The seventh TNM staging system for gastric cancer of the American Joint Committee on Cancer/International Union Against Cancer (AJCC/UICC) had a more detailed classification than the sixth TNM staging system for both the tumor (T) and lymph nodes (N). The authors compared survival rates assessed by the seventh staging system with those by the sixth system.
The authors analyzed the prospectively collected database on patients with gastric cancer who underwent surgery at Seoul National University Hospital between 1986 and 2006, and calculated the survival rates of 9998 cases with primary cancer, R0 resection, and >14 retrieved lymph nodes.
The 5-year cumulative survival rates (5YSR) according to the seventh edition T or N classifications were significantly different. The 5YSR according to seventh edition of the TNM staging system were 95.1% (stage IA), 88.4% (stage IB), 84.0% (stage IIA), 71.7% (stage IIB), 58.4% (stage IIIA), 41.3% (stage IIIB), and 26.1% (stage IIIC), which were significantly different from each other. The 5YSR of the seventh edition T2 and T3 classifications had significant differences in patients with every N classification, and the 5YSR of seventh edition N1 and N2 classifications had significant differences in T2 patients, T3 patients, and T4 patients. Each stage in the sixth edition was divided into the seventh edition stage with different survival rates. In addition, the number of homogenous groupings in seventh edition TNM stages was increased from 1 to 2.
The TNM system for gastric cancer of the American Joint Committee on Cancer/International Union Against Cancer Classification (AJCC/UICC) has been revised, and the new system took effect for patients diagnosed on or after January 1, 2010.
The following changes had been adopted by the AJCC Task Force. In the new edition, 1) previous subgroups pT2a (muscularis propria) and pT2b (subserosa) are classified to pT2 (muscularis propria) and pT3 (subserosa) and 2) previous group pN1 (1-6 involved regional lymph nodes [LNs]) is amended into new groups pN1 (1-2 involved regional LNs) and pN2 (3-6 involved regional LNs; and 3) and previous groups pN2 (7-15 involved regional LNs) and pN3 (>15 involved regional LNs) are united into the new pN3 group (>6 involved regional LNs).1-3 These changes were made to predict accurately the different prognosis of tumors invading muscularis propria and those invading subserosa4, 5 and to harmonize the gastric staging systems with the esophageal system for easy use, especially with gastroesophageal junction tumors.
For finalizing TNM staging for gastric cancer, 2 designated members of the International Gastric Cancer Association (T.S. and H.-K.Y.) and other representatives of the UICC were invited to the meeting of the AJCC Task Force held in Buffalo, New York, on August 18, 2008. In making revisions to the TNM, the Task Force considered analyses of gastric cancer outcome data sets from North America, Japan, and Korea, including the Cancer Institute Hospital of the Japanese Federation for Research and the Seoul National University Hospital, Korea. The definitions for T and N and the final stage groupings for the sixth (sixth-TNM) and the seventh (seventh-TNM) editions of the TNM are shown in Figure 1. These revisions were based on review of the 5-year cumulative survival rates (5YSR) according to seventh edition T and N classifications and the sixth TNM classification from the US, Korean, and Japanese databases. In addition to the reclassification related to the T classification, a key change was the aggregation of all patients with >6 positive LNs into a single stage grouping. Overall analysis of these data did not demonstrate a statistically significant cutoff value for any number of positive LNs >6 so that the stage groupings do not include a separate group of those patients with ≥16 LNs. However, a separate N (N3b) classification was preserved for this group to foster continued data collection on this question moving forward.
In this study, we present a comparison of survival rates from the database of the Seoul National University Hospital assessed by the new seventh-TNM and by the sixth-TNM.
MATERIALS AND METHODS
A total of 12,687 patients underwent operation for gastric adenocarcinoma at the Department of Surgery, Seoul National University Hospital (SNUH) from 1986 through 2006 and were entered into the prospective database (Gastric Cancer Registry of the Seoul National University Hospital), which was approved by the Institutional Review Board of SNUH (H-0603-072-170). For survival analysis, we excluded the following: 1) patients with recurrent cancer or stump cancer after undergoing subtotal gastrectomy for gastric cancer; 2) patients with any other cancer detected at surgery; 3) patients who received preoperative chemotherapy; 4) patients with distant metastasis, residual macroscopic or microscopic tumor, or no history of resection; 5) patients with <15 LNs histologically examined; 6) patients with mortality within 30 days after surgery; and 7) patients with missed data regarding the T or N classification.
With regard to surgical method, en bloc resection of the primary tumor and its lymphatic drainage area was routinely performed after the evaluation to define that a patient had no distant metastases as well as the standard resection technique.
The database included items regarding demographics (age and sex), surgical characteristics (surgery and number of retrieved LNs), tumor characteristics (location, histology, Lauren classification, and TNM stage), and follow-up results. The location of the tumor was classified as the lower third, middle third, upper third, or whole stomach, and the histology was categorized as differentiated (papillary, well-differentiated, and moderately differentiated carcinoma) or undifferentiated (poorly differentiated, mucinous adenocarcinoma, and signet ring cell carcinoma). The Lauren classification was established as intestinal, diffuse, or mixed type as a reference.6 The T classification, N classification, and staging of each tumor were classified according to the sixth-TNM1 and the new seventh-TNM.3
Follow-up vital status was obtained from patient hospital records and the National Statistical Office (NSO). In Korea, death reports of all citizens should be sent to NSO. We received survival data from NSO as of December 31, 2007. Survival according to the sixth-TNM stage or seventh-TNM stage was analyzed using the Kaplan-Meier method. Differences in survival were determined by the log-rank test. A P <.050 (2-sided) was regarded as statistically significant. SPSS statistical software (version 12.0; SPSS Inc, Chicago, IL) was used to generate these analysis.
The stepwise process of data extraction from the SNUH gastric cancer registry is depicted in Figure 2. The characteristics of 9998 patients who satisfied the above-mentioned requirements are summarized in Table 1. The mean age of the patients was 55.6 years (standard deviation [SD], 11.7 years), and the male-to-female ratio was 2.1:1. The mean number of retrieved LNs was 32.2 (SD, 13.1 LNs; range, 15-126 LNs). The mean follow-up period was 2600.3 days (SD, 2132.7 days; median, 1862 days [range, 31-8026 days]).
Survival According to T Classification and N Classification
The 5YSR according to the sixth-TNM and seventh-TNM T classifications are shown in Figure 3 (Top and Bottom). According to seventh-TNM T classification, the 5YSR were 94.1% for T1 (n = 3832), 81.6% for T2 (n = 1247), 61.1% for T3 (n = 2368), 42.6% for T4a (n = 2448), and 17.9% for T4b (n = 103), respectively (P <.001).
The 5YSR according to the sixth-TNM and seventh-TNM N classifications are shown in Figure 4 (Top and Bottom). The 5YSR according to the seventh-TNM N classification were 90.0% for N0 (n = 5106), 76.5% for N1 (n = 1277), 58.0% for N2 (n = 1421), and 32.5% for N3 (n = 2194), respectively (P <.001). The 5YSR of tumors with 7 to 15 involved regional LNs (N3a; n = 1442) were significantly different from that of tumors with >15 involved regional LNs (N3b; n = 752) (38.4% vs 20.9%; P <.001).
The 5YSR according to the T and N classifications are shown in Figure 5 (Top and Bottom). The 5YSR for the seventh-TNM T2 and T3 (T2a and T2b according to sixth-TNM) classifications were significantly different in every seventh-TNM N classification patient (P = .031 in N0 patients and P <.001 in N1 patients, N2 patients, and N3 patients, respectively). The 5YSR for the seventh-TNM N1 and N2 (N1 according to sixth-TNM) were significantly different in seventh-TNM T2 patients, T3 patients, and T4a patients (P = .001 in T2 patients and P <.001 in T3 patients and T4a patients, respectively), but not in T1 patients and T4b patients (P = .063 and P = .409, respectively). The 5YSR for the seventh-TNM N3a and N3b classification demonstrated significant differences in T1 patients, T3 patients, and T4a patients (P = .016 in T1 patients, P <.001 in T3 patients, and P = .011 in T4a patients, respectively), but not in T2 patients and T4b patients (P = .063 and P = .978, respectively).
Survival According to the Sixth-TNM and Seventh-TNM and Stage Migration
The number of patients and the 5YSR according to the sixth-TNM and seventh-TNM stage classifications are shown in Figure 6. The 5YSR of 1 of the seventh-TNM stages were significantly different from each other.
The patients with sixth-TNM stage IB (n = 1668) were reclassified to the patients with seventh-TNM stage IB (n = 885) and stage IIA (n = 783), who had significantly different 5YSR (88.6% vs 82.4%; P = .002). The patients with sixth-TNM stage II (n = 1914) were classified to the patients with seventh-TNM stage IIA (n = 261), stage IIB (n = 1104), and stage IIIA (n = 549), who had significantly different 5YSR (88.7%, 71.9%, and 57.7%, respectively; P <.001). The patients with sixth-TNM stage IIIA (n = 1475) were classified to the patients with seventh-TNM stage IIIA (n = 417) and stage IIIB (n = 1058), who had significantly different 5YSR (60.9% vs 44.4%; P <.001). Finally, the patients with sixth-TNM stage IV (n = 810) were divided into the patients with seventh-TNM stage IIB (n = 8), stage IIIA (n = 35), stage IIIB (n = 198), and stage IIIC (n = 569), who had significantly different 5YSRs (35.0%, 40.0%, 24.3%, and 17.7%, respectively; P = .010).
Homogeneity Analysis Among Patients in Each TNM Stage
We compared the survival rates of the T, N, and M classification subgroups in each TNM stage as shown in Figure 5. According to the sixth-TNM T, N, and M classifications, only 1 sixth-TNM stage, stage II, demonstrated homogeneity, in which the survival rates of sixth-TNM T1N2M0, T2N1M0, and T3N0M0 were not significantly different (P = .41). In the sixth-TNM stage IB, stage IIIA, and stage IV, the survival rates of each sixth-TNM T, N, and M category subgroups were significantly different (P = .003, P = .002, and P = .007, respectively). According to the seventh-TNM T, N, and M classifications, all sixth-TNM stages demonstrated no homogeneity. The survival rates of the seventh-TNM T1N1M0, T1N2M0, T2N0M0, and T3N0M0 in sixth-TNM stage IB had significantly different survival rates (P = .001). In addition, the seventh-TNM T1N3M0, T2N1M0, T2N2M0, T3N1M0, T3N2M0, and T4aN0M0 in sixth-TNM stage II; the seventh-TNM T2N3M0, T3N3M0, T4aN1M0, T4aN2M0, and T4bN0M0 in sixth-TNM stage IIIA; and the seventh-TNM T1N3M0, T2N3M0, T3N3M0, T4aN3, T4bN1, T4bN2, and T4bN3 in sixth-TNM stage IV also had significantly different survival rates (P <.001, P <.001, and P = .003, respectively).
With regard to the seventh-TNM stages, seventh-TNM stage IIB and stage IIIA demonstrated homogeneity. The seventh-TNM T, N, and M subgroups of seventh-TNM stage IIB (T1N3M0, T2N2M0, T3N1M0, and T4aN0M0) and stage IIIA (T2N3M0, T3N2M0, and T4aN1M0) demonstrated no significantly different survival rates (P = .247 and P = 551, respectively). However, the subgroups of the seventh-TNM stage IB (T1N1M0 and T2N0M0), stage IIA (T1N2M0, T2N1M0, and T3N0M0), stage IIIB (T3N3M0, T4aN2M0, T4bN0M0, and T4bN1M0), and stage IIIC (T4aN3M0, T4bN2M0, and T4bN3M0) demonstrated significantly different survival rates (P = .013, P = .029, P <.001, and P <.001, respectively).
Ideal cancer staging provides the critical benchmark for defining prognosis. In addition, it provides a framework for deciding on appropriate treatment and the evaluation of results of management and clinical trials, and serves as the standard for local, regional, and international reporting of cancer incidence and outcomes.3, 7 Toward these ends, the fifth edition of the TNM staging system adapted the N classification to be determined by the number of metastatic LNs, rather than by the LN location. After the great change, there were many reports that it was effective and provided better prognostic stratification.8-10 In the sixth-TNM staging system, which had only small changes, muscularis propria tumors and subserosal tumors were amended into T2a and T2b. In the seventh-TNM staging system, 1) distant metastasis was considered as an incurable factor and was separated into stage IV, and 2) more precise classifications such as T2 (muscularis propria)/T3 (subserosa) and N1 (1-2)/N2 (3-6) were adopted.
In our study, the cumulative survival curves according to the seventh-TNM stage were well separated, which represented significantly different survival rates. The study findings also indicated that the 5YSR of the seventh-TNM T2 and T3 had significant differences in all N classifications, and those of the seventh-TNM N1 and N2 had significant difference in T2, T3, and T4 classifications. Each stage of the sixth-TNM was divided into the seventh-TNM stage with different survival rates. In addition, the number of homogenous groupings in the seventh-TNM stages was increased from 1 (sixth-TNM stage II) to 2 (seventh-TNM stage IIB and stage IIIA), although the seventh-TNM T, N, M classifications were more detailed than those of the sixth-TNM classification. All these changes meant that the seventh-TNM staging system could provide more detailed classification than sixth-TNM system. Although the analysis of survival for patients with >6 positive LNs did not demonstrate any cutoff point with a worse survival in the US and Japanese databases, the 5YSR for patients with seventh-TNM N3a tumors were significantly different from those of patients with seventh-TNM N3b tumors in our study. These results make it imperative that the number of positive LNs be recorded in cancer registries, and that the use of the N3b classification be encouraged. This ongoing data collection will allow future research to determine whether tumors with the seventh-TNM N3a and N3b stage groupings have different survival rates, and to analyze to what extent the number of metastatic LNs has prognostic significance in the patients with >6 metastatic LNs.
In addition, although the seventh-TNM staging system provides detailed classification and combines the gastric staging systems with the esophageal system for ease of use, it appeared to be more complex than the sixth-TNM system to apply and memorize. Therefore, the use of appropriate data collection tools and cancer registry tools is encouraged to ensure that accurate data are recorded for future revision.
The seventh-TNM staging system for gastric cancer appeared to provide better categorized grouping than the sixth-TNM, especially between T2 and T3 tumors and N1 and N2 tumors. It demonstrated relatively increased homogeneity in each TNM stage, although the seventh-TNM T, N, and M classifications were more detailed. Furthermore, the 5YSR of the seventh-TNM T, N, and M stage subgroups and TNM stages in this study could be reference data when deciding on proper treatment and the future clinical trials for gastric cancer. However, further studies are needed for the homogeneous grouping and to further understand the value of the N3a and N3b classifications.