|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
Journal of Clinical Oncology, Vol 26, No 1 (January 1), 2008: pp. 121-126 © 2008 American Society of Clinical Oncology. DOI: 10.1200/JCO.2007.12.9247 Lymph Node Density Is Superior to TNM Nodal Status in Predicting Disease-Specific Survival After Radical Cystectomy for Bladder Cancer: Analysis of Pooled Data From MDACC and MSKCC
From the Department of Urology and the Division of Quantitative Sciences, The University of Texas M.D. Anderson Cancer Center, Houston, TX; and the Department of Urology, Memorial Sloan-Kettering Cancer Center, New York, NY Corresponding author: Ashish M. Kamat, MD, Department of Urology, Unit 1373, The University of Texas M.D. Anderson Cancer Center, 1515 Holcombe Blvd, Houston, TX 77030; e-mail: akamat{at}mdanderson.org
Purpose To compare the utility of lymph node density (LND) with TNM nodal status in predicting disease-specific survival (DSS) after radical cystectomy. Patients and Methods We identified 248 patients with nodal metastasis after radical cystectomy (without neoadjuvant chemotherapy): 162 patients from Memorial Sloan-Kettering Cancer Center (MSKCC) and 86 patients from M.D. Anderson Cancer Center (MDACC). We assessed the effect of several variables on DSS.
Results After a median follow-up duration of 24 months, 134 patients died of their disease. The median DSS was 36 months, and the 1-year, 2-year, and 5-year DSS rates were 83.7%, 57.4%, and 36.6%, respectively. The median LND was 20%. The 5-year DSS rate was 54.6% for patients with LND Conclusion LND is superior to TNM nodal status in predicting DSS for patients with lymph node–positive disease after radical cystectomy, even in the context of adjuvant chemotherapy.
The gold-standard treatment for muscle-invasive bladder cancer is radical cystectomy with pelvic lymph node dissection (PLND). While regional lymph node metastases are associated with decreased recurrence-free and overall survival rates, in select individuals, pelvic lymphadenectomy can be curative.1 Several approaches have been used to correlate nodal burden of disease with disease-specific survival (DSS). One approach is the use of the 2002 TNM staging system in which nodal disease is quantified by the number and size of involved regional lymph nodes. In this staging system, pathologic nodal (pN) status is classified as pN1 (metastasis in one node,  2 cm in greatest dimension), pN2 (metastasis in one node, between 2 cm and 5 cm in greatest dimension, or in multiple nodes 5 cm in greatest dimension), and pN3 (metastasis in one or more nodes, > 5 cm in greatest dimension).2 Studies have validated the prognostic value of nodal burden by demonstrating decreased 3-year DSS rates3,4 and decreased 10-year recurrence-free survival rates1 with the presence of five or more positive lymph nodes. Recent studies suggest that lymph node density (LND), the ratio of positive lymph nodes to the total number of lymph nodes removed, is superior to TNM pathologic nodal staging5 and to the absolute number of positive lymph nodes5,6 in predicting DSS and recurrence-free survival. The purpose of this study was to compare the values of LND and TNM nodal status in predicting DSS after radical cystectomy for bladder cancer, using pooled data from two comprehensive cancer centers.
After obtaining institutional review board approval, we retrospectively reviewed the medical records of 150 patients who were found to have lymph node metastases at the time of radical cystectomy plus PLND for bladder cancer at M.D. Anderson Cancer Center (MDACC; Houston, TX) between 1993 and 2003. Forty-two patients received preoperative chemotherapy and, thus, were excluded from the study. Of the remaining 108 patients, complete information regarding pN status was available for 86, allowing for TNM classification and LND calculations. These 86 patients constituted the MDACC segment of the study population. Institutional review board approval was also obtained from Memorial Sloan-Kettering Cancer Center (MSKCC; New York, NY) to retrospectively review the records of the patients treated with radical cystectomy plus PLND at that institution between 1979 to 1999; 162 patients met our eligibility criteria. The MSKCC patients are unique in that they represent a single surgeon's experience (H.W.H.) with pathological analysis by at least 15 different pathologists. Thus our total study population included 248 patients. Sixty-one (71%) of 86 patients from MDACC received adjuvant systemic chemotherapy, which consisted primarily (93%) of platinum-based regimens. Of the 162 MSKCC patients with lymph node–positive disease, only five (3%) received adjuvant systemic chemotherapy (P < .001). Our analysis involved an assessment of pN status and LND. Nodal status was determined on the basis of the 2002 TNM staging system. The percentages chosen for the LND analysis (20%5,7 and 25%6) were based on previous analyses in similar patient populations. A standardized lymphadenctomy template was not used in these patients and may account for some surgical heterogeneity among patients. However, lymph node dissections performed by surgeons at both institutions have gradually evolved over the past 15 years so as to incorporate at least the nodes at the aortic bifurcation in the superior extent of the dissection. Other variables assessed were age, sex, clinical stage, histology, lymphovascular invasion, presence of carcinoma in situ, surgical margin status, and adjuvant chemotherapy. We used the Kaplan-Meier method8 to estimate the median DSS. The Cox proportional hazards regression model9 was used to determine the prognostic significance of each variable studied, including LND and TNM nodal status, in a univariate fashion. All potential prognostic factors with a P value less than .25 on univariate analysis were included in a saturated model. Backward elimination was used to remove factors from the model on the basis of the likelihood ratio test in the multiple regression analysis using the Cox proportional hazards regression model. The total number of lymph nodes excised and the number of positive lymph nodes were not included in these analyses but were used to determine LND, which was included as a single variable. For all statistical analyses, a P value less than .05 was deemed statistically significant.
The clinical characteristics of the 248 patients with node-positive disease at the time of radical cystectomy and PLND are listed in Table 1. The median follow-up duration was 24 months for all patients (range, 0.2 to 142 months) and 47 months for survivors (range, 0.2 to 142 months). At the time of analysis, 134 patients (54%) died of their disease: 27 (31%) from the MDACC patients and 107 (66%) from the MSKCC patients. Overall, the median DSS was 36 months, 30 months for MSKCC patients and median not reached for MDACC patients. Kaplan-Meier estimates of 1-year, 2-year, and 5-year DSS were 83.7%, 57.4%, and 36.6%, respectively.
On the basis of the 2002 TNM staging system, the pN status of lymph nodes recovered at the time of radical cystectomy was pN1 in 78 patients (32%), pN2 in 127 patients (51%), and pN3 in 43 patients (17%). The median number of lymph nodes removed was 12 (range, 2 to 58) and the median number of positive lymph nodes was two (range, 1 to 14). The median LND was 20%. The distribution of patients by pN status and LND is summarized in Table 2. A similar distribution was done for patients using a LND separation point of 25% (data not shown). In fact, two complete analyses using the two different LND separation points of 20% and 25% were performed with the data set; these analyses yielded similar results. The data using 20% is featured in this analysis because it has been used more commonly by other groups. The hazard ratios (HRs) for these two cutoffs were almost identical: HR for 20% = 2.95 (univariate analysis) and 2.75 (multivariate analysis); and HR for 25% = 2.93 (univariate analysis) and 2.68 (multivariate analysis).
The results of the association of several variables with DSS by univariate and multivariate analyses are summarized in Table 3. On univariate analysis, the factors found to be statistically significant predictors of longer DSS were excision of greater than 12 lymph nodes (HR, 0.41; 95% CI, 0.29 to 0.58; P < .01) and the use of adjuvant chemotherapy (HR, 0.50; 95% CI, 0.31 to 0.79; P < .01). Variables associated with decreased DSS were study center (HR, 1.89; 95% CI, 1.24 to 2.88; P < .01), presence of more than two positive lymph nodes (HR, 1.73; 95% CI, 1.23 to 2.44; P < .01), a LND higher than 20% (HR, 2.95; 95% CI, 2.06 to 4.22; P < .01), positive surgical margins (HR, 1.69; 95% CI, 1.02 to 2.82; P = .043), nonorgan-confined disease (HR, 2.70; 95% CI, 1.64 to 4.45; P < .01), local recurrence of disease (HR, 3.20; 95% CI, 2.20 to 4.65; P < .01), and increasing pN status (HR, 1.80; 95% CI, 1.19 to 2.74; P < .01 for pN2; HR, 2.47; 95% CI, 1.47 to 4.17; P < .01 for pN3). When patients with increasing pN status were compared with patients with pN1 disease, a statistically significant (P < .001) association with decreased DSS was noted (Fig 1A). In addition, combining pN subgroups also revealed statistically significant differences in DSS. This relationship held true for combining the pN1 and pN2 subgroups and comparing with the pN3 subgroup (P = .017) and for combining the pN2 and pN3 subgroups and comparing with the pN1 subgroup (P < .001). DSS was also intimately related to LND as 5-year DSS for patients with LND 20% was 54.6% v 15.3% for patients with LND higher than 20% (P < .01), as demonstrated by the Kaplan-Meier curves in Figure 1B.
On multivariate analysis, only three variables retained statistical significance with DSS: LND higher than 20% (HR, 2.63; 95% CI, 1.83 to 3.79; P < .01), adjuvant chemotherapy (HR, 0.47; 95% CI, 0.30 to 0.75; P < .01), and nonorgan-confined primary tumor (HR, 2.40; 95% CI, 1.45 to 3.97; P < .01). We did not consider local recurrence in the multivariate analysis because we do not have data on time to local recurrence, so we could not model this outcome as a time varying covariate, as would be appropriate. Unlike LND, the pN status determined using TNM staging did not remain statistically significant on multivariate analysis. Furthermore, when LND and pN status were compared jointly in a multivariate model, only LND higher than 20% (HR, 2.75; 95% CI, 1.80 to 4.20; P < .01) and study center (HR, 1.87; 95% CI, 1.23 to 2.86; P < .01) remained significantly associated with decreased DSS. When the analysis was performed again by comparing pN status and LND jointly in a multivariate model in only those patients who did not receive adjuvant chemotherapy, study center disappeared as a statistically significant variable on multivariate analysis (P = .79). Notably, when adjuvant chemotherapy was factored in, pN status again was not statistically significant (P = .42 for pN3 and P = .85 for pN2), whereas LND higher than 20% again retained a significant association with decreased DSS (HR, 3.70; 95% CI, 2.22 to 6.14; P < .01). The inclusion of an interaction term in the model combining TNM pN status and LND does not alter the conclusions regarding the utility of LND.
Kaplan-Meier analysis was performed in all patients with LND
In our analysis of a pooled population of patients from two comprehensive cancer centers (MDACC and MSKCC), we found LND lower than 20% to be superior to TNM nodal status in predicting DSS for patients with node-positive bladder cancer after radical cystectomy and PLND. We also found that LND retained superiority even in the context of adjuvant chemotherapy. Nodal metastases at the time of radical cystectomy and PLND portend a poor prognosis with 5-year overall survival rates as low as 7% in some studies.10 Even in contemporary series, 5-year survival rates are approximately 30%.1,11 These more recent survival statistics are more sobering as they are from institutions that practice extended lymphadenectomy at the time of radical cystectomy. Risk stratification of recurrence and survival after PLND has traditionally been based on TNM pN staging. Nevertheless, the use of this approach in predicting recurrence-free survival and overall survival has led to conflicting results. Vieweg and colleagues12 demonstrated the value of pN status in stratifying 5-year DSS with rates of 44.2% in pN1 disease, 26.6% in pN2 disease, and 0% in pN3 disease (P = .0006). Other studies, however, have failed to confirm the prognostic ability of pN status in predicting recurrence-free and overall survival on multivariate analysis.5,11 Other groups have demonstrated decreased recurrence-free and overall survival with an increasing absolute number of positive lymph nodes on univariate analysis.1,7,11 Most of these studies used a cutoff point of more than five positive lymph nodes, but not all of the studies have confirmed these findings on multivariate analysis. Other studies have demonstrated better prognostic significance of the total number of lymph nodes removed irrespective of lymph node positivity, suggesting that more than eight to 15 lymph nodes should be removed to improve outcomes.7,13 On the shortfall of these approaches emerged the concept of LND, which incorporated the number of positive nodes removed in relation to the total number of lymph nodes removed.5,7 Also referred to as ratio-based lymph node staging and percentage of positive nodes removed, LND (the ratio of positive lymph nodes removed/total number of nodes removed) has been shown to be predictive of survival in contemporary series of patients and can be used to reliably risk stratify patients with bladder cancer in whom positive lymph nodes are found at surgery.5-7 A recent report by Herr5 suggests that LND is superior to TNM-based pN status in predicting prognosis after radical cystectomy. Herr found that a LND cutoff of 20% was superior to the 1997 TNM staging system on multivariate analysis in predicting DSS (P = .002 v P = .21) and local recurrence (P = .01 v P = .06). However, this study represented a single-center experience, it did not account for the potential impact of adjuvant chemotherapy, and it may have diluted the predictive value of TNM pN status by making comparisons between three individual groups (pN1 v pN2 v pN3) as opposed to pooling groups to test the predictive capability of a single group (eg, pN1 and pN2 v pN3 or pN1 v pN2 and pN3). A recent review of the literature demonstrated that only five studies have assessed the ability of LND in predicting survival after radical cystectomy for bladder cancer.14 Although four of these studies confirmed that LND was predictive of survival on multivariate analysis, three of these reports were from single-institution series and one was from the Surveillance, Epidemiology, and End Results population database. Moreover, only one study compared the predictive ability of LND to pN staging under the TNM system. We pooled radical cystectomy data for patients with lymph node–positive bladder cancer from two large comprehensive cancer centers to verify the conclusions reached in the aforementioned studies and to answer their unaddressed concerns.
Our results confirm that the following variables are significantly associated with DSS on univariate analysis: total number of lymph nodes excised, number of positive lymph nodes excised, LND, surgical margin status, use of adjuvant chemotherapy, organ-confined disease, local recurrence, and pN status. However, only LND, the use of adjuvant chemotherapy, and organ-confined primary tumor retained statistical significance on multivariate analysis. When pN status and LND were jointly analyzed in a multivariate model with study center, only LND higher than 20% and study center remained significantly associated with decreased DSS, even when the pN staging groups were combined to increase predictive capability. Furthermore, joint analysis of pN status and LND in a series of Kaplan-Meier curves demonstrates the ability of LND to predict DSS in pN subgroups and the inability of pN subgroups to predict DSS in different LND subgroups (Figs 2 and 3). Of course, the relationship was not statistically significant in the pN3 subgroup stratified by LND but the curves are widely separated and the lack of significance most likely represents small numbers in the pN3 subgroup. From a practical management standpoint, the number of patients one sees who would be classified as pN3 but have LND Our finding that study center is associated with decreased DSS on multivariate analysis is likely due to the effect of adjuvant chemotherapy, as 71% of the MDACC patients received such therapy compared with only 3% of the MSKCC patients. Adjuvant chemotherapy emerged as a prognostic factor on multivariate analysis when study center was eliminated from the multivariate model. Because adjuvant chemotherapy is intimately related to study center, no difference in DSS between study centers emerged when the analysis was repeated after removing the patients who received adjuvant chemotherapy. Of note, when adjuvant chemotherapy was included in the multivariate analysis, LND higher than 20% remained significantly associated with decreased DSS. This finding suggests that LND can still serve as a risk stratification tool even in patients who receive adjuvant chemotherapy. pN status based on the TNM system remains statistically insignificant in such an analysis (P = .24 to .58). As typical of retrospective studies, our study is limited by biases such as lack of random assignment, patient selection, and incomplete data acquisition. The use of patients from two large cancer centers may reflect a cohort with more aggressive disease, which is the referral pattern for such centers. Furthermore, patients receiving adjuvant chemotherapy may introduce selection bias as well. Interestingly, we found that adjuvant chemotherapy was associated with improved survival. This finding is in agreement with a recent study that has demonstrated that patients with lymph node–positive transitional cell carcinoma of the bladder treated by radical cystectomy were 2.1 times less likely to die of their disease if they received adjuvant chemotherapy (P = .005).3 We propose that LND is superior to pN status determined using the TNM system in predicting DSS after radical cystectomy and PLND. Furthermore, LND is still superior whether or not adjuvant chemotherapy is administered. As a result, LND may be useful to better risk stratify patients when entering them in clinical trials designed to study the role of adjuvant chemotherapy. However, in the absence of a prospectively evaluated standardized template for lymphadenectomy, we hesitate to suggest using LND as a criterion for not recommending adjuvant chemotherapy. Although relatively easy to calculate, LND is a multifactorial entity. Due to interindividual variability in pelvic lymph node anatomy,15 it may not be prognostic in patients with little lymphatic tissue. In addition, LND is intimately related to histologic processing and the extent of review by the genitourinary pathologist. In most cases, however, LND is probably a reflection of the quality and extent of pelvic lymphadenectomy. Since the extent of lymphadenectomy will clearly affect the LND, for prospective studies, we recommend that a minimum consensus number be established. For our study, we iteratively eliminated from analysis patients with small numbers of lymph nodes sampled, until the factors for pN status were statistically significant. This happened when we eliminated all patients with fewer than nine lymph nodes sampled. Even at this point, LND retained its superiority. Finally, the predictive power of LND might be enhanced by other variables, such as volume of micrometastases, presence of extracapsular extension, and anatomic location of positive lymph nodes.
The author(s) indicated no potential conflicts of interest.
Conception and design: Wassim Kassouf, Piyush K. Agarwal, Harry W. Herr, Colin P. Dinney, Ashish M. Kamat Administrative support: Ashish M. Kamat Provision of study materials or patients: Wassim Kassouf, Harry W. Herr, Mark F. Munsell, Louis L. Pisters, H. Barton Grossman, Ashish M. Kamat Collection and assembly of data: Wassim Kassouf, Piyush K. Agarwal, Mark F. Munsell, Philippe E. Spiess, Gordon A. Brown, Ashish M. Kamat Data analysis and interpretation: Wassim Kassouf, Piyush K. Agarwal, Mark F. Munsell, Ashish M. Kamat Manuscript writing: Wassim Kassouf, Piyush K. Agarwal, Ashish M. Kamat Final approval of manuscript: Wassim Kassouf, Piyush K. Agarwal, Harry W. Herr, Mark F. Munsell, Philippe E. Spiess, Gordon A. Brown, Louis L. Pisters, H. Barton Grossman, Colin P. Dinney, Ashish M. Kamat
Supported by the M.D. Anderson Cancer Center Bladder SPORE (5P50CA091846-03) and a Department of Urology NIH T32 Training Grant CA079449. W.K. and P.K.A. contributed equally to this work. Presented at the 2007 Annual American Urological Association meeting, May 22, 2007, Anaheim, CA. Authors' disclosures of potential conflicts of interest and author contributions are found at the end of this article.
1. Stein JP, Lieskovsky G, Cote R, et al: Radical cystectomy in the treatment of invasive bladder cancer: Long-term results in 1,054 patients. J Clin Oncol 19:666-675, 2001 2. Greene FL, Page DL, Fleming ID, et al: AJCC Cancer Staging Manual (ed 6). New York, NY, Springer-Verlag, 2002 3. Frank I, Cheville JC, Blute ML, et al: Transitional cell carcinoma of the urinary bladder with regional lymph node involvement treated by cystectomy: Clinicopathologic features associated with outcome. Cancer 97:2425-2431, 2003[CrossRef][Medline] 4. Abdel-Latif M, Abol-Enein H, El-Baz M, et al: Nodal involvement in bladder cancer cases treated with radical cystectomy: Incidence and prognosis. J Urol 172:85-89, 2004[CrossRef][Medline] 5. Herr HW: Superiority of ratio based lymph node staging for bladder cancer. J Urol 169:943-945, 2003[CrossRef][Medline] 6. Kassouf W, Leibovici D, Munsell MF, et al: Evaluation of the relevance of lymph node density in a contemporary series of patients undergoing radical cystectomy. J Urol 176:53-57, 2006[CrossRef][Medline] 7. Stein JP, Cai J, Groshen S, et al: Risk factors for patients with pelvic lymph node metastases following radical cystectomy with en bloc pelvic lymphadenectomy: The concept of lymph node density. J Urol 170:35-41, 2003[CrossRef][Medline] 8. Kaplan EL, Meier P: Nonparametric estimation from incomplete observations. J Am Statist Assoc 53:457-481, 1958[CrossRef] 9. Cox DR: Regression models and life tables (with discussion). J Roy Statist Soc B 34:187-220, 1972 10. Smith JA Jr, Whitmore WF Jr: Regional lymph node metastasis from bladder cancer. J Urol 126:591-593, 1981[Medline] 11. Fleischmann A, Thalmann GN, Markwalder R, et al: Extracapsular extension of pelvic lymph node metastases from urothelial carcinoma of the bladder is an independent prognostic factor. J Clin Onc 23:2358-2365, 2005[CrossRef] 12. Vieweg J, Gschwend JE, Herr HW, et al: Pelvic lymph node dissection can be curative in patients with node positive bladder cancer. J Urol 161:449-454, 1999[CrossRef][Medline] 13. 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[CrossRef][Medline] 14. Herr HW: The concept of lymph node density –is it ready for clinical practice? J Urol 177:1273-1276, 2007[CrossRef][Medline] 15. Weingartner K, Ramaswamy A, Bittinger A, et al: Anatomical basis for pelvic lymphadenectomy in prostate cancer: Results of an autopsy study and implications for the clinic. J Urol 156:1969-1971, 1996[CrossRef][Medline] Submitted June 5, 2007; accepted September 27, 2007.
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2008 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
TOP
ABSTRACT
INTRODUCTION
