Lymphovascular Invasion Is Independently Associated With Overall Survival, Cause-Specific Survival, and Local and Distant Recurrence in Pati
http://www.100md.com
《临床肿瘤学》
the Department of Urology, University of Texas Southwestern Medical Center, Dallas
Baylor College of Medicine, Houston, TX
The James Buchanan Brady Urological Institute, The Johns Hopkins School of Medicine, Baltimore, MD
Centre Hospitalier de l'Université de Montréal, Campus St-Luc, Montreal, Quebec, Canada
ABSTRACT
PURPOSE: We hypothesized that bladder cancer patients with associated lymphovascular invasion (LVI) are at increased risk of occult metastases.
METHODS: A multi-institutional group (University of Texas Southwestern [Dallas, TX], Baylor College of Medicine [Houston, TX], Johns Hopkins University [Baltimore, MD]) carried out a retrospective study of 958 patients who underwent cystectomy for bladder cancer between 1984 and 2003. Of patients with transitional-cell carcinoma (n = 776), LVI status was available for 750. LVI was defined as the presence of tumor cells within an endothelium-lined space.
RESULTS: LVI was present in 36.4% (273 of 750) overall, involving 26% (151 of 581) and 72% (122 of 169) of node-negative and node-positive patients, respectively. Prevalence of LVI increased with higher pathologic stage (9.0%, 23%, 60%, and 78%, for T1, T2, T3, and T4, respectively; P < .001). Using multivariate Cox regression analyses including age, stage, grade, and number of pelvic lymph nodes removed, LVI was an independent predictor of local (HR = 2.03, P = .049), distant (HR = 2.60, P = .0011), and overall (HR = 2.02, P = .0003) recurrence in node-negative patients. LVI was an independent predictor of overall (HR = 1.84, P = .0002) and cause-specific (HR = 2.07, P = .0012) survival in node-negative patients. LVI maintained its independent predictor status in competing risks regression models (P = .013), where other-cause mortality was considered as a competing risk. LVI was not a predictor of recurrence or survival in node-positive patients.
CONCLUSION: LVI is an independent predictor of recurrence and decreased cause-specific and overall survival in patients who undergo cystectomy for invasive bladder cancer and are node-negative. These patients represent a high risk group that may benefit from integrated therapy with cystectomy and perioperative systemic chemotherapy.
INTRODUCTION
Bladder cancer is the fourth most common cancer in men (6%) and the 10th most common cancer in women (2%), accounting in men for 3% of cancer deaths in the year 2004 in the United States.1 On average, 15% to 30% of all patients with bladder cancer are diagnosed with muscle-invasive tumors, for which radical cystectomy is the current gold standard therapy.2 Unfortunately, as many as 40% of patients with organ-confined disease at the time of cystectomy subsequently have recurrence.3-5 Once bladder cancer metastasizes, treatments such as radiotherapy and systemic chemotherapy do not significantly change the overall survival rates.6,7
Several studies have evaluated the risk factors for recurrence and survival after cystectomy.8,9 Advanced pathologic stage,3,8-10 nodal involvement,3,9,10 tumor size greater than 3 cm,3 elevated creatinine,3 and vascular invasion10 have been found to be independent risk factors for recurrence. Advanced pathologic stage and nodal involvement have also been found to be independent risk factors for survival.9,11,12 The role of lymphovascular invasion (LVI) as a prognostic factor for progression and survival is controversial. Early studies identified LVI as a poor prognostic feature.13,14 Vascular but not lymphatic invasion was identified as an independent predictor of recurrence after cystectomy in one study.10 In other studies, however, LVI was not a predictor of lymph node metastases,15 relapse,3 or survival.9,12 To test the hypothesis that LVI is a predictor of recurrence and survival, we performed a retrospective multi-institutional study (University of Texas Southwestern [Dallas, TX], Baylor College of Medicine [Houston, TX], Johns Hopkins University [Baltimore, MD]) of 958 patients who underwent cystectomy for bladder cancer between 1984 and 2003.
PATIENTS AND METHODS
Patient Population
All studies were undertaken with the approval and institutional oversight of the institutional review board for the Protection of Human Subjects. A retrospective multi-institutional study (University of Texas Southwestern, Baylor College of Medicine, Johns Hopkins University) was performed on 958 patients who underwent cystectomy with bilateral lymphadenectomy for bladder cancer between 1984 and 2003. For each patient, we recorded sociodemographic, precystectomy, cystectomy, and postcystectomy variables. Sociodemographics consisted of patient age at cystectomy, the institution where cystectomy was performed, and the year of surgery. Precystectomy variables included transurethral resection (TUR) 1997 TNM stage, WHO grade, presence of carcinoma-in-situ, preoperative intravesical chemotherapy or immunotherapy, and neoadjuvant chemotherapy. Cystectomy variables included pathologic 1997 TNM stage, WHO grade, presence of coexisting carcinoma-in-situ, LVI, total number of resected lymph nodes, and number of positive lymph nodes. Postcystectomy variables included adjuvant chemotherapy and radiotherapy. There were 838 patients with complete datasets who were eligible for our analysis. Our analysis was limited to those patients who had transitional cell carcinoma. Thus, 62 patients with non–transitional cell carcinoma histology (26 squamous, 12 adenocarcinomas, 16 other, five small-cell, one spindle cell, and two undifferentiated) were excluded from analysis. Of the remaining 776 patients, information on LVI was not available for 26 patients. Thus, the analyses were based on 750 patients.
Of the patients in the study sample, 612 were men and 138 were women. The mean age at cystectomy was 64.8 years (standard deviation, 10.3; median, 65.9). The mean number of lymph nodes per patient removed at the time of cystectomy was 20.1 ± 10.2 (median, 17; interquartile range, 12). The number of nodes removed during pelvic lymphadenectomy was divided into quartiles for this analysis. The operative surgeon assigned the preoperative clinical stage according to the 1997 American Joint Committee on Cancer /TNM system. Final pathologic staging was defined as the higher of the preoperative clinical and the pathologic stages for patients with organ-confined disease. For clinically and/or pathologically non–organ-confined disease in patients who had not received neoadjuvant systemic chemotherapy or radiation therapy, the consensus stage for this analysis was based on the true postoperative pathologic stage. For clinically and pathologically non–organ-confined disease in patients who had received neoadjuvant systemic chemotherapy or radiation therapy, the consensus stage was highest of the two. The consensus grade was the higher of the preoperative and postoperative pathologic grades.
Preoperative and postoperative radiation was delivered to 2.3% and 4.7% of patients, respectively. Preoperative and postoperative chemotherapy was given to 4.9% and 26.7% of patients, respectively. For patients who were alive at the last follow-up and whose LVI status was known (N = 450), the mean follow-up was 3.9 years, and the median was 3.1 years (range, 13 days to 18 years). For patients who had died during the follow-up period and had known LVI status (n = 293), the mean follow-up was 2.4 years. The median follow-up was 1.3 years (range, 2 days to 14.3 years). Overall, 191 (22%) and 45 (5%) patients had more than 5- and 10-year follow-up data, respectively. Recurrences developed in 33.6% of patients (n = 252). Patients were classified as having either distant or local only, or both distant and local recurrence. Patients were classified as having simultaneous distant and local recurrence if the distant and local recurrences were detected within 30 days for each other. Overall, 39.7% of patients (n = 298) died, and 66% of all deaths (n = 198) were due to bladder cancer. The remaining 100 deaths (33%) were considered as other-cause mortality.
Pathologic Examination
The extent and depth of the TUR was defined by each of the participating surgeons. Standard radical cystectomy was performed at each institution, though specific technique and extent of lymphadenectomy varied between surgeons and institutions. Pathologic TUR, cystectomy, and lymphadenectomy specimens were processed according to institutional protocols. The 1997 TNM and WHO classifications were used for tissue staging and grading. Lymphovascular invasion was defined as the presence of tumor cells within an endothelium-lined space without underlying muscular walls. The presence of a clear-cut endothelial lining was an important requirement in that retraction space artifact is especially common in invasive urothelial carcinoma of the bladder. Immunohistochemistry for endothelial cells was not done, more in keeping with standard practice of pathologists. Any equivocal foci and foci in which tumor cells merely encroached on a vascular lumen were considered negative. No attempt was made to differentiate between vascular and lymphatic vessels because of the difficulty and lack of reproducibility when using routine light microscopic examination. Hospital charts, physician records, and approved cancer registries were used to determine cause-specific mortality.
Statistical Analysis
2 tests were used to evaluate the association between categorical variables. The Kaplan-Meier method was used to calculate survival functions, and differences were assessed with the log-rank statistic. Univariate and multivariate survival analyses were performed using the Cox proportional hazard regression model. Competing risks regression was used to test the significance of predictor variables after accounting for other-cause mortality. Statistical significance in this study was set as P .05. All reported P values are two-sided. Analyses were performed with SAS version 8.02 (SAS Institute Inc, Cary, NC) and S-PLUS Professional, version 1 (MathSoft Inc, Seattle, Washington).
RESULTS
LVI was present in 36.4% of patients (273 of 750) overall, involving 26% (151 of 581) and 72% (122 of 169) of node-negative and node-positive patients, respectively. Prevalence of LVI increased with higher pathologic stage (9.0%, 23%, 60%, and 78%, for T1, T2, T3, and T4, respectively; P < .001). LVI was associated with advanced pathologic features including grade, stage, and nodal status (Table 1) . When patients were stratified on the basis of lymph node status after pelvic lymphadenectomy, LVI was associated on univariate analysis with overall and cause-specific survival, as well as with recurrence (overall, local, and distant) in node-negative patients (Table 2). These associations were absent in node-positive patients.
On multivariate Cox regression analyses that included age, stage, grade, and number of pelvic lymph nodes removed, LVI was an independent predictor of local (HR = 2.03, P = .049), distant (HR = 2.60, P = .0011), and overall (HR = 2.02, P = .0003) recurrence (Table 3) in node-negative patients. On multivariate Cox regression analyses that included age, stage, grade, and number of pelvic lymph nodes removed, LVI was an independent predictor of overall (HR = 1.84, P = .0002) and cause-specific (HR = 2.07, P = .0012) survival (Table 4) in node-negative patients. In competing-risks regression models, which included age, stage, grade, and quartiles of the number of pelvic lymph nodes removed, LVI represented an independent predictor of local recurrence (P = .013), distant recurrence (P = .0095), overall recurrence (P = .0071), and cause-specific survival (P = .014).
In addition to LVI in node-negative patients, stage and number of lymph nodes removed during lymphadenectomy were independent predictors of overall and distant recurrence using multivariate Cox regression analysis (Table 3). Stage and number of lymph nodes removed during lymphadenectomy were also independent predictors of overall and cause-specific survival using multivariate Cox regression analysis (Table 4).
LVI status, lymph node status, and whether patients with positive nodes received postoperative chemotherapy impacted cause-specific survival in patients with varying pathologic stages of bladder cancer (Figs 1, 2, and 3). The 5-year actuarial cause-specific survival for patients with negative nodes was 86.9% ± 3.2 (SE) for patients without LVI and 72.1% ± 8.8 (SE) for patients with LVI and stage T2 disease; 65.1% ± 7.7 (SE) for patients without LVI, and 53.9% ± 7.5 (SE) for patients with LVI and stage T3 disease; 65.6% ±14 (SE) for patients without LVI; and 36.1% ± 11.3 (SE) for patients with LVI and stage T4 disease.
DISCUSSION
A major difficulty with treating bladder cancer is the high recurrence rate in patients with "organ-confined" disease after cystectomy.3,8 Once bladder cancer has metastasized, treatment options such as radiotherapy and systemic chemotherapy do not significantly change the overall survival rates.6,7 Consequently, identifying patients at high risk for recurrence and administering adjuvant therapies may offer improved survival. Our multi-institutional study found that LVI predicts advanced stage, grade, and nodal status. LVI was also an independent predictor of overall and cause-specific survival, as well as overall, local, and distant recurrence in patients who had negative lymph nodes on lymphadenectomy at the time of cystectomy. Moreover, we tested the contribution of LVI in competing risks regression models. This modeling technique accounts for the effect of other-cause mortality, which may preclude the occurrence of a bladder cancer–specific event (recurrence or cause-specific mortality). Competing risks regression provides a more conservative estimate of the effect relative to Cox regression or Kaplan-Meier models. In competing-risks regression analyses, LVI was an independent predictor of local recurrence, distant recurrence, overall recurrence, and of cause-specific survival.
In Cox regression models, LVI was not an independent predictor of recurrence or survival in those patients found to have lymph node metastases. While most patients with positive lymph nodes at the time of cystectomy would be given adjuvant chemotherapy anyway if medically tolerable, this is an important finding because it may help us identify select patients who are lymph node–negative but might benefit from adjuvant therapies. This feature may also be useful in choosing high-risk patients for future clinical trials or in analyzing future trial results.
Results of our study were consistent with those of others. We found that 36.4% of patients had LVI. We also noted that the incidence of LVI increased with higher pathologic tumor stage. This corresponds to rates of LVI found in previous studies, which ranged from 35% to 54%.3,9,10,12 Early studies identified LVI as a poor prognostic feature but did not control for possible confounding factors by using multivariate analyses.13,14 Bassi et al evaluated 535 patients with bladder cancer treated with cystectomy, and found that while LVI was significantly associated with overall survival in univariate analysis, it was not an independent predictor on multivariate analysis. Only stage and lymph node status were predictive of survival in this study.9 In another study, Ennis et al evaluated 85 patients who underwent radical cystectomy and did not receive chemotherapy or radiation therapy. They found that those patients with and without LVI had 3-year relapse-free survival rates of 53% and 73%, respectively. This result did not reach statistical significance, however, because of the small patient sample (P = .06), and it was not an independent prognostic variable in multivariate analysis.3 In a study of 154 patients with bladder cancer who underwent cystectomy, Hara et al similarly found an association between LVI and survival on univariate, but not on multivariate analysis.12 Notably, Leissner et al found that vascular but not lymphatic invasion was identified as an independent predictor of recurrence after cystectomy in an evaluation of 283 patients.10 Our study had the advantage of evaluating LVI in a large number of patients from multiple institutions, and thus allowed more powerful statistical analyses and provided more generalizable findings. The present study demonstrated an association of LVI with advanced stage, high grade, and lymph node metastases. Furthermore, the strong association between LVI and the rate of metastases to regional lymph nodes in the current study supports the hypothesis that lymphatic vessel invasion precedes, or occurs concurrently with lymph node metastasis. Once the patient has lymph node metastases, the LVI status no longer adds predictive value for survival or recurrence. However, the LVI status in patients with negative lymph nodes at the time of cystectomy and lymphadenectomy is an important predictor of survival and recurrence, as it seems able to identify a subgroup of patients with micrometastases or false-negative lymph node status. Support for this contention is the finding that an increased number of nodes removed during lymphadenectomy was an independent predictor of overall and cause-specific survival, as well as distant and overall recurrence in node-negative patients. If patients with "negative" nodes have micrometastases within lymph nodes that were not identified by a pathologist, and a greater number of these nodes are removed, then the chance of achieving a surgical cure increases. The association between increased survival and an increased number of nodes removed has been reported previously.16
One may extrapolate that patients with LVI may benefit from adjuvant systemic agents as they are at high risk for recurrence. To date, there has been at least one study that evaluated the use of adjuvant cyclophosphamide, doxorubicin, and cisplatin chemotherapy in patients with high risk for relapse after cystectomy.14 In this study comparing outcomes for 62 high-risk controls and 71 patients receiving adjuvant therapy, patients with resected nodal metastases, extravesicular involvement of tumor, and direct invasion of pelvic viscera benefited from adjuvant therapy, but those with LVI as a sole manifestation of their high risk status did not. While conclusions cannot be reached on the basis of this one study, it is apparent that careful evaluation is required before use of adjuvant therapies in patients solely because of LVI as a risk factor.
One avenue for improving patient outcomes is to improve predictions of which patients may best benefit from neoadjuvant chemotherapy. Grossman et al found that neoadjuvant chemotherapy (methotrexate, vinblastine, doxorubicin, and cisplatin) followed by radical cystectomy increased the likelihood of eliminating residual tumor and was associated with improved survival compared with cystectomy alone.17 LVI status on transurethral resection of bladder tumor may identify those patients who are best suited for neoadjuvant chemotherapy. Millikan et al found that LVI was associated with a high degree of pathological upstaging after cystectomy from clinically organ-confined bladder cancer to either extravesical extension (36%) or lymph node positivity (27%).5 This observation supports our finding of a significant association between high-stage disease, advanced lymph node status, and LVI.
Several studies have identified markers that are predictive of increased risk of recurrence and metastases in patients undergoing radical cystectomy. Preoperative measures of plasma soluble E-cadherin,18 interleukin-6,19 urokinase plasminogen activator,20 and transforming growth factor-beta(1)21 were predictive of poor outcomes after cystectomy, features of advanced pathologic stage, lymph node involvement, and LVI. Therefore, it is possible that a combination of preoperative serum or urine markers and features such as LVI could improve identification of patients who would benefit from neoadjuvant or adjuvant chemotherapy.
Several limitations of our study need to be considered. The population underwent radical cystectomy and bilateral pelvic lymphadenectomy by multiple surgeons, and their specimens were evaluated by multiple pathologists. Most studies evaluating LVI have come from single-institution experiences. While prognostic indicators may perform well in a select group of patients treated at such centers, it remains to be determined whether these are applicable to the greater population of patients with bladder cancer. Multicenter studies provide the advantage of limiting the biases of single institution studies. Similarly, while it may be preferable for a single pathologist specializing in genitourinary pathology to review each cystectomy specimen, the set-up of this multi-institutional study reflects a real-world practice in which multiple pathologists review tissue specimens, and their interpretation is then used in clinical decision-making with the patient. It should be noted that in our study, surgery was performed primarily by urologic oncologists at their respective institutions. Furthermore, pathologists at our institutions evaluate a higher volume of urologic cancer than pathologists at most community hospitals.
Authors' Disclosures of Potential Conflicts of Interest
The authors indicated no potential conflicts of interest.
Acknowledgment
We acknowledge Dr Thomas Wheeler for his assistance with this manuscript.
NOTES
Drs Lotan and Gupta contributed equally to this study.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Jemal A, Tiwari RC, Murray T, et al: Cancer statistics, 2004. CA Cancer J Clin 54:8-29, 2004
Lerner SP, Skinner E, Skinner DG: Radical cystectomy in regionally advanced bladder cancer. Urol Clin North Am 19:713-723, 1992
Ennis RD, Petrylak DP, Singh P, et al: The effect of cystectomy, and perioperative methotrexate, vinblastine, doxorubicin and cisplatin chemotherapy on the risk and pattern of relapse in patients with muscle invasive bladder cancer. J Urol 163:1413-1418, 2000
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
Millikan R, Dinney C, Swanson D, et al: Integrated therapy for locally advanced bladder cancer: Final report of a randomized trial of cystectomy plus adjuvant M-VAC versus cystectomy with both preoperative and postoperative M-VAC. J Clin Oncol 19:4005-4013, 2001
Borden LS Jr, Clark PE, Hall MC: Bladder cancer. Curr Opin Oncol 15:227-233, 2003
Pashos CL, Botteman MF, Laskin BL, et al: Bladder cancer: Epidemiology, diagnosis, and management. Cancer Pract 10:311-322, 2002
Slaton JW, Swanson DA, Grossman HB, et al: A stage specific approach to tumor surveillance after radical cystectomy for transitional cell carcinoma of the bladder. J Urol 162:710-714, 1999
Bassi P, Ferrante GD, Piazza N, et al: Prognostic factors of outcome after radical cystectomy for bladder cancer: A retrospective study of a homogeneous patient cohort. J Urol 161:1494-1497, 1999
Leissner J, Koeppen C, Wolf HK: Prognostic significance of vascular and perineural invasion in urothelial bladder cancer treated with radical cystectomy. J Urol 169:955-960, 2003
Frazier HA, Robertson JE, Dodge RK, et al: The value of pathologic factors in predicting cancer-specific survival among patients treated with radical cystectomy for transitional cell carcinoma of the bladder and prostate. Cancer 71:3993-4001, 1993
Hara S, Miyake H, Fujisawa M, et al: Prognostic variables in patients who have undergone radical cystectomy for transitional cell carcinoma of the bladder. Jpn J Clin Oncol 31:399-402, 2001
Anderstrom C, Johansson S, Nilsson S: The significance of lamina propria invasion on the prognosis of patients with bladder tumors. J Urol 124:23-26, 1980
Logothetis CJ, Johnson DE, Chong C, et al: Adjuvant cyclophosphamide, doxorubicin, and cisplatin chemotherapy for bladder cancer: An update. J Clin Oncol 6:1590-1596, 1988
Jaeger TM, Weidner N, Chew K, et al: Tumor angiogenesis correlates with lymph node metastases in invasive bladder cancer. J Urol 154:69-71, 1995
Sanderson KM, Stein JP, Skinner DG: The evolving role of pelvic lymphadenectomy in the treatment of bladder cancer. Urol Oncol 22:205-211, 2004
Grossman HB, Natale RB, Tangen CM, et al: Neoadjuvant chemotherapy plus cystectomy compared with cystectomy alone for locally advanced bladder cancer. N Engl J Med 349:859-866, 2003
Matsumoto K, Shariat SF, Casella R, et al: Preoperative plasma soluble E-cadherin predicts metastases to lymph nodes and prognosis in patients undergoing radical cystectomy. J Urol 170:2248-2252, 2003
Andrews B, Shariat SF, Kim JH, et al: Preoperative plasma levels of interleukin-6 and its soluble receptor predict disease recurrence and survival of patients with bladder cancer. J Urol 167:1475-1481, 2002
Shariat SF, Monoski MA, Andrews B, et al: Association of plasma urokinase-type plasminogen activator and its receptor with clinical outcome in patients undergoing radical cystectomy for transitional cell carcinoma of the bladder. Urology 61:1053-1058, 2003
Shariat SF, Kim JH, Andrews B, et al: Preoperative plasma levels of transforming growth factor beta(1) strongly predict clinical outcome in patients with bladder carcinoma. Cancer 92:2985-2992, 2001(Yair Lotan, Amit Gupta, S)
Baylor College of Medicine, Houston, TX
The James Buchanan Brady Urological Institute, The Johns Hopkins School of Medicine, Baltimore, MD
Centre Hospitalier de l'Université de Montréal, Campus St-Luc, Montreal, Quebec, Canada
ABSTRACT
PURPOSE: We hypothesized that bladder cancer patients with associated lymphovascular invasion (LVI) are at increased risk of occult metastases.
METHODS: A multi-institutional group (University of Texas Southwestern [Dallas, TX], Baylor College of Medicine [Houston, TX], Johns Hopkins University [Baltimore, MD]) carried out a retrospective study of 958 patients who underwent cystectomy for bladder cancer between 1984 and 2003. Of patients with transitional-cell carcinoma (n = 776), LVI status was available for 750. LVI was defined as the presence of tumor cells within an endothelium-lined space.
RESULTS: LVI was present in 36.4% (273 of 750) overall, involving 26% (151 of 581) and 72% (122 of 169) of node-negative and node-positive patients, respectively. Prevalence of LVI increased with higher pathologic stage (9.0%, 23%, 60%, and 78%, for T1, T2, T3, and T4, respectively; P < .001). Using multivariate Cox regression analyses including age, stage, grade, and number of pelvic lymph nodes removed, LVI was an independent predictor of local (HR = 2.03, P = .049), distant (HR = 2.60, P = .0011), and overall (HR = 2.02, P = .0003) recurrence in node-negative patients. LVI was an independent predictor of overall (HR = 1.84, P = .0002) and cause-specific (HR = 2.07, P = .0012) survival in node-negative patients. LVI maintained its independent predictor status in competing risks regression models (P = .013), where other-cause mortality was considered as a competing risk. LVI was not a predictor of recurrence or survival in node-positive patients.
CONCLUSION: LVI is an independent predictor of recurrence and decreased cause-specific and overall survival in patients who undergo cystectomy for invasive bladder cancer and are node-negative. These patients represent a high risk group that may benefit from integrated therapy with cystectomy and perioperative systemic chemotherapy.
INTRODUCTION
Bladder cancer is the fourth most common cancer in men (6%) and the 10th most common cancer in women (2%), accounting in men for 3% of cancer deaths in the year 2004 in the United States.1 On average, 15% to 30% of all patients with bladder cancer are diagnosed with muscle-invasive tumors, for which radical cystectomy is the current gold standard therapy.2 Unfortunately, as many as 40% of patients with organ-confined disease at the time of cystectomy subsequently have recurrence.3-5 Once bladder cancer metastasizes, treatments such as radiotherapy and systemic chemotherapy do not significantly change the overall survival rates.6,7
Several studies have evaluated the risk factors for recurrence and survival after cystectomy.8,9 Advanced pathologic stage,3,8-10 nodal involvement,3,9,10 tumor size greater than 3 cm,3 elevated creatinine,3 and vascular invasion10 have been found to be independent risk factors for recurrence. Advanced pathologic stage and nodal involvement have also been found to be independent risk factors for survival.9,11,12 The role of lymphovascular invasion (LVI) as a prognostic factor for progression and survival is controversial. Early studies identified LVI as a poor prognostic feature.13,14 Vascular but not lymphatic invasion was identified as an independent predictor of recurrence after cystectomy in one study.10 In other studies, however, LVI was not a predictor of lymph node metastases,15 relapse,3 or survival.9,12 To test the hypothesis that LVI is a predictor of recurrence and survival, we performed a retrospective multi-institutional study (University of Texas Southwestern [Dallas, TX], Baylor College of Medicine [Houston, TX], Johns Hopkins University [Baltimore, MD]) of 958 patients who underwent cystectomy for bladder cancer between 1984 and 2003.
PATIENTS AND METHODS
Patient Population
All studies were undertaken with the approval and institutional oversight of the institutional review board for the Protection of Human Subjects. A retrospective multi-institutional study (University of Texas Southwestern, Baylor College of Medicine, Johns Hopkins University) was performed on 958 patients who underwent cystectomy with bilateral lymphadenectomy for bladder cancer between 1984 and 2003. For each patient, we recorded sociodemographic, precystectomy, cystectomy, and postcystectomy variables. Sociodemographics consisted of patient age at cystectomy, the institution where cystectomy was performed, and the year of surgery. Precystectomy variables included transurethral resection (TUR) 1997 TNM stage, WHO grade, presence of carcinoma-in-situ, preoperative intravesical chemotherapy or immunotherapy, and neoadjuvant chemotherapy. Cystectomy variables included pathologic 1997 TNM stage, WHO grade, presence of coexisting carcinoma-in-situ, LVI, total number of resected lymph nodes, and number of positive lymph nodes. Postcystectomy variables included adjuvant chemotherapy and radiotherapy. There were 838 patients with complete datasets who were eligible for our analysis. Our analysis was limited to those patients who had transitional cell carcinoma. Thus, 62 patients with non–transitional cell carcinoma histology (26 squamous, 12 adenocarcinomas, 16 other, five small-cell, one spindle cell, and two undifferentiated) were excluded from analysis. Of the remaining 776 patients, information on LVI was not available for 26 patients. Thus, the analyses were based on 750 patients.
Of the patients in the study sample, 612 were men and 138 were women. The mean age at cystectomy was 64.8 years (standard deviation, 10.3; median, 65.9). The mean number of lymph nodes per patient removed at the time of cystectomy was 20.1 ± 10.2 (median, 17; interquartile range, 12). The number of nodes removed during pelvic lymphadenectomy was divided into quartiles for this analysis. The operative surgeon assigned the preoperative clinical stage according to the 1997 American Joint Committee on Cancer /TNM system. Final pathologic staging was defined as the higher of the preoperative clinical and the pathologic stages for patients with organ-confined disease. For clinically and/or pathologically non–organ-confined disease in patients who had not received neoadjuvant systemic chemotherapy or radiation therapy, the consensus stage for this analysis was based on the true postoperative pathologic stage. For clinically and pathologically non–organ-confined disease in patients who had received neoadjuvant systemic chemotherapy or radiation therapy, the consensus stage was highest of the two. The consensus grade was the higher of the preoperative and postoperative pathologic grades.
Preoperative and postoperative radiation was delivered to 2.3% and 4.7% of patients, respectively. Preoperative and postoperative chemotherapy was given to 4.9% and 26.7% of patients, respectively. For patients who were alive at the last follow-up and whose LVI status was known (N = 450), the mean follow-up was 3.9 years, and the median was 3.1 years (range, 13 days to 18 years). For patients who had died during the follow-up period and had known LVI status (n = 293), the mean follow-up was 2.4 years. The median follow-up was 1.3 years (range, 2 days to 14.3 years). Overall, 191 (22%) and 45 (5%) patients had more than 5- and 10-year follow-up data, respectively. Recurrences developed in 33.6% of patients (n = 252). Patients were classified as having either distant or local only, or both distant and local recurrence. Patients were classified as having simultaneous distant and local recurrence if the distant and local recurrences were detected within 30 days for each other. Overall, 39.7% of patients (n = 298) died, and 66% of all deaths (n = 198) were due to bladder cancer. The remaining 100 deaths (33%) were considered as other-cause mortality.
Pathologic Examination
The extent and depth of the TUR was defined by each of the participating surgeons. Standard radical cystectomy was performed at each institution, though specific technique and extent of lymphadenectomy varied between surgeons and institutions. Pathologic TUR, cystectomy, and lymphadenectomy specimens were processed according to institutional protocols. The 1997 TNM and WHO classifications were used for tissue staging and grading. Lymphovascular invasion was defined as the presence of tumor cells within an endothelium-lined space without underlying muscular walls. The presence of a clear-cut endothelial lining was an important requirement in that retraction space artifact is especially common in invasive urothelial carcinoma of the bladder. Immunohistochemistry for endothelial cells was not done, more in keeping with standard practice of pathologists. Any equivocal foci and foci in which tumor cells merely encroached on a vascular lumen were considered negative. No attempt was made to differentiate between vascular and lymphatic vessels because of the difficulty and lack of reproducibility when using routine light microscopic examination. Hospital charts, physician records, and approved cancer registries were used to determine cause-specific mortality.
Statistical Analysis
2 tests were used to evaluate the association between categorical variables. The Kaplan-Meier method was used to calculate survival functions, and differences were assessed with the log-rank statistic. Univariate and multivariate survival analyses were performed using the Cox proportional hazard regression model. Competing risks regression was used to test the significance of predictor variables after accounting for other-cause mortality. Statistical significance in this study was set as P .05. All reported P values are two-sided. Analyses were performed with SAS version 8.02 (SAS Institute Inc, Cary, NC) and S-PLUS Professional, version 1 (MathSoft Inc, Seattle, Washington).
RESULTS
LVI was present in 36.4% of patients (273 of 750) overall, involving 26% (151 of 581) and 72% (122 of 169) of node-negative and node-positive patients, respectively. Prevalence of LVI increased with higher pathologic stage (9.0%, 23%, 60%, and 78%, for T1, T2, T3, and T4, respectively; P < .001). LVI was associated with advanced pathologic features including grade, stage, and nodal status (Table 1) . When patients were stratified on the basis of lymph node status after pelvic lymphadenectomy, LVI was associated on univariate analysis with overall and cause-specific survival, as well as with recurrence (overall, local, and distant) in node-negative patients (Table 2). These associations were absent in node-positive patients.
On multivariate Cox regression analyses that included age, stage, grade, and number of pelvic lymph nodes removed, LVI was an independent predictor of local (HR = 2.03, P = .049), distant (HR = 2.60, P = .0011), and overall (HR = 2.02, P = .0003) recurrence (Table 3) in node-negative patients. On multivariate Cox regression analyses that included age, stage, grade, and number of pelvic lymph nodes removed, LVI was an independent predictor of overall (HR = 1.84, P = .0002) and cause-specific (HR = 2.07, P = .0012) survival (Table 4) in node-negative patients. In competing-risks regression models, which included age, stage, grade, and quartiles of the number of pelvic lymph nodes removed, LVI represented an independent predictor of local recurrence (P = .013), distant recurrence (P = .0095), overall recurrence (P = .0071), and cause-specific survival (P = .014).
In addition to LVI in node-negative patients, stage and number of lymph nodes removed during lymphadenectomy were independent predictors of overall and distant recurrence using multivariate Cox regression analysis (Table 3). Stage and number of lymph nodes removed during lymphadenectomy were also independent predictors of overall and cause-specific survival using multivariate Cox regression analysis (Table 4).
LVI status, lymph node status, and whether patients with positive nodes received postoperative chemotherapy impacted cause-specific survival in patients with varying pathologic stages of bladder cancer (Figs 1, 2, and 3). The 5-year actuarial cause-specific survival for patients with negative nodes was 86.9% ± 3.2 (SE) for patients without LVI and 72.1% ± 8.8 (SE) for patients with LVI and stage T2 disease; 65.1% ± 7.7 (SE) for patients without LVI, and 53.9% ± 7.5 (SE) for patients with LVI and stage T3 disease; 65.6% ±14 (SE) for patients without LVI; and 36.1% ± 11.3 (SE) for patients with LVI and stage T4 disease.
DISCUSSION
A major difficulty with treating bladder cancer is the high recurrence rate in patients with "organ-confined" disease after cystectomy.3,8 Once bladder cancer has metastasized, treatment options such as radiotherapy and systemic chemotherapy do not significantly change the overall survival rates.6,7 Consequently, identifying patients at high risk for recurrence and administering adjuvant therapies may offer improved survival. Our multi-institutional study found that LVI predicts advanced stage, grade, and nodal status. LVI was also an independent predictor of overall and cause-specific survival, as well as overall, local, and distant recurrence in patients who had negative lymph nodes on lymphadenectomy at the time of cystectomy. Moreover, we tested the contribution of LVI in competing risks regression models. This modeling technique accounts for the effect of other-cause mortality, which may preclude the occurrence of a bladder cancer–specific event (recurrence or cause-specific mortality). Competing risks regression provides a more conservative estimate of the effect relative to Cox regression or Kaplan-Meier models. In competing-risks regression analyses, LVI was an independent predictor of local recurrence, distant recurrence, overall recurrence, and of cause-specific survival.
In Cox regression models, LVI was not an independent predictor of recurrence or survival in those patients found to have lymph node metastases. While most patients with positive lymph nodes at the time of cystectomy would be given adjuvant chemotherapy anyway if medically tolerable, this is an important finding because it may help us identify select patients who are lymph node–negative but might benefit from adjuvant therapies. This feature may also be useful in choosing high-risk patients for future clinical trials or in analyzing future trial results.
Results of our study were consistent with those of others. We found that 36.4% of patients had LVI. We also noted that the incidence of LVI increased with higher pathologic tumor stage. This corresponds to rates of LVI found in previous studies, which ranged from 35% to 54%.3,9,10,12 Early studies identified LVI as a poor prognostic feature but did not control for possible confounding factors by using multivariate analyses.13,14 Bassi et al evaluated 535 patients with bladder cancer treated with cystectomy, and found that while LVI was significantly associated with overall survival in univariate analysis, it was not an independent predictor on multivariate analysis. Only stage and lymph node status were predictive of survival in this study.9 In another study, Ennis et al evaluated 85 patients who underwent radical cystectomy and did not receive chemotherapy or radiation therapy. They found that those patients with and without LVI had 3-year relapse-free survival rates of 53% and 73%, respectively. This result did not reach statistical significance, however, because of the small patient sample (P = .06), and it was not an independent prognostic variable in multivariate analysis.3 In a study of 154 patients with bladder cancer who underwent cystectomy, Hara et al similarly found an association between LVI and survival on univariate, but not on multivariate analysis.12 Notably, Leissner et al found that vascular but not lymphatic invasion was identified as an independent predictor of recurrence after cystectomy in an evaluation of 283 patients.10 Our study had the advantage of evaluating LVI in a large number of patients from multiple institutions, and thus allowed more powerful statistical analyses and provided more generalizable findings. The present study demonstrated an association of LVI with advanced stage, high grade, and lymph node metastases. Furthermore, the strong association between LVI and the rate of metastases to regional lymph nodes in the current study supports the hypothesis that lymphatic vessel invasion precedes, or occurs concurrently with lymph node metastasis. Once the patient has lymph node metastases, the LVI status no longer adds predictive value for survival or recurrence. However, the LVI status in patients with negative lymph nodes at the time of cystectomy and lymphadenectomy is an important predictor of survival and recurrence, as it seems able to identify a subgroup of patients with micrometastases or false-negative lymph node status. Support for this contention is the finding that an increased number of nodes removed during lymphadenectomy was an independent predictor of overall and cause-specific survival, as well as distant and overall recurrence in node-negative patients. If patients with "negative" nodes have micrometastases within lymph nodes that were not identified by a pathologist, and a greater number of these nodes are removed, then the chance of achieving a surgical cure increases. The association between increased survival and an increased number of nodes removed has been reported previously.16
One may extrapolate that patients with LVI may benefit from adjuvant systemic agents as they are at high risk for recurrence. To date, there has been at least one study that evaluated the use of adjuvant cyclophosphamide, doxorubicin, and cisplatin chemotherapy in patients with high risk for relapse after cystectomy.14 In this study comparing outcomes for 62 high-risk controls and 71 patients receiving adjuvant therapy, patients with resected nodal metastases, extravesicular involvement of tumor, and direct invasion of pelvic viscera benefited from adjuvant therapy, but those with LVI as a sole manifestation of their high risk status did not. While conclusions cannot be reached on the basis of this one study, it is apparent that careful evaluation is required before use of adjuvant therapies in patients solely because of LVI as a risk factor.
One avenue for improving patient outcomes is to improve predictions of which patients may best benefit from neoadjuvant chemotherapy. Grossman et al found that neoadjuvant chemotherapy (methotrexate, vinblastine, doxorubicin, and cisplatin) followed by radical cystectomy increased the likelihood of eliminating residual tumor and was associated with improved survival compared with cystectomy alone.17 LVI status on transurethral resection of bladder tumor may identify those patients who are best suited for neoadjuvant chemotherapy. Millikan et al found that LVI was associated with a high degree of pathological upstaging after cystectomy from clinically organ-confined bladder cancer to either extravesical extension (36%) or lymph node positivity (27%).5 This observation supports our finding of a significant association between high-stage disease, advanced lymph node status, and LVI.
Several studies have identified markers that are predictive of increased risk of recurrence and metastases in patients undergoing radical cystectomy. Preoperative measures of plasma soluble E-cadherin,18 interleukin-6,19 urokinase plasminogen activator,20 and transforming growth factor-beta(1)21 were predictive of poor outcomes after cystectomy, features of advanced pathologic stage, lymph node involvement, and LVI. Therefore, it is possible that a combination of preoperative serum or urine markers and features such as LVI could improve identification of patients who would benefit from neoadjuvant or adjuvant chemotherapy.
Several limitations of our study need to be considered. The population underwent radical cystectomy and bilateral pelvic lymphadenectomy by multiple surgeons, and their specimens were evaluated by multiple pathologists. Most studies evaluating LVI have come from single-institution experiences. While prognostic indicators may perform well in a select group of patients treated at such centers, it remains to be determined whether these are applicable to the greater population of patients with bladder cancer. Multicenter studies provide the advantage of limiting the biases of single institution studies. Similarly, while it may be preferable for a single pathologist specializing in genitourinary pathology to review each cystectomy specimen, the set-up of this multi-institutional study reflects a real-world practice in which multiple pathologists review tissue specimens, and their interpretation is then used in clinical decision-making with the patient. It should be noted that in our study, surgery was performed primarily by urologic oncologists at their respective institutions. Furthermore, pathologists at our institutions evaluate a higher volume of urologic cancer than pathologists at most community hospitals.
Authors' Disclosures of Potential Conflicts of Interest
The authors indicated no potential conflicts of interest.
Acknowledgment
We acknowledge Dr Thomas Wheeler for his assistance with this manuscript.
NOTES
Drs Lotan and Gupta contributed equally to this study.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Jemal A, Tiwari RC, Murray T, et al: Cancer statistics, 2004. CA Cancer J Clin 54:8-29, 2004
Lerner SP, Skinner E, Skinner DG: Radical cystectomy in regionally advanced bladder cancer. Urol Clin North Am 19:713-723, 1992
Ennis RD, Petrylak DP, Singh P, et al: The effect of cystectomy, and perioperative methotrexate, vinblastine, doxorubicin and cisplatin chemotherapy on the risk and pattern of relapse in patients with muscle invasive bladder cancer. J Urol 163:1413-1418, 2000
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
Millikan R, Dinney C, Swanson D, et al: Integrated therapy for locally advanced bladder cancer: Final report of a randomized trial of cystectomy plus adjuvant M-VAC versus cystectomy with both preoperative and postoperative M-VAC. J Clin Oncol 19:4005-4013, 2001
Borden LS Jr, Clark PE, Hall MC: Bladder cancer. Curr Opin Oncol 15:227-233, 2003
Pashos CL, Botteman MF, Laskin BL, et al: Bladder cancer: Epidemiology, diagnosis, and management. Cancer Pract 10:311-322, 2002
Slaton JW, Swanson DA, Grossman HB, et al: A stage specific approach to tumor surveillance after radical cystectomy for transitional cell carcinoma of the bladder. J Urol 162:710-714, 1999
Bassi P, Ferrante GD, Piazza N, et al: Prognostic factors of outcome after radical cystectomy for bladder cancer: A retrospective study of a homogeneous patient cohort. J Urol 161:1494-1497, 1999
Leissner J, Koeppen C, Wolf HK: Prognostic significance of vascular and perineural invasion in urothelial bladder cancer treated with radical cystectomy. J Urol 169:955-960, 2003
Frazier HA, Robertson JE, Dodge RK, et al: The value of pathologic factors in predicting cancer-specific survival among patients treated with radical cystectomy for transitional cell carcinoma of the bladder and prostate. Cancer 71:3993-4001, 1993
Hara S, Miyake H, Fujisawa M, et al: Prognostic variables in patients who have undergone radical cystectomy for transitional cell carcinoma of the bladder. Jpn J Clin Oncol 31:399-402, 2001
Anderstrom C, Johansson S, Nilsson S: The significance of lamina propria invasion on the prognosis of patients with bladder tumors. J Urol 124:23-26, 1980
Logothetis CJ, Johnson DE, Chong C, et al: Adjuvant cyclophosphamide, doxorubicin, and cisplatin chemotherapy for bladder cancer: An update. J Clin Oncol 6:1590-1596, 1988
Jaeger TM, Weidner N, Chew K, et al: Tumor angiogenesis correlates with lymph node metastases in invasive bladder cancer. J Urol 154:69-71, 1995
Sanderson KM, Stein JP, Skinner DG: The evolving role of pelvic lymphadenectomy in the treatment of bladder cancer. Urol Oncol 22:205-211, 2004
Grossman HB, Natale RB, Tangen CM, et al: Neoadjuvant chemotherapy plus cystectomy compared with cystectomy alone for locally advanced bladder cancer. N Engl J Med 349:859-866, 2003
Matsumoto K, Shariat SF, Casella R, et al: Preoperative plasma soluble E-cadherin predicts metastases to lymph nodes and prognosis in patients undergoing radical cystectomy. J Urol 170:2248-2252, 2003
Andrews B, Shariat SF, Kim JH, et al: Preoperative plasma levels of interleukin-6 and its soluble receptor predict disease recurrence and survival of patients with bladder cancer. J Urol 167:1475-1481, 2002
Shariat SF, Monoski MA, Andrews B, et al: Association of plasma urokinase-type plasminogen activator and its receptor with clinical outcome in patients undergoing radical cystectomy for transitional cell carcinoma of the bladder. Urology 61:1053-1058, 2003
Shariat SF, Kim JH, Andrews B, et al: Preoperative plasma levels of transforming growth factor beta(1) strongly predict clinical outcome in patients with bladder carcinoma. Cancer 92:2985-2992, 2001(Yair Lotan, Amit Gupta, S)