- Email: [email protected]
Increasing the Number of Lymph Nodes Examined after Colectomy Does Not Improve Colon Cancer Staging
Increasing the Number of Lymph Nodes Examined after Colectomy Does Not Improve Colon Cancer Staging Cristina N Budde, MD, Vassiliki L Tsikitis, MD, FACS, Karen E Deveney, MD, FACS, Brian S Diggs, PhD, Kim C Lu, MD, FACS, Daniel O Herzig, MD, FACS Current quality initiatives call for examination of at least 12 lymph nodes in curative colon cancer resections. The aim of this study was to determine if the number of nodes harvested has increased, and if the increased number nodes correlates with improved staging or overall survival. STUDY DESIGN: A review of the Surveillance, Epidemiology, and End Results program database from 20042010 was performed. All patients who underwent colon cancer resection during this date range were analyzed. Number of nodes retrieved, patient stage, overall survival, and overall survival by stage were examined. Multivariable analysis controlled for stage, cancer site, age, year of diagnosis, and number of nodes retrieved. Improved staging was defined as increased detection of stage III patients. RESULTS: A total of 147,076 patients met inclusion criteria. Median number of nodes analyzed increased sequentially with each year examined, from 12 in 2004 to 17 in 2010. Despite greater number of total nodes obtained and analyzed, there was no increase in the percentage of patients with positive nodes (stage III disease). On multivariable analysis, after controlling for stage, site of disease, age, and year of diagnosis, there was a slight overall survival benefit with increasing nodal retrieval (hazard ratio ¼ 0.987 for each additional node removed; 95% CI, 0.9860.988; p < 0.001). CONCLUSIONS: Since quality initiatives have been put in place, there has been an increase in the number of nodes examined in colon cancer resections, but no improvement in staging. The improved survival seen with higher node counts was independent of stage, site of disease, patient age, and year of diagnosis. (J Am Coll Surg 2014;218:1004e1011. 2014 by the American College of Surgeons)
BACKGROUND:
Colon cancer is the third most common malignancy in both sexes and the third most common cause of cancer death in the United States.1 Identification of all positive lymph nodes is essential for accurate staging. Hospital system, surgeon operative resection, pathologist review, and individual patient factors all play a role in the number of nodes analyzed. Accurate staging is particularly important for colon cancer because the presence of lymph Disclosure Information: Nothing to disclose. Presented at Digestive Disease Week, Orlando, FL, May 2013. Received September 19, 2013; Revised December 31, 2013; Accepted January 22, 2014. From the Division of Gastrointestinal and General Surgery, Department of Surgery, Oregon Health and Science University, Portland, OR. Correspondence address: Cristina N Budde, MD, Division of Gastrointestinal and General Surgery, Department of Surgery, Oregon Health and Science University, 3181 SW Sam Jackson Park Rd, Mail Code L223A, Portland, OR 97239. email: [email protected]
ª 2014 by the American College of Surgeons Published by Elsevier Inc.
node disease is the most important determinant of who receives adjuvant chemotherapy.2 The 2002 American Joint Committee on Cancer (AJCC) decision to establish the identification of a minimum of 12 lymph nodes as standard of care in colon cancer resection created a nationwide incentive to increase the number of lymph nodes obtained and analyzed.3 This incentive was reinforced with the National Quality Forum (NQF) endorsement of retrieval and pathologic examination of at least 12 lymph nodes in colon cancer resections as a quality performance measure.4 It is well established that in nonmetastatic colon cancer, nodal status is the best predictor of survival in colorectal cancer.5-7 If positive nodes are missed, a patient will be understaged, and will not likely receive appropriate life-prolonging adjuvant treatment.2 The quality measure of examining at least 12 nodes is based on the observation that patients with more nodes examined demonstrated an
1004
ISSN 1072-7515/14/$36.00 http://dx.doi.org/10.1016/j.jamcollsurg.2014.01.039
Vol. 218, No. 5, May 2014
Abbreviations and Acronyms
AJCC HR NQF SEER
¼ ¼ ¼ ¼
American Joint Committee on Cancer hazard ratio National Quality Forum Surveillance, Epidemiology, and End Results program
increased survival.5-8 This led to the assumption that retrieval and examination of more lymph nodes will result in higher rates of positive nodes and, therefore, more patients receiving adjuvant treatment.6,8-10 Wong and colleagues, after examining the National Cancer Data Base, concluded that stage migration did play a role in the increased survival seen with increasing nodal retrieval and evaluation.8 When examining the California Cancer Registry, Namm and colleagues concluded that there was substantial understaging occurring due to the retrieval and analysis of lower numbers of lymph nodes.6 Despite these observations, this concept has been challenged recently.11-14 Bui and colleagues examined the Ontario Cancer Registry and found that increasing nodal retrieval did not correlate with higher rates of positive nodal disease.13 Likewise, before the AJCC and NQF initiatives, 2 large Surveillance, Epidemiology, and End Results program (SEER)-based studies demonstrated that an increase in nodal retrieval did not correlate with upstaging of the disease.11,12 The aim of our study was to determine if, since the AJCC and NQF initiatives have been in place, the number of nodes resected has been increasing, and if increasing number of resected nodes correlates with improved staging, defined as increased number of nodepositive (stage III) patients, and/or overall survival within the SEER database from 2004 to 2010.
METHODS Data were obtained from the SEER database from the years 2004 to 2010. Colon cancer was defined as adenocarcinoma that had been resected at any of the following sites: cecum (ICD-O-3 code C180), ascending colon (C182), hepatic flexure (C183), transverse colon (C184), splenic flexure (C185), descending colon (C186), sigmoid colon (C187), or large intestine not otherwise specified (C188, C189, or C260). Cases before 2004 were excluded, as well as colectomy cases that took place for cancers where the colon was the secondary metastatic site. Those with missing staging information (no data for all 4 categories: T, N, M and stage) were excluded from the analysis. For each individual comparison, those with missing data were eliminated, including those with Tx, Nx, or Mx.
Budde et al
Lymphadenectomy and Colon Cancer
1005
Table 1. Demographic and Tumor Information for Patients Examined Variable
Value
Age, y, median (IQR) Sex, n (%) Male Female Race, n (%) White Black Asian American Indian Pacific Islander Other T stage, n (%) Tx T0 Tis T1 T2 T3 T4 N stage, n (%) Nx N0 N1 N2 M stage, n (%) Mx M0 M1 AJCC stage, n (%) I II III IV Unknown Tumor size, n (%) <1 cm 1e2 cm 2e3 cm 3e4 cm 4e5 cm >5 cm Tumor site, n (%) Cecum Ascending colon Hepatic flexure Transverse colon Splenic flexure Descending colon Not specified
71 (60e80) 71,729 (49) 75,347 (51) 119,015 16,886 9,309 830 514 522
(81) (11) (6) (1) (<1) (<1)
1,472 24 140 19,288 21,315 82,979 21,858
(1) (<1) (<1) (13) (14) (56) (15)
1,363 84,554 35,791 25,368
(1) (57) (24) (17)
2,055 (1) 123,469 (84) 21,552 (15) 33,843 45,978 43,204 21,552 2,499
(23) (31) (29) (15) (2)
5,533 11,550 21,516 25,982 23,257 44,926
(4) (9) (16) (20) (18) (34)
35,820 30,216 7,943 14,895 5,529 9,000 3,228
(24) (21) (5) (10) (4) (6) (2)
AJCC, American Joint Committee on Cancer; IQR, interquartile range.
1006
Budde et al
J Am Coll Surg
Lymphadenectomy and Colon Cancer
Outcomes assessed include distribution of number of nodes examined each year, proportions of cases by stage and staging components each year, and overall survival. Improved staging was defined as increase in percent of node-positive patients (stage III disease). Comparisons of continuous variables across time or stage were made using the nonparametric JonchkeereTerpstra test for trends across levels. Categorical variables were compared using a chi-square test and trends in category membership were assessed with a linear regression on the percentage. Overall survival over time was estimated with Kaplan-Meier curves and compared with a log-rank test for trend. A Cox proportional hazard model was used to determine the effect on survival of various predictors and adjusted for other confounders. The multivariable analysis included site of disease, cancer stage, and year of diagnosis as discrete variables, as well as patient age and number of lymph nodes examined as continuous variables.
RESULTS A total of 147,076 patients with colon cancer met inclusion criteria. There were 33,843 patients with stage I; 45,978 patients with stage II; 43,204 with stage III; and 21,552 patients with stage IV disease. The remaining 2,499 patients had incomplete staging. Patient demographics and tumor characteristics are listed in Table 1. There was an observed increase in the median number of nodes analyzed sequentially per each year examined (Fig. 1). The median number of nodes examined in colectomy pathology specimens in 2004 was 12 (interquartile range 7 to 18 nodes), and sequentially increased during the subsequent 6-year period to 17 (interquartile range 12 to 23 nodes) (p < 0.001). With respect to the primary tumor, the percentage of T1 disease rose by a mean of 0.18 percentage points each year (95% CI, 0.060.31; p ¼ 0.01), and the percentage of T3 disease decreased by a mean of 0.43 percentage points each year (95% CI, 0.73 to 0.13;
Figure 1. Number of nodes examined each year; dark bar marks median, box indicates interquartile range, symbol around median and boxes represents symmetric density (violin) plot.
Vol. 218, No. 5, May 2014
Budde et al
Lymphadenectomy and Colon Cancer
1007
Figure 2. Colon cancer N stage expressed as a percent of cases each year. Number above each column represents number of examined cases each year. Dark gray, N0; medium gray, N1; light gray, N2.
p ¼ 0.02). T2 disease and T4 disease remained unchanged (p ¼ 0.27, p ¼ 0.39, respectively). With respect to nodal disease, the percent of N0 and N1 disease remained unchanged (p ¼ 0.31, p ¼ 0.16, respectively), but there was a slight increase in N2 disease, a mean of 0.34 percentage points per year (95% CI, 0.080.59; p ¼ 0.02) (Fig. 2). With respect to metastatic disease, M0 disease increased by a mean of 0.18 percentage points per year (95% CI, 0.050.32; p ¼ 0.02) and M1 disease decreased correspondingly. Most notably, even though the number of nodes retrieved has clearly been increasing with each year examined, the percent of patients with stage III disease has not had any statistically significant change. There
was a slight decrease in stage II (mean of 0.22 percentage points per year; 95% CI, 0.38 to 0.06; p ¼ 0.02) and stage IV disease (mean of 0.19 percentage points per year; 95% CI, 0.32 to 0.05; p ¼ 0.02). Percent of stage I and stage III disease was unchanged (p ¼ 0.06 and p ¼ 0.07, respectively) (Fig. 3). Overall 5-year survival percentage of the cohort was 56.1%, with a median follow-up of 26 months. Overall survival increased slightly with each year examined (p < 0.001), with an overall survival difference of 4.1% at 1 year when comparing 2010 with 2004 (Fig. 4A). Likewise, overall survival by stage increased slightly with each year examined (p < 0.001) (Fig. 4B). The increases in survival at 1 year in stages I, II, III, and IV were 2.0%,
1008
Budde et al
J Am Coll Surg
Lymphadenectomy and Colon Cancer
Figure 3. Colon cancer stage expressed as a percent of cases each year. Number above each column represents number of examined cases each year.
3.1%, 3.1%, and 11%, respectively, in patients who underwent colectomy in 2010 compared with those who underwent colectomy in 2004. Multivariable Cox proportional hazards analysis demonstrated statistically significant decreased overall survival with increasing stage of disease (with stage I as a reference, stage II hazard ratio [HR] ¼ 1.587; 95% CI, 1.537e1.638; p < 0.001; stage III HR ¼ 2.641; 95% CI, 2.561e2.723; p < 0.001; and stage IV HR ¼ 9.694; 95% CI, 9.395e10.002; p < 0.001). Using cecal location as a reference, there was no difference in survival if cancer location was hepatic flexure or transverse colon (p ¼ 0.70 and p ¼ 0.48, respectively), but there was improved survival with location in ascending colon (HR ¼ 0.970; 95% CI, 0.945e0.996; p ¼ 0.02), splenic flexure (HR ¼ 0.951; 95% CI, 0.905e0.998; p ¼ 0.04), descending colon (HR ¼ 0.902; 95% CI, 0.865e0.940; p < 0.001), or sigmoid
colon (HR ¼ 0.846; 95% CI, 0.825e0.868; p < 0.001). There was a statistically significant decreased survival with each additional year in age of patient at time of diagnosis (HR ¼ 1.041; 95% CI, 1.040e1.042; p < 0.001). Using 2004 as a reference, there was no statistically significant difference in survival if diagnosis was made in 2005, 2006, or 2007 (p ¼ 0.49, p ¼ 0.31, and p ¼ 0.08, respectively); however, if diagnosed in 2008 or after, there was a statistically significant improvement in survival (year 2008 HR ¼ 0.962; 95% CI, 0.931e0.995; p ¼ 0.02; year 2009 HR ¼ 0.945; 95% CI, 0.909e0.982; p 0.001; year 2010 HR ¼ 0.842; 95% CI, 0.798e0.888; p < 0.001). When controlling for stage, site of disease, age, and year of diagnosis, there was a small overall survival benefit with higher number of nodes retrieved (HR ¼ 0.987 for each additional node removed; 95% CI, 0.986e0.988; p < 0.001) (Table 2).
Vol. 218, No. 5, May 2014
Budde et al
Lymphadenectomy and Colon Cancer
Figure 4. (A) Overall survival by year showing increasing survival with each year examined; p < 0.001. (B) Stage-specific overall survival by year showing increasing survival with each year examined within each stage; p < 0.001. Number at risk each year demonstrated in box below each graph.
1009
1010
Budde et al
Lymphadenectomy and Colon Cancer
Table 2. Multivariable Cox Proportional Hazards Analysis for Overall Survival, Adjusting for Stage, Site of Disease, Age, Year of Diagnosis, and Number of Nodes Examined Variable
Stage (reference: stage I) II III IV Nodes (per node examined) Site (reference: cecum) Ascending colon Hepatic flexure Transverse colon Splenic flexure Descending colon Sigmoid colon Age (per year) Year of diagnosis (reference: 2004) 2005 2006 2007 2008 2009 2010
Hazard ratio (95% CI)
p Value
1.587 2.641 9.694 0.987
(1.537e1.638) (2.561e2.723) (9.395e10.002) (0.986e0.988)
<0.001 <0.001 <0.001 <0.001
0.970 0.992 1.012 0.951 0.902 0.846 1.041
(0.945e0.996) (0.951e1.034) (0.979e1.045) (0.905e0.998) (0.865e0.940) (0.825e0.868) (1.040e1.042)
0.023 0.699 0.481 0.042 <0.001 <0.001 <0.001
0.990 0.985 0.973 0.962 0.945 0.842
(0.963e1.018) (0.957e1.014) (0.943e1.004) (0.931e0.995) (0.909e0.982) (0.798e0.888)
0.493 0.308 0.085 0.024 0.004 <0.001
There is a statistically significant increased survival with increased node retrieval and examination.
DISCUSSION In this population-based study using SEER data, we have shown that, in more recent years, after the quality initiative established by the AJCC and the NQF mandated retrieving and analyzing at least 12 nodes in each colon specimen, the number of nodes examined in colon cancer resections in the United States has increased. In fact, in contrast to some other recent studies,10,14 we have shown that most colon resections are now exceeding the benchmark of the recommended 12 nodes retrieved and examined in the specimen. The decision to set this quality initiative was based on the assumption that retrieving and analyzing more nodes will lead to more accurate identification of node-positive patients. As a result, they would receive appropriate adjuvant chemotherapy that would have been omitted had understaging occurred. Some studies have suggested that stage migration is responsible for the increased survival seen with more nodes examined.6,13,15 In contrast, our results suggest that there is not a relationship between an increase in nodes examined leading to an increase in node-positive patients. There was no statistically significant change in N1 disease during the 7 years examined,
J Am Coll Surg
and there was no statistically significant change in stage III disease during the 7 years examined. This is congruent with other population-based studies.11-13 There was a small statistically significant increase in N2 disease during the years examined; therefore, if obtaining and examining more nodes after colectomy is changing staging at all, it is merely finding more N2 disease, which would not change adjuvant treatment, as the adjuvant treatment for N1 and N2 disease is the same. It is possible that increasing nodal retrieval and examination is upstaging some patients, and our data fail to show the increase because of a similar increase in earlier-stage diagnosis of some patients over time. For example, improvements in colorectal cancer screening and early detection over time might have occurred, leading to a group of patients who were diagnosed at an earlier stage than in previous years. If something like this occurred, it could offset the upstaging that might have occurred with improved nodal retrieval and examination over time. Our study did note a small increased overall survival and overall survival by stage throughout the 7 years examined. This is consistent with national trends, which have seen colon cancer survival improving during the last 20 years.1 In our analysis, the largest driver of increased survival is within the stage IV disease cohort (11% 1-year survival advantage during the 7 years examined). This improved survival is most likely secondary to improvements in chemotherapeutics with targeted regimens and advances in successful metastasectomies.16 Improvement in patient overall medical conditions can also play a role. The overall survival benefits observed throughout all other stages are small (2.0% to 3.1% 1-year survival advantage during the 7 years examined for stage I, II, and III disease) (Fig. 4). The drivers behind this survival improvement are most likely multifactorial and include improvements in vigilant screening and advancements in chemotherapy regimens.1,16-19 Our results suggest that examining more nodes does not seem to achieve the desired quality measure of more accurately staging colon cancer; yet, on multivariable analysis, there is a small, but statistically significant overall survival advantage with each additional node removed and examined. These results do confirm the association between higher lymph node counts and better survival, but suggest it is not because of upstaging. It is unlikely that more extensive resections of noncancerous nodes would improve outcomes, which suggests that there is another unmeasured factor besides better staging that is driving this survival improvement. Some authors have proposed that an increased number of nodes are found in patients with better immunologic response to the tumor, enabling these patients to better control their disease and might be responsible for the observed increased survival.11,13
Vol. 218, No. 5, May 2014
Although we cannot measure immunologic function in our study, the improved survival without upstaging seen in our study would support such a theory. This study has certain limitations inherent to all studies that rely on retrospective data. For example, this study is limited by differences in data input over time, and it is subject to clerical errors. Incomplete data might have had the greatest impact on our results. There are unmeasured variables that can affect our outcomes in the total lymph node yield. For example, we lack information on variables including surgeon, pathologist, and hospital surgical volume. Another limitation of the database is its lack of data on adjuvant therapies. All these variables, as well as the patient as an individual, play important roles in the total number of lymph nodes retrieved in a surgical specimen. Patients with stage IV disease in this study are also a heterogeneous group, as resection can be palliative vs with curative attempt if metastatectomy is feasible. We cannot differentiate these 2 groups within the SEER database; we have, however, controlled for stage within our multivariable analysis.
CONCLUSIONS Since the establishment of the current quality initiative, the number of nodes examined in colon cancer resections has increased, but this has not improved staging. There is a small survival benefit seen with incremental increases in lymph node counts, which is not related to upstaging. This suggests that the relationship between lymph node counts and survival is not modifiable by more extensive surgery or pathologic evaluation. Author Contributions Study conception and design: Tsikitis, Lu, Herzig Acquisition of data: Budde, Tsikitis, Diggs, Herzig Analysis and interpretation of data: Budde, Tsikitis, Diggs, Lu, Herzig Drafting of manuscript: Budde, Tsikitis, Diggs, Herzig Critical revision: Deveney, Lu Acknowledgment: The authors would like to thank Mary Kwatkosky-Lawlor for her assistance in the editing and preparation of the bibliography of this article. REFERENCES 1. American Cancer Society. Colorectal Cancer Facts and Figures 20112013. Atlanta, GA: American Cancer Society Inc. 2013. Available at: http://www.cancer.org/acs/groups/content/@ epidemiologysurveilance/documents/document/acspc-028323.pdf. Accessed October 24, 2012.
Budde et al
Lymphadenectomy and Colon Cancer
1011
2. Moertel CG, Fleming TR, Macdonald JS, et al. Levamisole and fluorouracil for adjuvant therapy of resected colon carcinoma. N Engl J Med 1990;322:352e358. 3. Edge S, Byrd DR, Compton CC, eds. Colon and rectum. In: AJCC Cancer Staging Manual. 7th ed. New York: SpringerVerlag; 2010:143e164. 4. National Quality Forum. NQF Endorses Additional Cancer Measures. Washington, DC: National Quality Forum; 2012. Available at: http://www.qualityforum.org/News_And_Resources/Press_ Releases/2012/NQF_Endorses_Additional_Cancer_Measures. aspx. Accessed October 24, 2012. 5. Le Voyer TE, Sigurdson ER, Hanlon AL, et al. Colon cancer survival is associated with increasing number of lymph nodes analyzed: a secondary survey of intergroup trial INT-0089. J Clin Oncol 2003;21:2912e2919. 6. Namm J, Ng M, Roy-Chowdhury S, et al. Quantitating the impact of stage migration on staging accuracy in colorectal cancer. J Am Coll Surg 2008;207:882e887. 7. Cserni G, Vinh-Hung V, Burzykowski T. Is there a minimum number of lymph nodes that should be histologically assessed for a reliable nodal staging of T3N0M0 colorectal carcinomas? J Surg Oncol 2002;81:63e69. 8. Wong JH, Severino R, Honnebier MB, et al. Number of nodes examined and staging accuracy in colorectal carcinoma. J Clin Oncol 1999;17:2896e2900. 9. Goldstein NS, Sanford W, Coffey M, Layfield LJ. Lymph node recovery from colorectal resection specimens removed for adenocarcinoma. Trends over time and a recommendation for a minimum number of lymph nodes to be recovered. Am J Clin Pathol 1996;106:209e216. 10. Baxter NN, Virnig DJ, Rothenberger DA, et al. Lymph node evaluation in colorectal cancer patients: a population-based study. J Natl Cancer Inst 2005;97:219e225. 11. Parsons HM, Tuttle TM, Kuntz KM, et al. Association between lymph node evaluation for colon cancer and node positivity over the past 20 years. JAMA 2011;306: 1089e1097. 12. Wong SL, Ji H, Hollenbeck BK, et al. Hospital lymph node examination rates and survival after resection for colon cancer. JAMA 2007;298:2149e2154. 13. Bui L, Rempel E, Reeson D, Simunovic M. Lymph node counts, rates of positive lymph nodes, and patient survival for colon cancer surgery in Ontario, Canada: a populationbased study. J Surg Oncol 2006;93:439e445. 14. Wong JH, Lum SS, Morgan JW. Lymph node counts as an indicator of quality at the hospital level in colorectal surgery. J Am Coll Surg 2011;213:226e230. 15. Sjo OH, Merok MA, Svindland A, Nesbakken A. Prognostic impact of lymph node harvest and lymph node ratio in patients with colon cancer. Dis Colon Rectum 2012;55: 307e315. 16. Cunningham D, Atkin W, Lenz HJ, et al. Colorectal cancer. Lancet 2010;375:1030e1047. 17. Trombold J, Farmer RW, McCafferty M. The impact of colorectal cancer screening in a veteran hospital population. Am Surg 2013;79:296e300. 18. Jemal A, Siegel R, Ward E, et al. Cancer statistics, 2008. CA Cancer J Clin 2008;58:71e96. 19. Prenen H, Vecchione L, Van Cutsem E. Role of targeted agents in metastatic colorectal cancer. Targeted Oncol 2013;8:83e96.
BACKGROUND:
Colon cancer is the third most common malignancy in both sexes and the third most common cause of cancer death in the United States.1 Identification of all positive lymph nodes is essential for accurate staging. Hospital system, surgeon operative resection, pathologist review, and individual patient factors all play a role in the number of nodes analyzed. Accurate staging is particularly important for colon cancer because the presence of lymph Disclosure Information: Nothing to disclose. Presented at Digestive Disease Week, Orlando, FL, May 2013. Received September 19, 2013; Revised December 31, 2013; Accepted January 22, 2014. From the Division of Gastrointestinal and General Surgery, Department of Surgery, Oregon Health and Science University, Portland, OR. Correspondence address: Cristina N Budde, MD, Division of Gastrointestinal and General Surgery, Department of Surgery, Oregon Health and Science University, 3181 SW Sam Jackson Park Rd, Mail Code L223A, Portland, OR 97239. email: [email protected]
ª 2014 by the American College of Surgeons Published by Elsevier Inc.
node disease is the most important determinant of who receives adjuvant chemotherapy.2 The 2002 American Joint Committee on Cancer (AJCC) decision to establish the identification of a minimum of 12 lymph nodes as standard of care in colon cancer resection created a nationwide incentive to increase the number of lymph nodes obtained and analyzed.3 This incentive was reinforced with the National Quality Forum (NQF) endorsement of retrieval and pathologic examination of at least 12 lymph nodes in colon cancer resections as a quality performance measure.4 It is well established that in nonmetastatic colon cancer, nodal status is the best predictor of survival in colorectal cancer.5-7 If positive nodes are missed, a patient will be understaged, and will not likely receive appropriate life-prolonging adjuvant treatment.2 The quality measure of examining at least 12 nodes is based on the observation that patients with more nodes examined demonstrated an
1004
ISSN 1072-7515/14/$36.00 http://dx.doi.org/10.1016/j.jamcollsurg.2014.01.039
Vol. 218, No. 5, May 2014
Abbreviations and Acronyms
AJCC HR NQF SEER
¼ ¼ ¼ ¼
American Joint Committee on Cancer hazard ratio National Quality Forum Surveillance, Epidemiology, and End Results program
increased survival.5-8 This led to the assumption that retrieval and examination of more lymph nodes will result in higher rates of positive nodes and, therefore, more patients receiving adjuvant treatment.6,8-10 Wong and colleagues, after examining the National Cancer Data Base, concluded that stage migration did play a role in the increased survival seen with increasing nodal retrieval and evaluation.8 When examining the California Cancer Registry, Namm and colleagues concluded that there was substantial understaging occurring due to the retrieval and analysis of lower numbers of lymph nodes.6 Despite these observations, this concept has been challenged recently.11-14 Bui and colleagues examined the Ontario Cancer Registry and found that increasing nodal retrieval did not correlate with higher rates of positive nodal disease.13 Likewise, before the AJCC and NQF initiatives, 2 large Surveillance, Epidemiology, and End Results program (SEER)-based studies demonstrated that an increase in nodal retrieval did not correlate with upstaging of the disease.11,12 The aim of our study was to determine if, since the AJCC and NQF initiatives have been in place, the number of nodes resected has been increasing, and if increasing number of resected nodes correlates with improved staging, defined as increased number of nodepositive (stage III) patients, and/or overall survival within the SEER database from 2004 to 2010.
METHODS Data were obtained from the SEER database from the years 2004 to 2010. Colon cancer was defined as adenocarcinoma that had been resected at any of the following sites: cecum (ICD-O-3 code C180), ascending colon (C182), hepatic flexure (C183), transverse colon (C184), splenic flexure (C185), descending colon (C186), sigmoid colon (C187), or large intestine not otherwise specified (C188, C189, or C260). Cases before 2004 were excluded, as well as colectomy cases that took place for cancers where the colon was the secondary metastatic site. Those with missing staging information (no data for all 4 categories: T, N, M and stage) were excluded from the analysis. For each individual comparison, those with missing data were eliminated, including those with Tx, Nx, or Mx.
Budde et al
Lymphadenectomy and Colon Cancer
1005
Table 1. Demographic and Tumor Information for Patients Examined Variable
Value
Age, y, median (IQR) Sex, n (%) Male Female Race, n (%) White Black Asian American Indian Pacific Islander Other T stage, n (%) Tx T0 Tis T1 T2 T3 T4 N stage, n (%) Nx N0 N1 N2 M stage, n (%) Mx M0 M1 AJCC stage, n (%) I II III IV Unknown Tumor size, n (%) <1 cm 1e2 cm 2e3 cm 3e4 cm 4e5 cm >5 cm Tumor site, n (%) Cecum Ascending colon Hepatic flexure Transverse colon Splenic flexure Descending colon Not specified
71 (60e80) 71,729 (49) 75,347 (51) 119,015 16,886 9,309 830 514 522
(81) (11) (6) (1) (<1) (<1)
1,472 24 140 19,288 21,315 82,979 21,858
(1) (<1) (<1) (13) (14) (56) (15)
1,363 84,554 35,791 25,368
(1) (57) (24) (17)
2,055 (1) 123,469 (84) 21,552 (15) 33,843 45,978 43,204 21,552 2,499
(23) (31) (29) (15) (2)
5,533 11,550 21,516 25,982 23,257 44,926
(4) (9) (16) (20) (18) (34)
35,820 30,216 7,943 14,895 5,529 9,000 3,228
(24) (21) (5) (10) (4) (6) (2)
AJCC, American Joint Committee on Cancer; IQR, interquartile range.
1006
Budde et al
J Am Coll Surg
Lymphadenectomy and Colon Cancer
Outcomes assessed include distribution of number of nodes examined each year, proportions of cases by stage and staging components each year, and overall survival. Improved staging was defined as increase in percent of node-positive patients (stage III disease). Comparisons of continuous variables across time or stage were made using the nonparametric JonchkeereTerpstra test for trends across levels. Categorical variables were compared using a chi-square test and trends in category membership were assessed with a linear regression on the percentage. Overall survival over time was estimated with Kaplan-Meier curves and compared with a log-rank test for trend. A Cox proportional hazard model was used to determine the effect on survival of various predictors and adjusted for other confounders. The multivariable analysis included site of disease, cancer stage, and year of diagnosis as discrete variables, as well as patient age and number of lymph nodes examined as continuous variables.
RESULTS A total of 147,076 patients with colon cancer met inclusion criteria. There were 33,843 patients with stage I; 45,978 patients with stage II; 43,204 with stage III; and 21,552 patients with stage IV disease. The remaining 2,499 patients had incomplete staging. Patient demographics and tumor characteristics are listed in Table 1. There was an observed increase in the median number of nodes analyzed sequentially per each year examined (Fig. 1). The median number of nodes examined in colectomy pathology specimens in 2004 was 12 (interquartile range 7 to 18 nodes), and sequentially increased during the subsequent 6-year period to 17 (interquartile range 12 to 23 nodes) (p < 0.001). With respect to the primary tumor, the percentage of T1 disease rose by a mean of 0.18 percentage points each year (95% CI, 0.060.31; p ¼ 0.01), and the percentage of T3 disease decreased by a mean of 0.43 percentage points each year (95% CI, 0.73 to 0.13;
Figure 1. Number of nodes examined each year; dark bar marks median, box indicates interquartile range, symbol around median and boxes represents symmetric density (violin) plot.
Vol. 218, No. 5, May 2014
Budde et al
Lymphadenectomy and Colon Cancer
1007
Figure 2. Colon cancer N stage expressed as a percent of cases each year. Number above each column represents number of examined cases each year. Dark gray, N0; medium gray, N1; light gray, N2.
p ¼ 0.02). T2 disease and T4 disease remained unchanged (p ¼ 0.27, p ¼ 0.39, respectively). With respect to nodal disease, the percent of N0 and N1 disease remained unchanged (p ¼ 0.31, p ¼ 0.16, respectively), but there was a slight increase in N2 disease, a mean of 0.34 percentage points per year (95% CI, 0.080.59; p ¼ 0.02) (Fig. 2). With respect to metastatic disease, M0 disease increased by a mean of 0.18 percentage points per year (95% CI, 0.050.32; p ¼ 0.02) and M1 disease decreased correspondingly. Most notably, even though the number of nodes retrieved has clearly been increasing with each year examined, the percent of patients with stage III disease has not had any statistically significant change. There
was a slight decrease in stage II (mean of 0.22 percentage points per year; 95% CI, 0.38 to 0.06; p ¼ 0.02) and stage IV disease (mean of 0.19 percentage points per year; 95% CI, 0.32 to 0.05; p ¼ 0.02). Percent of stage I and stage III disease was unchanged (p ¼ 0.06 and p ¼ 0.07, respectively) (Fig. 3). Overall 5-year survival percentage of the cohort was 56.1%, with a median follow-up of 26 months. Overall survival increased slightly with each year examined (p < 0.001), with an overall survival difference of 4.1% at 1 year when comparing 2010 with 2004 (Fig. 4A). Likewise, overall survival by stage increased slightly with each year examined (p < 0.001) (Fig. 4B). The increases in survival at 1 year in stages I, II, III, and IV were 2.0%,
1008
Budde et al
J Am Coll Surg
Lymphadenectomy and Colon Cancer
Figure 3. Colon cancer stage expressed as a percent of cases each year. Number above each column represents number of examined cases each year.
3.1%, 3.1%, and 11%, respectively, in patients who underwent colectomy in 2010 compared with those who underwent colectomy in 2004. Multivariable Cox proportional hazards analysis demonstrated statistically significant decreased overall survival with increasing stage of disease (with stage I as a reference, stage II hazard ratio [HR] ¼ 1.587; 95% CI, 1.537e1.638; p < 0.001; stage III HR ¼ 2.641; 95% CI, 2.561e2.723; p < 0.001; and stage IV HR ¼ 9.694; 95% CI, 9.395e10.002; p < 0.001). Using cecal location as a reference, there was no difference in survival if cancer location was hepatic flexure or transverse colon (p ¼ 0.70 and p ¼ 0.48, respectively), but there was improved survival with location in ascending colon (HR ¼ 0.970; 95% CI, 0.945e0.996; p ¼ 0.02), splenic flexure (HR ¼ 0.951; 95% CI, 0.905e0.998; p ¼ 0.04), descending colon (HR ¼ 0.902; 95% CI, 0.865e0.940; p < 0.001), or sigmoid
colon (HR ¼ 0.846; 95% CI, 0.825e0.868; p < 0.001). There was a statistically significant decreased survival with each additional year in age of patient at time of diagnosis (HR ¼ 1.041; 95% CI, 1.040e1.042; p < 0.001). Using 2004 as a reference, there was no statistically significant difference in survival if diagnosis was made in 2005, 2006, or 2007 (p ¼ 0.49, p ¼ 0.31, and p ¼ 0.08, respectively); however, if diagnosed in 2008 or after, there was a statistically significant improvement in survival (year 2008 HR ¼ 0.962; 95% CI, 0.931e0.995; p ¼ 0.02; year 2009 HR ¼ 0.945; 95% CI, 0.909e0.982; p 0.001; year 2010 HR ¼ 0.842; 95% CI, 0.798e0.888; p < 0.001). When controlling for stage, site of disease, age, and year of diagnosis, there was a small overall survival benefit with higher number of nodes retrieved (HR ¼ 0.987 for each additional node removed; 95% CI, 0.986e0.988; p < 0.001) (Table 2).
Vol. 218, No. 5, May 2014
Budde et al
Lymphadenectomy and Colon Cancer
Figure 4. (A) Overall survival by year showing increasing survival with each year examined; p < 0.001. (B) Stage-specific overall survival by year showing increasing survival with each year examined within each stage; p < 0.001. Number at risk each year demonstrated in box below each graph.
1009
1010
Budde et al
Lymphadenectomy and Colon Cancer
Table 2. Multivariable Cox Proportional Hazards Analysis for Overall Survival, Adjusting for Stage, Site of Disease, Age, Year of Diagnosis, and Number of Nodes Examined Variable
Stage (reference: stage I) II III IV Nodes (per node examined) Site (reference: cecum) Ascending colon Hepatic flexure Transverse colon Splenic flexure Descending colon Sigmoid colon Age (per year) Year of diagnosis (reference: 2004) 2005 2006 2007 2008 2009 2010
Hazard ratio (95% CI)
p Value
1.587 2.641 9.694 0.987
(1.537e1.638) (2.561e2.723) (9.395e10.002) (0.986e0.988)
<0.001 <0.001 <0.001 <0.001
0.970 0.992 1.012 0.951 0.902 0.846 1.041
(0.945e0.996) (0.951e1.034) (0.979e1.045) (0.905e0.998) (0.865e0.940) (0.825e0.868) (1.040e1.042)
0.023 0.699 0.481 0.042 <0.001 <0.001 <0.001
0.990 0.985 0.973 0.962 0.945 0.842
(0.963e1.018) (0.957e1.014) (0.943e1.004) (0.931e0.995) (0.909e0.982) (0.798e0.888)
0.493 0.308 0.085 0.024 0.004 <0.001
There is a statistically significant increased survival with increased node retrieval and examination.
DISCUSSION In this population-based study using SEER data, we have shown that, in more recent years, after the quality initiative established by the AJCC and the NQF mandated retrieving and analyzing at least 12 nodes in each colon specimen, the number of nodes examined in colon cancer resections in the United States has increased. In fact, in contrast to some other recent studies,10,14 we have shown that most colon resections are now exceeding the benchmark of the recommended 12 nodes retrieved and examined in the specimen. The decision to set this quality initiative was based on the assumption that retrieving and analyzing more nodes will lead to more accurate identification of node-positive patients. As a result, they would receive appropriate adjuvant chemotherapy that would have been omitted had understaging occurred. Some studies have suggested that stage migration is responsible for the increased survival seen with more nodes examined.6,13,15 In contrast, our results suggest that there is not a relationship between an increase in nodes examined leading to an increase in node-positive patients. There was no statistically significant change in N1 disease during the 7 years examined,
J Am Coll Surg
and there was no statistically significant change in stage III disease during the 7 years examined. This is congruent with other population-based studies.11-13 There was a small statistically significant increase in N2 disease during the years examined; therefore, if obtaining and examining more nodes after colectomy is changing staging at all, it is merely finding more N2 disease, which would not change adjuvant treatment, as the adjuvant treatment for N1 and N2 disease is the same. It is possible that increasing nodal retrieval and examination is upstaging some patients, and our data fail to show the increase because of a similar increase in earlier-stage diagnosis of some patients over time. For example, improvements in colorectal cancer screening and early detection over time might have occurred, leading to a group of patients who were diagnosed at an earlier stage than in previous years. If something like this occurred, it could offset the upstaging that might have occurred with improved nodal retrieval and examination over time. Our study did note a small increased overall survival and overall survival by stage throughout the 7 years examined. This is consistent with national trends, which have seen colon cancer survival improving during the last 20 years.1 In our analysis, the largest driver of increased survival is within the stage IV disease cohort (11% 1-year survival advantage during the 7 years examined). This improved survival is most likely secondary to improvements in chemotherapeutics with targeted regimens and advances in successful metastasectomies.16 Improvement in patient overall medical conditions can also play a role. The overall survival benefits observed throughout all other stages are small (2.0% to 3.1% 1-year survival advantage during the 7 years examined for stage I, II, and III disease) (Fig. 4). The drivers behind this survival improvement are most likely multifactorial and include improvements in vigilant screening and advancements in chemotherapy regimens.1,16-19 Our results suggest that examining more nodes does not seem to achieve the desired quality measure of more accurately staging colon cancer; yet, on multivariable analysis, there is a small, but statistically significant overall survival advantage with each additional node removed and examined. These results do confirm the association between higher lymph node counts and better survival, but suggest it is not because of upstaging. It is unlikely that more extensive resections of noncancerous nodes would improve outcomes, which suggests that there is another unmeasured factor besides better staging that is driving this survival improvement. Some authors have proposed that an increased number of nodes are found in patients with better immunologic response to the tumor, enabling these patients to better control their disease and might be responsible for the observed increased survival.11,13
Vol. 218, No. 5, May 2014
Although we cannot measure immunologic function in our study, the improved survival without upstaging seen in our study would support such a theory. This study has certain limitations inherent to all studies that rely on retrospective data. For example, this study is limited by differences in data input over time, and it is subject to clerical errors. Incomplete data might have had the greatest impact on our results. There are unmeasured variables that can affect our outcomes in the total lymph node yield. For example, we lack information on variables including surgeon, pathologist, and hospital surgical volume. Another limitation of the database is its lack of data on adjuvant therapies. All these variables, as well as the patient as an individual, play important roles in the total number of lymph nodes retrieved in a surgical specimen. Patients with stage IV disease in this study are also a heterogeneous group, as resection can be palliative vs with curative attempt if metastatectomy is feasible. We cannot differentiate these 2 groups within the SEER database; we have, however, controlled for stage within our multivariable analysis.
CONCLUSIONS Since the establishment of the current quality initiative, the number of nodes examined in colon cancer resections has increased, but this has not improved staging. There is a small survival benefit seen with incremental increases in lymph node counts, which is not related to upstaging. This suggests that the relationship between lymph node counts and survival is not modifiable by more extensive surgery or pathologic evaluation. Author Contributions Study conception and design: Tsikitis, Lu, Herzig Acquisition of data: Budde, Tsikitis, Diggs, Herzig Analysis and interpretation of data: Budde, Tsikitis, Diggs, Lu, Herzig Drafting of manuscript: Budde, Tsikitis, Diggs, Herzig Critical revision: Deveney, Lu Acknowledgment: The authors would like to thank Mary Kwatkosky-Lawlor for her assistance in the editing and preparation of the bibliography of this article. REFERENCES 1. American Cancer Society. Colorectal Cancer Facts and Figures 20112013. Atlanta, GA: American Cancer Society Inc. 2013. Available at: http://www.cancer.org/acs/groups/content/@ epidemiologysurveilance/documents/document/acspc-028323.pdf. Accessed October 24, 2012.
Budde et al
Lymphadenectomy and Colon Cancer
1011
2. Moertel CG, Fleming TR, Macdonald JS, et al. Levamisole and fluorouracil for adjuvant therapy of resected colon carcinoma. N Engl J Med 1990;322:352e358. 3. Edge S, Byrd DR, Compton CC, eds. Colon and rectum. In: AJCC Cancer Staging Manual. 7th ed. New York: SpringerVerlag; 2010:143e164. 4. National Quality Forum. NQF Endorses Additional Cancer Measures. Washington, DC: National Quality Forum; 2012. Available at: http://www.qualityforum.org/News_And_Resources/Press_ Releases/2012/NQF_Endorses_Additional_Cancer_Measures. aspx. Accessed October 24, 2012. 5. Le Voyer TE, Sigurdson ER, Hanlon AL, et al. Colon cancer survival is associated with increasing number of lymph nodes analyzed: a secondary survey of intergroup trial INT-0089. J Clin Oncol 2003;21:2912e2919. 6. Namm J, Ng M, Roy-Chowdhury S, et al. Quantitating the impact of stage migration on staging accuracy in colorectal cancer. J Am Coll Surg 2008;207:882e887. 7. Cserni G, Vinh-Hung V, Burzykowski T. Is there a minimum number of lymph nodes that should be histologically assessed for a reliable nodal staging of T3N0M0 colorectal carcinomas? J Surg Oncol 2002;81:63e69. 8. Wong JH, Severino R, Honnebier MB, et al. Number of nodes examined and staging accuracy in colorectal carcinoma. J Clin Oncol 1999;17:2896e2900. 9. Goldstein NS, Sanford W, Coffey M, Layfield LJ. Lymph node recovery from colorectal resection specimens removed for adenocarcinoma. Trends over time and a recommendation for a minimum number of lymph nodes to be recovered. Am J Clin Pathol 1996;106:209e216. 10. Baxter NN, Virnig DJ, Rothenberger DA, et al. Lymph node evaluation in colorectal cancer patients: a population-based study. J Natl Cancer Inst 2005;97:219e225. 11. Parsons HM, Tuttle TM, Kuntz KM, et al. Association between lymph node evaluation for colon cancer and node positivity over the past 20 years. JAMA 2011;306: 1089e1097. 12. Wong SL, Ji H, Hollenbeck BK, et al. Hospital lymph node examination rates and survival after resection for colon cancer. JAMA 2007;298:2149e2154. 13. Bui L, Rempel E, Reeson D, Simunovic M. Lymph node counts, rates of positive lymph nodes, and patient survival for colon cancer surgery in Ontario, Canada: a populationbased study. J Surg Oncol 2006;93:439e445. 14. Wong JH, Lum SS, Morgan JW. Lymph node counts as an indicator of quality at the hospital level in colorectal surgery. J Am Coll Surg 2011;213:226e230. 15. Sjo OH, Merok MA, Svindland A, Nesbakken A. Prognostic impact of lymph node harvest and lymph node ratio in patients with colon cancer. Dis Colon Rectum 2012;55: 307e315. 16. Cunningham D, Atkin W, Lenz HJ, et al. Colorectal cancer. Lancet 2010;375:1030e1047. 17. Trombold J, Farmer RW, McCafferty M. The impact of colorectal cancer screening in a veteran hospital population. Am Surg 2013;79:296e300. 18. Jemal A, Siegel R, Ward E, et al. Cancer statistics, 2008. CA Cancer J Clin 2008;58:71e96. 19. Prenen H, Vecchione L, Van Cutsem E. Role of targeted agents in metastatic colorectal cancer. Targeted Oncol 2013;8:83e96.