Lack of pathologic down-staging with neoadjuvant chemotherapy for muscle-invasive urothelial carcinoma of the bladder

A contemporary series

Authors


Abstract

BACKGROUND:

The postcystectomy survival benefit associated with the combination of methotrexate, vinblastine, doxorubicin, and cisplatin (MVAC) neoadjuvant chemotherapy (NC) for muscle-invasive bladder cancer has been most evident in patients who achieve a pathologic complete response. The outcome of NC and open radical cystectomy (RC) was evaluated in a contemporary cohort of patients in a tertiary referral setting.

METHODS:

From January 2006 to November 2007, 117 patients underwent open RC at Cleveland Clinic for muscle-invasive bladder cancer, 29 (25%) of whom received NC. Patient information was obtained from a prospective database.

RESULTS:

Clinical stage at the time of diagnosis in the NC cohort was T2 in 23 (79%) and T3-4a in 6 (21%) patients. A total of 20 (69%) patients received the combination of gemcitabine and cisplatin (GC), 4 (14%) received MVAC, and 5 (17%) received other regimens. The median interval from the time of diagnosis of muscle-invasive bladder cancer to RC was 208 days (interquartile range, 149 days -327 days) in the NC cohort. Overall, only 2 patients (7%; 95% confidence interval [95% CI], 0 patients-17 patients) achieved a pathologic complete response, 18 (62%; 95% CI, 43 patients-81 patients) had nonorgan-confined residual cancer, and the overall median progression-free survival was 10.5 months (95% CI, 7 months -14 months).

CONCLUSIONS:

Few RC patients in these investigators' recent experience achieved a pathologic complete response with NC, and most experienced rapid disease progression. These poor outcomes may be related to the use of non-MVAC-based regimens or excessive delay in performing RC. In the absence of supportive data for GC in the neoadjuvant setting, MVAC remained the preferred regimen. Excessive delays in performing RC may negate the benefit of NC. Cancer 2009. © 2009 American Cancer Society.

The use of neoadjuvant chemotherapy (NC) before open radical cystectomy (RC) for invasive urothelial carcinoma of the bladder is supported by level I evidence demonstrating a survival benefit compared with RC alone. A phase 3 randomized trial by the Southwest Oncology Group (SWOG 8710) showed a 25% and 40% relative risk reduction in all-cause and cancer-specific mortality, respectively, in patients receiving 3 cycles of the neoadjuvant combination of methotrexate, vinblastine, doxorubicin, and cisplatin (MVAC).1 This survival benefit was observed across all clinical stages. A meta-analysis of 11 other phase 3 randomized trials demonstrated a significant benefit with NC, equivalent to a 5% absolute improvement in survival.2, 3 Similar evidence in support of adjuvant chemotherapy is not as robust because these trials were flawed by poor design, protocol violations, and inadequate sample sizes. As such, NC before RC is considered by many to be the standard of care for muscle-invasive bladder cancer.4

Neoadjuvant MVAC is associated with substantial pathologic down-staging as pathologic complete responses (pCR; defined as no viable cancer within the cystectomy specimen) have been reported in 33% to 40% of patients receiving 2 to 3 cycles of cisplatin-based chemotherapy compared with a rate of 6% to 15% of pathologic stage T0 (pT0) in patients treated with RC alone.1, 5-10 In SWOG 8710, a pCR with neoadjuvant MVAC was observed in 50% and 30% of patients with clinical stage T2 and T3-4a bladder cancer, respectively.1 In a secondary analysis, the survival benefit associated with NC was principally observed in patients achieving a pCR; 85% of these patients were alive at 5 years, whereas the median survival of patients with residual cancer after NC was between 2.4 and 3.8 years. The favorable outcome of patients with a pCR after neoadjuvant MVAC has been reported by others.6 Hence, the ability to achieve a pCR may be considered a reasonable substitute endpoint for overall survival to assess the efficacy of NC.

Over the past decade, the combination of gemcitabine and cisplatin (GC) has supplanted MVAC as the standard regimen for advanced bladder cancer due to an improved toxicity profile without substantial differences in overall or progression-free survival.11, 12 Based on this evidence, GC is often substituted for MVAC for patients receiving NC before RC, despite the lack of evidence indicating a similar benefit in this setting. The logic of improved toxicity is not justification for the use of neoadjuvant GC given the lack of evidence that neoadjuvant MVAC precludes or delays RC in a substantial number of patients.1, 6

In a contemporary cohort of patients with clinical stage T2 through T4a bladder cancer, the pathologic responses to NC and postcystectomy survival were analyzed and compared with published clinical trials. We hypothesize that, if GC is equivalent to MVAC in the neoadjuvant setting, pCR rates should be similar.

MATERIALS AND METHODS

From January 2006 through November 2007, 244 consecutive patients with localized urothelial carcinoma of the bladder underwent RC at Cleveland Clinic (Cleveland, Ohio). Patients who underwent a laparoscopic RC (n = 79), had nonmuscle-invasive bladder cancer (n = 43), underwent salvage RC after chemoradiation (n = 3), or participated in a phase 2 neoadjuvant protocol (n = 2) were excluded. Of the 117 patients with clinical stage T2 through T4a, N0-2, M0 bladder cancer who underwent open RC, 29 (25%) received NC. The decision to administer NC in the absence of advanced clinical features varied according to patient and surgeon preference. All patients were staged clinically with computed tomography (CT) scans of the abdomen and pelvis and either CT scan of the chest or chest x-ray.

All patients underwent open RC. The choice of urinary diversion was left to surgeon and/or patient preference. The extent of pelvic lymph node dissection (PLND) varied by surgeon; all patients had at a minimum a standard PLND (level I—hypogastric, external iliac, obturator, and distal common iliac lymph nodes), and some surgeons routinely performed extended PLND (level I and level II—presacral and common iliac lymph nodes to aortic bifurcation) with or without removal of lymph nodes in level III (para-aortic, paracaval, and interaortocaval retroperitoneal lymph nodes to the inferior mesenteric artery).13

All patients had muscle-invasive urothelial carcinoma of the bladder confirmed by review of transurethral resection (TUR) specimens by genitourinary pathologists at our institution before treatment. Pathologic analysis of the RC specimen included extensive macroscopic and histologic evaluation. Specifically, residual macroscopic tumor or identifiable scar from prior transurethral resection were sampled at 1 to 2 sections/cm, which included complete transmural sections of the bladder wall to accurately determine pathologic stage. In addition, several histologic sections from various anatomic sites within the bladder, including the dome, anterior wall, lateral walls, posterior wall, trigone, urethra, and bilateral ureters were sampled for superficial disease or a second primary tumor. In cystoprostatectomy specimens, the prostate and the prostatic urethra were extensively sampled to identify the presence of superficial or invasive disease into the prostatic stroma. All tumors were staged according to the criteria set forth in the American Joint Committee on Cancer staging manual.14 All lymph nodes were entirely submitted for analysis per designated site and representative sections of surrounding fibroadipose tissue were in addition sampled. Review of all slides was performed by an experienced genitourinary pathologist. No evidence of bladder cancer in the RC specimen or lymph nodes was classified as pCR (or pT0 in patients not receiving NC).

Patient clinical information and follow-up data were obtained from a prospectively maintained RC database after patient consent. Data were analyzed using SPSS 12.0 statistical software (SPSS Inc, Chicago, Ill). Proportions were analyzed with the chi-square test. Continuous variables were analyzed with the Mann-Whitney U test. Postcystectomy survival was analyzed using the Kaplan-Meier method. The association of parameters with pathologic outcomes was assessed using logistic regression analysis. All P values resulted from the use of 2-sided statistical tests, and the level of significance was set at .05. The study was conducted under Health Insurance Portability and Accountability Act guidelines and received institutional review board approval.

RESULTS

Patient preoperative characteristics for the NC cohort are presented in Table 1. Clinical stage at diagnosis was T2 in 23 patients (79%), T3 in 5 patients (17%), and T4a in 1 patient (3%). Four patients (14%) were found to have enlarged lymph nodes (>1.5 cm) on pretreatment imaging. Chemotherapy received was GC (±paclitaxel) in 20 patients (69%), MVAC in 4 patients (14%), gemcitabine and carboplatin in 3 patients (10%), single-agent gemcitabine in 1 patient, and etoposide and cisplatin in 1 patient for presumed small cell carcinoma (later reclassified as urothelial carcinoma on the restaging TUR and RC specimen). The median number of cycles of chemotherapy received was 3 (range, 1 cycle-8 cycles).

Table 1. Preoperative Characteristics of Patients Receiving Neoadjuvant Chemotherapy
  1. MVAC indicates methotrexate, vinblastine, doxorubicin, and cisplatin.

Median age, y (interquartile range)63 (56-73)
Male gender (%)25 (86)
Neoadjuvant chemotherapy regimen (%)
 Gemcitabine/cisplatin (+paclitaxel in 1)20 (69)
 MVAC4 (14)
 Gemcitabine/carboplatin3 (10)
 Gemcitabine alone1 (3)
 Etoposide/cisplatin1 (3)
Clinical T classification (%)
 T223 (79)
 T35 (17)
 T4a1 (3)
Radiographic lymph nodes >1.5 cm (%)4 (14)

Pathologic outcomes are summarized in Table 2 according to NC status. Overall, only 2 patients (7%; 95% confidence interval [95% CI], 0 patient-17 patients) in the NC cohort achieved a pCR, both of whom received GC (±paclitaxel) compared with 8 (9%; 95% CI, 3 patients -15 patients) who were pT0 in the immediate RC cohort (P = .9). Both the NC and immediate RC cohorts were comparable with regard to pathologic stage, soft tissue surgical margins, number of lymph nodes removed, and lymph node density. Overall, 62% (95% CI, 43 patients -81 patients) versus 60% (95% CI, 50 patients -70 patients) of patients had nonorgan-confined cancer (pT3-4 or pN1-2) in the NC and immediate RC cohorts, respectively (P = .9). Of the 20 NC patients with clinical stage T2N0, 1 (5%; 95% CI, 0 patients -15 patients) achieved a pCR and 11 (55%; 95% CI, 31 patients -79 patients) had nonorgan-confined residual cancer. Comparing MVAC and GC with other chemotherapy regimens, there was a trend toward improved pathologic stage with MVAC and GC (nonorgan-confined cancer, 44% vs 66%) that did not reach statistical significance (P = .3).

Table 2. Perioperative Characteristics of Patients Undergoing Neoadjuvant Chemotherapy and Immediate Radical Cystectomy
 Neoadjuvant ChemotherapyImmediate Radical CystectomyP Value
  1. IQR indicates interquartile range; 95% CI, 95% confidence interval.

Median interval from diagnosis of muscle-invasive cancer to radical cystectomy, d (IQR)208 (144-327)48 (35-70)<.001
Pathologic T classification (%)  .9
 pT0-T19 (31)24 (27) 
 pT25 (17)17 (19) 
 pT3-T415 (52)47 (53) 
Median lymph nodes removed (IQR)20 (13-28)19 (11-27).7
Positive lymph nodes (%)11 (38)38 (43).8
Median lymph node density (IQR)10% (5-15)13% (6-43).4
Positive soft tissue margins (%)1 (3)8 (9).4
Pathologic complete response (%; 95% CI)2 (7%; 0-17)8 (9%; 2-15).9
Nonorgan-confined cancer (%; 95% CI)18 (62%; 43-81)53 (60%; 50-70).9

Over a median follow-up of 12 months (interquartile range [IQR], 6 months-16 months) in the NC cohort, 13 (45%) patients experienced cancer recurrence and 7 (24%) died of bladder cancer. The median recurrence-free and overall survivals were 10.5 months (95% CI, 7 months-14 months) and 20 months (95% CI, 19 months-21 months), respectively. In patients receiving immediate RC, the median survival was not reached and the 2-year survival rate was 59% (95% CI, 45%-63%) (Fig. 1).

Figure 1.

Kaplan-Meier estimates of overall survival for patients receiving neoadjuvant chemotherapy before radical cystectomy and those receiving immediate radical cystectomy.

A delay in RC has been previously reported to be associated with advanced pathologic stage and diminished postcystectomy survival.15, 16 The median interval from the time of the diagnosis of muscle-invasive bladder cancer to RC was 208 days (IQR, 149 days-327 days) and 48 days (IQR, 35 days-70 days) for the NC and immediate RC cohorts, respectively. Given that our patients are frequently referred from locations remote to Cleveland Clinic, NC is often administered under the care of patients' local oncologists. Subdividing patients into those that received NC at our institution (n = 6) versus other centers (n = 23), a significant delay in the interval from diagnosis of muscle-invasive bladder cancer to RC was observed for the latter group (155 vs 222 days; P = .046), and these patients also tended to have higher pathologic stage (nonorgan-confined cancer, 33% vs 56%; P = .4). A significant association with nonorgan-confined cancer was not observed for the interval from diagnosis to RC (P = .6), although the statistical power was small and 27 patients had a delay of >120 days. When comparing standard with nonstandard regimens, there was a significant delay in performing RC (median of 182 days vs 332 days; P = .016).

DISCUSSION

NC before RC has been increasingly used at our institution for the treatment of muscle-invasive bladder cancer based on level I evidence of a survival benefit. This improvement in survival was mainly observed among patients who achieved a pCR to NC.1 In our recent experience with NC, we failed to observe a substantial benefit with NC; few patients achieved a pCR and the rate of nonorgan-confined cancer was similar to that of patients undergoing RC without NC. Most patients experienced rapid disease progression and death from bladder cancer despite NC. Selection bias may explain these poor outcomes, although the clinical stage of our patients at diagnosis was similar to those in the SWOG 8710 study. Alternatively, these poor outcomes may be related to the use of non-MVAC NC regimens and/or substantial delays in performing RC. These factors may negate the small absolute survival benefit of NC. As such, MVAC should be considered the standard NC regimen in the absence of supporting data. Likewise, immediate RC (±adjuvant chemotherapy based on the pathology of the RC specimen) may be a reasonable approach if NC will lead to excessive delays in performing RC.

The most compelling evidence in support of NC is from the SWOG 8710 study, in which 3 cycles of MVAC conferred a 14% absolute improvement in overall survival at 5 years.1 As observed in our study, GC is now frequently substituted for MVAC in the neoadjuvant setting based on data demonstrating similar survival and an improved toxicity profile in patients with advanced bladder cancer.12 Extrapolation of these results to the neoadjuvant setting is potentially flawed, because these cohorts are substantially different with regard to performance status and potentially tumor biology.

In the absence of prospective, randomized data, we must rely on retrospective surgical series comparing pCR rates and survival among patients receiving MVAC and GC to determine whether benefits in the neoadjuvant setting are similar. A pCR appears to be a reasonable substitute endpoint with which to assess the efficacy of various NC regimens because the survival benefit of NC is largely restricted to these patients. In the SWOG 8710 study, the pCR rate was 38% (95% CI, 29%-47%) with MVAC, which was similar to the 32% (95% CI, 26%-39%) rate reported in a Medical Research Council NC trial of cisplatin, methotrexate, and vinblastine.5 In our study of patients largely treated with GC, the pCR rate was 7%, which is substantially lower than that reported in these trials and closer to the pT0 rates reported in RC series without NC.1, 5, 7-10

One possible explanation for these poor results is the use of non-MVAC regimens. There is sufficient evidence in the literature that NC regimens using single-agent cisplatin or carboplatin-based combination chemotherapy regimens produce inferior outcomes compared with MVAC. A recent meta-analysis reported a trend toward decreased survival among patients receiving NC with single-agent cisplatin compared with immediate RC.3 An NC trial of 3 cycles of the combination of paclitaxel, gemcitabine, and carboplatin reported a clinical CR of 46%, although only 40% of these patients were confirmed to have a pCR at RC.17 This finding translates into an overall pCR of 15%, which is well below the pCR rate reported in the SWOG 8710 study. A 32% pCR was reported among 22 patients who were evaluable for response in a similar NC trial evaluating 3 cycles of this regimen, although the pCR rate was only 22% by intent-to-treat analysis.18 To our knowledge to date, direct comparisons between MVAC and GC in a randomized trial have not been conducted in localized bladder cancer. A retrospective series of 42 patients receiving neoadjuvant GC before RC at Memorial Sloan-Kettering Cancer Center reported a 26% (95% CI, 14%-42%) rate of pCR, which was similar to a historical cohort of patients receiving neoadjuvant MVAC on protocol before RC or partial cystectomy.19 However, limitations of this study were that patients who received NC at an outside institution were not included and that the number of patients who underwent immediate RC during the same time period was not reported, suggesting a potential selection bias.

A review of phase 2 of 3 trials of MVAC and GC in patients with advanced bladder cancer demonstrated similar overall and complete clinical response rates in the majority, but the 3 highest clinical CR rates (23%-50%) have been observed in patients receiving MVAC (Table 3).12, 20-29 This finding suggests potentially greater activity for MVAC in achieving a complete CR and potential superiority over GC as a NC regimen. However, differences in overall and complete responses in phase 2 trials may simply be due to case-mix and methods of evaluation. Indeed, the clinical CR in a phase 3 trial comparing MVAC and GC in patients with advanced bladder cancer was similar and overall survival was not found to be significantly different. However, these results were achieved despite MVAC patients receiving less therapy than GC patients in terms of number of cycles of chemotherapy (median, 4 cycles vs 6 cycles) and dose reductions (63% vs 37%) due to toxicity.

Table 3. Overall and Complete Clinical Response Rates to Induction Chemotherapy for Patients with Advanced Urothelial Carcinoma Receiving GC and Those Receiving MVAC in Phase 2 and 3 Clinical Trials.
 Overall Response RateComplete Response Rate
GCMVACGCMVAC
  1. GC indicates gemcitabine and cisplatin; MVAC, methotrexate, vinblastine, doxorubicin, and cisplatin; —, not available.

von der Maase 20001249%46%12%12%
Bellmunt 20072346%
Stadler 20022544%20%
Kaufman 20002441%22%
Moore 19992657%19%
von der Maase 19992942%18%
Bajorin 19992764%23%
Loehrer 19922238%13%
Logothetis 19902165%35%
Sternberg 19882071%50%
Herr 20012816%

In the neoadjuvant setting, in which the patients are healthier and the performance status is better, one would anticipate that proportionately more patients would receive all planned cycles of chemotherapy without substantial dose reductions. For example, before RC, most patients (82%) have Karnofsky performance status scores of >90%,30 whereas in the metastatic setting, >40% of patients have a Karnofsky performance status scores of <80%.27 In NC trials with MVAC, 80% completed the prescribed regimen, and treatment-related mortality due to chemotherapy is reported in 0% to 1% of patients.1, 5 Thus, the aforementioned differences in clinical CR rates between GC and MVAC in the phase 2 trials may indeed become significant in a neoadjuvant setting. MVAC may be the more effective regimen if it can be tolerated. Ideally, the use of GC as a standard NC regimen for muscle-invasive bladder cancer should be based on evidence from randomized trials demonstrating noninferiority or equivalence with MVAC. However, a noninferiority trial with 90% power to rule out a difference in overall survival >10% at 5 years associated with neoadjuvant GC (assuming a 5-year overall survival rate of 58% with neoadjuvant MVAC) would require >5000 patients. A trial of this size is unlikely to be successful when one considers the historical accrual rates to randomized trials of localized muscle-invasive bladder cancer worldwide; 317 patients were accrued to the SWOG 8710 study over 11 years through the intergroup mechanism in the US.

Another possible explanation for the poor results observed with NC in the current study is selection bias; it is possible that only the more aggressive cases were selected for NC in our cohort. However, our cohort compares favorably with patients in the SWOG 8710 study and patients undergoing immediate RC at our institution; 79% of NC patients in the current study had clinical stage T2 disease compared with 40% in the SWOG 8710 study, and the median age in both studies was 63 years. We included 4 patients in our analysis with clinically positive lymph nodes and patients with this feature were excluded from the SWOG 8710 study. However, if we exclude these patients, the pCR rate on the remaining 25 patients is 4%. A higher rate of lymph node-positive disease was observed in the current series, which may be due to the referral of patients with less complicated, lower stage bladder cancer for laparoscopic RC.

In our NC patients, we observed considerable delays in performing RC, which may have contributed to the disappointing outcomes. Several authors have recognized that a delay in RC results in worse pathologic stage and diminished survival.15, 16 NC undoubtedly delays definitive therapy, but current evidence suggests any adverse effect of this delay is offset by the benefit of neoadjuvant MVAC. It must be emphasized that the median interval to RC in the SWOG 8710 study for patients receiving NC was 115 days, which was a delay in RC of 98 days on average compared with patients in the control arm (median interval to RC of 17 days). In our NC cohort, the median time to RC from the time of diagnosis of muscle-invasive cancer was 208 days. This delay is substantially longer than that reported in the NC arm of the SWOG 8710 study and 160 days longer than patients at our institution undergoing immediate RC.1 Although definitive conclusions cannot be made from this study, it is plausible that an excessive delay (≥120 days) in performing RC may negate the survival benefit of NC. Thus, if NC cannot be administered in a timely manner, it may be prudent to proceed directly with RC and reserve perioperative chemotherapy for the adjuvant setting. The interval to RC observed in the current series may more closely represent the typical time frame for a patient receiving NC and RC at a tertiary referral center compared with that reported in patients treated on-protocol in a phase 3 randomized trial. These practical issues should be considered when counseling patients on the benefits and risks of NC.

Difficulties in executing a coordinated treatment plan between urologists and medical oncologists at remote locations may be the cause of the delay observed in the current study. In the US, muscle-invasive bladder cancer is often diagnosed in the community setting, but RC is increasingly being performed at high-volume tertiary referral hospitals; perioperative chemotherapy is frequently administered by local medical oncologists. Few patients in the current study received NC at our institution, and these patients received NC and proceeded to RC in a timely manner (median of 155 days from the time of diagnosis to RC) and the median delay in RC compared with patients undergoing immediate RC (107 days) was similar to that of patients in the SWOG 8710 study (98 days). In contrast, the median time from diagnosis to RC for those patients who received chemotherapy from outside oncologists was 222 days. Although the numbers are small and preclude definitive conclusions, the 6 patients receiving NC at our institution demonstrated a trend toward a lower pathologic stage.

The current study is limited by its retrospective nature, small patient numbers, and lack of standardization regarding the administration of NC and the regimens used. Nevertheless, the 95% CIs of the pCR rate and median survival estimates do not approach those reported in the SWOG 8710 study.

We believe the combination of NC and RC is the optimal approach to the treatment of patients with muscle-invasive bladder cancer given the level I evidence demonstrating a reduction in all-cause and cancer-specific mortality, and the lack of similar evidence in support of adjuvant chemotherapy. There is also evidence that patients are more likely to receive both RC and the planned number of cycles of chemotherapy if it is administered in the neoadjuvant setting due to the frequent complications and prolonged convalescence associated with RC.4, 6 However, the results of the current study call into question the benefit of NC if non-MVAC regimens are administered or if NC leads to excessive delays in performing RC. These deviations from the optimal approach (3 cycles of neoadjuvant MVAC and timely RC) may negate the benefit of NC. We believe these issues are important for physicians and patients to be aware of when considering the use of non-MVAC neoadjuvant chemotherapy regimens or when substantial delays (≥120 days) in performing RC are anticipated.

Conclusions

Few RC patients in our recent experience achieved a pCR with NC and most experienced rapid disease progression. These poor outcomes may be related to the use of non-MVAC-based regimens or excessive delays in performing RC. In the absence of supportive data for GC in the neoadjuvant setting, MVAC remains the preferred regimen. Excessive delays in performing RC may negate the benefit of NC.

Conflict of Interest Disclosures

The authors made no disclosures.

Ancillary