|
|
||
 |
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
Journal of Clinical Oncology, Vol 24, No 22 (August 1), 2006: pp. 3570-3575 © 2006 American Society of Clinical Oncology. DOI: 10.1200/JCO.2006.06.8866 Increasing Negative Lymph Node Count Is Independently Associated With Improved Long-Term Survival in Stage IIIB and IIIC Colon Cancer
From the Department of Surgery, Dalhousie University, Halifax, Nova Scotia; Department of Surgery, St Michael's Hospital, University of Toronto, Toronto, Ontario, Canada; Department of Surgery, University of Minnesota, Minneapolis, MN Address reprint requests to Nancy Baxter, MD, PhD, Division of General Surgery, St Michael's Hospital, 30 Bond St, CC16-040, Toronto, Ontario, Canada, M5B 1W8; e-mail: baxtern{at}smh.toronto.on.ca
PURPOSE: The purpose of this study was to examine the impact of the number of negative lymph nodes on survival in patients with stage III colon cancer. PATIENTS AND METHODS: Patients who underwent surgery for stage III colon cancer between January 1988 and December 1997 were identified from the Surveillance, Epidemiology and End Results cancer registry. The number of negative and positive nodes was determined for 20,702 eligible patients. Disease-specific survival was examined by substage according to the number of negative nodes identified. A proportional hazards model was constructed to determine the effect of the number of negative nodes on survival. RESULTS: For stage IIIB and IIIC patients, there was a significant decrease in disease-specific mortality as the number of negative nodes increased; cumulative 5-year cancer mortality was 27% in stage IIIB patients with 13 or more negative nodes identified versus 45% in those with three or fewer negative lymph nodes evaluated (P < .0001). In patients with stage IIIC cancer, those with 13 or more negative nodes had a 5-year mortality of 42% versus 65% in those with three or fewer negative lymph nodes evaluated (P < .0001). There was no association between the number of negative nodes identified and disease-specific survival for patients with stage IIIA disease. After controlling for the number of positive nodes, a higher number of negative nodes was found to be independently associated with improved disease-specific survival. CONCLUSION: The number of negative nodes is an important independent prognostic factor for patients with stage IIIB and IIIC colon cancer.
The presence of lymph node (LN) metastases in colorectal cancer (CRC) has important implications regarding prognosis,1 may determine the use of adjuvant therapies which are proven to increase disease-free and overall survival,2,3 and may be an indicator of quality of care. The number of LNs that must be assessed to correctly determine a patient's LN status is controversial, with recommendations in the literature ranging from six to more than 40 LNs.4-10 Most of these recommendations have been made in the context of the number of LNs needed to accurately determine that a patient has node-negative disease. Studies of patients with node-negative colon cancer have consistently reported that disease-free5,9,11,12 and overall survival5,8-14 improve as the number of assessed LNs increases. Comparatively little research has sought to define the relationship between the number of LNs identified and long-term survival in patients with stage III colon cancer. Most studies that have examined this issue have suggested no association between the total number of LNs assessed and long-term survival in patients with stage III colon cancer4,5,9,14; however there are conflicting results.11,13 The relationship between number of LNs and survival is confounded by the prognostic effect of increasing number of positive LNs, and this may explain the inconsistent findings. The effect of number of negative LNs on survival has not been evaluated in stage III patients. To improve the understanding of the relationship between LN counts and survival in patients with stage III colon cancer, we designed this study specifically to examine the impact of the number of negative LNs on survival using population-based data.
Data Data from the Surveillance, Epidemiology and End Results (SEER) cancer registry were utilized for this study. SEER, a population-based registry sponsored by the National Cancer Institute (Bethesda, MD), collects information on cancer incidence and survival from 11 population-based cancer registries, including approximately 14% of the US population. The information collected by SEER includes patient characteristics, primary tumor site, tumor grade, stage at diagnosis (formal American Joint Committee on Cancer [AJCC] staging has only been available in SEER since 1988), first course of treatment (radiation and surgical treatment only), and follow-up for vital status. SEER routinely collects data on the number of nodes evaluated and the number of positive nodes.15 SEER does not report use of chemotherapy, and therefore this could not be evaluated in our study.
Patients The vital status for all patients in our cohort was determined using SEER data through December, 2002. For those patients who died, the cause of death was attributed to colon cancer if CRC was listed as the cause of death.
Statistical Analysis To control for potential confounders and to determine the effect of the number of negative LNs on survival, a proportional hazards model was constructed adjusting for the number of positive LNs, age at diagnosis, sex, tumor grade, race, tumor site (proximal to the splenic flexure/distal to the splenic flexure), and geographic location by registry. Because tumor grade was missing for 5% of the patients, we conducted multivariate analyses with and without the tumor grade variable and compared the results for consistency. The total negative LN number was treated as a categoric variable (four categories corresponding to quartiles). Age was evaluated in 10-year increments. The number of positive LNs was treated as a categoric variable (one LN positive, two LN positive, and three LN positive for patients with stage IIIA and stage IIIB, in quartiles based on the number of positive LN for patients with stage IIIC cancers). We tested the model for possible interaction effects between the number of negative and positive LN to determine if these variables independently influenced survival. Data were analyzed using SAS version 9.1 (SAS Institute, Cary, NC). All statistical tests were two sided, and P values over .05 were considered significant. The Human Subjects Committee of the University of Minnesota's institutional review board determined this study was exempt from review.
Of the 20,702 patients meeting selection criteria (48% male, 52% female) 8.3% had stage IIIA cancers, 60.4% had stage IIIB cancers, and 31.3% had stage IIIC cancers (Table 1). Right-sided (proximal to the splenic flexure) cancers were found in 50% of patients and the majority of patients (74%) had well- or moderately differentiated tumors. The median number of negative LNs for the cohort was seven and the median number of positive LNs was two. As expected, patients with stage IIIC cancer had more positive LNs (median of six) than patients with stage IIIB (median of two) or stage IIIA cancer (median of one; P < .0001). Correlations between number of negative LNs and number of positive LNs were weak or negligible16 (correlation coefficient, .03 to –.16), although due to the large sample size, the correlations were statistically significant.
The mean follow-up time for the cohort was 5.1 years. In total, 65.3% of patients died during follow-up; 44.3% of stage IIIA patients, 63.4% of stage IIIB patients, and 74.6% of stage IIIC patients. The cause of death was attributed to CRC in 60% of patients who died. Disease-specific survival for stage IIIA, IIIB, and IIIC patients stratified by the number of negative LNs is presented in Figures 1A, 1B, and 1C, respectively. For all three substages, patients who had 13 or more negative LNs identified had lower disease-specific mortality than patients who had three or fewer negative LNs identified. For patients with stage IIIB and IIIC cancers, this difference was statistically significant (P < .0001). The 5-year disease-specific survival rates for patients with stage III colon cancer according to the number of negative LNs identified are reported in Table 2. For patients with stage IIIB cancer, those with 13 or more negative LNs evaluated were less likely to die of CRC within 5 years of diagnosis (27%) than those with three or fewer negative LN evaluated (45%), a relative reduction of 40% in deaths from CRC. Similarly, for patients with stage IIIC cancer, those with 13 or more negative LNs evaluated were less likely to die of CRC within 5 years of diagnosis (42%) than those with three or fewer negative LNs evaluated (65%), a relative reduction of 35% in deaths from CRC. For patients with stage IIIA cancer, overall CRC mortality was low and little difference was found between those with 13 or more negative LNs (14%) and those with three or fewer negative LNs (14.5%).
The relationship between the number of positive LNs identified and the risk of CRC death is reported in Table 3. For all three substages, the risk of death from CRC increased as the number of positive LNs increased. Therefore, the effect of the number of negative LNs controlling for the number of positive LN was evaluated in multivariable analyses (Table 4). For stage IIIB and stage IIIC patients, both the number of negative and the number of positive nodes were highly statistically associated with CRC mortality. For stage IIIB and IIIC patients, those with 13 or more negative nodes identified had half the mortality from CRC than those with three or fewer negative nodes evaluated (stage IIIB hazard ratio, 0.53; 95% CI, 0.49 to 0.58; P < .0001; stage IIIC hazard ratio, 0.51; 95% CI, 0.46 to 0.57; P < .0001) regardless of number of positive LNs present. For patients with stage IIIA disease, a lower number of positive LNs was associated with a statistically significantly lower mortality from CRC, while the number of negative LNs did not predict outcome. No statistical interactions between the number of negative and positive LNs were found, indicating that the number of negative LNs and the number of positive LNs influenced mortality independently.
Total number of LNs evaluated has been consistently associated with disease-specific survival in patients with stage II CRC; the more nodes identified the better the long-term survival5,8-14 (although until now, the relationship between number of negative LNs and survival in stage III patients was unknown). Despite this, the mechanism underlying the relationship between number of negative LNs and survival has not been determined. There are several proposed mechanisms. The evaluation of high numbers of LNs in patients with stage II CRC may decrease misclassification of node-positive (stage III) patients as node-negative.10,17,18 Thus poor survival associated with low LN harvest in stage II patients may be due to stage-migration (where survival by stage of disease improves because of better classification, though no change occurs in outcome overall or for a given individual). In addition, in the past, understaged patients would be less likely to receive adjuvant therapies19,20 and this may have resulted in actual worsening of outcome. If stage-migration was the only mechanism, improvement in survival should be primarily found in patients with extremely low LN counts where there is a high likelihood of missing positive LN, and therefore the proportion of patients who have been understaged greatest. However, many studies find the effect of number of nodes increases even at relatively high (> 12) LN counts,4,8,21 and some studies have identified a continuous relationship between number of LNs and survival.21 The number of LNs may be a marker for the adequacy of surgical, pathological, and/or institutional care. Although there is no evidence that more extensive procedures, such as high ligation of the internal mesenteric artery22 or lateral pelvic node dissection, directly result in improved outcome,23 physician interest, experience, and expertise may be associated with more extensive LN harvest, or more intensive evaluation of the surgical specimen for LNs. These factors may relate to improved outcome and may reflect institutional interest in the treatment of colon cancer patients. Previous studies have identified factors such as volume of patients treated to be important predictors of outcome for CRC cancer patients.24,25 Thus, number of LNs may be associated with better care overall and may not influence outcome directly. However, even single center studies identify a relationship between number of LNs and outcome, indicating that when other factors are kept constant, LN number is still important.5,6,26 Based on anatomic studies, the number of LNs present is known to vary between patients, and is influenced by disease state.27-29 It is therefore plausible that interactions between tumor and host might influence the number of assessable LNs; tumor factors may stimulate LNs to increase in size, facilitating the ability of pathologists to detect nodes, germinal centers may be stimulated by tumor to develop into LN, increasing the actual number present. If true, the number of negative LNs would be a marker of underlying tumor-host interactions, which may have an independent effect on survival. More than likely, the consistent and strong relationship demonstrated between LN number and survival in stage II colorectal cancers is influenced by all three factors (stage migration, quality care, and tumor-host interactions)—however understanding the mechanism is important to determining what action, if any, is necessary to improve patient outcomes. If stage migration is the primary mechanism, interventions to increase LN counts, while improving the ability to predict outcome, will not result in any other patient benefit—particularly with the increasing use of adjuvant therapy in stage II patients. If the mechanism is primarily due to better care, ensuring all patients have access to quality cancer care should be the target of intervention; attempting to achieve a given LN count in each patient will be unlikely to result in benefit. If tumor-host interactions are important, the number of LNs would be an independent prognostic factor that should be considered when determining the need for adjuvant therapy. In addition, a better understanding of this mechanism might lead to novel therapeutic interventions for future patients. Our study supports the extension of findings regarding the relationship of number of negative LNs and survival to stage III patients. Other authors, considering total number of LNs, have produced conflicting results when relating the effect of LN harvest to survival.4,5,9,11,13,14 This is not unexpected, as the total number of LNs is correlated to both the number of positive LNs (where an increased number is associated with worse prognosis) and the number of negative LNs (where, as demonstrated by our study, an increased number is associated with better prognosis). Our study, specifically evaluating the effect of the number of negative nodes (versus the total number of nodes), demonstrates an independent effect of both number of negative and number of positive LNs in patients with stage IIIB and IIIC colon cancer. As the number of negative LNs identified increased, disease-specific survival improved. The effect was substantial; after controlling for the number of positive LNs, patients with 13 or more LNs identified had twice the likelihood of surviving 5 years as those with three or fewer LNs identified. The effect was independent of the number of positive LNs, indicating that the number of negative LNs has independent prognostic implications for stage III patients. The results of this study suggest, that when considered together, the number of positive LNs and negative LNs will provide better prognostic information than the number of positive LNs alone. Our study does not clarify the mechanism underlying the relationship between negative node number and survival, but does, to some extent, question the prevailing dogma that the relationship is simply due to stage migration or quality of care. Stage migration likely plays a role, however the number of negative nodes appears important even in patients who have a comparatively large number of total nodes evaluated (ie, stage IIIC patients). The number of negative LNs may be a marker of the adequacy of LN evaluation, which in turn may reflect the quality of pathologic and surgical care. While the influence of processes of care could not be evaluated using the SEER data, they are clearly important, and may influence LN number and prognosis. However, particularly as our cohort excluded patients with rectal cancer, the magnitude of improvement in outcome demonstrated in our study is unlikely to be related to variations in surgical and pathologic standards only. This study does not provide evidence to support or refute the hypothesis that tumor-host interactions are important; however, recent evidence supports the plausibility of this hypothesis. Pages et al30 evaluated 959 colorectal cancer specimens and found that the immune response (in particular effector memory T cells) within the tumors was associated with a less advanced pathologic stage and prolonged survival. There was no association between the number of negative LNs identified and disease-specific survival for patients with stage IIIA disease. The reason for this is unclear. Overall prognosis for these patients was excellent, and thus our study, although large, may have insufficient power. In addition, patients with T1 and T2 CRC have been found to have, on average, fewer LNs evaluated than patients with T3 or T4 tumors.31 The relationship between negative LN number and prognosis may differ between patients with superficial involvement of the bowel wall and patients with full thickness involvement. We did not evaluate lymph node ratio (the proportion of positive LN) as a prognostic variable in our study. Lymph node ratio and number of negative LNs are clearly related and both appear to be important in terms of prognosis. Berger et al32 have demonstrated that LN ratio is an important predictor of survival in a group of patients with stage II and stage III colon cancer, however this was true only when at least 10 nodes recovered. We did not identify any particular cutoff—rather the effect of number of negative and number of positive LNs was seen at low numbers and at high numbers. Developing the optimal prognostic model was beyond the scope of this study, but should be the focus of future research. This study used cancer registry data; detailed patient or tumor information, data regarding surgical and pathologic technique, or the use of adjuvant chemotherapy were not available. However, SEER is population based and includes patients from diverse locations and practices in the United States, making our findings broadly applicable. In addition, the large numbers of stage III colon cancer patients included in SEER permitted the evaluation of patients by stage III substage. Although chemotherapy data are unavailable, as the relationship between number of negative LNs and survival in stage III colon cancer has not been previously described, it is unlikely that treatment was given on the basis of negative LN number. In addition, if chemotherapy was given on the basis of low negative LN counts, this would tend to bias the study towards the null hypothesis. Importantly, the limitations of SEER data do not effect the main conclusion, that the number of negative LNs is an important independent prognostic factor for patients with stage IIIB and stage IIIC colon cancer. The mechanism underlying this relationship requires further study.
The authors indicated no potential conflicts of interest.
Supported by an American Society of Clinical Oncology career development award (N.N.B.). Presented in part at the American Society of Clinical Oncology Gastrointestinal Cancers Symposium, San Francisco, CA, January 26-28, 2006. Authors' disclosures of potential conflicts of interest and author contributions are found at the end of this article.
Related Correspondence
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2006 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
TOP
ABSTRACT
INTRODUCTION
