥ grade 2 gastric toxicity. Patients in the largefield group received a greater range of radiation exposure with the same prescribed dose, meaning that the volume of the intrathoracic stomach irradiated was higher than that in the smallfield group. Thus, the largefield group had a higher risk of radiation toxicity than the smallfield group, though statistical significance was not reached. In fact, literature on RT-induced stomach toxicity is relatively sparse, with insufficient data to arrive at firm dosevolume constraints for partial volume irradiation.
There were several limitations in our study. First, the volume of the stomach is variable; therefore, errors in our data were inevitable. However, patients were requested to avoid large meals or carbonated beverages before simulation and treatment to minimize variability in the volume and location of the stomach. Second, we record the toxicities mainly based on patients' symptoms rather than objective examinations. Third, selection bias may have influenced the results. If more patients were enrolled and observed for a longer time period, the values might change. Thus, a larger study is necessary to verify our results.
Therefore, it is acceptable to keep the maximum point dose to the intrathoracic stomach at more than 60 Gy or less for acute and late gastric toxicity for patients with esophageal cancer treated with radiotherapy after esophagectomy. A strong dose-volume relationship exists for the development of Grade 2 acute intrathoracic stomach toxicity in our study. Further studies are necessary to clarify the dose-volume relationship for intrathoracic stomach toxicity and to determine its dose constraint.