Ann Surg Oncol (2015) 22:11–16 DOI 10.1245/s10434-014-4004-z
ORIGINAL ARTICLE – HEALTHCARE POLICY AND OUTCOMES
Incorporating Surgical Quality into the AJCC 7th Edition Improves Staging Accuracy in Gastric Cancer Gary B. Deutsch, MD1, Victoria O’Connor, MD1, Myung-Shin Sim, PhD2, Ji Hey Lee, PhD2, and Anton J. Bilchik, MD, PhD1,3 1
Gastrointestinal Research Program, John Wayne Cancer Institute at Providence Saint John’s Health Center, Santa Monica, CA; 2Department of Biostatistics, John Wayne Cancer Institute at Providence Saint John’s Health Center, Santa Monica, CA; 3California Oncology Research Institute, Los Angeles, CA
ABSTRACT Background. The staging of gastric cancer has become increasingly complex. With an emerging 15-node quality measure and a revised American Joint Committee on Cancer (AJCC) staging system, we evaluated the need for more intricate staging systems to predict survival outcomes in gastric cancer. Methods. The Surveillance, Epidemiology and End Results Program (SEER) database was used to identify 124,972 patients with gastric cancer between 2000 and 2010. Primary endpoints were 5-year disease-specific survival (DSS) and overall survival (OS). Analysis was performed on patients with C15 nodes evaluated. Multivariable regression with/without the inclusion of lymph node (LN) assessment and LN ratio were compared using the Akaike information criterion. Results. The number of patients included in the final analysis was 12,096. The proportion of patients with an adequate lymphadenectomy increased markedly from 27 % in 2000 to 52 % in 2010. Overall 5-year DSS and OS was 61.9 and 48.8 %, respectively, for patients with C15 nodes examined, versus 57.7 and 39.9 %, respectively, for those with \15 sampled nodes (p \ 0.0001). In patients with C15 nodes evaluated, the addition of LN evaluation and LN ratio to the existing staging model improved its ability
This work was presented as an oral presentation at the annual meeting of the Society of Surgical Oncology, Phoenix, AZ, USA, 12–15 March 2014. Ó Society of Surgical Oncology 2014 First Received: 20 April 2014; Published Online: 6 September 2014 A. J. Bilchik, MD, PhD e-mail: [email protected]
to predict 5-year DSS and OS (p \ 0.0001). LN evaluation and LN ratio were comparable in their ability to supplement the existing AJCC 7th edition (AJCC7) staging system. Conclusion. The inclusion of a minimum 15-LN quality measure improves the prognostic ability of the AJCC7 staging system, without adding significant complexity.
The staging of gastric cancer has become increasingly complex without clearly improving prediction of overall survival (OS).1 The most recent edition of the American Joint Committee on Cancer (AJCC) staging system, the AJCC 7th edition (AJCC7), was initiated in 2010 and heralded as a major improvement over previous editions.2 This version offered several modifications, including the division of T1 into T1a (lamina propria) and T1b (submucosa), bi-directional changes to stage groupings (i.e. Stage 1B ? 2A, 3A ? 3B, 4 ? 3B/3C, etc.), and a greater emphasis on node positivity.3 As an example, three positive nodes were formerly considered N1 disease, whereas they are now classified in the N2 category, potentially upstaging a tumor with the same TNM characteristics. This seems to place a greater emphasis on finding additional involved nodes. However, the AJCC7 does not have any minimum lymph node (LN) evaluation requirements and the most recent edition of the Union for International Cancer Control (UICC) staging system has eliminated any requirement for a minimum 15-LN examination.4 It is well established that OS in gastric cancer shows a correlation with the number of LNs examined.5,6 Historically though, gastric cancer LN yields in the US have been quite poor, with a median of less than 10 nodes examined. Several modifications have been proposed in order to more
12
G. B. Deutsch et al.
Percentage of Patients (%)
1000%
n = 4699
n = 6092
n = 8420
72.1%
66.9%
MATERIALS AND METHODS
n = 8253
80% 54.3%
60%
73.3%
1 - 14 LN 15 + LN
40% 20% 26.7%
27.9%
1988 - 1994
1995 - 2000
33.1%
45.7%
0% 2001 - 2005
2006 - 2010
Years of treatment
FIG. 1 Proportion of patients undergoing an adequate lymphadenectomy (C15 LN) has nearly doubled in the last decade. LN lymph nodes
accurately stage these patients, utilizing combinations of the AJCC 6th and 7th editions, or models with LN ratio and negative LN number, amongst others.7–10 While these approaches have their strengths, they complicate an already obscure staging system, making patient discussions about prognosis even more challenging. Additionally, most of these studies are based on international experiences, which may vary significantly from the US. Many of them utilize historical populations of gastric cancer patients, including cases from the 1980 s and 1990 s, when LN evaluation had not yet been discussed as a quality measure or shown to have survival benefit. The long-term results from the Dutch D1D2 gastric cancer trial confirm both a benefit in decreased local recurrence as well as lower gastric cancer-specific mortality for patients who undergo a D2 lymphadenectomy.11 At the same time, increasing numbers of American surgeons have adopted the Japanese approach to gastric cancer surgery, performing a D2 lymphadenectomy for all cases of adenocarcinoma.12 In colon cancer, the American College of Surgeons Commission on Cancer (CoC) published a report in 2005 proposing a minimum of 12 LNs evaluated and pathologically examined for resected colon cancer. Since endorsement by the National Quality Forum in 2007, compliance has grown to more than 75 % of patients.13 Nationwide quality assurance programs abroad have produced similar improvements.14 The overall proportion of patients with an adequate lymphadenectomy (minimum 15LN evaluation) in gastric cancer resections has improved from 27 % in 2000 to 52 % in 2010 (Fig. 1). With the expected introduction of a 15-node quality measure by the CoC, it is likely that nodal yield for gastric cancer resections in the US will follow the same trend as in colon cancer. We evaluated whether the inclusion of a minimum 15LN quality measure improves the prognostic accuracy of the AJCC7 staging system, while limiting any further complexity.
The Surveillance, Epidemiology and End Results (SEER) database was reviewed to identify all patients who underwent surgical resection for gastric adenocarcinoma. Cases with multiple primaries, in situ (Tis) and metastatic disease (M1), and those with inadequate or discordant staging information were excluded. We used extent of disease codes (EOD10) for 1988–2003 and collaborative staging (CS—a unified data collection system) for 2004– 2010 to stage gastric cancers according to the AJCC7. A contemporary cohort from 2001–2010 was selected for the final analysis, in order to better represent the current population of patients being treated for gastric cancer. Prior to selecting the final group for analysis, data from all gastric cancer cases were compiled to evaluate trends in LN evaluation over time. Additionally, demographic information was collected. Primary endpoints were 5-year disease-specific survival (DSS) and OS. Further analyses were carried out on patients with C15 nodes assessed. Three regression models were created, including the following baseline variables: age at diagnosis (continuous), race, primary site, grade, radiation therapy, and the AJCC7 stage (categorical). Model 1 did not have any additional elements, whereas models 2 and 3 included the number of LNs evaluated and LN ratio (LNs positive/LNs evaluated), respectively. Five-year DSS and OS were estimated using Kaplan– Meier survival curves. Univariable and multivariable regression analyses with/without the inclusion of LN assessment and LN ratio were performed and compared with the likelihood ratio test. Goodness of fit was assessed by the Akaike information criterion (AIC). The AIC is a means of establishing the quality of a model for a given set of data by balancing the goodness of fit with its statistical complexity.15 This study was approved by the Institutional Review Board.
RESULTS Of 124,972 patients with gastric cancer, 12,096 were included in the final analysis (2001–2010). Demographic information and tumor/treatment characteristics are shown in Table 1. There was no significant difference in patients treated during the first part of the decade (2001–2005) and the second half (2006–2010). Over half of the patients were men (56.5 %) and over the age of 70 years (54.0 %). More than three-quarters of cases were moderately or poorly differentiated adenocarcinoma (88.6 %), and a little less than half involved the antrum (42.0 %). Median follow-up for the study population was 22 months (23 months for C15 nodes examined, 21 months for\15 nodes examined). Overall 5-year DSS and OS was 61.9 and 48.8 %,
Incorporating Quality—AJCC Gastric Cancer
13
TABLE 1 Profile of 12,096 patients who underwent surgical treatment of gastric adenocarcinoma between 2001 and 2010
TABLE 2 Five-year rates of DSS and OS by stage and number of lymph nodes evaluated
2001–2010 [n (%)]
2001–2005 [n (%)]
2006–2010 [n (%)]
Stage
Female Age (years)
5-year OS
6,832 (56.5)
3,427 (56.5)
3,405 (56.5)
IA
88.0
94.8
67.3
5,264 (43.5)
2,638 (43.5)
2,626 (43.5)
IB
80.1
82.8*
57.0
67.4
IIA
64.8
77.6
44.6
66.7
AJCC 7 \ 15 AJCC 7 C 15 AJCC 7 \ 15 AJCC 7 C 15 nodes (%) nodes (%) nodes (%) nodes (%)
Sex Male
5-year DSS
81.5
B49
1,242 (10.3)
634 (10.5)
608 (10.1)
IIB
47.1
63.7
31.3
49.5
50–59
1,657 (13.7)
778 (12.8)
879 (14.6)
IIIA
34.6
51.6
21.9
36.4
60–69
2,662 (22.0)
1,307 (21.6)
1,355 (22.5)
IIIB
25.1
38.4
16.1
27.9
70–79
3,737 (30.9)
1,929 (31.8)
1,808 (30.0)
IIIC
16.5
23.2
9.3
16.7
C80
2,798 (23.1)
1,417 (23.4)
1,381 (22.9)
Overall 57.7
61.9
39.9
48.8
7,192 (59.5)
3,687 (60.8)
3,505 (58.1)
DSS disease-specific survival, OS overall survival, AJCC American Joint Committee on Cancer
Race White Black
1,974 (16.3)
987 (16.3)
987 (16.4)
Other
2,893 (23.9)
1,378 (22.7)
1,515 (25.1)
Unknown
37 (0.3)
13 (0.2)
24 (0.4)
490 (4.0)
245 (4.0)
245 (4.1)
Body
4,231 (35.0)
2,108 (34.8)
2,123 (35.2)
Antrum Unknown
5,082 (42.0) 2,293 (19.0)
2,534 (41.8) 1,178 (19.4)
2,548 (42.2) 1,115 (18.5)
Primary site Fundus
Grade Well-differentiated
612 (5.1)
293 (4.8)
319 (5.3)
Moderatelydifferentiated
3,174 (26.2)
1,618 (26.7)
1,556 (25.8)
Poorlydifferentiated
7,553 (62.4)
3,775 (62.2)
3,778 (62.6)
267 (2.2)
130 (2.1)
137 (2.3)
490 (4.1)
249 (4.1)
241 (4.0)
Undifferentiated Unknown Tumor size, cm 0–2
2,062 (17.1)
916 (15.1)
1,146 (19.0)
2–4
3,305 (27.3)
1,651 (27.2)
1,654 (27.4)
4–6
2,566 (21.2)
1,253 (20.7)
1,313 (21.8)
6?
2,029 (16.8)
1,130 (18.7)
899 (15.0)
Unknown
2,134 (17.6)
1,115 (18.4)
1,019 (16.9)
Yes
3,507 (29.0)
1,788 (29.5)
1,719 (28.5)
No
8,589 (71.0)
4,277 (70.5)
4,312 (71.5)
Radiation
respectively, for patients with C15 nodes examined, versus 57.7 and 39.9 %, respectively, for those with \15 sampled nodes (p \ 0.0001). The improvement in DSS and OS in the adequate lymphadenectomy group occurred independent of the AJCC7 stage (Table 2) or time period (Fig. 2), although the largest changes were seen in Stage IIB and IIIA. We performed multivariable regression analyses for all staging models. In model 1 (AJCC7), age at diagnosis,
*p = 0.003; all other rows are p \ 0.0001
adjuvant radiation treatment, and AJCC7 staging all significantly impacted 5-year DSS and OS (Table 3). In addition to the above variables, the number of LNs examined and LN ratio, in models 2 and 3, respectively, were also significant. Each additional LN examined (HR 0.978; p \ 0.0001) and radiation treatment given after surgical resection (HR 0.665; p \ 0.0001) resulted in a lower risk of death. Increasing AJCC7 stage (HR 1.474; p \ 0.0001), age at diagnosis (HR 1.024; p \ 0.0001), and LN ratio (HR 3.141; p \ 0.0001) resulted in an incrementally worse outcome. Tumor grade, specifically poorly differentiated (HR 1.765; p = 0.0370) or undifferentiated (HR 2.083; p = 0.0190) tumors, lowered the 5-year DSS compared with well-differentiated tumors. In a direct comparison of the three models, the AJCC7 ? LN evaluation and AJCC7 ? LN ratio regression models were superior to the AJCC7 alone, as demonstrated by lower AICs. For 5-year DSS, the AIC for model 2 was 18,849.9 and for model 3 was 18,829.8 compared with 18,892.5 for model 1. Similarly, for 5-year OS, the AIC for model 2 was 28,257.1 and for model 3 was 28,259.9 compared with 28,326.3 for model 1. The addition of LN evaluation to the existing staging model improved its goodness of fit for both 5-year DSS and OS. LN evaluation had a better fit for predicting 5-year OS, whereas LN ratio was slightly better for 5-year DSS. Overall, the LN evaluation and LN ratio models were comparable. DISCUSSION Based on trends in LN evaluation in the US over the last decade and the most recently available data, increasing numbers of patients are expected to undergo a more
14
G. B. Deutsch et al.
2001 – 2005
0.8 0.6 0.4 0.2
0.6 0.4 0.2
5YR OS = 43.0% 0.0
0.0 0
10
1A 3C
20
30 sv_5yr_month
40
AJCC7_substage 2A 2B
1B
50
60
3A
0
3B
Survival Probability
0.8 0.6 0.4 0.2 0.0 30
40
50
60
1B
AJCC7_substage 2A 2B
AJCC7_substage 2A 2B
1B
50
3A
60
3B
2006 – 2010 Censored
0.4 0.2
5YR OS = 48.9% 0
10
20
sv_5yr_month
1A 3C
40
0.6
0.0 20
30 sv_5yr_month
0.8
5YR OS = 46.8% 10
20
1.0
Censored
0
10
1A 3C
2001 – 2005
1.0
Survival Probability
Censored
0.8
5YR OS = 38.7%
≥ 15 nodes examined
2006 – 2010
1.0
Censored Survival Probability
< 15 nodes examined
Survival Probability
1.0
30
40
50
60
sv_5yr_month
3A
3B
1A 3C
1B
AJCC7_substage 2A 2B
3A
3B
FIG. 2 Superior 5-year OS for the C15 LN cohort maintained over time. Top two panels show 5-year OS over two time periods (2001–2005 and 2006–2010) for patients with \15 LN assessed.
Bottom two panels show 5-year OS over two time periods (2001–2005 and 2006–2010) for patients with C15 LN assessed. AJCC American Joint Committee on Cancer, LN lymph nodes, OS overall survival
extensive lymphadenectomy in the coming years.16 Some have attributed differences in LN yield to disparities in hospital volume and patient age, amongst other variables.17–19 With the introduction of a new quality measure in gastric cancer, it is likely that soon all centers will move to become compliant. Furthermore, having a nationally accepted guideline for LN assessment will not only help improve surgical outcomes, but also quality of pathologic analysis.20 While previous studies have performed similar investigations, the study populations examined had included patients treated during previous decades.21 Fortunately, much has changed over the last several years, including a renewed focus on surgical quality (number and location of LNs removed), more effective adjuvant therapies, and a higher frequency of endoscopic procedures identifying earlier gastric cancers. As a result, we have focused on a more contemporary cohort of resectable gastric cancer (2001–2010) in order to better represent the current population being treated for gastric cancer. Recent publications
have also argued for simplifying the AJCC7 and have done so by selecting a more modern patient cohort for analysis.22 We believe that having an adequate LN evaluation may in fact represent a surrogate marker for the oncologic quality of an operation.23 Increasing LN counts not only correlate with a superior survival outcome5 but may also be a potential marker for completeness of surgery.24 Performing a D2 dissection has been shown to improve DSS and locoregional control in the long-term, both in Eastern and Western experiences. Our analysis demonstrates a significant survival benefit with the incorporation of an LN quality measure into the current staging system. Resected cases with a minimum yield of 15 LNs had an improved 5-year DSS and OS. The difference was most pronounced in Stage IIB and IIIA, with at least a 15 % absolute improvement in 5-year DSS and OS in patients who had an adequate lymphadenectomy. This may in part be related to upstaging, since finding even one more positive node increases stage. More adequate staging allows patients to receive more appropriate adjuvant therapy. However, it is
Incorporating Quality—AJCC Gastric Cancer
15
TABLE 3 Multivariable analysis for model 1 (AJCC7 alone) Parameter
Age at diagnosis
5-year OS
5-year DSS
HR
95 % HR CL
p-Value
HR
95 % HR CL
p-Value
1.025
1.021–1.028
\0.0001
1.013
1.008–1.017
\0.0001
0.751
0.589–0.958
NS
0.864
0.636–1.173
NS
Primary site (ref ? fundus) Body Antrum
0.741
0.596–0.921
0.861
0.654–1.135
Unknown
0.866
0.677–1.109
0.953
0.697–1.303
Moderately differentiated Poorly differentiated
0.763 0.951
0.562–1.037 0.706–1.281
1.389 1.771
0.808–2.390 1.039–3.019
NS 0.0358
Undifferentiated
1.031
0.693–1.534
2.150
1.165–3.969
0.0144
Radiation after surgery
0.664
0.598–0.736
\0.0001
0.699
0.619–0.790
\0.0001
Radiation before and after surgery
1.228
0.548–2.753
NS
1.579
0.653–3.817
NS
Radiation prior to surgery
0.900
0.496–1.632
NS
1.022
0.528–1.977
NS
Grade (ref ? well differentiated) NS
Radiation (ref ? no radiation)
Sequence unknown AJCC 7th stage
2.687
0.670–10.780
NS
3.455
0.854–13.973
NS
1.470
1.429–1.512
\0.0001
1.600
1.543–1.660
\0.0001
AJCC American Joint Committee on Cancer, OS overall survival, DSS disease-specific survival, HR hazard ratio, CL confidence limits, NS nonsignificant
also possible these patients received ‘better’ operations that more sufficiently cleared their locoregional disease. Studies comparing the current AJCC/UICC staging system with previous editions have also utilized the AIC statistic to determine if the modifications were justified. The AJCC/UICC 7th editions showed a statistical, but not clinically significant, improvement in predicting survival outcomes, at the expense of significantly higher complexity.1 While several authors have proposed modifications to the most recent AJCC7 staging system with the inclusion of LN ratio or number of negative LNs, there may be limited clinical value to further complicating an already complex system.8, 9 Furthermore, when specifically evaluating the most recent changes in N staging, the AJCC/ UICC 7th edition has failed to significantly improve the prognostic relevance of LN status, even with the inclusion of LN ratio.25 Our study demonstrates that incorporating a 15-LN quality measure influences survival without overcomplicating the existing staging system. Our study is not without limitations, as our results are based on retrospective data from the SEER database. We were unable to control for any discrepancies in confounding variables or selection bias that may exist between the patients who had an adequate lymphadenectomy and those who did not. Additionally, the SEER database has several known limitations and biases.26 There is no information available on recurrences, an important outcome in the assessment of surgical treatment for various cancers, specifically gastric cancer. All results and conclusions are based on the coding accuracy of those inputting the data,
which can potentially be less reliable in less common histologies, such as neuroendocrine tumors. Furthermore, specific information regarding the performance of a D1 versus D2 dissection is not available. We believe that C15 LNs evaluated is a good surrogate for the performance of a more oncologically sound operation. The utilization of adjuvant chemotherapy is another unknown in our SEER population, which may have had an impact on survival outcomes. The so-called ‘stage migration’ effect has been hypothesized as the reason patients with an adequate lymphadenectomy have better survival outcomes. By incorporating a 15-LN quality measure into the current staging system, patients are more likely to be appropriately staged, improving the selection of adjuvant treatment. In our study, knowing the proportion of patients who went on to receive further therapy, and controlling for it, would strengthen our conclusions. Future studies analyzing other national databases that include information about chemotherapy would add valuable knowledge to management decisions. With continued advancements in personalized medicine for patients with gastric cancer, future efforts to predict outcomes will incorporate various molecular and immunologic markers. Simpler staging systems based on cancer biology will likely replace the current schemes that are dependent solely on the anatomic extent of cancer.27 However, we believe quality measures will continue to be valuable, as a means of assuring adequacy and standardization of surgical resection, independent of any other potential influences.
16
G. B. Deutsch et al.
CONCLUSION In the current era of cancer quality measures and the adoption of a 15-LN minimum guideline in gastric cancer, significantly more complex staging schemes will become unnecessary. The inclusion of an LN quality measure into the AJCC7 will serve to standardize important elements of gastric cancer surgery and thereby improve staging accuracy. ACKNOWLEDGMENT Supported by funding from the California Oncology Research Institute, Los Angeles, CA, USA, and by fellowship funding from the Harold McAlister Charitable Foundation, Los Angeles, CA, USA (Dr. O’Connor). DISCLOSURE Gary B. Deutsch, Victoria O’Connor, Myung-Shin Sim, Ji Hey Lee, and Anton J. Bilchik have no conflicts of interest to disclose.
11.
12. 13.
14.
15.
16.
17.
REFERENCES 1. Reim D, Loos M, Vogl F, et al. Prognostic implications of the seventh edition of the International Union Against Cancer classification for patients with gastric cancer: the Western experience of patients treated in a single-center European institution. J Clin Oncol. 2013;31(2):263–71. 2. McGhan LJ, Pockaj BA, Gray RJ, Bagaria SP, Wasif N. Validation of the updated 7th edition AJCC TNM staging criteria for gastric adenocarcinoma. J Gastrointest Surg. 2012;16(1):53–61; discussion 61. 3. Washington K. 7th edition of the AJCC cancer staging manual: stomach. Ann Surg Oncol. 2010;17(12):3077–9. 4. Kwon SJ. Evaluation of the 7th UICC TNM staging system of gastric cancer. J Gastric Cancer. 2011;11(2):78–85. 5. 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 US-population database. J Clin Oncol. 2005;23(28):7114–24. 6. Schwarz RE, Smith DD. Clinical impact of lymphadenectomy extent in resectable gastric cancer of advanced stage. Ann Surg Oncol. 2007;14(2):317–28. 7. Wang J, Dang P, Raut CP, et al. Comparison of a lymph node ratio–based staging system with the 7th AJCC system for gastric cancer. Ann Surg. 2012;255:478–85. 8. Wang W, Xu DZ, Li YF, et al. Tumor-ratio-metastasis staging system as an alternative to the 7th edition UICC TNM system in gastric cancer after D2 resection–results of a single-institution study of 1343 Chinese patients. Ann Oncol. 2011;22(9):2049–56. 9. Deng J, Zhang R, Zhang L, Liu Y, Hao X, Liang H. Negative node count improvement prognostic prediction of the seventh edition of the TNM classification for gastric cancer. PLoS One. 2013;8(11):e80082. 10. Zhang J, Niu Z, Zhou Y, Cao S. A comparison between the seventh and sixth editions of the American Joint Committee on
18.
19.
20.
21.
22.
23.
24. 25.
26.
27.
Cancer/International Union Against classification of gastric cancer. Ann Surg. 2013;257(1):81–6. Songun I, Putter H, Kranenbarg EM-K, Sasako M, van de Velde CJH. Surgical treatment of gastric cancer: 15-year follow-up results of the randomised nationwide Dutch D1D2 trial. Lancet Oncol. 2010;11:439–49. Strong VE, Yoon SS. Extended lymphadenectomy in gastric cancer is debatable. World J Surg. 2013;37(8):1773–7. Parsons HM, Begun JW, Kuntz KM, Tuttle TM, McGovern PM, Virnig BA. Lymph node evaluation for colon cancer in an era of quality guidelines: who improves? J Oncol Pract. 2013;9(4): e164–71. Dikken JL, Cats A, Verheij M, van de Velde CJH. Randomized trials and quality assurance in gastric cancer surgery. J Surg Oncol. 2013;107(3):298–305. Yoon HM, Ryu KW, Nam BH, et al. Is the new seventh AJCC/ UICC staging system appropriate for patients with gastric cancer? J Am Coll Surg. 2012;214(1):88–96. Dubecz A, Solymosi N, Schweigert M, et al. Time trends and disparities in lymphadenectomy for gastrointestinal cancer in the United States: a population-based analysis of 326,243 patients. J Gastrointest Surg. 2013;17(4):611–8; discussion 618-9. Enzinger PC, Benedetti JK, Meyerhardt JA, et al. Impact of hospital volume on recurrence and survival after surgery for gastric cancer. Ann Surg. 2007;245(3):426–34. Dudeja V, Gay G, Habermann EB, et al. Do hospital attributes predict guideline-recommended gastric cancer care in the United States? Ann Surg Oncol. 2012;19(2):365–72. Dudeja V, Habermann EB, Zhong W, et al. Guideline recommended gastric cancer care in the elderly: insights into the applicability of cancer trials to real world. Ann Surg Oncol. 2011;18(1):26–33. Qureshi AP, Ottensmeyer CA, Mahar AL, et al. Quality indicators for gastric cancer surgery: a survey of practicing pathologists in Ontario. Ann Surg Oncol. 2009;16(7):1883–9. Smith DD, Nelson RA, Schwarz RE. A comparison of five competing lymph node staging schemes in a cohort of resectable gastric cancer patients. Ann Surg Oncol. 2014;21(3):875–82. Patel MI, Rhoads KF, Ma Y, et al. Seventh edition (2010) of the AJCC/UICC staging system for gastric adenocarcinoma: is there room for improvement? Ann Surg Oncol. 2013;20(5):1631–8. Kong S-H, Lee H-J, Ahn HS, et al. Stage migration effect on survival in gastric cancer surgery with extended lymphadenectomy: the reappraisal of positive lymph node ratio as a proper N-staging. Ann Surg. 2012;255(1):50–8. Gilbert SM, Hollenbeck BK. Limitations of lymph node counts as a measure of therapy. J Natl Compr Canc Netw. 2009;7(1):58–61. Rausei S, Dionigi G, Ruspi L, et al. Lymph node staging in gastric cancer: new criteria, old problems. Int J Surg. 2013;11 Suppl 1:S90–4. Park HS, Lloyd S, Decker RH, Wilson LD, Yu JB. Limitations and biases of the Surveillance, Epidemiology, and End Results database. Curr Probl Cancer. 2012;36(4):216–24. Edge SB, Compton CC. The American Joint Committee on Cancer: the 7th edition of the AJCC cancer staging manual and the future of TNM. Ann Surg Oncol. 2010;17(6):1471–4.
ORIGINAL ARTICLE – HEALTHCARE POLICY AND OUTCOMES
Incorporating Surgical Quality into the AJCC 7th Edition Improves Staging Accuracy in Gastric Cancer Gary B. Deutsch, MD1, Victoria O’Connor, MD1, Myung-Shin Sim, PhD2, Ji Hey Lee, PhD2, and Anton J. Bilchik, MD, PhD1,3 1
Gastrointestinal Research Program, John Wayne Cancer Institute at Providence Saint John’s Health Center, Santa Monica, CA; 2Department of Biostatistics, John Wayne Cancer Institute at Providence Saint John’s Health Center, Santa Monica, CA; 3California Oncology Research Institute, Los Angeles, CA
ABSTRACT Background. The staging of gastric cancer has become increasingly complex. With an emerging 15-node quality measure and a revised American Joint Committee on Cancer (AJCC) staging system, we evaluated the need for more intricate staging systems to predict survival outcomes in gastric cancer. Methods. The Surveillance, Epidemiology and End Results Program (SEER) database was used to identify 124,972 patients with gastric cancer between 2000 and 2010. Primary endpoints were 5-year disease-specific survival (DSS) and overall survival (OS). Analysis was performed on patients with C15 nodes evaluated. Multivariable regression with/without the inclusion of lymph node (LN) assessment and LN ratio were compared using the Akaike information criterion. Results. The number of patients included in the final analysis was 12,096. The proportion of patients with an adequate lymphadenectomy increased markedly from 27 % in 2000 to 52 % in 2010. Overall 5-year DSS and OS was 61.9 and 48.8 %, respectively, for patients with C15 nodes examined, versus 57.7 and 39.9 %, respectively, for those with \15 sampled nodes (p \ 0.0001). In patients with C15 nodes evaluated, the addition of LN evaluation and LN ratio to the existing staging model improved its ability
This work was presented as an oral presentation at the annual meeting of the Society of Surgical Oncology, Phoenix, AZ, USA, 12–15 March 2014. Ó Society of Surgical Oncology 2014 First Received: 20 April 2014; Published Online: 6 September 2014 A. J. Bilchik, MD, PhD e-mail: [email protected]
to predict 5-year DSS and OS (p \ 0.0001). LN evaluation and LN ratio were comparable in their ability to supplement the existing AJCC 7th edition (AJCC7) staging system. Conclusion. The inclusion of a minimum 15-LN quality measure improves the prognostic ability of the AJCC7 staging system, without adding significant complexity.
The staging of gastric cancer has become increasingly complex without clearly improving prediction of overall survival (OS).1 The most recent edition of the American Joint Committee on Cancer (AJCC) staging system, the AJCC 7th edition (AJCC7), was initiated in 2010 and heralded as a major improvement over previous editions.2 This version offered several modifications, including the division of T1 into T1a (lamina propria) and T1b (submucosa), bi-directional changes to stage groupings (i.e. Stage 1B ? 2A, 3A ? 3B, 4 ? 3B/3C, etc.), and a greater emphasis on node positivity.3 As an example, three positive nodes were formerly considered N1 disease, whereas they are now classified in the N2 category, potentially upstaging a tumor with the same TNM characteristics. This seems to place a greater emphasis on finding additional involved nodes. However, the AJCC7 does not have any minimum lymph node (LN) evaluation requirements and the most recent edition of the Union for International Cancer Control (UICC) staging system has eliminated any requirement for a minimum 15-LN examination.4 It is well established that OS in gastric cancer shows a correlation with the number of LNs examined.5,6 Historically though, gastric cancer LN yields in the US have been quite poor, with a median of less than 10 nodes examined. Several modifications have been proposed in order to more
12
G. B. Deutsch et al.
Percentage of Patients (%)
1000%
n = 4699
n = 6092
n = 8420
72.1%
66.9%
MATERIALS AND METHODS
n = 8253
80% 54.3%
60%
73.3%
1 - 14 LN 15 + LN
40% 20% 26.7%
27.9%
1988 - 1994
1995 - 2000
33.1%
45.7%
0% 2001 - 2005
2006 - 2010
Years of treatment
FIG. 1 Proportion of patients undergoing an adequate lymphadenectomy (C15 LN) has nearly doubled in the last decade. LN lymph nodes
accurately stage these patients, utilizing combinations of the AJCC 6th and 7th editions, or models with LN ratio and negative LN number, amongst others.7–10 While these approaches have their strengths, they complicate an already obscure staging system, making patient discussions about prognosis even more challenging. Additionally, most of these studies are based on international experiences, which may vary significantly from the US. Many of them utilize historical populations of gastric cancer patients, including cases from the 1980 s and 1990 s, when LN evaluation had not yet been discussed as a quality measure or shown to have survival benefit. The long-term results from the Dutch D1D2 gastric cancer trial confirm both a benefit in decreased local recurrence as well as lower gastric cancer-specific mortality for patients who undergo a D2 lymphadenectomy.11 At the same time, increasing numbers of American surgeons have adopted the Japanese approach to gastric cancer surgery, performing a D2 lymphadenectomy for all cases of adenocarcinoma.12 In colon cancer, the American College of Surgeons Commission on Cancer (CoC) published a report in 2005 proposing a minimum of 12 LNs evaluated and pathologically examined for resected colon cancer. Since endorsement by the National Quality Forum in 2007, compliance has grown to more than 75 % of patients.13 Nationwide quality assurance programs abroad have produced similar improvements.14 The overall proportion of patients with an adequate lymphadenectomy (minimum 15LN evaluation) in gastric cancer resections has improved from 27 % in 2000 to 52 % in 2010 (Fig. 1). With the expected introduction of a 15-node quality measure by the CoC, it is likely that nodal yield for gastric cancer resections in the US will follow the same trend as in colon cancer. We evaluated whether the inclusion of a minimum 15LN quality measure improves the prognostic accuracy of the AJCC7 staging system, while limiting any further complexity.
The Surveillance, Epidemiology and End Results (SEER) database was reviewed to identify all patients who underwent surgical resection for gastric adenocarcinoma. Cases with multiple primaries, in situ (Tis) and metastatic disease (M1), and those with inadequate or discordant staging information were excluded. We used extent of disease codes (EOD10) for 1988–2003 and collaborative staging (CS—a unified data collection system) for 2004– 2010 to stage gastric cancers according to the AJCC7. A contemporary cohort from 2001–2010 was selected for the final analysis, in order to better represent the current population of patients being treated for gastric cancer. Prior to selecting the final group for analysis, data from all gastric cancer cases were compiled to evaluate trends in LN evaluation over time. Additionally, demographic information was collected. Primary endpoints were 5-year disease-specific survival (DSS) and OS. Further analyses were carried out on patients with C15 nodes assessed. Three regression models were created, including the following baseline variables: age at diagnosis (continuous), race, primary site, grade, radiation therapy, and the AJCC7 stage (categorical). Model 1 did not have any additional elements, whereas models 2 and 3 included the number of LNs evaluated and LN ratio (LNs positive/LNs evaluated), respectively. Five-year DSS and OS were estimated using Kaplan– Meier survival curves. Univariable and multivariable regression analyses with/without the inclusion of LN assessment and LN ratio were performed and compared with the likelihood ratio test. Goodness of fit was assessed by the Akaike information criterion (AIC). The AIC is a means of establishing the quality of a model for a given set of data by balancing the goodness of fit with its statistical complexity.15 This study was approved by the Institutional Review Board.
RESULTS Of 124,972 patients with gastric cancer, 12,096 were included in the final analysis (2001–2010). Demographic information and tumor/treatment characteristics are shown in Table 1. There was no significant difference in patients treated during the first part of the decade (2001–2005) and the second half (2006–2010). Over half of the patients were men (56.5 %) and over the age of 70 years (54.0 %). More than three-quarters of cases were moderately or poorly differentiated adenocarcinoma (88.6 %), and a little less than half involved the antrum (42.0 %). Median follow-up for the study population was 22 months (23 months for C15 nodes examined, 21 months for\15 nodes examined). Overall 5-year DSS and OS was 61.9 and 48.8 %,
Incorporating Quality—AJCC Gastric Cancer
13
TABLE 1 Profile of 12,096 patients who underwent surgical treatment of gastric adenocarcinoma between 2001 and 2010
TABLE 2 Five-year rates of DSS and OS by stage and number of lymph nodes evaluated
2001–2010 [n (%)]
2001–2005 [n (%)]
2006–2010 [n (%)]
Stage
Female Age (years)
5-year OS
6,832 (56.5)
3,427 (56.5)
3,405 (56.5)
IA
88.0
94.8
67.3
5,264 (43.5)
2,638 (43.5)
2,626 (43.5)
IB
80.1
82.8*
57.0
67.4
IIA
64.8
77.6
44.6
66.7
AJCC 7 \ 15 AJCC 7 C 15 AJCC 7 \ 15 AJCC 7 C 15 nodes (%) nodes (%) nodes (%) nodes (%)
Sex Male
5-year DSS
81.5
B49
1,242 (10.3)
634 (10.5)
608 (10.1)
IIB
47.1
63.7
31.3
49.5
50–59
1,657 (13.7)
778 (12.8)
879 (14.6)
IIIA
34.6
51.6
21.9
36.4
60–69
2,662 (22.0)
1,307 (21.6)
1,355 (22.5)
IIIB
25.1
38.4
16.1
27.9
70–79
3,737 (30.9)
1,929 (31.8)
1,808 (30.0)
IIIC
16.5
23.2
9.3
16.7
C80
2,798 (23.1)
1,417 (23.4)
1,381 (22.9)
Overall 57.7
61.9
39.9
48.8
7,192 (59.5)
3,687 (60.8)
3,505 (58.1)
DSS disease-specific survival, OS overall survival, AJCC American Joint Committee on Cancer
Race White Black
1,974 (16.3)
987 (16.3)
987 (16.4)
Other
2,893 (23.9)
1,378 (22.7)
1,515 (25.1)
Unknown
37 (0.3)
13 (0.2)
24 (0.4)
490 (4.0)
245 (4.0)
245 (4.1)
Body
4,231 (35.0)
2,108 (34.8)
2,123 (35.2)
Antrum Unknown
5,082 (42.0) 2,293 (19.0)
2,534 (41.8) 1,178 (19.4)
2,548 (42.2) 1,115 (18.5)
Primary site Fundus
Grade Well-differentiated
612 (5.1)
293 (4.8)
319 (5.3)
Moderatelydifferentiated
3,174 (26.2)
1,618 (26.7)
1,556 (25.8)
Poorlydifferentiated
7,553 (62.4)
3,775 (62.2)
3,778 (62.6)
267 (2.2)
130 (2.1)
137 (2.3)
490 (4.1)
249 (4.1)
241 (4.0)
Undifferentiated Unknown Tumor size, cm 0–2
2,062 (17.1)
916 (15.1)
1,146 (19.0)
2–4
3,305 (27.3)
1,651 (27.2)
1,654 (27.4)
4–6
2,566 (21.2)
1,253 (20.7)
1,313 (21.8)
6?
2,029 (16.8)
1,130 (18.7)
899 (15.0)
Unknown
2,134 (17.6)
1,115 (18.4)
1,019 (16.9)
Yes
3,507 (29.0)
1,788 (29.5)
1,719 (28.5)
No
8,589 (71.0)
4,277 (70.5)
4,312 (71.5)
Radiation
respectively, for patients with C15 nodes examined, versus 57.7 and 39.9 %, respectively, for those with \15 sampled nodes (p \ 0.0001). The improvement in DSS and OS in the adequate lymphadenectomy group occurred independent of the AJCC7 stage (Table 2) or time period (Fig. 2), although the largest changes were seen in Stage IIB and IIIA. We performed multivariable regression analyses for all staging models. In model 1 (AJCC7), age at diagnosis,
*p = 0.003; all other rows are p \ 0.0001
adjuvant radiation treatment, and AJCC7 staging all significantly impacted 5-year DSS and OS (Table 3). In addition to the above variables, the number of LNs examined and LN ratio, in models 2 and 3, respectively, were also significant. Each additional LN examined (HR 0.978; p \ 0.0001) and radiation treatment given after surgical resection (HR 0.665; p \ 0.0001) resulted in a lower risk of death. Increasing AJCC7 stage (HR 1.474; p \ 0.0001), age at diagnosis (HR 1.024; p \ 0.0001), and LN ratio (HR 3.141; p \ 0.0001) resulted in an incrementally worse outcome. Tumor grade, specifically poorly differentiated (HR 1.765; p = 0.0370) or undifferentiated (HR 2.083; p = 0.0190) tumors, lowered the 5-year DSS compared with well-differentiated tumors. In a direct comparison of the three models, the AJCC7 ? LN evaluation and AJCC7 ? LN ratio regression models were superior to the AJCC7 alone, as demonstrated by lower AICs. For 5-year DSS, the AIC for model 2 was 18,849.9 and for model 3 was 18,829.8 compared with 18,892.5 for model 1. Similarly, for 5-year OS, the AIC for model 2 was 28,257.1 and for model 3 was 28,259.9 compared with 28,326.3 for model 1. The addition of LN evaluation to the existing staging model improved its goodness of fit for both 5-year DSS and OS. LN evaluation had a better fit for predicting 5-year OS, whereas LN ratio was slightly better for 5-year DSS. Overall, the LN evaluation and LN ratio models were comparable. DISCUSSION Based on trends in LN evaluation in the US over the last decade and the most recently available data, increasing numbers of patients are expected to undergo a more
14
G. B. Deutsch et al.
2001 – 2005
0.8 0.6 0.4 0.2
0.6 0.4 0.2
5YR OS = 43.0% 0.0
0.0 0
10
1A 3C
20
30 sv_5yr_month
40
AJCC7_substage 2A 2B
1B
50
60
3A
0
3B
Survival Probability
0.8 0.6 0.4 0.2 0.0 30
40
50
60
1B
AJCC7_substage 2A 2B
AJCC7_substage 2A 2B
1B
50
3A
60
3B
2006 – 2010 Censored
0.4 0.2
5YR OS = 48.9% 0
10
20
sv_5yr_month
1A 3C
40
0.6
0.0 20
30 sv_5yr_month
0.8
5YR OS = 46.8% 10
20
1.0
Censored
0
10
1A 3C
2001 – 2005
1.0
Survival Probability
Censored
0.8
5YR OS = 38.7%
≥ 15 nodes examined
2006 – 2010
1.0
Censored Survival Probability
< 15 nodes examined
Survival Probability
1.0
30
40
50
60
sv_5yr_month
3A
3B
1A 3C
1B
AJCC7_substage 2A 2B
3A
3B
FIG. 2 Superior 5-year OS for the C15 LN cohort maintained over time. Top two panels show 5-year OS over two time periods (2001–2005 and 2006–2010) for patients with \15 LN assessed.
Bottom two panels show 5-year OS over two time periods (2001–2005 and 2006–2010) for patients with C15 LN assessed. AJCC American Joint Committee on Cancer, LN lymph nodes, OS overall survival
extensive lymphadenectomy in the coming years.16 Some have attributed differences in LN yield to disparities in hospital volume and patient age, amongst other variables.17–19 With the introduction of a new quality measure in gastric cancer, it is likely that soon all centers will move to become compliant. Furthermore, having a nationally accepted guideline for LN assessment will not only help improve surgical outcomes, but also quality of pathologic analysis.20 While previous studies have performed similar investigations, the study populations examined had included patients treated during previous decades.21 Fortunately, much has changed over the last several years, including a renewed focus on surgical quality (number and location of LNs removed), more effective adjuvant therapies, and a higher frequency of endoscopic procedures identifying earlier gastric cancers. As a result, we have focused on a more contemporary cohort of resectable gastric cancer (2001–2010) in order to better represent the current population being treated for gastric cancer. Recent publications
have also argued for simplifying the AJCC7 and have done so by selecting a more modern patient cohort for analysis.22 We believe that having an adequate LN evaluation may in fact represent a surrogate marker for the oncologic quality of an operation.23 Increasing LN counts not only correlate with a superior survival outcome5 but may also be a potential marker for completeness of surgery.24 Performing a D2 dissection has been shown to improve DSS and locoregional control in the long-term, both in Eastern and Western experiences. Our analysis demonstrates a significant survival benefit with the incorporation of an LN quality measure into the current staging system. Resected cases with a minimum yield of 15 LNs had an improved 5-year DSS and OS. The difference was most pronounced in Stage IIB and IIIA, with at least a 15 % absolute improvement in 5-year DSS and OS in patients who had an adequate lymphadenectomy. This may in part be related to upstaging, since finding even one more positive node increases stage. More adequate staging allows patients to receive more appropriate adjuvant therapy. However, it is
Incorporating Quality—AJCC Gastric Cancer
15
TABLE 3 Multivariable analysis for model 1 (AJCC7 alone) Parameter
Age at diagnosis
5-year OS
5-year DSS
HR
95 % HR CL
p-Value
HR
95 % HR CL
p-Value
1.025
1.021–1.028
\0.0001
1.013
1.008–1.017
\0.0001
0.751
0.589–0.958
NS
0.864
0.636–1.173
NS
Primary site (ref ? fundus) Body Antrum
0.741
0.596–0.921
0.861
0.654–1.135
Unknown
0.866
0.677–1.109
0.953
0.697–1.303
Moderately differentiated Poorly differentiated
0.763 0.951
0.562–1.037 0.706–1.281
1.389 1.771
0.808–2.390 1.039–3.019
NS 0.0358
Undifferentiated
1.031
0.693–1.534
2.150
1.165–3.969
0.0144
Radiation after surgery
0.664
0.598–0.736
\0.0001
0.699
0.619–0.790
\0.0001
Radiation before and after surgery
1.228
0.548–2.753
NS
1.579
0.653–3.817
NS
Radiation prior to surgery
0.900
0.496–1.632
NS
1.022
0.528–1.977
NS
Grade (ref ? well differentiated) NS
Radiation (ref ? no radiation)
Sequence unknown AJCC 7th stage
2.687
0.670–10.780
NS
3.455
0.854–13.973
NS
1.470
1.429–1.512
\0.0001
1.600
1.543–1.660
\0.0001
AJCC American Joint Committee on Cancer, OS overall survival, DSS disease-specific survival, HR hazard ratio, CL confidence limits, NS nonsignificant
also possible these patients received ‘better’ operations that more sufficiently cleared their locoregional disease. Studies comparing the current AJCC/UICC staging system with previous editions have also utilized the AIC statistic to determine if the modifications were justified. The AJCC/UICC 7th editions showed a statistical, but not clinically significant, improvement in predicting survival outcomes, at the expense of significantly higher complexity.1 While several authors have proposed modifications to the most recent AJCC7 staging system with the inclusion of LN ratio or number of negative LNs, there may be limited clinical value to further complicating an already complex system.8, 9 Furthermore, when specifically evaluating the most recent changes in N staging, the AJCC/ UICC 7th edition has failed to significantly improve the prognostic relevance of LN status, even with the inclusion of LN ratio.25 Our study demonstrates that incorporating a 15-LN quality measure influences survival without overcomplicating the existing staging system. Our study is not without limitations, as our results are based on retrospective data from the SEER database. We were unable to control for any discrepancies in confounding variables or selection bias that may exist between the patients who had an adequate lymphadenectomy and those who did not. Additionally, the SEER database has several known limitations and biases.26 There is no information available on recurrences, an important outcome in the assessment of surgical treatment for various cancers, specifically gastric cancer. All results and conclusions are based on the coding accuracy of those inputting the data,
which can potentially be less reliable in less common histologies, such as neuroendocrine tumors. Furthermore, specific information regarding the performance of a D1 versus D2 dissection is not available. We believe that C15 LNs evaluated is a good surrogate for the performance of a more oncologically sound operation. The utilization of adjuvant chemotherapy is another unknown in our SEER population, which may have had an impact on survival outcomes. The so-called ‘stage migration’ effect has been hypothesized as the reason patients with an adequate lymphadenectomy have better survival outcomes. By incorporating a 15-LN quality measure into the current staging system, patients are more likely to be appropriately staged, improving the selection of adjuvant treatment. In our study, knowing the proportion of patients who went on to receive further therapy, and controlling for it, would strengthen our conclusions. Future studies analyzing other national databases that include information about chemotherapy would add valuable knowledge to management decisions. With continued advancements in personalized medicine for patients with gastric cancer, future efforts to predict outcomes will incorporate various molecular and immunologic markers. Simpler staging systems based on cancer biology will likely replace the current schemes that are dependent solely on the anatomic extent of cancer.27 However, we believe quality measures will continue to be valuable, as a means of assuring adequacy and standardization of surgical resection, independent of any other potential influences.
16
G. B. Deutsch et al.
CONCLUSION In the current era of cancer quality measures and the adoption of a 15-LN minimum guideline in gastric cancer, significantly more complex staging schemes will become unnecessary. The inclusion of an LN quality measure into the AJCC7 will serve to standardize important elements of gastric cancer surgery and thereby improve staging accuracy. ACKNOWLEDGMENT Supported by funding from the California Oncology Research Institute, Los Angeles, CA, USA, and by fellowship funding from the Harold McAlister Charitable Foundation, Los Angeles, CA, USA (Dr. O’Connor). DISCLOSURE Gary B. Deutsch, Victoria O’Connor, Myung-Shin Sim, Ji Hey Lee, and Anton J. Bilchik have no conflicts of interest to disclose.
11.
12. 13.
14.
15.
16.
17.
REFERENCES 1. Reim D, Loos M, Vogl F, et al. Prognostic implications of the seventh edition of the International Union Against Cancer classification for patients with gastric cancer: the Western experience of patients treated in a single-center European institution. J Clin Oncol. 2013;31(2):263–71. 2. McGhan LJ, Pockaj BA, Gray RJ, Bagaria SP, Wasif N. Validation of the updated 7th edition AJCC TNM staging criteria for gastric adenocarcinoma. J Gastrointest Surg. 2012;16(1):53–61; discussion 61. 3. Washington K. 7th edition of the AJCC cancer staging manual: stomach. Ann Surg Oncol. 2010;17(12):3077–9. 4. Kwon SJ. Evaluation of the 7th UICC TNM staging system of gastric cancer. J Gastric Cancer. 2011;11(2):78–85. 5. 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 US-population database. J Clin Oncol. 2005;23(28):7114–24. 6. Schwarz RE, Smith DD. Clinical impact of lymphadenectomy extent in resectable gastric cancer of advanced stage. Ann Surg Oncol. 2007;14(2):317–28. 7. Wang J, Dang P, Raut CP, et al. Comparison of a lymph node ratio–based staging system with the 7th AJCC system for gastric cancer. Ann Surg. 2012;255:478–85. 8. Wang W, Xu DZ, Li YF, et al. Tumor-ratio-metastasis staging system as an alternative to the 7th edition UICC TNM system in gastric cancer after D2 resection–results of a single-institution study of 1343 Chinese patients. Ann Oncol. 2011;22(9):2049–56. 9. Deng J, Zhang R, Zhang L, Liu Y, Hao X, Liang H. Negative node count improvement prognostic prediction of the seventh edition of the TNM classification for gastric cancer. PLoS One. 2013;8(11):e80082. 10. Zhang J, Niu Z, Zhou Y, Cao S. A comparison between the seventh and sixth editions of the American Joint Committee on
18.
19.
20.
21.
22.
23.
24. 25.
26.
27.
Cancer/International Union Against classification of gastric cancer. Ann Surg. 2013;257(1):81–6. Songun I, Putter H, Kranenbarg EM-K, Sasako M, van de Velde CJH. Surgical treatment of gastric cancer: 15-year follow-up results of the randomised nationwide Dutch D1D2 trial. Lancet Oncol. 2010;11:439–49. Strong VE, Yoon SS. Extended lymphadenectomy in gastric cancer is debatable. World J Surg. 2013;37(8):1773–7. Parsons HM, Begun JW, Kuntz KM, Tuttle TM, McGovern PM, Virnig BA. Lymph node evaluation for colon cancer in an era of quality guidelines: who improves? J Oncol Pract. 2013;9(4): e164–71. Dikken JL, Cats A, Verheij M, van de Velde CJH. Randomized trials and quality assurance in gastric cancer surgery. J Surg Oncol. 2013;107(3):298–305. Yoon HM, Ryu KW, Nam BH, et al. Is the new seventh AJCC/ UICC staging system appropriate for patients with gastric cancer? J Am Coll Surg. 2012;214(1):88–96. Dubecz A, Solymosi N, Schweigert M, et al. Time trends and disparities in lymphadenectomy for gastrointestinal cancer in the United States: a population-based analysis of 326,243 patients. J Gastrointest Surg. 2013;17(4):611–8; discussion 618-9. Enzinger PC, Benedetti JK, Meyerhardt JA, et al. Impact of hospital volume on recurrence and survival after surgery for gastric cancer. Ann Surg. 2007;245(3):426–34. Dudeja V, Gay G, Habermann EB, et al. Do hospital attributes predict guideline-recommended gastric cancer care in the United States? Ann Surg Oncol. 2012;19(2):365–72. Dudeja V, Habermann EB, Zhong W, et al. Guideline recommended gastric cancer care in the elderly: insights into the applicability of cancer trials to real world. Ann Surg Oncol. 2011;18(1):26–33. Qureshi AP, Ottensmeyer CA, Mahar AL, et al. Quality indicators for gastric cancer surgery: a survey of practicing pathologists in Ontario. Ann Surg Oncol. 2009;16(7):1883–9. Smith DD, Nelson RA, Schwarz RE. A comparison of five competing lymph node staging schemes in a cohort of resectable gastric cancer patients. Ann Surg Oncol. 2014;21(3):875–82. Patel MI, Rhoads KF, Ma Y, et al. Seventh edition (2010) of the AJCC/UICC staging system for gastric adenocarcinoma: is there room for improvement? Ann Surg Oncol. 2013;20(5):1631–8. Kong S-H, Lee H-J, Ahn HS, et al. Stage migration effect on survival in gastric cancer surgery with extended lymphadenectomy: the reappraisal of positive lymph node ratio as a proper N-staging. Ann Surg. 2012;255(1):50–8. Gilbert SM, Hollenbeck BK. Limitations of lymph node counts as a measure of therapy. J Natl Compr Canc Netw. 2009;7(1):58–61. Rausei S, Dionigi G, Ruspi L, et al. Lymph node staging in gastric cancer: new criteria, old problems. Int J Surg. 2013;11 Suppl 1:S90–4. Park HS, Lloyd S, Decker RH, Wilson LD, Yu JB. Limitations and biases of the Surveillance, Epidemiology, and End Results database. Curr Probl Cancer. 2012;36(4):216–24. Edge SB, Compton CC. The American Joint Committee on Cancer: the 7th edition of the AJCC cancer staging manual and the future of TNM. Ann Surg Oncol. 2010;17(6):1471–4.
