ORIGINAL ARTICLE. Proposal to Subclassify Stage IV Gastric Cancer Into IVA, IVB, and IVM

Transcription

1 ORIGINAL ARTICLE roposal to Subclassify Stage IV Gastric Cancer Into IVA, IVB, and IVM Ji Yeong An, MD; Tae Kyung Ha, MD; Jae Hyung Noh, MD; Tae Sung Sohn, MD; Sung Kim, MD Objective: This study examined the prognosis of patients with stage IV gastric cancer by subgroups after surgical treatment. Design: Retrospective study. Setting: Tertiary care referral center. atients: A total of 1056 patients with stage IV gastric cancer who underwent gastrectomy from January 1, 1995, through December 31, 2006, were divided into the following 3 groups: T4N1-3M0 (group 1), T1-3N3M0 (group 2), and T(any)N(any)M1 (group 3). Main Outcome Measures: The clinicopathological characteristics, recurrence pattern, and survival were compared among the 3 groups. Results: There was a significant difference in the surgical curability, operation type, Lauren classification, histological differentiation, lymphatic invasion, number of lymph nodes retrieved, and adjuvant therapy among the 3 groups. The 5-year survival rates in groups 1, 2, and 3 were 18.3%, 27.1%, and 9.3%, respectively (.001). After R0 resection, locoregional recurrence (40.9%) followed by peritoneal recurrence (27.3%) was most common in group 1, whereas distant (30.2%) and peritoneal recurrence (26.7%) were most common in group 2. Multivariate analysis showed the following significant prognostic factors for survival: surgical curability and adjuvant therapy for group 1; surgical curability, surgical extent, adjuvant therapy, and number of retrieved lymph nodes for group 2; and surgical extent and chemotherapy for group 3. Conclusions: Each subgroup of stage IV gastric cancer had different clinical outcomes, including histological behavior, recurrence pattern, survival, and prognostic factors. Therefore, subclassification of stage IV gastric cancer into IVA (T1-3N3M0), IVB (T4N1-3M0), and IVM (T[any]N[any]M1) might be useful for a more accurate prediction of prognosis and selection of appropriate therapeutic options. Arch Surg. 2009;144(1):38-45 Author Affiliations: Department of Surgery, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, Korea. THE SURVIVAL OF ATIENTS with gastric cancer is associatedwiththetumorstage, including the depth of tumor invasion, the number of lymph node metastases, and the presence of a distant metastasis. Stage IV gastric cancer is defined as T4N1-3M0, T1-3N3M0, or T(any)N(any)M1bytheInternationalUnion See Invited Critique at end of article Against Cancer classification. 1 This system definesstageivgastriccancerasadiseaseaccompanied by advanced locoregional invasion, extensivelymphnodemetastasis, ordistant metastasis. Therefore, the prognosis of patients with stage IV gastric cancer is generally poor. Most previous studies have focused on tracing the clinical course of stage IV gastric cancer according to the pathological features and treatment modalities. 2-5 Although the overall 5-year survival rate of patients with stage IV gastric cancer is reported to be less than 10%, the prognosis of stage IV gastric cancer can vary depending on disease progression, the extent of disease, and the potential for cure with treatment. 5,6 Therefore, a more detailed classification of stage IV gastric cancer, according to differences in clinical outcome, would provide useful information for selecting more appropriate therapeutic modalities to achieve a better survival outcome. The aim of this study was to evaluate the prognosis of stage IV gastric cancer subgroups after surgical treatment and to identify the characteristics associated with prognosis for each group. METHODS ATIENTS From January 1, 1995, through December 31, 2006, a total of patients with gastric adenocarcinoma underwent surgical treatment at the Samsung Medical Center. Among them, 1720 patients were diagnosed as having stage IV gastric cancer after a laparotomy accord- 38

2 ing to the International Union Against Cancer classification. 1 All of the patients had a histologically confirmed adenocarcinoma of the stomach with or without evidence of distant metastasis. atients who had undergone previous gastric surgery or neoadjuvant chemotherapy were excluded. Among the 1720 patients, 664 (38.6%) underwent a bypass procedure or biopsy only and 1056 (61.4%) underwent curative or palliative gastric resection. atients with localized resectable gastric cancer received curative surgical resection if possible, and, in these cases, the presence of multiple-organ direct invasion and major vessel encasement was a key factor making surgical resection impossible. In M1 disease, the extent of metastatic lesions and symptoms such as bleeding and obstruction as well as the resectability of the primary lesion are important for determining operation methods. Therefore, gastric cancer with single-organ metastasis and localized peritoneal seeding, when complete gross resection is possible, is indicated for surgical resection. Symptomatic patients with resectable primary cancer and more advanced metastatic lesions underwent palliative gastrectomy. However, all of the operation methods and the extent of the operation have been determined by considering the patients general condition and tolerance. The 1056 patients who underwent a gastric resection were enrolled in this study and divided into the following 3 groups according to the surgical and pathological staging: T4N1-3M0 (group 1), T1-3N3M0 (group 2), and T(any)N(any)M1 (group 3). SURGICAL TREATMENT An R0 resection was defined as a complete resection of the localized tumors without a distant metastasis. An R1 resection was defined as the gross removal of a microscopic residual tumor or an accompanying peritoneal or distant metastasis. An R2 resection was defined grossly as leaving residual tumors. Therefore, the microscopic resection margin involvement was regarded as an R1 resection in groups 1 and 2. atients with M1 disease, including single liver metastasis, ovary metastasis without peritoneal seeding nodules, or localized tumor seeding to the greater and lesser omentum, were classified as having undergone an R1 resection when there was no gross residual lesion after the removal of the metastatic lesions. The group 1 and 2 patients routinely received a gastrectomy with D2 lymph node dissection because the operation was performed with curative intent, whereas the group 3 patients underwent D2 lymph node dissection in R1 resection and D1 or D2 lymph node dissection in R2 resection cases. An extended gastrectomy included a resection of the adjacent organs such as the spleen, colon, pancreas, small bowel, liver, and kidney in addition to a subtotal or total gastrectomy. ADJUVANT THERAY Adjuvant chemoradiotherapy based on leucovorin calcium and fluorouracil with radiotherapy was administered in patients with a potentially curative resection. atients with metastatic lesions received taxane- and cisplatin-based chemotherapy or irinotecan hydrochloride, leucovorin-, and fluorouracil-based chemotherapy. INITIAL RECURRENCE ATTERN DETERMINATION AFTER R0 RESECTION During the routine postoperative follow-up, results of an endoscopic examination, abdominal computed tomography, complete blood cell count, chemistry studies, measurement of levels of biological markers such as carcinoembryonic antigen and cancer antigens 19-9 and 72-4, and chest radiography were evaluated every 6 months. The tumor recurrence pattern was evaluated in the patients who underwent an R0 resection and was classified as locoregional, peritoneal, distant, or combined. It was difficult to determine the initial recurrence mode because this study was performed retrospectively using a review of the medical records, and most patients did not undergo a biopsy to confirm the recurrence. Therefore, 517 of 631 patients with R0 resection who underwent evaluation every 3 to 6 months serially and had sufficient information to discern the presence and site of the initial relapse were examined to decide the initial recurrence pattern. The other 114 patients were also followed up, but they occasionally missed examinations. An endoscopic examination, abdominal computed tomography, cytological evaluation of peritoneal fluid, computed tomography of the chest, bone scan, or positron emission tomography was performed to confirm recurrence. Locoregional recurrence included tumor reappearance and progression in the gastric bed, the anastomotic site, or the upper abdominal lymph nodes. atients with disease progression in the peritoneal nodules, peritoneal wall thickening, or ascites with positive cytological findings were considered to have a peritoneal recurrence. A distant recurrence was defined when specific intra-abdominal or extraabdominal organs such as the liver, lung, bone, brain, adrenal glands, or cervical lymph nodes were involved. 7 EVALUATION The clinical features and variables for each group, including sex, age, tumor location, tumor size, type of operation, histological type, recurrence, and survival, were analyzed on the basis of information in the medical records. The histological characteristics of the tumor were classified into the following 2 groups: the differentiated type, which included papillary, welldifferentiated, and moderately differentiated adenocarcinoma, and the undifferentiated type, which included poorly differentiated and undifferentiated adenocarcinoma, signet ring cell carcinoma, and mucinous carcinoma. STATISTICAL ANALYSIS The categorical variables were compared using a 2 test. The continuous data are presented as mean (SD), and we used the Kruskal-Wallis test or Mann-Whitney test for statistical comparisons. The survival distributions of the 3 groups were calculated in months from the primary surgical treatment to the final follow-up or death of the patient using the Kaplan-Meier method. The univariate association of the various factors with survival was performed using a log-rank test. We performed the examination for independent prognostic factors using the Cox proportional hazards regression model. In all statistical analyses,.05 was considered significant. RESULTS CLINICAL AND ATHOLOGICAL FEATURES IN EACH GROU OF STAGE IV GASTRIC CANCER Table 1 shows the characteristics of each group. Sex, age, tumor location, primary tumor size, and gross appearance had a similar distribution in all 3 groups. The frequency of extended gastrectomy procedures was higher in group 1 (54.1%) compared with groups 2 (30.4%) and 3 (34.3%). An R0 resection was performed in 54.1% and 90.7% of patients in groups 1 and 2, respectively. As described in the Surgical Treatment subsection of the 39

3 Table 1. Clinicopathological Features of atients With Stage IV Gastric Cancer According to Their Subgroup a Group 1 (n=157) Group 2 (n=602) Group 3 (n=297) Value b Sex.08 Male 109 (69.4) 378 (62.8) 174 (58.6) Female 48 (30.6) 224 (37.2) 123 (41.4) Age, y.12 c Mean (SD) 58.2 (12.0) 55.7 (11.3) 50.1 (14.8) Range Tumor location, No. of patients.28 Upper Middle Lower Entire Operation type.001 Subtotal gastrectomy 50 (31.8) 239 (39.7) 122 (41.1) Total gastrectomy 22 (14.0) 180 (29.9) 73 (24.6) Extended resection 85 (54.1) 183 (30.4) 102 (34.3) Curability.001 R0 85 (54.1) 546 (90.7) 0 R1 67 (42.7) 51 (8.5) 59 (19.9) R2 5 (3.2) 5 (0.8) 238 (80.1) Tumor size, cm.40 c Mean (SD) 7.4 (2.6) 7.5 (2.6) 7.7 (3.2) Range Borrmann classification.31 I 0 10 (1.7) 4 (1.3) II 26 (16.6) 63 (10.5) 41 (13.8) III 92 (58.6) 360 (59.8) 165 (55.6) IV 37 (23.6) 165 (27.4) 85 (28.6) V 2 (1.3) 4 (0.7) 2 (0.7) Lauren classification.003 Intestinal 62 (39.5) 155 (25.7) 100 (33.7) Diffuse 86 (54.8) 409 (67.9) 186 (62.6) Mixed 9 (5.7) 38 (6.3) 11 (3.7) Histological type.001 Differentiated 51 (32.5) 116 (19.3) 73 (24.6) Undifferentiated 106 (67.5) 486 (80.7) 224 (75.4) Lymphatic invasion.001 Absent 56 (35.7) 133 (22.1) 109 (36.7) resent 101 (64.3) 469 (77.9) 188 (63.3) Vascular invasion.49 Absent 132 (84.1) 491 (81.6) 251 (84.5) resent 25 (15.9) 111 (18.4) 46 (15.5) erineural invasion.07 Absent 97 (61.8) 345 (57.3) 152 (51.2) resent 60 (38.2) 257 (42.7) 145 (48.8) No. of retrieved lymph nodes, mean (SD) 38.1 (16.6) 43.8 (13.6) 30.4 (14.6).001 c Adjuvant therapy.001 None 66 (42.0) 212 (35.2) 106 (35.7) Chemotherapy 59 (37.6) 192 (31.9) 173 (58.2) Chemoradiotherapy 32 (20.4) 198 (32.9) 18 (6.1) a Groups are described in the atients subsection of the Methods section. Unless otherwise specified, data are expressed as number (percentage) of patients. ercentages have been rounded and may not total 100. b Calculated by means of the 2 test unless otherwise indicated. c Calculated by means of the Kruskal-Wallis test. Methods section, patients with M1 disease were classified as having undergone an R1 resection when there was no gross residual lesion after the removal of the metastatic lesions. There were significant differences in the distribution of the Lauren classification, histological differentiation, lymphatic invasion, number of retrieved lymph nodes, and type of adjuvant therapy in comparisons among the 3 groups. EVALUATION FOR RECURRENCE ATTERN We examined 517 patients in groups 1 and 2 to determine the sites of relapse after R0 resection (Table 2). During follow-up, 65.7% and 70.7% of patients in groups 1 and 2, respectively, had a recurrence. In group 1, locoregional recurrence (40.9%) was the most common, followed by peritoneal (27.3%), combined (18.2%), and 40

4 Table 2. Recurrence atterns of Stage IV Gastric Cancer After R0 Resection a Group 1 (n=67) distant (13.6%) recurrences, whereas distant (30.2%) and peritoneal (26.7%) recurrences were the most common in group 2. SURVIVAL ANALYSIS Group 2 (n=450) The median follow-up was 13.5 months. At the last followup, 42.8% of patients were still alive. The overall 1-, 3-, and 5-year survival rates of the patients with stage IV gastric cancer were 69.1%, 31.0%, and 21.5%, respectively. The median survival time was 18.9 months. The 5-year survival rates in groups 1, 2, and 3 were 18.3%, 27.1%, and 9.3%, respectively (.001). The median survival times in each group were 17.6, 23.1, and 14.3 months, respectively. There was a significant difference in the plotted survival curve among the 3 groups (Figure). The rate of postoperative major morbidity requiring a reoperation, intervention, or intensive care unit stay was 3.8%, and the postoperative mortality rate was 0.4%. Information on quality of life, including the patient s level of satisfaction, hospital stay, and symptomatic or functional aspects, was unavailable for this study. ROGNOSTIC FACTORS FOR SURVIVAL Value b Recurrence.40 No 23 (34.3) 132 (29.3) Yes 44 (65.7) 318 (70.7) Mean (SD) time to recurrence, mo 10.9 (9.7) 13.6 (12.4).13 c Recurrence pattern.008 Locoregional 18 (40.9) 63 (19.8) eritoneal 12 (27.3) 85 (26.7) Distant 6 (13.6) 96 (30.2) Combined 8 (18.2) 74 (23.3) a Groups are described in the atients subsection of the Methods section. Unless otherwise specified, data are expressed as number (percentage) of patients. b Calculated by means of the 2 test unless otherwise indicated. c Calculated by means of the Kruskal-Wallis test. The 5-year survival rate and median survival time according to the factors studied are summarized in Table 3, and the results of the multivariate analysis are shown in Table 4. In group 1 (T4N1-3M0), the univariate analysis showed that the surgical curability, the Borrmann type, adjuvant therapy, and recurrence pattern were associated with patient survival. However, the multivariate analysis showed that the surgical curability and adjuvant therapy were independent prognostic factors for survival. The relative hazard value of the R1 group was 1.92 (=.002). atients without adjuvant therapy had a higher relative hazard value (2.67 [=.001]) than those with adjuvant therapy. In group 2 (T1-3N3M0), the surgical curability, operation type, adjuvant therapy, number of retrieved lymph nodes, and recurrence pattern were associated with survival. Of these 5 factors, multivariate analysis showed that the surgical curability, operation Cumulative Survival, % Follow-up, mo type, adjuvant therapy, and number of retrieved lymph nodes were significant independent predictive factors. atients in whom 30 or fewer lymph nodes were retrieved had a poor survival compared with those in whom 31 or more lymph nodes were examined. In group 3 (T[any]N[any]M1), the type of surgery and adjuvant therapy were determined to be prognostic factors by univariate and multivariate analysis. COMMENT Group 1 Group 2 Group 3 Figure. Cumulative survival curves for patients with stage IV gastric cancer according to their subgroups. The 5-year survival rate for group 1 (T4N1-3M0) was 18.3%; for group 2 (T1-3N3M0), 27.1%; and for group 3 (T[any]N[any]M1), 9.3% (.001). Because the prognosis of stage IV gastric cancer has been considered to be poor and the benefits of surgery are controversial, many reports have focused on selecting the appropriate therapeutic options and determining the prognostic factors for survival. 5,8-10 A radical gastrectomy is the treatment for gastric cancer that is most highly associated with cure. However, patients with stage IV gastric cancer often receive other palliative surgical or nonsurgical procedures owing to their advanced disease status. Some studies have demonstrated that a palliative gastrectomy is associated with a survival benefit as a result of removal of gross disease; this procedure in these patients has been shown to improve function and to achieve a better response to adjuvant therapy. 3 Others have reported long-term survival after surgical resection in selected patients with stage IV gastric cancer. 11 Moreover, only after a surgical resection can the evaluation and comparison of tumor progression status, such as the depth of tumor invasion, status of lymph node metastasis, and distant metastasis (the most important prognostic factors for gastric cancer), be interpreted with a high degree of accuracy. Therefore, it may be reasonable to divide stage IV gastric cancer, confirmed by a surgical resection, into 3 groups according to the characteristics of disease progression. In this study, group 1 represented stage IV gastric cancer associated with adjacent organ invasion. Therefore, 41

5 Table 3. rognostic Variables and 5-Year Survival Rates in atients With Stage IV Gastric Cancer According to Their Subgroup a Group 1 Group 2 Group 3 5-y Survival, % 5-y Survival, % 5-y Survival, % Variable (Median, mo) Value b (Median, mo) Value b (Median, mo) Value b Sex Male 21.1 (17.7) 28.2 (22.8) 7.9 (15.6) Female 11.6 (15.8) 22.9 (25.6) 5.6 (12.5) Age, y (22.6) 28.5 (22.2) 11.3 (14.8) (15.7) 25.4 (23.1) 7.3 (14.0) Tumor location Upper 32.3 (17.6) 26.1 (20.1) 5.8 (19.1) Middle 18.9 (15.7) 27.9 (23.0) 5.3 (12.2) Lower 15.8 (17.9) 28.4 (27.5) 7.8 (14.5) Entire... c 0.0 (32.9)... (11.2) c Operation type Subtotal 13.3 (19.0) 34.8 (33.8) 15.9 (16.7) Total 11.6 (19.2) 27.1 (27.5) 0.0 (15.2) Combined 15.2 (14.9) 15.5 (17.2) 0.0 (12.4) Curability R (25.5) 28.8 (25.8)... d R (13.3) 0.0 (15.0) 0.0 (17.3) R2... (5.4) c... (9.3) c 8.3 (13.8) Borrmann classification I NA d 90.0 (37.1)... (11.5) c II 16.5 (15.7) 38.6 (32.4) 5.6 (14.8) III 21.5 (18.4) 25.6 (23.2) 7.2 (14.5) IV 12.7 (17.6) 13.0 (19.6) 7.8 (13.8) V 0.0 (4.6) e... (11.5) c Lauren classification Intestinal 0.0 (15.7) 30.0 (27.0) 4.6 (14.9) Diffuse 20.9 (17.7) 26.0 (20.6) 11.1 (14.2) Mixed 21.9 (29.1) 32.8 (27.5) 0.0 (10.6) Histological type Differentiated 0.0 (18.5) 29.5 (28.4) 7.8 (15.6) Undifferentiated 17.1 (15.8) 26.5 (21.8) 7.2 (14.2) Lymphatic invasion Absent 26.6 (14.0) 23.0 (21.5) 12.9 (14.3) resent 9.4 (18.9) 28.7 (23.1) 5.8 (14.2) Vascular invasion Absent 17.9 (15.8) 27.5 (24.9) 10.3 (14.8) resent 16.8 (19.3) 20.1 (16.9) 0.0 (11.8) erineural invasion Absent 19.0 (19.7) 30.8 (25.3) 13.6 (14.8) resent 7.8 (14.8) 21.0 (21.8) 4.8 (13.9) No. of retrieved lymph nodes (14.8) 7.4 (14.7) 8.1 (13.7) (14.5) 26.2 (25.5) 10.8 (14.8) (20.5) 31.1 (25.8) 0.0 (13.8) Adjuvant therapy None 0.0 (11.3) 10.2 (15.1) 2.1 (9.7) Chemotherapy 20.3 (19.8) 23.3 (21.2) 9.1 (17.4) Chemoradiotherapy 32.4 (25.7) 41.9 (43.8) 40.7 (18.3) Recurrence pattern f NA Locoregional 0.0 (32.0) 6.0 (17.5) NA eritoneal 0.0 (9.1) 0.0 (15.5) NA Hematogenous... (15.7) 11.1 (22.0) NA Combined 0.0 (12.7) 11.7 (19.2) NA Abbreviations: NA, not applicable; ellipses, not calculated. a Groups are described in the atients subsection of the Methods section. b Calculated by means of the log-rank test. c Because of the small number of patients and/or short survival time, it was impossible to calculate the 5-year survival rate and/or median survival time. d Because there were no patients, calculation of the 5-year survival rate and median survival time was not applicable. e Because all of the patients were censored, median survival time could not be calculated. f Recurrence pattern analysis was performed only for patients undergoing R0 resection (ie, groups 1 and 2). 42

6 Table 4. Cox roportional Hazards Analysis of Factors Affecting Survival of Stage IV Gastric Cancer a Variable Stage IV Total Group 1 Group 2 Group 3 Curability R0 RH 1 [Reference] 1 [Reference] 1 [Reference] 1 [Reference] R1 RH (95% CI) 1.88 ( ) 1.92 ( ) 1.92 ( ) value R2 RH (95% CI) 1.98 ( ) 1.91 ( ) 1.17 ( ) value Operation type Subtotal RH 1 [Reference] 1 [Reference] 1 [Reference] 1 [Reference] Total RH (95% CI) 1.10 ( ) 1.14 ( ) 1.06 ( ) value Extended RH (95% CI) 1.65 ( ) 1.97 ( ) 1.48 ( ) value Adjuvant therapy Chemoradiotherapy RH 1 [Reference] 1 [Reference] 1 [Reference] 1 [Reference] Chemotherapy RH (95% CI) 1.42 ( ) 1.31 ( ) 1.60 ( ) 1.17 ( ) value None RH (95% CI) 2.66 ( ) 2.67 ( ) 2.61 ( ) 2.46 ( ) value No. of retrieved lymph nodes 51 RH 1 [Reference] 1 [Reference] 1 [Reference] 1 [Reference] RH (95% CI) 1.20 ( ) value RH (95% CI) 1.74 ( ) value.004 Abbreviations: CI, confidence interval; RH, relative hazard. a values are calculated by means of the Cox proportional hazards model. Groups are described in the atients subsection of the Methods section. adjacent organ resection combined with gastrectomy was performed more frequently in group 1 (54.1%) with curative intent than in groups 2 (30.4%) and 3 (34.3%). The group 1 patients showed locoregional disease spread most frequently (40.9%) as the initial recurrence pattern after an R0 resection; this was followed by peritoneal (27.3%), combined multiple (18.2%), and distant (13.6%) recurrence patterns. The analysis showed that the independent prognostic factors for survival were whether the R0 resection and adjuvant therapy were performed. The relative hazard value of an R1 resection was 1.92, which was significant (=.002). The relative hazard value of an R2 resection was 1.91, but this was not significant, probably owing to the small number of patients. These findings suggest that an aggressive surgical approach for locally advanced primary cancer should be considered to achieve a better outcome and is consistent with other reports that patients with T4 gastric carcinoma benefit from aggressive surgery with curative intent. 9,12,13 However, there is some debate as to the status of lymph node metastasis. 9,12,13 Adjuvant therapy was the strongest prognostic factor in patients with T4 gastric cancer (relative hazard, 2.67 [=.001]), although there was no difference between the chemotherapy and chemoradiotherapy groups. This suggests that, even if T4 gastric cancer shows no distant metastasis during surgery and shows a locoregional aggressive pattern as the initial recurrence, this tumor has already become a systemically advanced disease rather than remaining a local disease. Group 2 patients had disease associated with extensive lymph node metastasis. Because most patients with N3 gastric cancer received the diagnosis after surgery, they had a higher rate of R0 resections (90.7%) than did patients in groups 1 and 3. Tumors of the diffuse type with undifferentiated histological findings or with lymphatic involvement were more common in group 2 patients than in groups 1 and 3, which indicates that these tumors have unfavorable histological characteristics. The patients in group 2 showed distant (30.2%) and peritoneal (26.7%) recurrence more frequently as their initial relapse site, and locoregional recurrence was present in 19.8% of patients, in contrast to group 1. Univariate analysis showed that the prognostic factors for survival were the type of surgery, surgical curability, Borrmann classification, number of retrieved lymph nodes, adjuvant therapy, and recurrence pattern. How- 43

7 ever, the type of surgery, surgical curability, number of retrieved lymph nodes, and adjuvant therapy remained independent prognostic variables after multivariate analysis. An incomplete resection or positive margin status (relative hazard, 1.92) and surgery combined with adjacent organ resection (relative hazard, 1.97) were associated with a less favorable prognosis. The number of lymph nodes retrieved had a prognostic effect in group 2, in contrast to groups 1 and 3. The overall 5-year survival rates were 31.1%, 26.2%, and 7.4% in the patients with more than 51, 31 to 50, and 16 to 30 retrieved lymph nodes, respectively. Considering that the relative hazard ratio was 1.74 (=.004) in patients in whom 30 or fewer lymph nodes were retrieved, the cutoff point for an appropriate lymph node dissection is believed to be more than 30. Adjuvant chemoradiotherapy had better outcomes than did chemotherapy or no adjuvant therapy in patients with N3 gastric cancer. The overall 5-year survival rate was 41.9% in patients who received chemoradiotherapy, 23.3% in patients who received chemotherapy, and 10.2% in patients who underwent surgery alone (.001). This result highlights the role of adjuvant chemoradiotherapy after a gastric resection, consistent with previous reports. 14 A combination of radiotherapy for local control and chemotherapy for systemic control makes combined chemoradiotherapy a logical postoperative adjuvant treatment for patients with gastric cancer. 14 Group 3 included patients with gastric cancer and a distant metastasis, regardless of the TN status. Adjuvant therapy consisting of chemotherapy or chemoradiotherapy had the most important effect on patient prognosis. Although the role of a palliative gastrectomy is still a matter of debate, the median survival time was 14.3 months after gastrectomy and 6 months after a bypass procedure or biopsy only. This result is in agreement with a previous report 15 that found that a primary gastric cancer with a distant metastasis should be resected in cases where it can be resected. However, combined resection was an independent poor prognostic factor in gastric cancer with distant metastasis. This suggests that aggressive surgical treatment for patients with metastatic disease does not offer any survival benefit; rather, it is associated with a poor survival outcome. In this study, the overall 5-year survival rates in groups 1, 2, and 3 were 18.3%, 27.1%, and 9.3%, respectively. atients with T1-3N3M0 gastric cancer showed a much better survival rate than did patients with T4N1-3M0 and T(any)N(any)M1 cancer. The high rate of curative resections and the better response to adjuvant therapy in the T1-3N3M0 gastric cancer group appear to be related to a favorable survival outcome, despite their unfavorable histological features. The survival time for the patients with T4N1-3M0 gastric cancer was also much better than the survival time for patients with M1 gastric cancer. In a recent study, ark et al 16 suggested a subclassification to stages IVA (T1-3N3M0 and T4N1-2M0) and IVB (T4N3M0-1). These authors reported that the survival outcome of the T4N3M0 group was similar to that of the T4N3M1 group because the 5-year survival rates in T1-3N3M0, T4N1-2M0, T4N3M0, and T(any)N(any)M1 groups were 10.5%, 9.7%, 0%, and 1.4%, respectively. However, in our study, the survival curve of the T4N1-2M0 and T4N3M0 groups showed no significant difference, and the T1-3N3M0 group showed a much better survival rate compared with other groups. The subclassification of stage IV gastric cancer is still a matter worthy of consideration. This study was based on an 11-year experience at a single center and demonstrated retrospectively the clinical and pathological characteristics and prognosis of patients with stage IV gastric cancer. However, there was some selection bias because only patients with operable disease at the preoperative evaluations were enrolled in this study. In addition, the patients received different treatments, including operation type and adjuvant therapy, because there is no internationally accepted standard of care. Even if these differences are probably associated with patient prognosis, they appear to represent the clinical characteristics of each stage IV gastric cancer subgroup. Overall, the subclassification of stage IV gastric cancer into groups IVA, IVB, and IVM might offer more useful and detailed information for determining disease characteristics, predicting patient prognosis, and determining therapeutic options. Stage IVA (T1-3N3M0) gastric cancer represents disease with extensive lymph node metastasis, unfavorable histological findings, the need for more radical lymph dissection, a high probability of a curative resection, and the possibility of effective adjuvant therapy and a favorable survival outcome. Stage IVA gastric cancer tends to recur initially as distant and peritoneal disease rather than locoregional disease. Stage IVB (T4N1-3M0) gastric cancer is disease with adjacent organ invasion, requiring R0 resection and adjuvant chemotherapy, and was associated with an intermediate survival outcome. The initial relapse pattern of this tumor is a locoregional recurrence. Stage IVM (T[any]N[any]M1) gastric cancer with distant metastasis has the worst prognosis, and aggressive surgical treatment does not seem to offer any survival benefit, which shifts the treatment option to adjuvant chemotherapy. Therefore, the findings of this study support subclassifying stage IV gastric cancer into stages IVA (T1-3N3M0), IVB (T4N1-3M0), and IVM (T[any]N[any]M1) for more accurate disease identification and treatment decisions. Accepted for ublication: December 15, Correspondence: Tae Sung Sohn, MD, Department of Surgery, Samsung Medical Center, Sungkyunkwan University School of Medicine, 50 Ilwon-dong, Gangnam-gu, Seoul, Korea (ts.sohn@samsung.com). Author Contributions: Study concept and design: An and Sohn. Acquisition of data: An. Analysis and interpretation of data: An, Ha, Noh, Sohn, and Kim. Drafting of the manuscript: An and Noh. Critical revision of the manuscript for important intellectual content: An, Ha, Sohn, and Kim. Statistical analysis: An. Administrative, technical, and material support: An. Study supervision: An, Noh, Sohn, and Kim. Financial Disclosure: None reported. 44

8 REFERENCES 1. Sobin LH, Wittekind CH, eds; International Union Against Cancer. TNM Classification of Malignant Tumors. 6th ed. New York, NY: Wiley-Liss; 2002: Lee JH, Noh SH, Lah KH, Choi SH, Min JS. The prognosis of stage IV gastric carcinoma patients after curative resection. Hepatogastroenterology. 2001; 48(42): Saidi RF, ReMine SG, Dudrick S, Hanna NN. Is there a role for palliative gastrectomy in patients with stage IV gastric cancer? World J Surg. 2006;30(1): Medina-Franco H, Contreras-Saldivar A, Ramos-De La Medina A, alacios- Sanchez, Cortes-Gonzalez R, Ugarte JA. Surgery for stage IV gastric cancer. Am J Surg. 2004;187(4): Yagi Y, Seshimo A, Kameoka S. rognostic factors in stage IV gastric cancer: univariate and multivariate analyses. Gastric Cancer. 2000;3(2): Choi SB, Hong KD, Cho JS, et al. rognostic factors of resected stage IV gastric cancer patients. J Korean Gastric Cancer Assoc. 2006;6(1): D Angelica M, Gonen M, Brennan MF, Turnbull AD, Bains M, Karpeh MS. atterns of initial recurrence in completely resected gastric adenocarcinoma. Ann Surg. 2004;240(5): Lee JH, Noh SH, Choi SH, Min JS. The prognosis of patients with stage IV gastric carcinoma without distant metastasis. J Korean Gastric Cancer Assoc. 2001; 1(2): Saito H, Tsujitani S, Maeda Y, et al. Combined resection of invaded organs in patients with T4 gastric carcinoma. Gastric Cancer. 2001;4(4): Shitara K, Ishiguro A, Munakata M, Wada R, Sakata Y. Retrospective analysis of stage IV advanced gastric cancer treated with S-1 or other chemotherapy. Int J Clin Oncol. 2006;11(5): Lim S, Muhs BE, Marcus SG, Newman E, Berman RS, Hiotis S. Results following resection for stage IV gastric cancer: are better outcomes observed in selected patient subgroups? J Surg Oncol. 2007;95(2): Kitamura K, Tani N, Koike H, et al. Combined resection of the involved organs in T4 gastric cancer. Hepatogastroenterology. 2000;47(36): Kobayashi A, Nakagohri T, Konishi M, et al. Aggressive surgical treatment for T4 gastric cancer. J Gastrointest Surg. 2004;8(4): Lim L, Michael M, Mann GB, Leong T. Adjuvant therapy in gastric cancer. J Clin Oncol. 2005;23(25): Moriwaki Y, Kunisaki C, Kobayashi S, Harada H, Imai S, Kasaoka C. Does the surgical stress associated with palliative resection for patients with incurable gastric cancer with distant metastasis shorten their survival? Hepatogastroenterology. 2004;51(57): ark JM, ark SS, Mok YJ, Kim CS. pn3m0 Gastric cancer: the category that allows the sub-classification of stage-iv gastric cancer (IVA and IVB). Ann Surg Oncol. 2007;14(9): INVITED CRITIQUE T he efficacy of resection in properly selected patients with stage IV gastric cancer is established. An et al convincingly propound subclassification of advanced gastric cancer based on outcomes from a high-volume center. They demonstrate improved outcome for T1-3N3M0 cancers resected with D2 lymphadenectomy (5-year survival, 27.1%) compared with T(any)N(any)M1 (5-year survival, 9.3%) and T4N1-3M0 disease (5-year survival, 18.3%). I agree with the authors, but 27.1% long-term survival does not to my sentiment warrant classification as stage IV disease. I would thus advocate that the more favorable T1-3N3M0 group (group 2) be classified as stage IIIC (not unlike colorectal cancer staging); the T4N1-3M0, stage IVA; and T(any) N(any)M1, stage IVB. Diagnosis of N3 disease can only be made if adequate lymphadenectomy is performed, which is not achieved in up to a quarter of US patients. Although the Medical Research Council 1 and Dutch trials 2,3 fail to show survival benefit for D2 nodal dissection in western populations, clear data have emerged correlating increased total lymph node harvest with prognosis regardless of stage. 4 At The Johns Hopkins University, we favor D1 over or D1 (stations 1-6 appropriate to the level of cancer) plus stations 7 to 9 and 12, leaving stations 10 and 11 with the spleen intact unless gross nodal involvement is found. Subclassification of advanced gastric cancer emphasizes the rationale for more aggressive and optimistic treatment, supporting the role of surgical exploration and resection for patients with nodal disease in the absence of distal disease demonstrated on imaging, with the expectation of better results than historical western controls staged conventionally in aggregate as stage IV would suggest. The surgical public and the American Joint Committee on Cancer in considering the seventh edition of the staging manual may take up the challenge of subclassification; not all stage IV gastric cancer is the same. Mark D. Duncan, MD Correspondence: Dr Duncan, Department of Surgery, The Johns Hopkins University, Johns Hopkins Bayview, 4940 Eastern Ave, Baltimore, MD (mduncan@jhmi.edu). Financial Disclosure: None reported. 1. Cuschieri A, Weeden S, Fielding J, et al; Surgical Co-operative Group. atient survival after D1 and D2 resections for gastric cancer: long-term results of the MRC randomised surgical trial. Br J Cancer. 1999;79(9-10): Bonenkamp JJ, Hermans J, Sasako M, et al; Dutch Gastric Cancer Group. Extended lymph-node dissection for gastric cancer. N Engl J Med. 1999;340 (12): Hartgrink HH, van de Velde CJ, utter H, et al. Extended lymph node dissection for gastric cancer: who may benefit? final results of the randomized Dutch gastric cancer group trial. J Clin Oncol. 2004;22(11): Smith DD, Schwarz RR, Schwarz RE. Impact of total lymph node count on staging and survival after gastrectomy for gastric cancer: data from a large USpopulation database. J Clin Oncol. 2005;23(28):