Presented in part as a podium presentation at the American Urological Association (AUA) Annual Conference May 2010, San Francisco, CA, USA.
Prognostic risk stratification of pathological stage T2N0 bladder cancer after radical cystectomy
Version of Record online: 19 NOV 2010
© 2010 THE AUTHORS. BJU INTERNATIONAL © 2010 BJU INTERNATIONAL
Volume 108, Issue 5, pages 687–692, September 2011
How to Cite
Sonpavde, G., Khan, M. M., Svatek, R. S., Lee, R., Novara, G., Tilki, D., Lerner, S. P., Amiel, G. E., Skinner, E., Karakiewicz, P. I., Bastian, P. J., Kassouf, W., Fritsche, H.-M., Izawa, J. I., Ficarra, V., Dinney, C. P., Lotan, Y., Fradet, Y. and Shariat, S. F. (2011), Prognostic risk stratification of pathological stage T2N0 bladder cancer after radical cystectomy. BJU International, 108: 687–692. doi: 10.1111/j.1464-410X.2010.09902.x
- Issue online: 19 AUG 2011
- Version of Record online: 19 NOV 2010
- Accepted for publication 1 September 2010
- pathological stage T2N0 bladder cancer;
- risk stratification;
- radical cystectomy
Study Type – Therapy (individual cohort)
Level of Evidence 2b
What’s known on the subject? and What does the study add?
Patients with urothelial carcinoma of the bladder (UCB) and pathological (p) stage T2N0 disease exhibit a range of clinical outcomes with an overall estimated 10–25% experiencing recurrence and death after radical cystectomy (RC). Nomograms to prognosticate UCB post-RC have been developed in heterogeneous datasets of patients across different stages and do not address factors unique to pT2N0 disease.
A user-friendly prognostic risk model was devised for patients with pT2N0 UCB undergoing RC based on residual pathological stage at RC (pT2a, pT2b, <pT2), grade (high, low) and lymphovascular invasion (present or absent). The three risk groups exhibited five-year recurrence-free survivals of 95%, 86% and 62%, respectively. These data warrant validation and may help design adjuvant therapy trials as well as tailor the intensity of post-operative monitoring.
• To stratify risk of pathological (p) T2N0 urothelial carcinoma of the bladder after radical cystectomy (RC) based on pathological factors to facilitate the development of adjuvant therapy trials for high-risk patients.
PATIENTS AND METHODS
• The study comprised 707 patients from a database of patients with pT2N0 urothelial carcinoma of the bladder who had undergone RC and not received perioperative chemotherapy.
• The effect of residual pT-stage at RC, age, grade, lymphovascular invasion and number of lymph nodes removed on recurrence-free survival was evaluated using Cox regression analyses. A weighted prognostic model was devised with significant variables.
• The median follow up was 60.9 months. In multivariable analyses, residual disease at RC (pT2a: hazard ratio (HR) 1.740, P = 0.03; for pT2b: HR 3.075, P < 0.001; both compared with <pT2), high-grade (HR 2.127, P = 0.09) and lymphovascular invasion (HR 2.234, P < 0.001) were associated with recurrence-free survival (c = 0.70).
• Three risk groups were devised based on weighted variables with 5-year recurrence-free survival of 95% (95% CI 87–98), 86% (95% CI 81–90) and 62% (95% CI 54–69) in the good-risk, intermediate-risk and poor-risk groups, respectively (c = 0.68). The primary limitation is the retrospective and multicenter feature.
• A prognostic risk model for patients with pT2N0 bladder cancer undergoing RC with generally adequate lymph node dissection was constructed based on residual pathological stage at RC, grade and lymphovascular invasion.
• These data warrant validation and may enable the selection of patients with high-risk pT2N0 urothelial carcinoma of the bladder for adjuvant therapy trials.
urothelial carcinoma of the bladder
Patients with pathological (p) extravesical (T3 or T4) and/or lymph node (LN) -positive urothelial carcinoma of the urinary bladder (UCB) have a high risk of disease recurrence and are routinely included in trials of adjuvant chemotherapy [1–3]. In contrast, patients with UCB with organ-confined muscle-invasive disease (i.e. pT2N0) have a lower risk of recurrence and are not generally included in adjuvant therapy trials. However, pT2N0 patients exhibit a wide range of clinical outcomes with an overall estimated 10–25% experiencing disease recurrence and eventual death after radical cystectomy (RC). Besides stage, pathological variables known to impact outcomes and constituting components of nomograms developed in heterogeneous datasets of patients across stages include grade, lymphovascular invasion (LVI) and age [4–6]. Residual pathological stage and depth of muscle invasion at RC have been recently recognized as major prognostic factors in T2N0 UCB [7,8]. The quality of LN dissection may also have an impact, although definitive data are lacking [9–11].
Recurrent disease generally manifests as distant metastasis and is usually incurable. Identification of pT2N0 patients at high risk of disease recurrence could improve counselling of these patients regarding prognosis and possible inclusion into clinical trials of adjuvant therapy, which might improve outcomes and cure rates. Most recent trials of adjuvant chemotherapy have included patients with at least pT3 and/or node-positive disease and have been unable to complete the target accrual. The inclusion of high-risk pT2N0 patients may also confer the added secondary benefit of enabling more rapid accrual to trials of adjuvant therapy and facilitate advances in the management of patients with high-risk UCB. We hypothesized that patients with pT2N0 stage can be risk-stratified into prognostically different groups based on the aforementioned readily available pathological variables. The purpose of this study was to conduct a retrospective analysis of a large cohort of patients with pT2N0 stage UCB who underwent RC without perioperative therapy, to test our hypothesis.
MATERIALS AND METHODS
We studied a large international cohort (nine academic institutions) of patients with pT2N0 stage UCB treated with RC with curative intent between 1971 and 2008. Substaging information (pT2a or pT2b) was required to be available. The stage was the higher of either the clinical or pathological stage including pathological data from both transurethral resection of bladder tumour and RC. The study was performed with the approval of the Institutional Review Board for the protection of Human Subjects at each institution. None of the patients received perioperative radiation therapy or systemic chemotherapy. Patients were generally seen postoperatively at least every 3 to 4 months for the first year, semiannually for the second year, and annually thereafter. Follow-up visits consisted of a physical examination and serum chemistry evaluation. Diagnostic imaging of the upper tract (e.g. ultrasonography and/or intravenous pyelography, computed tomography of the abdomen/pelvis with contrast) and chest radiography were performed at least annually or more frequently if clinically indicated. Additional radiographic evaluations (e.g. bone scan) were taken at the discretion of the treating physician. Detection of cancer in the ureter and/or urethra was recorded as a second (metachronous) primary and not as a recurrence.
Genitourinary pathologists examined all the RC specimens according to institutional protocols. Multiple, well-oriented quadrant sections from the tumour, adjacent and distal bladder wall, ureters and urethra were processed. Pelvic LN dissections were examined grossly and subjected to histological examination. Tumours were staged according to the fifth edition of the American Joint Committee of Cancer staging manual and graded according to the 1973 WHO classification [12,13]. The pT2a and pT2b sub-stages were defined by the invasion of superficial muscle (inner half) or deep muscle (outer half), respectively. Patients with grade 0 and 1 disease were grouped together in the low-grade group and those with grades 2 and 3 disease were grouped in the high-grade group. LVI was defined as the unequivocal presence of tumour cells within an endothelium-lined space without underlying muscular walls [14–16].
Continuous, normally distributed variables were reported as the mean value with sd. Continuous non-normal variables were presented as the median values and interquartile range. Recurrence-free survival (RFS) was defined as time to recurrence or death from any cause, and 5-year RFS rates were calculated using the Kaplan–Meier method. Differences in RFS rates were evaluated using the log-rank test. Univariable and multivariable Cox regression analyses tested the effect of residual pT-stage at RC (pT2b, pT2a or <pT2), LVI (positive or negative), grade (high vs low), number of LNs dissected (continuous variable) and age (continuous variable) on RFS. All statistical tests were performed using SAS statistical software, version 9.2 (SAS Institute, Cary, NC, USA) and Stata, version 11 (StataCorp, College Station, TX, USA). All tests were two-sided with a significance level set at P < 0.05. We also constructed a model to risk-stratify pT2N0 patients into different risk groups based on factors that remained significant on multivariable analysis. We applied a weighting method to each variable that has been employed in previous studies by other investigators . The β-regression coefficients from the final model applied to the original sample were used to develop prognostic scores (with low scores reflecting a greater probability of RFS) for each variable in the model. The score was calculated by dividing the regression coefficients by 0.554 (the smallest β coefficient), multiplying by 2.0 and rounding to the nearest whole number. A concordance (c) index of 0.5 implied a model no better than chance and c > 0.5 implied better than chance .
A total of 707 patients with pT2N0 UCB were eligible for this analysis. The clinical and pathological characteristics of this cohort are shown in Table 1. Mean patient age was 66 years (sd 10.3 years) and 78% were men. Pathological stage at RC was pT2aN0 in 170 patients, pT2bN0 in 173, and 364 had <pT2N0 UCB. Of patients with RC stage <pT2N0, 102 had pT0N0 UCB, 138 had pT1N0 UCB, 21 had pTaN0 UCB and 103 had pTis UCB. Most patients had high-grade disease (n = 588) as opposed to low-grade disease (n = 117). LVI was present in 104 patients (14.7%) and the median number of LNs removed was 19 (interquartile range 21). The median follow up of all patients for RFS analysis was 60.9 months (95% CI 52.0–66.5).
|Characteristic||Radical cystectomy stage|
|All patients n (%)||Stage <T2 n (%)||Stage T2a n (%)||Stage T2b n (%)||P value (variance)|
|Number of patients||707 (100)||364 (51.5)||170 (24.0)||173 (24.5)|
|Male||551 (77.9)||288 (79.1)||139 (81.8)||124 (71.7)|
|Female||156 (22.1)||76 (20.9)||31 (18.2)||49 (28.3)||0.06|
|Low||117 (16.5)||111 (30.5)||4 (2.4)||2 (1.2)|
|High||588 (83.2)||251 (69.0)||166 (97.6)||171 (98.8)||<0.001|
|Lymphovascular invasion||104 (14.7)||10 (2.8)||45 (26.5)||49 (28.3)||<0.001|
|Number of lymph nodes removed*|
|0–10||148 (20.9)||66 (18.4)||33 (19.4)||49 (28.3)|
|11–20||217 (30.7)||118 (33.0)||44 (32.9)||55 (31.2)|
|>20||320 (45.3)||174 (48.6)||87 (51.2)||59 (34.1)||0.006|
|≤66 years||358 (50.6)||194 (53.3)||86 (50.6)||78 (45.1)||0.20|
Multivariable analysis of prognostic factors for long-term outcomes in patients with consensus stage T2N0 UCB found that the presence of LVI, high-grade disease and residual RC stage were significantly associated with RFS on univariable analysis, whereas age and number of LNs dissected were not (Table 2). Hazard ratios (HR) for LVI (HR 2.234, P < 0.001), high-grade disease (HR 2.127, P = 0.09) and residual muscle-invasive disease compared with non-muscle invasive disease (for pT2a: HR = 1.740, P ≤ 0.03; for pT2b: HR = 3.075, P < 0.001) were associated with RFS on multivariable analysis (Harrell’s c-statistic 0.70) (Table 3).
|Variable||Hazard ratio||95% CI||P value|
|Number of lymph nodes removed||0.997||0.987–1.008||0.59|
|Radical cystectomy stage|
|Variable||Hazard ratio||95% CI||P value||β coefficient||Prognostic score points*|
|Radical cystectomy stage|
A prognostic risk model was constructed to predict RFS based on residual RC pathological stage, LVI and grade (Table 3, Fig. 1). The range of the weighted score was 0–4 for each variable with a maximum total of 10 when combining the variables. The total of point scores for each variable was added to give a total risk stratification score of 0–10. Three risk groups were constructed based on the presence of 0 (good risk), 2–5 (intermediate risk) and 6–10 (poor risk) points, with a c-statistic of 0.68 (Table 4). The median RFS for the good-, intermediate- and poor-risk groups was not reached, and the estimated 5-year RFS was 95% (95% CI 87–98%), 86% (95% CI 81–90%) and 62% (95% CI 54–69%), respectively. The corresponding 5-year overall survivals for the risk groups were 89% (95% CI 80–94%), 78% (95% CI 73–82%) and 58% (95% CI 50–65%) (Fig. 2). The number of patients in the good-, intermediate- and poor-risk groups was 113, 367 and 227, respectively.
|Total number of points||Risk group||n (%)||5-year RFS, % (95% CI)||5-year OS, %, (95% CI)|
|0||Good||113 (16.0)||95 (87–98)||89 (80–94)|
|2–5||Intermediate||367 (51.9)||86 (81–90)||78 (73–82)|
|6–10||Poor||227 (32.1)||62 (54–69)||58 (50–65)|
Residual pathological stage at RC, LVI status and high grade were strong independent predictors of RFS in patients with pT2N0 stage UCB undergoing RC with generally adequate LN dissection and without perioperative chemotherapy. We were able to categorize relatively homogeneous patients with pT2N0 stage into three prognostically different risk groups with substantial accuracy (Fig. 1). This simple and user-friendly risk stratification based on readily available pathological features may aid in clinical decision-making regarding the eligibility of pT2N0 patients for trials of adjuvant therapy, e.g. outcomes for the high-risk group appear similar to outcomes for pT3N0 disease, which has traditionally been eligible for trials of adjuvant chemotherapy. Additionally, the rigor of post-RC clinical monitoring may be tailored depending on the risk group with more frequent monitoring for the high-risk group. Incorporation of these data may improve the accuracy of current predictive tools such as nomograms [4,6]. The Bladder Cancer Research Consortium and International Bladder Cancer Consortium nomograms included a large but heterogeneous group of patients. In addition, the International Bladder Cancer Consortium nomogram did not include LVI; and the Bladder Cancer Research Consortium nomogram included perioperative chemotherapy as a variable, which may confound an assessment of prognosis. Finally, artifical neural networks and other tools require computational support, which may be a barrier to their employment.
Our multivariable analyses validate our previous data that residual pathological stage at RC exerts a powerful effect on outcomes . Downstaging to residual pT0 from transurethral resection (TUR) to RC with or without neoadjuvant chemotherapy has been shown to be associated with better outcomes [2,7,19–21]. In addition, we recently validated the prognostic importance of the depth of muscle-invasion at RC in patients without LN involvement . Therefore, a continuum of risk appears to exist for residual pathological stage at RC with an increase in risk of recurrence from no muscle-invasive disease to pT2a to pT2b disease. The prognostically favourable outcome of lower stage at RC may be the result of a more complete TUR based on the surgeon’s skills and dedication as well as tumour characteristics such as extent, location, multifocality, growth rate and biology . Despite the inherent pitfalls in assessing TUR specimens and differing institutional standards for extent of TUR, an accurate pathological evaluation depends on the quality of tumour and tissue specimens provided by the urologist. Therefore, we believe that a thorough TUR should be attempted even if RC is planned, because a complete TUR may abrogate tumour progression in the time from TUR to RC and allow better down-staging . Our model also validates LVI and high-grade disease as powerful predictors of prognosis specifically in patients with T2N0 stage disease, which have already been shown to be generally important across different stages .
Age did not have an independently significant association with RFS, although one study showed that being older may confer a higher risk of recurrence . The lack of independent favourable prognostic impact of the number of LNs removed is somewhat surprising. However, the benefit from better LN dissection may not have been identifiable in the context of generally adequate LN dissection in the setting of bladder-confined disease. Additionally, the extent of LN dissection (i.e. extended dissection vs standard pelvic dissection) rather than the number of LNs identified by the pathologist may be more relevant, but the extent and template of dissection was not available from our database. Other retrospective studies have shown a benefit for extended lymphadenectomy in node-negative patients and for a higher node density (positive nodes as a proportion of total number removed) [9–11,25,26]. In addition, LN metastases are detected more often with extended dissection, which leads to stage migration and enhanced outcomes resulting from better staging. A prospective trial is ongoing in Germany and another trial is being planned by the Southwest Oncology Group that will compare standard with extended lymph node dissection.
Despite strong evidence supporting the use of neoadjuvant chemotherapy in patients with clinical stage T2–4aN0M0 disease, very few patients actually receive it, which underscores the reluctance of urologists to recommend neoadjuvant chemotherapy when only a proportion of patients will experience disease recurrence [20,27,28]. Conversely, initial RC allows clinical decision-making regarding adjuvant chemotherapy based on the prognostically more accurate pathological stage. Therefore, the adjuvant chemotherapy paradigm may be justified in clinical T2N0 patients in an effort to complete the pathological staging and enhance risk-stratification of such patients. Notably, approximately 40% of patients with clinical T2N0 disease are upstaged to extravesical or LN-positive disease at RC, which illustrates suboptimal current clinical staging . Optimizing clinical staging using emerging novel technologies so that it correlates better with pathological stage may assist in choosing between an initial RC approach for clinical T2N0 and reserving neoadjuvant chemotherapy for clinical ≥ T3–T4aN0M0 disease.
Several limitations inherent in retrospective studies may be operative. Although there may be no single optimal clinical endpoint, the employment of RFS as the primary clinical endpoint may have attenuated the impact of pathological variables because of confounding non-cancer mortality in this generally elderly population with comorbidities. We could not employ cancer-specific survival because this was not available in this database (cancer-specific survival also has a subjective component), and we did not employ overall survival as the primary clinical endpoint because it does not capture disease recurrence in addition to survival. Our c-statistic of 0.68 for the model may indicate the confounding influence of such mortality and the potential for substantial further refinement. We only studied patients who did not receive perioperative chemotherapy to isolate the prognostic impact of pathological variables (to help make clinical decisions regarding adjuvant chemotherapy) and avoid the confounding impact of chemotherapy on outcomes; this may have led to some bias in the selection of patients. We could not account for the impact of histological heterogeneity, tumour focality, location and size, time from last TUR to RC, the number of TURs performed, stage and/or grade progression and use of intravesical therapy. Another limitation is the heterogeneity in clinica decision-making, surgical intervention and pathological evaluation. The RC were performed by multiple surgeons and the specimens were evaluated by multiple pathologists, although these were academic centres of excellence and the data are probably valid. The database spanned 37 years and used the 1973 grading system. A central pathology review will probably enhance the quality of pathology data, but central review is currently beyond the scope of this study and will be considered for future studies. We studied node-negative patients who had muscle-invasive and bladder-confined disease (pT2) after combining pathological data from TUR and RC. Although this inclusion criterion enabled us to identify the predictive power of residual pathological stage at RC, most studies have only included pT2N0 disease at RC when studying pT2N0 UCB. The high proportion of patients (51.5%) with <pT2 disease at RC suggests that some patients may have been misclassified as having muscle-invasive disease at transurethral resection of bladder tumour, i.e. muscularis mucosa and not muscularis propria invasion may have been present. However, in a previous study of patients with baseline T2N0 disease, we found that a similar ∼43% of patients exhibited <pT2 disease at RC . Additionally, we only studied patients with pT2N0 disease, a group that is likely to contain a substantially higher proportion with <pT2 disease at RC than a database of all operable patients undergoing RC. It is known that among patients with clinical T2–T4aN0 disease undergoing RC without previous neoadjuvant chemotherapy in a prospective phase III trial setting, approximately 15% has pT0 disease at RC . A larger proportion with low grade disease (117 of 707), most of whom had <pT2 disease at RC, may also be attributed to the fact that our study included only pT2N0 patients. We could not identify differential outcomes based on RC stage pT0 versus presence of non-invasive disease, probably because there were too few patients in each subset (data not shown), although the excellent overall outcomes for the group with RC stage <pT2 suggests the lack of significantly different outcomes for different subsets within this group. We also excluded patients who did not have recorded depth of muscle-invasion at RC, although an examination of such patients revealed a similar distribution of residual non-muscle invasive disease at RC, LVI and grade (data not shown). This multi-institutional study reflects a real-world academic practice involving experienced urological oncologists and pathologists. Although it involved established academic centres with excellent patient follow-up data, there may be some differences in the rigor of follow up, which can influence RFS.
To conclude, our risk stratification of pT2N0 UCB is extremely user friendly and consequently is likely to be used in clinical practice. However, the data remain hypothesis-generating and further external validation in modern datasets is warranted. The model includes readily available pathological data and further refinement may be possible by the incorporation of biomarkers [29,30]. If validated, we propose risk-adapted adjuvant therapy with the routine consideration of adjuvant chemotherapy (with or without biological agents) on or off trials for the poor-risk group, whereas the good-risk group may be observed without adjuvant therapy. Those with intermediate-risk disease may warrant adjuvant therapy trials employing highly tolerable biological agents.
CONFLICT OF INTEREST
- 28Neoadjuvant cisplatin, methotrexate, and vinblastine chemotherapy for muscle-invasive bladder cancer: a randomized controlled trial. International collaboration of trialists. Lancet 1999; 354: 533–40