Fax: (813) 745-7231
Increased survival associated with surgery and radiation therapy in metastatic gastric cancer
A Surveillance, Epidemiology, and End Results database analysis
Version of Record online: 29 JAN 2013
Copyright © 2013 American Cancer Society
Volume 119, Issue 9, pages 1636–1642, 1 May 2013
How to Cite
Shridhar, R., Almhanna, K., Hoffe, S. E., Fulp, W., Weber, J., Chuong, M. D. and Meredith, K. L. (2013), Increased survival associated with surgery and radiation therapy in metastatic gastric cancer. Cancer, 119: 1636–1642. doi: 10.1002/cncr.27927
- Issue online: 22 APR 2013
- Version of Record online: 29 JAN 2013
- Manuscript Revised: 13 NOV 2012
- Manuscript Accepted: 13 NOV 2012
- Manuscript Received: 3 AUG 2012
- metastatic gastric cancer;
- End Results
Patients with metastatic gastric cancer have poor survival. The purpose of this study was to compare outcomes of metastatic gastric cancer patients stratified by surgery and radiation therapy.
The Surveillance, Epidemiology, and End Results (SEER) database was accessed to identify patients with AJCC M1 stage IV gastric cancer (based on the American Joint Committee on Cancer Cancer Staging Manual, 6th edition) between 2004 thru 2008. Patients were divided into 4 groups: group 1, no surgery or radiation; group 2, radiation alone; group 3, surgery alone; group 4, surgery and radiation. Survival analysis was determined by Kaplan-Meier and log-rank analysis. Multivariate analysis (MVA) was analyzed by the Cox proportional hazard ratio model.
A total of 5072 patients were identified. Surgery and/or radiation were associated with a survival benefit. Median and 2-year survival for groups 1, 2, 3, and 4 was 7 months and 8.2%, 8 months and 8.9%, 10 months and 18.2%, and 16 months and 31.7%, respectively (P < .00001). MVA for all patients revealed that surgery and radiation were associated with decreased mortality whereas T-stage, N-stage, age, signet ring histology, and peritoneal metastases were associated with increased mortality. In patients treated with surgery, MVA showed that radiation was associated with decreased mortality, whereas T-stage, N-stage, age, removal of < 15 lymph nodes, signet ring histology, and peritoneal metastases was associated with increased mortality. Age was the only prognostic factor in patients who did not undergo surgery.
Surgery and radiation are associated with increased survival in a subset of patients with metastatic gastric cancer. Prospective trials will be needed to address the role and sequence of surgery and radiation in metastatic gastric cancer. Cancer 2013. © 2013 American Cancer Society.
Gastric cancer is the second leading cause of cancer mortality and the fourth most common cancer in the world with approximately 934,000 new cases diagnosed and an anticipated 700,000 deaths annually.1 In the United States, an estimated 21,000 new cases and 11,000 deaths were expected in 2010.2 However, most gastric cancers are diagnosed at advanced or metastatic stages when the tumor is considered unresectable.3 Several phase 3 randomized control trials of various chemotherapy regimens in metastatic gastric cancer have shown median survivals ranging from 7.9 to 13.8 months.4
There are few reports on the role of locoregional therapy in the management of metastatic gastric cancer. Small retrospective series have suggested a survival benefit to palliative resections.5-7 Although radiation therapy has been shown to be effective at palliating pain, obstruction, and bleeding from gastric tumors, there have been no reports on a potential survival benefit in metastatic gastric cancer.8-13
Several studies have been published concerning outcomes of radiation and surgery in the setting of localized gastric cancer from the Surveillance, Epidemiology, and End Results (SEER) database. However, there has been only one report on the outcomes of metastatic gastric cancer from SEER, and it did not discuss the role of radiation and surgery in the management of metastatic gastric cancer.14 In the current report, we present data on outcomes of metastatic gastric cancer patients treated between 2004 and 2008 who were treated with or without surgery and radiation therapy. This is the first report of its kind from the SEER database.
MATERIALS AND METHODS
Data were obtained from the SEER Limited Access database. Using the SEER 17 Registries, the 2008 release of the public-use data set 2004-2008 was queried to determine metastatic gastric cancer patients who did or did not undergo surgery and/or radiation therapy. Patients included in the analysis were aged 18 years or older, had American Joint Committee on Cancer (AJCC), 6th edition,15 stage IV (M1) disease, were treated with or without postoperative or preoperative external beam radiation therapy, and had histologic subtypes of adenocarcinoma (not otherwise specified [NOS], intestinal-type, mucinous or mucin-producing, mixed cell or with mixed subtypes, in adenoma or polyp, and with neuroendocrine differentiation) or carcinoma (NOS, diffuse type, and signet ring cell). Patients were excluded from the analysis if they had < 3 months survival, received intraoperative radiation, radioimplants, radioisotopes, if radiation status was unknown, if surgery status was unknown, and if they had histologic subtypes of squamous cell carcinoma, carcinoid, neuroendocrine tumor, small cell carcinoma, leiomyosarcoma, sarcoma, gastrointestinal stromal sarcoma, clear cell adenocarcinoma, adenosquamous carcinoma, melanoma, carcinosarcoma, rhabdoid tumors, and malignant peripheral nerve sheet tumors. Surgical patients underwent some form of gastrectomy with or without a lymph node dissection. Our initial query of M1 gastric cancer patients between 2004 and 2008 that were ≥ 18 years of age identified 8978 patients. After excluding patients with based on histology, 8409 patients remained. There were 3337 patients with < 3 months survival excluded from the analysis. The vast majority of these patients did not receive any locoregional therapy. Of these 3337 patients with < 3 months survival, 2750 patient did not receive radiation or surgery, 241 patients were treated with radiation alone, 330 patients treated with surgery alone, and 16 patients were treated with both surgery and radiation. These patients were excluded to weed out poorly performing patients who were probably not healthy enough to receive any therapy that could dramatically impact the analysis in favor of locoregional therapy.
Data not included in the SEER database include patient comorbidities, nutritional status, performance status, surgical margin status, postoperative complications, type of lymphadenectomy, chemotherapy and radiation dose, and field design.
The variable radiation/surgery status consisted of 4 levels: patients who did not receive radiation or surgery, patients treated with radiation alone, patients treated with surgery alone, and patients who were treated with both surgery and radiation. Descriptive statistics were presented overall and by radiation/surgery status. Comparisons between radiation/surgery status and demographic characteristics were made using the exact Wilcoxon rank-sum test for continuous characteristics, and the exact Pearson chi-square test for categorical characteristics, both using Monte Carlo estimation. Overall survival was examined using Kaplan-Meier survival curves with log-rank test. Multivariate analysis (MVA) of prognostic factors related to survival (radiation therapy, AJCC T-stage, number of lymph nodes removed, AJCC N-Stage, surgery, tumor location, histologic subtype, and tumor grade) was performed by the Cox proportional hazard model. All P values are 2-sided and considered statistically significant at the .05 level. All statistical analyses were performed using SAS (version 9.3; SAS Institute, Cary, NC).
We identified 5072 patients who met inclusion criteria. There were 3069 patients who did not receive radiation or surgery (group 1), 806 patients treated with radiation alone (group 2), 957 patients treated with surgery alone (group 3), and 240 patients treated with surgery and radiation (group 4). Patient characteristics are presented in Table 1.
|Variable||No Surg/No RT||No Surg/RT||Surg/No RT||Surg/RT||Pa|
|Mean age (± SD)||62.8 (14.7)||63.5 (13.1)||63.9 (14.2)||60 (13)||0.0012|
|Female||1206 (39.3%)||195 (24.2%)||400 (41.8%)||74 (30.8%)|
|Male||1863 (60.7%)||611 (75.8%)||557 (58.2%)||166 (69.2%)|
|Year of diagnosis||0.0164|
|2004||619 (20.2%)||172 (21.3%)||206 (21.5%)||58 (24.2%)|
|2005||578 (18.8%)||149 (18.5%)||212 (22.2%)||50 (20.8%)|
|2006||643 (21%)||160 (19.9%)||228 (23.8%)||52 (21.7%)|
|2007||704 (22.9%)||193 (23.9%)||184 (19.2%)||47 (19.6%)|
|2008||525 (17.1%)||132 (16.4%)||127 (13.3%)||33 (13.8%)|
|Metastases at diagnosis||<.0001|
|Distant nodes||2282 (74.4%)||511 (63.4%)||743 (77.6%)||136 (56.7%)|
|Carcinomatosis or NOS||487 (15.9%)||159 (19.7%)||92 (9.6%)||21 (8.8%)|
|Both||300 (9.8%)||136 (16.9%)||122 (12.7%)||83 (34.6%)|
|Cardia||853 (27.8%)||507 (62.9%)||112 (11.7%)||86 (35.8%)|
|Fundus||154 (5%)||24 (3%)||28 (2.9%)||6 (2.5%)|
|Body||313 (10.2%)||38 (4.7%)||96 (10%)||21 (8.8%)|
|Antrum||430 (14%)||53 (6.6%)||243 (25.4%)||57 (23.8%)|
|Pylorus||60 (2%)||9 (1.1%)||36 (3.8%)||10 (4.2%)|
|Lesser curve||165 (5.4%)||21 (2.6%)||115 (12%)||16 (6.7%)|
|Greater curve||96 (3.1%)||17 (2.1%)||69 (7.2%)||5 (2.1%)|
|Overlapping site||301 (9.8%)||53 (6.6%)||132 (13.8%)||12 (5%)|
|NOS||697 (22.7%)||84 (10.4%)||126 (13.2%)||27 (11.3%)|
|AC NOS||1900 (61.9%)||609 (75.6%)||505 (52.8%)||151 (62.9%)|
|Signet ring carcinoma||233 (7.6%)||52 (6.5%)||66 (6.9%)||21 (8.8%)|
|Carcinoma||33 (1.1%)||7 (0.9%)||23 (2.4%)||3 (1.3%)|
|Mucinous AC||41 (1.3%)||5 (0.6%)||30 (3.1%)||7 (2.9%)|
|Mixed cell AC||64 (2.1%)||19 (2.4%)||31 (3.2%)||12 (5%)|
|Linitis plastica||798 (26%)||114 (14.1%)||302 (31.6%)||46 (19.2%)|
|T0||6 (0.2%)||2 (0.2%)||1 (0.1%)||0 (0%)|
|T1||498 (16.2%)||168 (20.8%)||41 (4.3%)||11 (4.6%)|
|T2||338 (11%)||91 (11.3%)||312 (32.6%)||102 (42.5%)|
|T3||181 (5.9%)||86 (10.7%)||300 (31.3%)||65 (27.1%)|
|T4||641 (20.9%)||160 (19.9%)||277 (28.9%)||52 (21.7%)|
|Tx||1405 (45.8%)||299 (37.1%)||26 (2.7%)||10 (4.2%)|
|N0||976 (31.8%)||232 (28.8%)||128 (13.4%)||36 (15%)|
|N1||905 (29.5%)||354 (43.9%)||356 (37.2%)||114 (47.5%)|
|N2||14 (0.5%)||6 (0.7%)||255 (26.6%)||54 (22.5%)|
|N3||8 (0.3%)||4 (0.5%)||176 (18.4%)||29 (12.1%)|
|Nx||1166 (38%)||210 (26.1%)||42 (4.4%)||7 (2.9%)|
|Well||51 (1.7%)||17 (2.1%)||9 (0.9%)||4 (1.7%)|
|Moderate||516 (16.8%)||190 (23.6%)||140 (14.6%)||44 (18.3%)|
|Poor||1677 (54.6%)||414 (51.4%)||716 (74.8%)||157 (65.4%)|
|Undifferentiated||49 (1.6%)||19 (2.4%)||39 (4.1%)||11 (4.6%)|
|Unknown||776 (25.3%)||166 (20.6%)||53 (5.5%)||24 (10%)|
The overall survival (OS) for the 4 groups of patients listed above is displayed in Figure 1. Radiation and surgery were associated with increased survival. Median and 2-year OS for groups 1, 2, 3, and 4 was 7 months and 8.2%, 8 months and 8.9%, 10 months and 18.2%, and 16 months and 31.7%, respectively (log-rank P < .00001). All curves were statistically significant from one another (Fig. 1A). When patients with survival of < 3 months were included in the analysis, median and 2-year OS for groups 1, 2, 3, and 4 was 4 months and 5.1%, 6 months and 7.3%, 8 months and 14.9%, and 15 months and 31.2%, respectively (log-rank P < .00001) and all curves were statistically significant from one another (Fig. 1B). Given that majority of patients with < 3 months survival had not received radiation or surgery, we decided to proceed with rest of the analysis excluding those patients.
Multivariate analysis for all patients revealed that surgery (hazard ratio [HR] = 0.565; 95% CI = 0.495-0.645; P < .0001) and radiation (HR = 0.882; 95% CI = 0.781-0.995; P = .042) were associated with decreased mortality (Table 2). Increasing age, T4 tumors, N3 nodes, signet ring histology, and peritoneal carcinomatosis were associated with increased mortality. Sex, tumor grade, and tumor location were not prognostic for survival. MVA for patients who underwent surgical resection showed that radiation (HR = 0.733; 95% CI = 0.592-0.907; P = .004) was associated with decreased mortality, whereas increasing age, removal of < 15 lymph nodes, tumor stage, nodal stage, signet ring histology, and peritoneal carcinomatosis conferred higher mortality. There was a trend toward higher mortality as tumor and nodal stage increased. Sex, tumor location, and tumor grade were not prognostic for survival (Table 2). In patients who did not undergo surgical resection, MVA demonstrated that only increasing age was prognostic for higher mortality. There was a trend for worse mortality with signet ring histology. Sex, radiation, tumor stage, nodal stage, tumor location, tumor histology, tumor grade, and peritoneal carcinomatosis were not prognostic for survival (Table 2).
|All Patients||HR||95% CI||P|
|Sex (vs male)||1.05||0.946-1.166||0.362|
|Surgery (vs none)||0.565||0.495-0.645||<.0001|
|Radiation (vs none)||0.882||0.781-0.995||0.042|
|T-stage (vs T1)|
|N-stage (vs N0)|
|Grade (vs well)|
|Location of mets (vs nodes)|
|Peritoneal and nodes||1.398||1.169-1.671||<.0001|
|Site (vs Body)|
|Histology (vs AC NOS)|
|Surgery-Only Patients||HR||95% CI||P|
|Sex (vs male)||1.151||0.986-1.343||0.075|
|Radiation (vs none)||0.733||0.592-0.907||0.004|
|LN removed (vs ≥15)||1.239||1.031-1.490||0.022|
|T-stage (vs T1)|
|N-stage (vs N0)|
|Grade (vs well)|
|Location of mets (vs nodes)|
|Peritoneal and nodes||1.793||1.314-2.447||<.0001|
|Site (vs body)|
|Histology (vs AC NOS)|
|Nonsurgical Patients||HR||95% CI||P|
|Sex (vs male)||0.992||0.858-1.147||0.91|
|Radiation (vs none)||0.987||0.847-1.149||0.863|
|T-stage (vs T1)|
|N-stage (vs N0)|
|Grade (vs well)|
|Location of mets (vs nodes)|
|Peritoneal and nodes||1.162||0.932-1.450||0.183|
|Site (vs body)|
|Histology (vs AC NOS)|
This is the first report to suggest that surgery and radiation are associated with increased survival in metastatic gastric cancer. On univariate Kaplan-Meier analysis, patients who received radiation had a higher survival over patients not treated with radiation therapy, which was confirmed by MVA except in nonsurgical patients. Surgery also was associated with increased survival both on univariate and multivariate analysis. Interestingly, removal of > 15 lymph nodes, which has been shown to be associated with increased survival in the nonmetastatic setting,16, 17 was associated with higher survival, whereas there was a trend for worse survival with higher tumor and nodal stage. The location of metastases also affected survival. Patients with peritoneal carcinomatosis had worse survival compared with patients who had distant lymph node metastases. Signet ring histology also conferred worse outcomes. Sex, tumor location, and grade did not impact survival in metastatic gastric cancer.
Although systemic chemotherapy is by far the most effective modality for increasing survival in metastatic gastric cancer, it is not that effective in relieving symptoms of obstruction, pain, and bleeding caused by the primary tumor. Local treatment modalities such as partial gastrectomy, radiation therapy, or both are far more effective at managing symptoms caused by the primary tumor. Although radiation therapy has been shown to be effective at palliating pain, obstruction, and bleeding from gastric tumors, there have been no reports on a potential survival benefit in metastatic gastric cancer.8-13 There are studies of chemoradiation in the locally advanced setting that show a significant survival benefit. An early Mayo Clinic study randomized 48 patients with unresectable disease to concurrent chemoradiation with 35-40 Gy of supervoltage radiation and 5-fluorouracil.18 Median survival favored chemoradiation (13 months versus 6 months) with a 5-year OS of 12% for patients treated with chemoradiation. A North Central Cancer Treatment Group trial evaluated the role of 5-fluorouracil based chemoradiation with doses ranging from 45-54 Gy in unresectable gastrointestinal carcinomas.19 Included in the analysis were 6 gastric cancer patients, of whom 2 had no evidence of disease 38 to 50 months after completion of treatment.
Palliative gastrectomy has been associated with increased survival in patients with advanced gastric cancer.5, 6 Meijer et al6 showed that in 51 patients who underwent palliative resection or bypass, patients who underwent resection had prolonged survival (9.5 months versus 4.2 months) compared with patients who had gastric bypass. Similarly, a report from Bozzetti et al attempted to stratify patients with metastatic gastric cancer by disease extent.5 The median survival was higher for patients undergoing palliative resection compared with those who had bypass, exploratory laparotomy only, or no surgery at all (8 versus 3.5, 2.8, and 2.4 months, respectively). After stratification by disease extent, the survival advantage of palliative resection was still maintained. Suzuki et al reported on a trial from 2001 to 2009, where stage IV gastric cancer patients were treated with S-1/cisplatin or S-1/docetaxel chemotherapy.7 Adjuvant surgery of the primary and metastatic disease was performed on 37 patients. Surgical patients had a median survival of 855 days versus 277 days for those without an operation (P < .0001). In the Dutch Gastric Cancer Trial, 26% of the randomized patients were found to have incurable tumors at laparotomy, including unresectable tumor or metastases to liver, peritoneum, or distant lymph nodes.20 The patients underwent an explorative laparotomy, a gastroenterostomy, or a partial or total gastrectomy. For patients undergoing resection, median OS was greater compared with patients who were not resected (8.1 months versus 5.4 months; P < .001). The benefit was maintained in patients > 70 years, however there was a significant increase in perioperative mortality (50% versus 20%). In patients with one metastatic lesion, a significant survival benefit was noted in patients who underwent resection (median survival of 10.5 months versus 6.7 months; P = .034), but the benefit was lost if more than one metastasis was noted (median survival of 5.7 months versus 4.6 months; P = .084).
There are currently 2 prospective trials being conducted to address the role of palliative surgery in stage IV gastric cancer.7 The REGATTA trial being conducted in Japan and Korea will address the role of palliative surgery followed by chemotherapy in metastatic gastric cancer. In Japan, a phase 2/3 trial about to be conducted will treat patients with S-1 and docetaxel or cisplatin. Those patients achieving a partial or complete response will be randomized to either continued chemotherapy or will undergo surgical resection followed by additional chemotherapy.7
Other tumor sites have shown increased survival with the addition of locoregional therapy (LRT) in the treatment of metastatic disease.21-23 Nguyen et al reported on the outcomes of 733 metastatic breast cancer patients treated between 1996 and 2004 who underwent LRT with either surgery, radiation, or both from the British Columbia Cancer Agency.24 The 5-year OS for patients who had LRT was 21% compared with 14% for those patients not treated with LRT (P < .001). The survival benefit of LRT was also confirmed on MVA (HR = 0.78; 95% CI = 0.64-0.94; P = .009). Two randomized trials confirmed the benefit of surgery in metastatic renal cell carcinoma.21, 22 A study by Flanigan et al21 showed that metastatic patients who underwent radical nephrectomy followed by interferon had a median OS of 11.1 months compared with 8.1 months for patients treated with interferon alone (P = .05). Mikisch et al showed that in patients with metastatic kidney cancer, nephrectomy followed by interferon resulted in a median survival of 17 months compared with 7 months for patients treated with interferon alone (P = .03).22
Our study has limitations mainly associated with the use of the SEER database. Several aspects critical to the evaluation of survival are missing from the SEER database. Although one can assume the metastatic gastric cancer patients will be treated with systemic chemotherapy, the SEER database lacks this important piece of information. In addition, there is missing surgical data on resection margin status, extent of lymphadenectomy or information if resection was complete. From a radiation point of view, there is no available information if concurrent chemotherapy was used with radiation, the radiation field design and technique, or dose of radiation used. There is also missing information on patient's nutritional status and performance status. Although burden of disease was not quantified, staging and location of the primary tumor, lymph nodes, and metastases was provided and accounted for in the MVA. Finally, it is impossible to fully understand from SEER why some people may have received surgery and/or radiation therapy as part of their treatment management. It is possible that healthier patients, hence better prognosis patients, received these locoregional treatments. However, despite these inherent biases, SEER documents the real world outcomes of these patients. Although these limitations may confound the results, our findings do elucidate a possible association between surgery and radiation therapy and improved survival for metastatic gastric cancer patients.
Surgical resection and radiation therapy are associated with increased overall survival in metastatic gastric cancer patients from the SEER database. Caution must be exercised before locoregional treatments are recommended in the management of metastatic gastric cancer, because there are no randomized data to validate this finding and that there is significant data that are missing from the SEER to properly stratify patients. However, it does warrant the need for prospective clinical trials addressing the role of locoregional treatment in well-defined patient cohorts with metastatic gastric cancer.
No specific funding was disclosed.
CONFLICT OF INTEREST DISCLOSURE
The authors made no disclosure.
- 15Greene FL, Page DL, Fleming ID, et al, editors. AJCC Cancer Staging Handbook: TNM Classification of Malignant Tumors. 6th edition. New York, NY: Springer-Verlag; 2002.