Induction chemotherapy for unresectable urothelial carcinoma of the bladder


Urs E. Studer, Department of Urology, University Hospital Bern, 3010 Bern, Switzerland. e-mail:


Study Type – Therapy (case series) Level of Evidence 4

What’s known on the subject? and What does the study add?

In resectable muscle-invasive bladder cancer neoadjuvant chemotherapy followed by radical cystectomy confers to a significant 5% overall survival benefit. Less is known about induction chemotherapy followed by radical cystectomy in initially unresectable patients. Our retrospective analysis of a selected patient cohort suggests that patients with initially unresectable bladder cancer may benefit from this combined treatment approach.


• To analyse the outcome in selected patients with initially unresectable or minimally metastatic muscle-invasive urothelial bladder cancer who underwent induction chemotherapy (IC) followed by radical cystectomy (RC).


• Thirty patients with initially unresectable, locally advanced or minimally metastatic bladder cancer underwent platinum-based IC followed by RC with curative intent at our institution from 2000 to 2007.

• They received a median of four cycles (range 2–6 cycles) of cisplatin and gemcitabine (n= 19), carboplatin and gemcitabine (n= 9) or other platinum combinations (n= 2).

• We retrospectively analysed all 30 patients for complete pathological remission (pT0), disease free survival (DFS) and overall survival (OS). Chi-square tests, Kaplan–Meier analyses, and Cox univariate modelling were used.


• Before IC, 30 patients were deemed unresectable because of locally advanced tumour classification (cT4, 18/30) and/or clinically suspected lymph node (LN) metastasis (21/30) or suspected distant metastasis (3/30).

• At re-staging after IC there was a complete regression of all enlarged LN in 16/21 patients, a partial LN response in one patient or stable LN size in the remaining four patients.

• After RC, 9/30 (30%) of patients had attained pT0.

• The median follow-up was 28 months (range 4–97 months). The 5-year DFS and OS rates were 42% and 46%, respectively, for all patients.

• In the pT0 patients, the DFS (83%) and OS (71%) rates were significantly higher than in non-pT0 patients.


• Patients undergoing IC followed by RC showed encouraging response and survival rates, suggesting that selected patients with initially unresectable bladder cancer benefit from this combined regimen.


disease-free survival


gemcitabine and cisplatin


carboplatin AUC5 and gemcitabine


induction chemotherapy


lymph node


methotrexate, vinblastin, doxorubicin and cisplatin


neoadjuvant chemotherapy


overall survival


radical cystectomy


transurethral resection of the bladder.


In resectable muscle-invasive bladder cancer, platinum-based neoadjuvant chemotherapy (NC) followed by radical cystectomy (RC) confers a 5% overall survival (OS) benefit after 5 years compared with surgery alone, as shown in a meta-analysis of 11 phase III randomized trials [1]. The Southwest Oncology Group phase III randomized trial 8710 showed that disease-specific survival was superior in patients who received three cycles of NC with methotrexate, vinblastin, doxorubicin and cisplatin (MVAC) before RC (hazard ratio (HR) 1.66, P= 0.002, 95% CI, 1.22–2.45) [2]. They also found a trend toward increased OS in patients receiving MVAC but the trend was not statistically significant. In another study, four cycles of gemcitabine and cisplatin (GC) led to a 26% rate of complete pathological down-staging (pT0) [3], although a different study reported a pT0 rate after GC of only 7%[4]. The absolute OS benefit achieved by NC of 5% after 5 years is significant but small. The question arises as to which subgroup of patients may benefit most from NC. At first glance, this would appear to be patients who achieve pT0 [2]. These patients, however, may initially have had smaller tumours and/or their tumour was completely removed during the initial transurethral resection of the bladder (TUR-B) before chemotherapy. It has been reported that patients with advanced disease (cT3 or cT4a) benefit most from NC (65 months vs 24 months), although the difference does not attain statistical significance [2].

Given the effectivity of NC in patients with resectable, locally advanced bladder cancer, we hypothesized that patients with initially unresectable bladder cancer might benefit from induction chemotherapy (IC) followed by RC. We therefore reviewed the outcomes of 30 patients at our institution who initially did not qualify for RC because of unresectable locally advanced disease and/or clinically suspected LN metastasis or minimally metastatic disease and who were therefore given IC before RC.


From July 2000 to July 2007, 295 consecutive patients with muscle-invasive urothelial cancer of the bladder identified at TUR-B who had undergone bilateral pelvic lymphadenectomy and open RC at the Department of Urology, Bern University Hospital were retrospectively analysed. Thirty of these patients (10%) (median age at diagnosis 64 years; range 40–77 years) had received IC before RC having been deemed unresectable because of their locally advanced tumour classification (cT4, 18/30) and/or either biopsy proven or clinically suspected regional pelvic metastasis (18/30) and/or juxtaregional lymph node (LN) metastasis (12/30). Nine patients had both regional pelvic and juxtaregional LN metastasis, so a total of 21/30 patients had suspected LN metastasis, nine of which were biopsy proven. Of the remaining nine cN0 patients, six had a tumour classification cT4 and three patients had clinically suspected distant metastasis (confirmed by biopsy in one patient). Indications for IC are summarized in Table 1 and pretherapeutic patient characteristics are summarized in Table 2. All patients were staged by CT of the chest (or chest X-ray), abdomen and pelvis and nuclear bone scan. Radiological evidence of LN metastasis was considered in the presence of pathologically enlarged LN >1 cm in diameter. Enlarged pelvic LN were classified as regional pelvic LN metastases and enlarged inguinal or para-aortal LN were classified as juxtaregional LN metastases. After IC and RC the tumours were classified according to the 2002 Union Internationale Contre le Cancer TNM Classification of Malignant Tumours [5]. According to this classification, tumour invasion of the muscularis propria corresponds to pT2, invasion of the perivesical tissue corresponds to pT3, and invasion to the prostate, uterus, vagina and pelvic or abdominal wall corresponds to pT4 disease.

Table 1.  Indications for induction chemotherapy
Total patientsn (%)
  1. Total n = 30.

cT4, cN0, cM0 6 (20)
≤cT4, cN+, cM021 (70)
≤cT4, cN0, cM+ 3 (10)
Table 2.  Pretherapeutic patient characteristics
Total patientsn (%)
  1. Total n = 30. LN, lymph node. *Nine patients had both enlarged pelvic LN and juxtaregional LN, so in total, 21 patients had enlarged LN. In nine patients LN metastases were confirmed by biopsy. †The juxtaregional LN consisted of enlarged inguinal LN in one patient and enlarged para-aortal LN in 11 patients. ‡Of the patients with distant metastasis, one patient had bone metastasis and one patient had suspected lung and liver metastasis (none proven by biopsy). The remaining patient had an abdominal wall metastasis which was histologically confirmed.

 ≤ 64 years15 (50)
 >64 years15 (50)
 Female4 (13)
 Male26 (87)
Clinical tumour classification 
 cT24 (13)
 cT38 (28)
 cT418 (60)
Clinical nodal classification* 
 No enlarged LN9 (30)
 Single pelvic LN < 2 cm5 (17)
 Single pelvic LN 2–5 cm or multiple ≤ 5 cm13 (43)
 Enlarged juxtaregional LN12 (40)
Clinical metastasis classification 
 cM027 (90)
 cM13 (10)

In all cases the recommendation for IC was made after a multidisciplinary discussion. The standard treatment consisted of four cycles of cisplatin 70 mg/m2 on day 1 and gemcitabine 1000 mg/m2 on days 1, 8 and 15 every 4 weeks. In patients with cisplatin contraindications (creatinine clearance <60 mL/min, poor performance status, clinically relevant hearing loss or tinnitus, neuropathy), carboplatin AUC5 and gemcitabine (GCb) were used. After completion of IC, a re-staging CT and a bimanual physical examination along with a TUR-B were performed to assess the post-chemotherapy status. The decision for RC was made after a second multidisciplinary discussion. The choice of urinary diversion was based on the extent of the tumour, paracollicular biopsies, renal function, performance status and the patient’s preference. All patients underwent extended bilateral pelvic and, if necessary, retroperitoneal LN dissection along with the RC as previously described [6]. The cystectomy specimens were examined by an experienced genitourinary pathologist to assess the pathological stage and pT0 status. Clinical information and follow-up were entered into a prospective database after informed consent from all patients in accordance with the standards of the local ethics committee. Follow-up examinations were performed on a regular basis every 3 months for the first year, biannually for years 2 to 5, and yearly thereafter.

Endpoints were the achievement of pT0, disease-free survival (DFS) and OS. For DFS, local relapse, LN relapse and distant metastasis were considered as events. Patients were dichotomized according to age (≤64 vs >64 years), clinical tumour classification (≤cT3 vs >cT3), presence of clinically suspected pelvic LN metastasis (not present vs present), presence of clinically suspected juxtaregional LN metastasis (not present vs present), NC regimen (GC vs GCb) and pathological tumour classification (pT0 vs >pT0). The Pearson chi-squared test was used to analyse the association between categorical variables. DFS and OS were analysed by Kaplan–Meier analysis. Time to disease recurrence or death was calculated from the first day of IC to the last follow-up or disease recurrence for DFS or until death for OS. Clinical or pathological variables were compared with DFS and OS using univariate Cox regression analysis. Statistical significance was considered on a two-sided significance level (α) of 5%. Statistical analyses were performed using SPSS 16.0 software (SPSS Inc., Chicago, IL, USA).


The IC regimen consisted of GC in 19 (64%) patients, of GCb in nine (30%) patients, of MVAC in one (3%) patient and of gemcitabine, carboplatin and paclitaxel in one (3%) patient (Table 3). The most frequent reasons for using carboplatin instead of cisplatin were impaired renal function and/or poor performance status (performance status ≥2). The median number of chemotherapy cycles was four (range, 2–6). IC was administered without major toxicity in all patients. The median time from start of IC to RC was 175 days (range 100–365 days).

Table 3.  Patient characteristics after induction chemotherapy and cystectomy
Total patientsn (%)
  1. n = 30. IC, induction chemotherapy; LN, lymph node; MVAC, methotrexate, vinblastin, doxorubicin and cisplatin. *Gemcitabine/Carboplatin/Paclitaxel. †Median (range).

IC regimen 
 Gemcitabine/Cisplatin19 (64)
 Gemcitabine/Carboplatin9 (30)
 MVAC1 (3)
 Other*1 (3)
Median IC cycles (range)4 (2–6)
Urinary diversion 
 Ileal bladder substitute21 (70)
 Ileal conduit9 (30)
Median number of LN resected27 (11–49)
Median number of positive LN1 (0–17)
Pathological tumour classification
 pT09 (30)
 pT12 (6)
 pT28 (27)
 pT38 (27)
 pT43 (10)
Carcinoma in situ 
 pTis12 (40)
Pathological nodal classification 
 No LN metastases (pN0)15 (50)
 Single pelvic LN < 2 cm (pN1)8 (27)
 Single pelvic LN 2–5 cm or multiple ≤ 5 cm (pN2)7 (23)

At re-staging after IC there was a complete response of all radiologically enlarged LN in 16/21 patients, a partial LN response in one patient and LN showed stable size after IC in the remaining four patients.

After TUR-B, IC and RC, nine of the 30 patients (30%) were classified as pT0 (six of which received GC and three GCb) (Table 3). Five of these nine patients had a clinical tumour classification of cT4, six had clinically suspected pelvic LN metastases, two had suspected juxtaregional LN metastasis, none had suspected distant metastatic disease. No significant correlation was found between pT0 and the variables age (P= 0.4), clinical tumour classification (P= 0.7), clinical nodal classification (P= 0.7), presence of juxtaregional LN metastasis (P= 0.2) or IC regimen (P= 1.0). Two of the 30 patients (6%) were classified as pT1, both of whom received GCb. One of these two patients had a clinical tumour classification of cT4 and both patients had clinically suspected pelvic and juxtaregional LN metastasis. Distant metastasis was not suspected in these patients. All pT0 and pT1 patients were without evidence of LN metastasis (pN0). All 30 patients were resected with negative margins (R0 resection). One patient underwent adjuvant brachytherapy after intraoperative applicator placement because of a persistent enlarged LN after IC but final pathology showed no indication for persistent disease. Eight patients (27%) with persistent disease received additional adjuvant chemotherapy. Of these patients, four have died of disease and four remain alive.

After a median follow-up of 28 months (range 4–97 months), 16 of the 30 patients (53%) had experienced a cancer recurrence. Of these 16 patients, five developed distant metastasis only, four distant metastasis and local recurrence, four distant metastasis and a pelvic LN recurrence, two patients had a local recurrence only, and one patient had a local recurrence and a pelvic LN recurrence. Thirteen of the 15 observed deaths were the result of bladder cancer and two were from other causes. The 5-year DFS and OS rates were 42% and 46% for all patients, and 83% and 71% for patients with pT0, respectively (Fig. 1). In the Cox analysis the presence of pT0 was significantly associated with both increased DFS [hazard ratio (HR) 0.08, P= 0.015, 95% confidence interval (CI) 0.01–0.61] and increased OS (HR 0.21, P= 0.04, 95% CI 0.05–0.95). Age ≤64 years was significantly associated with increased OS (HR 0.27, P= 0.03, 95% CI 0.09–0.86) but not with increased DFS (HR 0.43, P= 0.1, 95% CI 0.16–1.19).

Figure 1.

Actuarial analysis of disease-free survival (A) and overall survival (C) stratified by the presence of a complete pathological down-staging of the primary tumour (pT0) in the resected bladder (B, D).


In patients with resectable, muscle-invasive bladder cancer, the Advanced Bladder Cancer meta-analysis (n > 3000) reports a 5-year survival of 45% for patients with surgery alone and 50% for patients with NC followed by surgery [1]. This benefit from NC was attributed to improved resectability and decreased recurrence rates. After NC, increased OS has been observed in patients who achieved a pT0 status in their resected bladder [2]. It cannot be ruled out, however, that a pT0 status after NC is also, or at least in part, the result of a previously performed TUR-B. Despite the known survival benefit of NC, an analysis by the National Bladder Cancer Database (n= 11 328) showed that only about 0.7% of patients with muscle-invasive bladder cancer received NC during the past decade [7]. Although the evidence in the second Advanced Bladder Cancer meta-analysis was limited by sample size [8] and the European Organization for Research and Treatment of Cancer 30994 trial results are still pending, many centres today, including our own, prefer to give adjuvant chemotherapy to selected bladder cancer patients at high risk (pT3-4 and/or pN+). For patients with initially unresectable bladder cancer, IC may be a reasonable approach that is also pursued with success in patients with advanced head-and-neck cancer.

In the year 2000, our institution began to use IC in selected patients with initially unresectable disease in an attempt to achieve resectability and proceed to RC. Because of the documented effectiveness and tolerability, the preferred combination was cisplatin (or carboplatin) and gemcitabine. The 5-year DFS and OS rates of our consecutive cohort of 30 patients treated by a median of four IC cycles (range 2–6 cycles) followed by RC were 42% and 46%, respectively; IC was administered without major toxicity. The 30% pT0 rate observed in our patients compares favourably with the 26% and 7% rates previously reported for GC, and with the 37% pT0 rate reported for MVAC [2–4]. Cisplatin (or carboplatin) and gemcitabine therefore remains our preferred combination for IC outside of a clinical trial.

This study has certain limitations. Interpretation of the results of our study is limited by sample size and patient selection. The inclusion of only platinum-eligible patients certainly contributed to the good outcome, because these patients have a better prognosis regardless of whether they receive chemotherapy or not, as we have shown previously [9]. Other patients may have been excluded from the study because of tumour progression during IC. As a tertiary referral centre for treatment of bladder cancer, we mostly see patients who have successfully finished IC and are eligible for RC so the reported long-term survival approaching 50% is a result not only of the applied treatment but also of patient selection. Nevertheless, the results do suggest that combined modality treatment with IC and RC can be successful in carefully selected patients with initially unresectable bladder cancer and poor prognosis. Certainly, this strategy warrants further investigation.

Also, our present results indirectly support the view that perioperative chemotherapy should be confined to patients with locally advanced (≥cT3 and/or N+) resectable bladder cancer. Despite the promising results attained with combined modality treatment described here, many patients did not achieve pT0 and died from disease progression. For the future, the development of new, more effective drugs is urgently needed.

Our results show that in selected patients with initially unresectable, locally advanced or minimally metastatic bladder cancer, IC followed by RC is both feasible and effective.


We thank M. Simcock, MSc, Swiss Group for Clinical Cancer Research Coordinating Center (SAKK CC), for performing the statistical calculations.


Oliver Gautschi has participated in advisory boards for and received honoraria from Eli Lilly Company.