Colon Cancer Survival Is Associated With Decreasing Ratio of Metastatic to Examined Lymph Nodes
http://www.100md.com
《临床肿瘤学》
the Department of Surgery, Thomas Jefferson University
Fox Chase Cancer Center
Eastern Cooperative Oncology Group, Philadelphia, PA
Brooke Army Medical Center, Fort Sam Houston, TX
Department of Surgical Oncology, Cancer and Leukemia Group B, Chicago, IL
Southwest Oncology Group, Ann Arbor, MI
ABSTRACT
PURPOSE: Colorectal cancer is the second leading cause of cancer deaths in the United States, with poor survival predicted by regional lymph node (LN) metastasis. The impact of LN ratio (LNR) on survival is unknown in this disease.
PATIENTS AND METHODS: We analyzed data from Intergroup trial 0089 of adjuvant chemotherapy for stage II and III patients with colon cancer, in which all patients received fluorouracil-based therapy. Survival was similar for all arms of the study, allowing us to evaluate all patients together. End points included overall survival (OS), cancer-specific survival (CSS), and disease-free survival (DFS). Multivariate analyses were performed on all patients and on groups according to LNR quartiles (LNR: < 0.05, 0.05 to 0.19, 0.2 to 0.39, and 0.4 to 1.0). Covariates included in the models were age, sex, tumor stage, grade, histology, number of positive LNs, number of LNs removed, and LNR.
RESULTS: The median age was 63.7 years, and the median number of LNs removed was 11. In the multivariate analysis, LNR was a significant factor for OS, DFS, and CSS in patients with 10 to 15 LN and more than 15 LN removed but not for patients with less than 10 LN removed. Using quartiles, LNR maintained its significance for all three end points when patients were grouped by node status.
CONCLUSION: After curative resection for colorectal cancer, the LNR is an important prognostic factor and should be used in stratification schemes for future clinical trials investigating adjuvant treatments.
INTRODUCTION
Adenocarcinoma of the colon is the fourth most common malignancy in humans. Surgery remains the primary treatment for patients with this disease. En-bloc removal of the colon with associated mesenteric lymph nodes (LN) is essential. Survival decreases with increasing LN involvement
Recent literature from many malignancies emphasizes the importance of the number of LNs examined in determining prognosis. In colon and rectal cancer, staging accuracy and survival are improved with increasing nodal examination and analysis.1-3 These analyses have also been borne out in patients with lung, breast, and bladder cancer.4-6 The number of LNs reported with colectomy varies widely and may be a result of variation in surgical technique, the thoroughness of the pathologist in finding the nodes in the specimen, or the actual number of regional LNs. Additionally, there are two opposing views on the importance of lymphadenectomy in determining survival; some investigators believe that a full lymphadenectomy has a therapeutic benefit, whereas others believe that it simply provides more accurate staging.7
There is recent evidence that the ratio of metastatic to examined LNs (LNR) is an important prognostic factor in malignancies such as gastric cancer, esophageal cancer, and breast cancer. Two recent reports from Japan and Spain have demonstrated that increasing LNR is a significant prognostic indicator for worse survival in gastric cancer.8,9 A recent publication also demonstrated the importance of LNR in pancreatic cancer.10
In patients, with LN metastases, adjuvant chemotherapy has been shown to improve survival. INT-0089, a large Intergroup trial of adjuvant chemotherapy, evaluated fluorouracil, levamisole, and leucovorin in patients with high-risk stage II or III colon cancer.11 Because survival was similar for all four arms of the study, this cohort of patients could be evaluated as a single group. We analyzed this population of patients to determine the relationship between the LNR and survival.
PATIENTS AND METHODS
Eligibility
INT-0089 opened for enrollment in August 1988 and closed in July 1992 after meeting its accrual goal. In all, 3,759 patients were entered onto the study, of whom 3,561 were eligible. Patients not receiving assigned therapy (50 patients) and who were otherwise eligible for entry onto the study were included in the final analyses on an intent-to-treat basis. The final analysis of the trial's arms was completed in November 1997, and the information was presented at the Annual Meeting of the American Society of Clinical Oncology in 1998.11
Patients were eligible for the adjuvant trial if they had histologic proof of adenocarcinoma of the colon and complete resection of the primary tumor with no gross or microscopic evidence of residual disease. The tumor must have been resected en bloc if it was adherent to any associated organs or structures. The gross inferior margin of the tumor must have been above the peritoneal reflection to exclude a rectal primary. The primary tumor must have shown evidence of poor prognosis, with either nodal involvement (Dukes' C) or Dukes' B2 with evidence of perforation or obstruction. Regional or mesenteric implants, if present, must have been resected en bloc. Patients with discontiguous or distant disease were ineligible.
Patients were considered to be ineligible if they received concurrent radiation or chemotherapy, prior fluorouracil therapy, or prior radiation or chemotherapy for the same malignant disease. Patients with a concurrent or previous malignant tumor within the previous 3 years (except for superficial squamous or basal cell carcinoma of the skin or carcinoma-in-situ of the cervix) were ineligible. Patients were entered onto the study and randomly assigned to a treatment arm between 21 and 35 days postoperatively. Patients were treated at institutions affiliated with the following cooperative groups: Eastern Cooperative Oncology Group, Southwest Oncology Group, Cancer and Leukemia Group B, and the Australasia Gastrointestinal Trials Group.
Random Assignment
Patients were stratified according to the extent of bowel-wall involvement, nodal stage, presence of perforation or obstruction, and presence of resected peritoneal implants. Patients with more than one simultaneously diagnosed colonic primary tumor were classified according to the most advanced-stage lesion.
Treatment
All patients underwent standard surgical resection to achieve a complete extirpation of the tumor; the operation performed was determined retrospectively from the operative report and the pathology report. Postoperatively, patients were randomly assigned to one of four fluorouracil-based chemotherapy arms depicted in Figure 1.
Follow-Up
Patients were followed up after completion of their chemotherapy every 3 months for 1 year after their resection, at which point they were followed up on a 6-month basis. After 5 years, patients were seen annually. During their follow-up visits, patients gave their histories and underwent physical examination. Blood work was obtained to evaluate their CBC count as well as to perform liver function tests and an optional carcinoembryonic antigen test. A chest x-ray was obtained every 6 months. A proctoscopic examination and barium enema or a colonoscopy were performed at the 6-month and 12-month visits and then at least every 24 months.
Statistical Methods
End points for our secondary analysis of INT-0089 were overall survival (OS), cause-specific survival (CSS), and disease-free survival (DFS). Rates for OS, CSS, and DFS were estimated using the Kaplan-Meier method12 measuring time from the date of surgery to the date of death, date of death secondary to colon cancer, and date of failure at any site, respectively. Censoring in accordance with the Kaplan-Meier method was performed for those patients who had not experienced the end point. The log-rank test13 was used to compare overall curves, and Cox proportional hazards regression14 was used to multivariately assess predictors of outcome. Covariates considered in this analysis were age (continuous), sex (dichotomous), the total number of LNs recovered (continuous), the total number of positive LNs (continuous), the LNR (continuous), tumor stage (T1 and T2 v T3 v T4), and adjuvant therapy regimen. The LNR, as determined by dividing the total positive LNs by the total examined LNs, was also divided into the following four groups based on quartiles (LNR with cut points based on 25%, median, and 75%): LNR less than 0.05 (n = 1,052), 0.05 to 0.19 (n = 707), 0.2 to 0.39 (n = 894), and 0.4 to 1.0 (n = 758).
RESULTS
Data from 3,557 patients from INT-0089 were available for review. Of these records, 146 were excluded because of missing data regarding the exact numbers of LNs removed or discrepancy with surgical dates or procedure, thus impacting survival calculations. The remaining 3,411 patients constitute this analysis. Median follow-up time of the population was 79 months (range, 1 to 131 months). The cohort analyzed consisted of 1,849 men and 1,562 women, with a median age of 63.7 years (range, 15 to 90 years). Table 1 lists the demographic characteristics according to tumor stage. Of the 3,411 assessable patients, 648 (19%) were N0, 1,857 (54%) were N1, and 906 (27%) were N2. As would be expected by the criterion for entry onto the adjuvant trial, the majority of patients were node positive and had an advanced tumor stage. The node-positive and node-negative groups were similar in regard to the mean and median number of LNs recovered (13 and 11 LNs, respectively).
In a prior subset analysis of this trial, we demonstrated that increasing number of nodes harvested as well as increasing number of positive nodes had a dramatic impact on survival; this impact was greatest in the patients with positive nodes.3 To determine the impact of LNR, we first examined the impact of LNR as a continuous variable on the entire population of patients from INT-0089. The multivariate stepwise Cox regression analysis is depicted in Table 2. For OS, age, sex, number of nodes removed, pathologic stage, number of positive nodes, and LNR were all highly significant predictors of survival, with the number of positive nodes being the most significant predictor of OS. In terms of DFS, the same variables were significant except for age. Finally, for CSS, all of the covariates maintained their significance; however, the most significant predictor of survival in this model was tumor stage.
To determine the impact of LNR within the confines of the current American Joint Committee on Cancer (AJCC) staging system, the multivariate analysis was performed after grouping patients into the N0, N1, and N2 categories. This analysis of the LNR as a continuous variable is shown in Table 3. As one can see, the greatest impact in node-negative patients came from patient age and tumor stage. As the node stage increased to N1 and N2, the number of nodes harvested and the number of positive nodes had increasing contribution to OS, DFS, and CSS. The only impact of the LNR was in CSS for patients with N2 disease. We also felt that it was important to analyze the impact of LNR based on the amount of LNs that were removed with the specimen to demonstrate the validity of this variable in patients with low, intermediate, and high node counts. Therefore, we repeated the stepwise multivariate analysis for patients with less than 10, 10 to 15, and more than 15 LNs removed (Table 4). This analysis demonstrated the number of positive nodes to be the most important prognostic factor for low LN count patients, whereas the LNR was the most important prognostic factor in intermediate and high LN count patients.
On the basis of previous work in other disease sites such as gastric cancer, we evaluated patients in LNR groups based on quartiles (LNR < 0.05, 0.05 to 0.19, 0.2 to 0.39, and 0.4 to 1.0). OS, DFS, and CSS were determined for all patients and N1 and N2 patients (Figs 2, 3, and 4). For all three survival end points examined, survival decreased with increasing LNR (P < .0001 for all curves). Therefore, when OS was examined for all patients in the trial, the 5-year survival rate was 79% in patients with an LNR less than 0.05, and this decreased to 52% (Table 5) in patients with a high LNR (> 0.39). There were no patients with N2 disease who had an LNR less than 0.05. However, within the N2 subgroup, one can see a dramatic difference in 5-year survival end points between patients with a lower LNR (73% OS rate for a patient with N2 disease and an LNR between 0.05 and 0.19) and patients with a higher LNR (45% OS rate for a patient with N2 disease and an LNR between 0.4 and 1.0). These differences are highlighted in Table 5 and show the importance of the stratification of patients by LNR.
DISCUSSION
There is increasing evidence in multiple organ sites that the number of LNs evaluated by the pathologist has an important impact on survival.7 The prevailing theory is that a thorough LN evaluation results in more accurate staging and, thus, better determination of prognosis. This concept of stage migration was first described, regarding emerging diagnostic techniques, by Feinstein et al15 in 1985 and was termed the Will Rogers Phenomenon. In general, there are two opposing views regarding the role of lymphadenectomy, particularly in breast cancer. The traditional Halstedian view claims that lymphadenectomy is important for staging and survival. Others feel that cancer is a systemic disease and that lymphadenectomy is useful only as a staging tool and does not affect survival.7 Many factors may contribute to the increasing number of LNs examined by the pathologist and its effect on survival. These include a more extensive lymphadenectomy performed by the surgeon and a more thorough examination of the specimen by the pathologist to help determine the node stage. In colon cancer operations, improved staging may be seen because the surgeon who performs a meticulous and extensive node dissection is usually one who thoroughly stages the patient preoperatively and intraoperatively extensively inspects for peritoneal implants and liver metastases and performs a better technical operation with a complete en-bloc resection of involved structures.
Along with the total number of LNs examined, another related measure is LNR. This measurement has gained prominence in the gastric cancer literature. In an examination of 475 Japanese gastric cancer patients, Inoue et al16 demonstrated that a ratio-based classification of LNs (< 25%, 25% to 50%, and > 50%) was the most significant prognostic factor and showed superiority to the AJCC classification with respect to stage migration. In another Japanese study, Bando et al9 divided 650 patients into four groups based on LNR. These ratios were 0, 0 to 10, 10 to 25, and more than 25; the 5-year survival rates were 86%, 68%, 35%, and 16%, respectively, and LNR was an independent prognostic factor.9 To our knowledge, this is the first report demonstrating a prognostic significance for the metastatic to examined LNR in patients with adenocarcinoma of the colon.
When one closely examines the data in Table 5, it is easy to see how dramatically survival and recurrence rates change within each category based on the LNR. This is certainly a reflection of the number of positive nodes and the total number of evaluated nodes, which have already been shown to be important factors for determining survival for colon and rectal cancer.1,3 Therefore, as the number of positive nodes increases (and the total nodes examined remains unchanged), the LNR will increase. As such, there are obviously clear subgroups within the N1 and N2 categories that must be considered. Within the N1 and N2 classifications, there are drastic changes in survival and recurrence rates based on whether the LNR is less than 5%, 5% to 20%, 20% to 40%, or more than 40%.
It is clear from our data and previous examinations of LN data from the Intergroup trial3 that the total number of positive nodes is one of the most important prognostic factors. However, it is interesting to evaluate the data in terms of stage migration. As noted by Inoue et al,16 the use of LNRs in the staging of gastric cancer decreases the incidence of this phenomenon. They compared the AJCC staging of LNs in patients to a ratio-based classification and found that there was less stage migration. Additionally, Bando et al9 demonstrated a 50% decrease in stage migration using a different ratio-based system. Thus, in examining our data, we see that the 5-year OS rate for an N1 patient with an LNR of more than 40% is much worse (60%) than the rate for an N2 patient with an LNR of less than 20% (73%). This difference is not currently reflected in the AJCC staging system.
It is possible that we can simplify stratifications of survival based on positive nodes and nodes examined by combining them into a single parameter of percentage of nodes involved. However, a ratio-based system is not a substitute for an adequate LN dissection; a minimum of 15 nodes should still be examined to obtain an accurate assessment of LN involvement. As we have demonstrated, for patients with fewer than 10 LNs recovered, LNR is not an important prognostic factor. Therefore, determination of the LNR cannot compensate for an inadequate LN dissection in colon cancer resections. In patients with low LN yields, the most important prognostic factor remains the total number of positive nodes, as demonstrated by Le Voyer et al.3
However, in patients with 10 to 15 LNs and more than 15 LNs removed, we have demonstrated that the LNR is the most significant prognostic factor for both OS and DFS. This demonstrates that, even in patients who are undergoing extensive resections with at least 10 LNs examined, the LNR has a dramatic impact on survival and recurrence. Finally, another potential role of a ratio-based classification is for stratification and prognostic information as part of a nomogram system to predict local and/or systemic recurrences, which could help determine the type and extent of adjuvant chemotherapy.
In conclusion, LNR is an important prognostic factor for adenocarcinoma of the colon. Evaluation of the LN data from a large Intergroup trial demonstrates the importance of the number of positive LNs, number of examined LNs, and the LNR. LNR may decrease the amount of stage migration, which can result when the pathologist examines an inadequate number of LNs. Future clinical trials in colon cancer should prospectively evaluate LNR as an important prognostic factor.
Authors' Disclosures of Potential Conflicts of Interest
The authors indicated no potential conflicts of interest.
NOTES
Presented at the 58th Annual Cancer Symposium of the Society of Surgical Oncology, Atlanta, GA, March 3-5, 2005.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Tepper JE, O'Connell MJ, Niedzwiecki D, et al: Impact of number of nodes retrieved on outcome in patients with rectal cancer. J Clin Oncol 19:157-163, 2001
Joseph NE, Sigurdson ER, Hanlon AL, et al: Accuracy of determining nodal negativity in colorectal cancer on the basis of the number of nodes retrieved on resection. Ann Surg Oncol 10:213-218, 2003
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 21:2912-2919, 2003
Gajra A, Newman N, Gamble GP, et al: Effect of number of lymph nodes sampled on outcome in patients with stage I non-small-cell lung cancer. J Clin Oncol 21:1029-1034, 2003
Herr HW, Bochner BH, Dalbagni G, et al: Impact of the number of lymph nodes retrieved on outcome in patients with muscle invasive bladder cancer. J Urol 167:1295-1298, 2002
Weir L, Speers C, D'yachkova Y, et al: Prognostic significance of the number of axillary lymph nodes removed in patients with node-negative breast cancer. J Clin Oncol 20:1793-1799, 2002
Sigurdson ER: Lymph node dissection: Is it diagnostic or therapeutic? J Clin Oncol 21:965-967, 2003
Sierra A, Regueira FM, Hernandez-Lizoain JL, et al: Role of the extended lymphadenectomy in gastric cancer surgery: Experience in a single institution. Ann Surg Oncol 10:219-226, 2003
Bando E, Yonemura Y, Taniguchi K, et al: Outcome of ratio of lymph node metastasis in gastric carcinoma. Ann Surg Oncol 9:775-784, 2002
Berger AC, Watson JC, Ross EA, et al: The metastatic/examined lymph node ratio is an important prognostic factor after pancreaticoduodenectomy for pancreatic adenocarcinoma. Am Surg 70:235-240, 2004
Haller DG, Catalano PJ, Macdonald JS, et al: Fluorouracil (FU), leucovorin (LV) and levamisole (LEV) adjuvant therapy for colon cancer: Five-year final report of INT-0089. Proc Am Soc Clin Oncol 17:256a, 1998 (abstr 962)
Kaplan EL, Meier P: Nonparametric estimation from incomplete observations. J Am Stat Assoc 53:457-481, 1958
Kalbfleisch JD, Prentice RL: The Statistical Analysis of Failure Time Data. New York, NY, John Wiley, 1980
Cox DR: Regression models and life tables. J R Stat Soc B 34:187-220, 1972
Feinstein AR, Sosin DM, Wells CK: The Will Rogers phenomenon: Stage migration and new diagnostic techniques as a source of misleading statistics for survival in cancer. N Engl J Med 312:1604-1608, 1985
Inoue K, Nakane Y, Iiyama H, et al: The superiority of ratio-based lymph node staging in gastric carcinoma. Ann Surg Oncol 9:27-34, 2002(Adam C. Berger, Elin R. S)
Fox Chase Cancer Center
Eastern Cooperative Oncology Group, Philadelphia, PA
Brooke Army Medical Center, Fort Sam Houston, TX
Department of Surgical Oncology, Cancer and Leukemia Group B, Chicago, IL
Southwest Oncology Group, Ann Arbor, MI
ABSTRACT
PURPOSE: Colorectal cancer is the second leading cause of cancer deaths in the United States, with poor survival predicted by regional lymph node (LN) metastasis. The impact of LN ratio (LNR) on survival is unknown in this disease.
PATIENTS AND METHODS: We analyzed data from Intergroup trial 0089 of adjuvant chemotherapy for stage II and III patients with colon cancer, in which all patients received fluorouracil-based therapy. Survival was similar for all arms of the study, allowing us to evaluate all patients together. End points included overall survival (OS), cancer-specific survival (CSS), and disease-free survival (DFS). Multivariate analyses were performed on all patients and on groups according to LNR quartiles (LNR: < 0.05, 0.05 to 0.19, 0.2 to 0.39, and 0.4 to 1.0). Covariates included in the models were age, sex, tumor stage, grade, histology, number of positive LNs, number of LNs removed, and LNR.
RESULTS: The median age was 63.7 years, and the median number of LNs removed was 11. In the multivariate analysis, LNR was a significant factor for OS, DFS, and CSS in patients with 10 to 15 LN and more than 15 LN removed but not for patients with less than 10 LN removed. Using quartiles, LNR maintained its significance for all three end points when patients were grouped by node status.
CONCLUSION: After curative resection for colorectal cancer, the LNR is an important prognostic factor and should be used in stratification schemes for future clinical trials investigating adjuvant treatments.
INTRODUCTION
Adenocarcinoma of the colon is the fourth most common malignancy in humans. Surgery remains the primary treatment for patients with this disease. En-bloc removal of the colon with associated mesenteric lymph nodes (LN) is essential. Survival decreases with increasing LN involvement
Recent literature from many malignancies emphasizes the importance of the number of LNs examined in determining prognosis. In colon and rectal cancer, staging accuracy and survival are improved with increasing nodal examination and analysis.1-3 These analyses have also been borne out in patients with lung, breast, and bladder cancer.4-6 The number of LNs reported with colectomy varies widely and may be a result of variation in surgical technique, the thoroughness of the pathologist in finding the nodes in the specimen, or the actual number of regional LNs. Additionally, there are two opposing views on the importance of lymphadenectomy in determining survival; some investigators believe that a full lymphadenectomy has a therapeutic benefit, whereas others believe that it simply provides more accurate staging.7
There is recent evidence that the ratio of metastatic to examined LNs (LNR) is an important prognostic factor in malignancies such as gastric cancer, esophageal cancer, and breast cancer. Two recent reports from Japan and Spain have demonstrated that increasing LNR is a significant prognostic indicator for worse survival in gastric cancer.8,9 A recent publication also demonstrated the importance of LNR in pancreatic cancer.10
In patients, with LN metastases, adjuvant chemotherapy has been shown to improve survival. INT-0089, a large Intergroup trial of adjuvant chemotherapy, evaluated fluorouracil, levamisole, and leucovorin in patients with high-risk stage II or III colon cancer.11 Because survival was similar for all four arms of the study, this cohort of patients could be evaluated as a single group. We analyzed this population of patients to determine the relationship between the LNR and survival.
PATIENTS AND METHODS
Eligibility
INT-0089 opened for enrollment in August 1988 and closed in July 1992 after meeting its accrual goal. In all, 3,759 patients were entered onto the study, of whom 3,561 were eligible. Patients not receiving assigned therapy (50 patients) and who were otherwise eligible for entry onto the study were included in the final analyses on an intent-to-treat basis. The final analysis of the trial's arms was completed in November 1997, and the information was presented at the Annual Meeting of the American Society of Clinical Oncology in 1998.11
Patients were eligible for the adjuvant trial if they had histologic proof of adenocarcinoma of the colon and complete resection of the primary tumor with no gross or microscopic evidence of residual disease. The tumor must have been resected en bloc if it was adherent to any associated organs or structures. The gross inferior margin of the tumor must have been above the peritoneal reflection to exclude a rectal primary. The primary tumor must have shown evidence of poor prognosis, with either nodal involvement (Dukes' C) or Dukes' B2 with evidence of perforation or obstruction. Regional or mesenteric implants, if present, must have been resected en bloc. Patients with discontiguous or distant disease were ineligible.
Patients were considered to be ineligible if they received concurrent radiation or chemotherapy, prior fluorouracil therapy, or prior radiation or chemotherapy for the same malignant disease. Patients with a concurrent or previous malignant tumor within the previous 3 years (except for superficial squamous or basal cell carcinoma of the skin or carcinoma-in-situ of the cervix) were ineligible. Patients were entered onto the study and randomly assigned to a treatment arm between 21 and 35 days postoperatively. Patients were treated at institutions affiliated with the following cooperative groups: Eastern Cooperative Oncology Group, Southwest Oncology Group, Cancer and Leukemia Group B, and the Australasia Gastrointestinal Trials Group.
Random Assignment
Patients were stratified according to the extent of bowel-wall involvement, nodal stage, presence of perforation or obstruction, and presence of resected peritoneal implants. Patients with more than one simultaneously diagnosed colonic primary tumor were classified according to the most advanced-stage lesion.
Treatment
All patients underwent standard surgical resection to achieve a complete extirpation of the tumor; the operation performed was determined retrospectively from the operative report and the pathology report. Postoperatively, patients were randomly assigned to one of four fluorouracil-based chemotherapy arms depicted in Figure 1.
Follow-Up
Patients were followed up after completion of their chemotherapy every 3 months for 1 year after their resection, at which point they were followed up on a 6-month basis. After 5 years, patients were seen annually. During their follow-up visits, patients gave their histories and underwent physical examination. Blood work was obtained to evaluate their CBC count as well as to perform liver function tests and an optional carcinoembryonic antigen test. A chest x-ray was obtained every 6 months. A proctoscopic examination and barium enema or a colonoscopy were performed at the 6-month and 12-month visits and then at least every 24 months.
Statistical Methods
End points for our secondary analysis of INT-0089 were overall survival (OS), cause-specific survival (CSS), and disease-free survival (DFS). Rates for OS, CSS, and DFS were estimated using the Kaplan-Meier method12 measuring time from the date of surgery to the date of death, date of death secondary to colon cancer, and date of failure at any site, respectively. Censoring in accordance with the Kaplan-Meier method was performed for those patients who had not experienced the end point. The log-rank test13 was used to compare overall curves, and Cox proportional hazards regression14 was used to multivariately assess predictors of outcome. Covariates considered in this analysis were age (continuous), sex (dichotomous), the total number of LNs recovered (continuous), the total number of positive LNs (continuous), the LNR (continuous), tumor stage (T1 and T2 v T3 v T4), and adjuvant therapy regimen. The LNR, as determined by dividing the total positive LNs by the total examined LNs, was also divided into the following four groups based on quartiles (LNR with cut points based on 25%, median, and 75%): LNR less than 0.05 (n = 1,052), 0.05 to 0.19 (n = 707), 0.2 to 0.39 (n = 894), and 0.4 to 1.0 (n = 758).
RESULTS
Data from 3,557 patients from INT-0089 were available for review. Of these records, 146 were excluded because of missing data regarding the exact numbers of LNs removed or discrepancy with surgical dates or procedure, thus impacting survival calculations. The remaining 3,411 patients constitute this analysis. Median follow-up time of the population was 79 months (range, 1 to 131 months). The cohort analyzed consisted of 1,849 men and 1,562 women, with a median age of 63.7 years (range, 15 to 90 years). Table 1 lists the demographic characteristics according to tumor stage. Of the 3,411 assessable patients, 648 (19%) were N0, 1,857 (54%) were N1, and 906 (27%) were N2. As would be expected by the criterion for entry onto the adjuvant trial, the majority of patients were node positive and had an advanced tumor stage. The node-positive and node-negative groups were similar in regard to the mean and median number of LNs recovered (13 and 11 LNs, respectively).
In a prior subset analysis of this trial, we demonstrated that increasing number of nodes harvested as well as increasing number of positive nodes had a dramatic impact on survival; this impact was greatest in the patients with positive nodes.3 To determine the impact of LNR, we first examined the impact of LNR as a continuous variable on the entire population of patients from INT-0089. The multivariate stepwise Cox regression analysis is depicted in Table 2. For OS, age, sex, number of nodes removed, pathologic stage, number of positive nodes, and LNR were all highly significant predictors of survival, with the number of positive nodes being the most significant predictor of OS. In terms of DFS, the same variables were significant except for age. Finally, for CSS, all of the covariates maintained their significance; however, the most significant predictor of survival in this model was tumor stage.
To determine the impact of LNR within the confines of the current American Joint Committee on Cancer (AJCC) staging system, the multivariate analysis was performed after grouping patients into the N0, N1, and N2 categories. This analysis of the LNR as a continuous variable is shown in Table 3. As one can see, the greatest impact in node-negative patients came from patient age and tumor stage. As the node stage increased to N1 and N2, the number of nodes harvested and the number of positive nodes had increasing contribution to OS, DFS, and CSS. The only impact of the LNR was in CSS for patients with N2 disease. We also felt that it was important to analyze the impact of LNR based on the amount of LNs that were removed with the specimen to demonstrate the validity of this variable in patients with low, intermediate, and high node counts. Therefore, we repeated the stepwise multivariate analysis for patients with less than 10, 10 to 15, and more than 15 LNs removed (Table 4). This analysis demonstrated the number of positive nodes to be the most important prognostic factor for low LN count patients, whereas the LNR was the most important prognostic factor in intermediate and high LN count patients.
On the basis of previous work in other disease sites such as gastric cancer, we evaluated patients in LNR groups based on quartiles (LNR < 0.05, 0.05 to 0.19, 0.2 to 0.39, and 0.4 to 1.0). OS, DFS, and CSS were determined for all patients and N1 and N2 patients (Figs 2, 3, and 4). For all three survival end points examined, survival decreased with increasing LNR (P < .0001 for all curves). Therefore, when OS was examined for all patients in the trial, the 5-year survival rate was 79% in patients with an LNR less than 0.05, and this decreased to 52% (Table 5) in patients with a high LNR (> 0.39). There were no patients with N2 disease who had an LNR less than 0.05. However, within the N2 subgroup, one can see a dramatic difference in 5-year survival end points between patients with a lower LNR (73% OS rate for a patient with N2 disease and an LNR between 0.05 and 0.19) and patients with a higher LNR (45% OS rate for a patient with N2 disease and an LNR between 0.4 and 1.0). These differences are highlighted in Table 5 and show the importance of the stratification of patients by LNR.
DISCUSSION
There is increasing evidence in multiple organ sites that the number of LNs evaluated by the pathologist has an important impact on survival.7 The prevailing theory is that a thorough LN evaluation results in more accurate staging and, thus, better determination of prognosis. This concept of stage migration was first described, regarding emerging diagnostic techniques, by Feinstein et al15 in 1985 and was termed the Will Rogers Phenomenon. In general, there are two opposing views regarding the role of lymphadenectomy, particularly in breast cancer. The traditional Halstedian view claims that lymphadenectomy is important for staging and survival. Others feel that cancer is a systemic disease and that lymphadenectomy is useful only as a staging tool and does not affect survival.7 Many factors may contribute to the increasing number of LNs examined by the pathologist and its effect on survival. These include a more extensive lymphadenectomy performed by the surgeon and a more thorough examination of the specimen by the pathologist to help determine the node stage. In colon cancer operations, improved staging may be seen because the surgeon who performs a meticulous and extensive node dissection is usually one who thoroughly stages the patient preoperatively and intraoperatively extensively inspects for peritoneal implants and liver metastases and performs a better technical operation with a complete en-bloc resection of involved structures.
Along with the total number of LNs examined, another related measure is LNR. This measurement has gained prominence in the gastric cancer literature. In an examination of 475 Japanese gastric cancer patients, Inoue et al16 demonstrated that a ratio-based classification of LNs (< 25%, 25% to 50%, and > 50%) was the most significant prognostic factor and showed superiority to the AJCC classification with respect to stage migration. In another Japanese study, Bando et al9 divided 650 patients into four groups based on LNR. These ratios were 0, 0 to 10, 10 to 25, and more than 25; the 5-year survival rates were 86%, 68%, 35%, and 16%, respectively, and LNR was an independent prognostic factor.9 To our knowledge, this is the first report demonstrating a prognostic significance for the metastatic to examined LNR in patients with adenocarcinoma of the colon.
When one closely examines the data in Table 5, it is easy to see how dramatically survival and recurrence rates change within each category based on the LNR. This is certainly a reflection of the number of positive nodes and the total number of evaluated nodes, which have already been shown to be important factors for determining survival for colon and rectal cancer.1,3 Therefore, as the number of positive nodes increases (and the total nodes examined remains unchanged), the LNR will increase. As such, there are obviously clear subgroups within the N1 and N2 categories that must be considered. Within the N1 and N2 classifications, there are drastic changes in survival and recurrence rates based on whether the LNR is less than 5%, 5% to 20%, 20% to 40%, or more than 40%.
It is clear from our data and previous examinations of LN data from the Intergroup trial3 that the total number of positive nodes is one of the most important prognostic factors. However, it is interesting to evaluate the data in terms of stage migration. As noted by Inoue et al,16 the use of LNRs in the staging of gastric cancer decreases the incidence of this phenomenon. They compared the AJCC staging of LNs in patients to a ratio-based classification and found that there was less stage migration. Additionally, Bando et al9 demonstrated a 50% decrease in stage migration using a different ratio-based system. Thus, in examining our data, we see that the 5-year OS rate for an N1 patient with an LNR of more than 40% is much worse (60%) than the rate for an N2 patient with an LNR of less than 20% (73%). This difference is not currently reflected in the AJCC staging system.
It is possible that we can simplify stratifications of survival based on positive nodes and nodes examined by combining them into a single parameter of percentage of nodes involved. However, a ratio-based system is not a substitute for an adequate LN dissection; a minimum of 15 nodes should still be examined to obtain an accurate assessment of LN involvement. As we have demonstrated, for patients with fewer than 10 LNs recovered, LNR is not an important prognostic factor. Therefore, determination of the LNR cannot compensate for an inadequate LN dissection in colon cancer resections. In patients with low LN yields, the most important prognostic factor remains the total number of positive nodes, as demonstrated by Le Voyer et al.3
However, in patients with 10 to 15 LNs and more than 15 LNs removed, we have demonstrated that the LNR is the most significant prognostic factor for both OS and DFS. This demonstrates that, even in patients who are undergoing extensive resections with at least 10 LNs examined, the LNR has a dramatic impact on survival and recurrence. Finally, another potential role of a ratio-based classification is for stratification and prognostic information as part of a nomogram system to predict local and/or systemic recurrences, which could help determine the type and extent of adjuvant chemotherapy.
In conclusion, LNR is an important prognostic factor for adenocarcinoma of the colon. Evaluation of the LN data from a large Intergroup trial demonstrates the importance of the number of positive LNs, number of examined LNs, and the LNR. LNR may decrease the amount of stage migration, which can result when the pathologist examines an inadequate number of LNs. Future clinical trials in colon cancer should prospectively evaluate LNR as an important prognostic factor.
Authors' Disclosures of Potential Conflicts of Interest
The authors indicated no potential conflicts of interest.
NOTES
Presented at the 58th Annual Cancer Symposium of the Society of Surgical Oncology, Atlanta, GA, March 3-5, 2005.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Tepper JE, O'Connell MJ, Niedzwiecki D, et al: Impact of number of nodes retrieved on outcome in patients with rectal cancer. J Clin Oncol 19:157-163, 2001
Joseph NE, Sigurdson ER, Hanlon AL, et al: Accuracy of determining nodal negativity in colorectal cancer on the basis of the number of nodes retrieved on resection. Ann Surg Oncol 10:213-218, 2003
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 21:2912-2919, 2003
Gajra A, Newman N, Gamble GP, et al: Effect of number of lymph nodes sampled on outcome in patients with stage I non-small-cell lung cancer. J Clin Oncol 21:1029-1034, 2003
Herr HW, Bochner BH, Dalbagni G, et al: Impact of the number of lymph nodes retrieved on outcome in patients with muscle invasive bladder cancer. J Urol 167:1295-1298, 2002
Weir L, Speers C, D'yachkova Y, et al: Prognostic significance of the number of axillary lymph nodes removed in patients with node-negative breast cancer. J Clin Oncol 20:1793-1799, 2002
Sigurdson ER: Lymph node dissection: Is it diagnostic or therapeutic? J Clin Oncol 21:965-967, 2003
Sierra A, Regueira FM, Hernandez-Lizoain JL, et al: Role of the extended lymphadenectomy in gastric cancer surgery: Experience in a single institution. Ann Surg Oncol 10:219-226, 2003
Bando E, Yonemura Y, Taniguchi K, et al: Outcome of ratio of lymph node metastasis in gastric carcinoma. Ann Surg Oncol 9:775-784, 2002
Berger AC, Watson JC, Ross EA, et al: The metastatic/examined lymph node ratio is an important prognostic factor after pancreaticoduodenectomy for pancreatic adenocarcinoma. Am Surg 70:235-240, 2004
Haller DG, Catalano PJ, Macdonald JS, et al: Fluorouracil (FU), leucovorin (LV) and levamisole (LEV) adjuvant therapy for colon cancer: Five-year final report of INT-0089. Proc Am Soc Clin Oncol 17:256a, 1998 (abstr 962)
Kaplan EL, Meier P: Nonparametric estimation from incomplete observations. J Am Stat Assoc 53:457-481, 1958
Kalbfleisch JD, Prentice RL: The Statistical Analysis of Failure Time Data. New York, NY, John Wiley, 1980
Cox DR: Regression models and life tables. J R Stat Soc B 34:187-220, 1972
Feinstein AR, Sosin DM, Wells CK: The Will Rogers phenomenon: Stage migration and new diagnostic techniques as a source of misleading statistics for survival in cancer. N Engl J Med 312:1604-1608, 1985
Inoue K, Nakane Y, Iiyama H, et al: The superiority of ratio-based lymph node staging in gastric carcinoma. Ann Surg Oncol 9:27-34, 2002(Adam C. Berger, Elin R. S)