Perioperative chemotherapy for muscle-invasive bladder cancer: A population-based outcomes study

Authors

  • Christopher M. Booth MD,

    Corresponding author
    1. Division of Cancer Care and Epidemiology, Queen's University Cancer Research Institute
    2. Department of Oncology, Queen's University, Kingston, Ontario, Canada
    • Corresponding author: Dr. Christopher Booth, Division of Cancer Care and Epidemiology, Queen's University Cancer Research Institute, 10 Stuart Street, Kingston, ON, K7L 3N6, Canada; Fax: (613) 533-6794; boothc@kgh.kari.net

    Search for more papers by this author
  • D. Robert Siemens MD,

    1. Department of Oncology, Queen's University, Kingston, Ontario, Canada
    2. Department of Urology, Queen's University, Kingston, Ontario, Canada
    Search for more papers by this author
  • Gavin Li MD,

    1. Division of Cancer Care and Epidemiology, Queen's University Cancer Research Institute
    Search for more papers by this author
  • Yingwei Peng PhD,

    1. Division of Cancer Care and Epidemiology, Queen's University Cancer Research Institute
    2. Department of Public Health Sciences, Queen's University, Kingston, Ontario, Canada
    Search for more papers by this author
  • Ian F. Tannock MD, PhD,

    1. Princess Margaret Hospital, Toronto, Ontario, Canada
    Search for more papers by this author
  • Weidong Kong MD,

    1. Division of Cancer Care and Epidemiology, Queen's University Cancer Research Institute
    Search for more papers by this author
  • David M. Berman MD, PhD,

    1. Department of Pathology and Molecular Medicine, Queen's University, Kingston, Ontario, Canada
    Search for more papers by this author
  • William J. Mackillop MB, ChB

    1. Division of Cancer Care and Epidemiology, Queen's University Cancer Research Institute
    2. Department of Oncology, Queen's University, Kingston, Ontario, Canada
    3. Department of Public Health Sciences, Queen's University, Kingston, Ontario, Canada
    Search for more papers by this author

Abstract

BACKGROUND

Practice guidelines recommend neoadjuvant chemotherapy (NACT) for bladder cancer. However, the evidence in support of adjuvant chemotherapy (ACT) is less robust. Here we describe whether the evidence of efficacy for NACT/ACT was sufficient to change clinical practice and whether the efficacy demonstrated in clinical trials was translated into effectiveness in the general population.

METHODS

Electronic records of treatment were linked to the population-based Ontario Cancer Registry to identify all patients with bladder cancer treated with cystectomy in Ontario 1994-2008. Utilization of NACT/ACT was compared across 1994-1998, 1999-2003, and 2004-2008. Logistic regression was used to analyze factors associated with NACT/ACT. Cox model and propensity score analyses were used to explore the association between ACT and survival.

RESULTS

Two thousand forty-four patients underwent cystectomy for muscle-invasive bladder cancer (MIBC). Use of NACT remained stable (mean, 4%), whereas utilization of ACT increased over time (16%, 18%, 22%; P = .001). Advanced stage (T3/T4; OR, 1.83; 95% CI, 1.38-2.46) and node-positive disease (OR, 8.10; 95% CI, 6.20-10.7) were associated with greater utilization of ACT. Five-year overall survival (OS) and cancer-specific survival (CSS) for all patients was 29% (95% CI, 28%-31%) and 33% (95% CI, 31%-35%), respectively. Utilization of ACT was associated with improved OS (HR, 0.71; 95% CI, 0.62-0.81) and CSS (HR, 0.73; 95% CI, 0.64-0.84). Results were consistent in propensity score analyses.

CONCLUSIONS

NACT remains substantially underutilized in routine clinical practice. Our results suggest that perioperative chemotherapy is associated with a substantial survival benefit in the general population. Patients who are planning to undergo cystectomy for bladder cancer should be reviewed by a multidisciplinary team. Cancer 2014;120:1630–1638. © 2013 American Cancer Society.

INTRODUCTION

Muscle-invasive bladder cancer (MIBC) may be locally controlled by cystectomy or radiotherapy (RT), but approximately 50% of patients ultimately die of distant metastases.

Randomized controlled trials and subsequent meta-analyses have demonstrated the efficacy of neoadjuvant (NACT)[1-3] and adjuvant chemotherapy (ACT)[4, 5] for bladder cancer. However, the evidence in support of ACT is less robust than that for NACT, and therefore its use is more controversial.[6] Despite this controversy, the level of evidence showing efficacy has proved sufficient to prevent accrual to 3 contemporary RCTs of ACT, which include an observation-only control arm.[7-9]

It is well known that efficacy may not translate into effectiveness in the general population.[10] This may relate to the finding that patients, practitioners, and health delivery systems in routine practice are very different from those in clinical trials.[11] Furthermore, when the evidence showing efficacy is controversial, its interpretation and subsequent adoption by practitioners may be suboptimal.

Population-based outcome studies may be useful in evaluating the external validity of the results of clinical trials.[12] We undertook a population-based study to describe adoption and outcomes of neoadjuvant and adjuvant chemotherapy for bladder cancer in the general population of Ontario. The study objectives were: 1) to explore whether the evidence of efficacy for NACT and ACT was sufficient to change clinical practice; and 2) to evaluate whether the efficacy of NACT and ACT demonstrated in the RCTs was translated into effectiveness in the general population.

MATERIALS AND METHODS

Study Design and Population

This is a population-based, retrospective cohort study to describe management and outcome of MIBC in the Canadian province of Ontario. Ontario has a population of approximately 13.5 million people and a single-payer universal health insurance program. All incident cases of bladder cancer in Ontario with transitional cell, adenocarcinoma, and squamous cell histology who underwent cystectomy or radical RT in 1994-2008 were identified using the Ontario Cancer Registry and linked treatment records. Stage of disease was not routinely available in the existing data sources; for this reason we obtained surgical pathology reports for all cystectomy cases. Patients with muscle-invasive urothelial cancer as per the cystectomy pathology report were included. The primary study objective was to describe practice patterns and outcomes associated with perioperative chemotherapy for MIBC in the general population. The study was approved by the Research Ethics Board of Queen's University.

Data Sources

The Ontario Cancer Registry (OCR) is a passive, population-based cancer registry that captures diagnostic and demographic information on at least 98% of all incident cases of cancer diagnosed in the province of Ontario.[13] The OCR provides information regarding cancer site/histology, date of diagnosis, demographics, and vital status. Complete information about vital status was available to December 31, 2010, and cause of death was available to December 31, 2008. The OCR does not compile information about extent of disease or treatment. Indicators of the socioeconomic status (SES) of the community in which patients resided at time of diagnosis were linked to the OCR as described previously to create SES quintiles.[14] Q1 represents the communities where the poorest 20% of the Ontario population resides.

A variety of electronic administrative health databases are linked to the OCR. Records of hospitalization from the Canadian Institute for Health Information provided information about surgical interventions and hospital care; collection of these records is known to be consistent and complete throughout Ontario.[15] Provincial physician billing records and treatment records from Ontario's regional cancer centers were used to identify utilization of chemotherapy. Surgical pathology reports were obtained from the OCR and reviewed by a team of trained data abstractors.

Definitions of Comorbidity and Management

Comorbidity was classified using the modified Charlson Index based on all noncancer diagnoses recorded during any hospital admission within 5 years prior to surgery.[16] NACT was defined as any chemotherapy administered within 16 weeks prior to surgery. ACT was defined as any chemotherapy administered within 16 weeks after surgery.

Statistical Analysis

Comparisons of proportions between study groups were made using the chi-square test. Survival was determined from date of surgery using the Kaplan-Meier technique, and comparisons between groups were made using the log-rank test. Factors associated with treatment practices were evaluated by logistic regression. The association between patient-, disease-, and treatment-related factors with overall/cancer-specific survival was evaluated using the Cox proportional hazards regression model. The survival analyses related to ACT were restricted to patients who did not receive NACT or preoperative radiotherapy because information about pathologic stage for these patients would not be reliable. To control for confounding variables when exploring the association between ACT and survival, we also employed the propensity score technique in the Cox proportional hazards model. The propensity score is the probability that a patient receives ACT, and it was modeled with observed confounding variables using the logistic regression model. The propensity scores allowed us to create 5 propensity strata with balanced confounding variables between ACT patients and non-ACT patients. Survival of patients treated with ACT was compared with those without ACT within each stratum using a Cox proportional hazards model; a summary HR combining the results across quintiles was calculated on the basis of the stratified Cox's PH model.[17, 18] Results were considered statistically significant at P < .05. All analyses were performed using SAS version 9.1 (SAS Institute, Cary, NC).

RESULTS

Study Population

During 1994-2008, 3879 patients in Ontario underwent cystectomy for bladder cancer, of whom 2944 patients had muscle-invasive urothelial carcinoma (Supplemental eFigure; see online supporting information). Characteristics of the study population are shown in Table 1. Most patients (60%) were older than age 70, and 75% were male. Locally advanced (T3/T4) tumors were present in 71% of patients, and 26% of patients had positive lymph nodes. No lymph nodes were removed in 31% of patients. Only 4% of patients (129 of 2944) received NACT; 19% of patients (572 of 2944) were treated with ACT. Thirty patients (1%) were treated with NACT and ACT.

Table 1. Characteristics of Patients With Muscle-Invasive Bladder Cancer (MIBC) Treated With Cystectomy in Ontario 1994-2008
 All Patients (n = 2944), n (%)No Perioperative Chemotherapya (n = 2267), n (%)Perioperative Chemotherapy
NACT (n = 129), n (%)ACTb (n = 542), n (%)
  1. Note: Percentages may not add to 100% because of rounding.

  2. Abbreviations: N/A, not available; N/R, not reportable because of small cell size.

  3. a

    Among the 2944 muscle-invasive patients, 2273 did not receive perioperative chemotherapy. Six of these cases received preoperative radiotherapy. These 6 cases were removed from subsequent analyses because pathologic stage would not reflect true stage of disease.

  4. b

    Five hundred seventy-two patients were treated with ACT. Thirty of these were also treated with NACT. These 30 cases were removed from the ACT analyses because pathologic stage would not reflect true stage of disease.

  5. c

    Quintile 1 includes patients who reside in neighborhoods with the lowest socioeconomic status (SES). SES data are missing for 2 patients.

Patient related    
Age, y    
20-49102 (3%)48 (2%)7 (5%)47 (9%)
50-59348 (12%)196 (9%)32 (25%)119 (22%)
60-69736 (25%)527 (23%)38 (29%)171 (32%)
70-791172 (40%)948 (42%)45 (35%)175 (32%)
80+586 (20%)548 (24%)7 (5%)30 (6%)
Sex    
Male2211 (75%)1694 (75%)92 (72%)420 (78%)
Female733 (25%)573 (25%)36 (28%)122 (23%)
Socioeconomic status by quintilec    
1583 (20%)477 (21%)17 (13%)88 (16%)
2653 (22%)506 (22%)38 (30%)108 (20%)
3662 (23%)507 (22%)29 (22%)126 (23%)
4562 (19%)428 (19%)25 (19%)107 (20%)
5482 (16%)347 (15%)20 (16%)113 (21%)
Charlson comorbidity score    
02046 (70%)1521 (67%)103 (80%)417 (77%)
1-2745 (25%)605 (27%)23 (18%)116 (21%)
3+153 (5%)141 (6%)N/R9 (2%)
Geographic region of Ontario    
A1303 (44%)961 (42%)68 (53%)272 (50%)
B428 (15%)363 (16%)13 (10%)51 (9%)
C217 (7%)169 (7%)12 (9%)34 (6%)
D172 (6%)128 (6%)N/R39 (7%)
E39 (1%)25 (1%)N/R12 (2%)
F107 (4%)81 (4%)N/R24 (4%)
G102 (7%)169 (7%)7 (5%)26 (5%)
H474 (16%)369 (16%)21 (16%)84 (16%)
Disease related    
Pathologic T stage    
<T3865 (29%)713 (31%)N/A96 (18%)
T3-T42079 (71%)1554 (69%)N/A446 (82%)
Pathologic N stage    
N negative1354 (46%)1167 (51%)N/A116 (21%)
N positive858 (29%)462 (20%)N/A363 (67%)
NX732 (25%)638 (28%)N/A63 (12%)
LVI    
No704 (24%)593 (26%)N/A67 (12%)
Yes1515 (51%)1087 (48%)N/A372 (68%)
Unstated725 (25%)587 (26%)N/A103 (19%)

Trends in the Use of Chemotherapy

Although use of NACT remained relatively static over time—1994-1998, 5%; 1999-2003, 3%; 2004-2008, 6% (P = .004)—use of ACT increased over the study period from 16% to 18% to 22% (P = .001); see Figure 1. The chemotherapy regimen used was identifiable for 308 of 671 patients (46%) treated with NACT/ACT. Chemotherapy included cisplatin or carboplatin in 82% (253 of 308) and 14% (43 of 308) of patients, respectively.

Figure 1.

Proportion of muscle-invasive bladder cancer patients treated with surgery in Ontario 1994-2008 (n = 2944) treated with perioperative chemotherapy.

Factors associated with use of perioperative chemotherapy are shown in Table 2. In adjusted analyses, younger age was associated with greater utilization of NACT and ACT. Patients residing in more affluent communities were more likely to have ACT than those residing in lower SES communities. Pathologic characteristics were also strongly associated with use of ACT: T3/T4 tumors (OR, 1.83; 95% CI, 1.38-2.46), node-positive disease (OR, 8.10; 95% CI, 6.20-10.7), and presence of lymphovascular invasion (OR, 1.53; 95% CI, 1.11-2.15) were all associated with greater utilization of ACT. There was greater utilization of NACT (OR, 1.73; 95% CI, 1.18-2.55) and ACT (OR, 1.43; 95% CI, 1.12-1.83) among patients whose surgery took place at a hospital associated with a regional cancer center.

Table 2. Variables Associated With Perioperative Chemotherapy Among 2944 Patients With Muscle-Invasive Bladder Cancer (MIBC) Treated With Cystectomy in Ontario 1994-2008
CharacteristicNeoadjuvant ChemotherapyAdjuvant Chemotherapya
Proportion NACTMultivariate AnalysisProportion ACTMultivariate Analysis
OR (95% CI)P TrendOR (95% CI)P Trend
  1. Abbreviation: N/R, not reportable because of small cell size.

  2. a

    Patients with NACT and/or preoperative radiotherapy (n = 135) were removed from the analysis of ACT because T stage, N stage, and LVI variables do not reflect actual surgical pathologic stage.

  3. b

    Quintile 1 includes patients who reside in neighborhoods with the lowest socioeconomic status (SES). SES data are missing for 2 patients.

  4. c

    T stage, N stage, and LVI are not reported for NACT patients because these pathologic variables (which are derived from surgical pathology reports) would not reflect the stage of disease at the time of initiation of NACT. Length of stay was also not known at the time of starting NACT and was therefore not included in this analysis.

Patient related      
Study period  .009  .337
1994-1998 (n = 705)5%Ref 15%Ref 
1999-2003 (n = 964)3%0.61 (0.36-1.04) 18%1.11 (0.80-1.53) 
2004-2008 (n = 1275)6%1.26 (0.82-1.98) 22%1.26 (0.92-1.73) 
Sex  .324  .596
Male (n = 2211)4%Ref 20%Ref 
Female (n = 733)5%1.23 (0.81-1.82) 18%0.93 (0.71-1.21) 
Age, y  < .001  <.001
20-49 (n = 102)7%Ref 50%Ref 
50-59 (n = 348)9%1.52 (0.67-3.90) 38%0.47 (0.27-0.83) 
60-69 (n = 736)5%0.87 (0.39-2.20) 25%0.27 (0.16-0.46) 
70-79 (n = 1172)4%0.65 (0.30-1.63) 16%0.15 (0.09-0.26) 
80+ (n = 586)1%0.19 (0.06-0.57)  5%0.04 (0.02-0.08) 
Socioeconomic status, quintileb  .204  .033
1 (n = 583)3%0.81 (0.41-1.59) 16%0.57 (0.39-0.84) 
2 (n = 653)6%1.60 (0.90-2.90) 18%0.60 (0.41-0.86) 
3 (n = 662)4%1.24 (0.68-2.29) 20%0.74 (0.52-1.06) 
4 (n = 562)4%1.14 (0.62-2.19) 20%0.71 (0.49-1.01) 
5 (n = 482)4%Ref 25%Ref 
Charlson comorbidity score  .211  < .001
0 (n = 2046)5%Ref 22%Ref 
1-2 (n = 745)3%0.71 (0.44-1.12) 16%0.86 (0.65-1.12) 
3+ (n = 153)N/R0.50 (0.12-1.37) 6%0.19 (0.08-0.39) 
Length of hospital stay (unit = 1 day0.95 (0.93-0.97)< .001
Disease relatedc      
T stage     < .001
<T3 (n = 809)12%Ref 
T3-T4 (n = 2000)22%1.83 (1.38-2.46) 
N stage     < .001
N negative (n = 1283) 9%Ref 
N positive (n = 825)44%8.10 (6.20-10.7) 
NX (n = 701) 9%1.45 (1.01-2.07) 
LVI     .029
No (n = 660)10%Ref 
Yes (n = 1459)26%1.53 (1.11-2.15) 
Unstated (n = 690)15%1.23 (0.84-1.80) 
System related      
Region  .140  .012
A (n = 1303)5%Ref 22%Ref 
B (n = 428)3%0.60 (0.31-1.08) 12%0.52 (0.35-0.76) 
C (n = 217)6%0.81 (0.40-1.52) 17%0.60 (0.37-0.95) 
D (n = 172)3%0.39 (0.13-0.93) 23%0.76 (0.46-1.21) 
E (n = 39)N/R0.33 (0.02-1.61) 32%1.42 (0.59-3.26) 
F (n = 107)N/R0.23 (0.04-0.75) 23%0.91 (0.50-1.60) 
G (n = 202)3%0.57 (0.23-1.22) 13%0.61 (0.36-1.01) 
H (n = 474)4%0.65 (0.37-1.09) 19%0.70 (0.50-0.99) 
Cancer center surgical hospital  .005  .005
No (n = 1594)4%Ref 16%Ref 
Yes (n = 1350)5%1.73 (1.18-2.55) 23%1.43 (1.12-1.83) 

Outcomes

Among the full study population (n = 2944), 5-year OS and CSS were 29% (95% CI, 28%-31%) and 33% (95% CI, 31%-35%), respectively (Supplemental eFigure; see online supporting information). Among NACT patients 5-year OS was 25% (95% CI, 17%-34%) and CSS was 28% (95% CI, 18%-39%); corresponding figures for ACT patients were 29% (95% CI, 25%-33%) and 28% (95% CI, 24%-33%), respectively.

Factors associated with OS and CSS among the 2809 ACT-eligible patients (ie, those not treated with NACT or preoperative RT) are shown in Table 3. In adjusted analyses greater comorbidity and advanced age were associated with reduced OS and CSS. Higher stage was strongly associated with inferior OS and CSS: T3/4 tumors (HR, 1.69; 95% CI, 1.52-1.88; and HR, 1.84; 95% CI, 1.62-2.08, respectively), node-positive disease (HR, 1.77; 95% CI, 1.58-1.98; and HR, 1.82; 95% CI, 1.60-2.07, respectively) or NX (HR, 1.39; 95% CI, 1.23-1.58; and HR, 1.39; 95% CI, 1.23-1.58, respectively), and presence of lymphovascular invasion (HR, 1.80; 95% CI, 1.59-2.03; and HR, 2.05; 95% CI, 1.77-2.37, respectively).

Table 3. Variables Associated With Overall Survival (OS) and Cancer-Specific Survival (CSS) Among 2809a Patients With Muscle-Invasive Bladder Cancer (MIBC) Treated With Cystectomy in Ontario 1994-2008
CharacteristicOverall SurvivalCancer-Specific Survival
5-Year OSMultivariate Analysis5-Year CSSMultivariate Analysis
HR (95% CI)P TrendHR (95% CI)P Trend
  1. Abbreviation: Ref, reference.

  2. a

    Patients with NACT and/or preoperative radiotherapy (n = 135) have been removed from this analysis because T stage, N stage, and LVI variables do not reflect actual surgical pathologic stage.

  3. b

    Quintile 1 includes patients who reside in neighborhoods with the lowest socioeconomic status (SES). SES data are missing for 2 patients.

Patient related      
Sex  .532  .778
Male (n = 2114)29%Ref 34%Ref 
Female (n = 695)29%0.97 (0.88-1.07) 33%0.98 (0.88-1.10) 
Age, y  <.001  .102
20-49 (n = 95)38%Ref 41%Ref 
50-59 (n = 315)38%1.01 (0.75-1.34) 37%1.08 (0.78-1.49) 
60-69 (n = 698)34%1.11 (0.85-1.45) 36%1.09 (0.80-1.47) 
70-79 (n = 1123)28%1.30 (1.00-1.70) 32%1.25 (0.93-1.69) 
80+ (n = 578)21%1.48 (1.12-1.94) 30%1.23 (0.90-1.67) 
Socioeconomic status, quintileb  .028  .107
1 (n = 565)26%1.13 (0.98-1.30) 31%1.14 (0.97-1.33) 
2 (n = 614)29%0.95 (0.83-1.10) 33%0.98 (0.83-1.14) 
3 (n = 633)29%1.05 (0.92-1.21) 34%1.03 (0.88-1.20) 
4 (n = 535)33%0.93 (0.80-1.07) 38%0.93 (0.79-1.10) 
5 (n = 460)29%Ref 32%Ref 
Charlson comorbidity score  <.001  .010
0 (n = 1938)32%Ref 35%Ref 
1-2 (n = 721)25%1.18 (1.07-1.30) 33%1.06 (0.95-1.19) 
3+ (n = 150)15%1.50 (1.24-1.80) 21%1.38 (1.12-1.71) 
Disease related      
Length of hospital stay (unit = 1 day)N/A1.01 (1.01-1.01)<.001N/A1.01 (1.01-1.01)<.001
T stage  < .001  <.001
<T3 (n = 809)49%Ref 53%Ref 
T3-T4 (n = 2000)22%1.69 (1.52-1.88) 25%1.84 (1.62-2.08) 
N stage  <.001   
N negative (n = 1283)40%Ref 45%Ref<.001
N positive (n = 825)18%1.77 (1.58-1.98) 20%1.82 (1.60-2.07) 
NX (n = 701)23%1.41 (1.27-1.57) 29%1.39 (1.23-1.58) 
LVI  <.001  <.001
No (n = 660)47%Ref 55%Ref 
Yes (n = 1459)18%1.80 (1.59-2.03) 19%2.05 (1.77-2.37) 
Unstated (n = 690)37%1.26 (1.10-1.44) 42%1.35 (1.15-1.59) 
Treatment related      
Cancer centre surgical hospital  .070  .164
No (n = 1536)27%Ref 32%Ref 
Yes (n = 1273)32%0.92 (0.84-1.01) 35%0.93 (0.84-1.03) 
ACT  <.001  <.001
Yes (n = 542)29%0.71 (0.62-0.81) 28%0.73 (0.64-0.84) 
No (n = 2267)30%Ref 35%Ref 

After controlling for relevant patient- and disease-related characteristics, utilization of ACT was associated with decreased risk of death from any cause (HR, 0.71; 95% CI, 0.62-0.81) and decreased risk of death from cancer (HR, 0.73; 95% CI, 0.64-0.84). This result was consistent in the propensity score analysis in which the combined HR across quintiles was 0.70 (95% CI, 0.62-0.80) for OS and 0.74 (95% CI, 0.64-0.85) for CSS (Table 4).

Table 4. Propensity Score Analysis Evaluating the Effect of Adjuvant Chemotherapy Compared With No Adjuvant Chemotherapy on Overall (OS) and Cancer-Specific Survival (CSS) Among 2809 Patients With Muscle-Invasive Bladder Cancer (MIBC) Resected in Ontario 1994-2008
QuintileCases Without ACTCases With ACTOverall SurvivalCancer-Specific Survival
Adjusted HR (95% CI)PAdjusted HR (95% CI)P
  1. Notes: Quintile 1 reflects the 20% of patients with the lowest propensity to have ACT.

  2. Four of 2809 cases (0.2%) were not included in the propensity score analysis because of missing data.

Quintile 1 (n = 561)551 100.76 (0.37-1.57).4570.68 (0.27-1.68).400
Quintile 2 (n = 561)537 240.96 (0.58-1.59).8831.00 (0.58-1.71).992
Quintile 3 (n = 561)507 540.68 (0.47-0.98).0390.75 (0.50-1.11).148
Quintile 4 (n = 561)4231380.83 (0.66-1.04).1060.87 (0.68-1.12).293
Quintile 5 (n = 561)2453160.62 (0.51-0.75)<.0010.66 (0.54-0.82)<.001
Combined HR across quintiles 1-5  0.70 (0.62-0.80)<.0010.74 (0.64-0.85)<.001

DISCUSSION

In this large population-based study we have explored utilization of perioperative chemotherapy and outcomes among patients with MIBC in the general population. Several important findings have emerged. First, contrary to existing evidence and published guidelines, utilization of NACT in routine clinical practice remains very low, whereas use of ACT is increasing over time. Despite the increased use of ACT, during the most recent study period (2004-2008), only 23% of patients with resected MIBC received any form of perioperative chemotherapy. Second, utilization of NACT/ACT was substantially lower among older patients, and utilization of ACT was strongly associated with poor prognostic pathologic findings. Third, 5-year overall survival of all NACT (25%) and ACT (29%) patients was substantially lower than outcomes reported in the relevant clinical trials. Finally, in adjusted analyses utilization of ACT was associated with substantial improvement in overall survival and cancer-specific survival in the general population.

The relative utilization of NACT and ACT in the general population of Ontario is not consistent with current evidence or treatment guidelines. The MRC/EORTC clinical trial (n = 976) evaluated 3 cycles of NACT or observation alone followed by each institution's choice of local management. The investigators recently reported updated results showing an absolute improvement in 5-year overall survival of 6% (P = .037).[1] A US Intergroup trial (n = 317) found that 3 cycles of NACT were associated with a 14% improvement in overall survival at 5 years with borderline significance compared with surgery alone (P = .06).[2] A meta-analysis reported a significant survival benefit associated with the use of NACT (5% improvement in OS).[3] On the basis of these data, multiple guidelines (including Cancer Care Ontario) have endorsed NACT.[19, 20]

The evidence in support of ACT for bladder cancer is less robust, with many of the RCTs plagued by small sample sizes and methodologic limitations. Two meta-analyses suggested improved outcomes with ACT. Using individual patient data from 6 RCTs (491 patients), the ABC Meta-Analysis Collaboration reported that ACT was associated with a 25% reduction in the risk of death (HR, 0.75; P = .019), translating into an absolute improvement in overall survival of 9% at 3 years.[4] Ruggeri et al performed a literature-based pooled analysis of 350 patients from RCTs and found improved OS and disease-free survival with ACT (RR, 0.74; P = .001; and RR, 0.65; P < 0.001, respectively).[5] Despite the more limited evidence and the lack of strong endorsement in published guidelines,[20, 21] the clinical community appears to have accepted the efficacy of ACT. This dichotomy is nowhere more apparent than in Ontario, where the provincial clinical practice guideline[21] and a commentary published from Ontario's preeminent cancer center[22] within 1 month of each other have diametrically opposed recommendations. Furthermore, all 3 large contemporary RCTs initiated to address the survival impact of ACT in MIBC have closed prematurely because of poor accrual.[7-9] It is very unlikely that any future RCT evaluating ACT in bladder cancer will have an observation-only treatment arm. The use of carboplatin in 14% of ACT-treated patients in our current study likely relates to postoperative renal dysfunction. It is worth noting that there are no data to support the use of carboplatin-based ACT in this population.

Although the observed practice patterns in Ontario are not congruent with treatment guidelines, they are consistent with other retrospective studies. Three reports have evaluated use of perioperative chemotherapy in patients with bladder cancer in large population-based samples. David et al used records from the National Cancer Data Base in the United States to describe treatment patterns for 7161 patients with stage III bladder cancer diagnosed between 1998 and 2003.[23] They found that 10% of patients received adjuvant chemotherapy, whereas only 1% received neoadjuvant chemotherapy, with a modest increase in use of perioperative chemotherapy over time, from 11% in 1998 to 17% in 2003. Using the Surveillance, Epidemiology, and End Results (SEER)-Medicare database, Schrag and colleagues identified 4664 patients aged 65 years and older with MIBC diagnosed between 1991 and 199924; 11% of patients with stage III disease had perioperative chemotherapy. In a related study by Porter and colleagues using SEER-Medicare data (1992-2002), perioperative chemotherapy was delivered in 18% and 23% of patients with stage II and III bladder cancer, respectively.[25] Although these studies provide insights into the use of chemotherapy for MIBC in the elderly population, our study is the first to evaluate use of perioperative chemotherapy for bladder cancer in a contemporary population-based sample. Svatek et al reported a retrospective series of 3497 patients treated with radical cystectomy at 11 academic centers during 1979-2008.26 ACT was used in 24% of patients and was associated with improved OS (HR, 0.83; 95% CI, 0.72-0.97). However, this study was limited by the lack of information regarding patient comorbidity and also by referral and selection biases inherent to institution-based case series.

In adjusted analyses we found a substantial and significant improvement in OS (HR, 0.71) and CSS (HR, 0.73) associated with ACT. These results were consistent in the Cox model as well as the propensity score analysis. Given the very high risk of death in this patient population, a HR of 0.71 translates to a very large absolute improvement in survival, as demonstrated for the following hypothetical patients. The 5-year OS of a 54-year-old man with no comorbidity and a pT3, node-positive tumor with surgery alone is 15% (95% CI, 10%-23%). His predicted survival increases to 26% (95% CI, 20%-34%) if treated with adjuvant chemotherapy. A 68-year-old woman with moderate comorbidity and a pT2, NX tumor has a 5-year OS of 49% (95% CI, 42%-58%) with surgery alone; this increases to 61% (95% CI, 54%-69%) with ACT. This magnitude of effect is comparable to the benefits associated with adjuvant chemotherapy for breast, colorectal, and non–small cell lung cancers which are accepted standards of care. The effect size of ACT on OS observed in our study (HR, 0.70) is comparable to the results reported in a meta-analysis,[4] a pooled report of RCTs,[5] and the institutional series reported by Svatek.[26]

The 5-year OS of NACT patients (26%) in Ontario is much lower than the survival reported from the pivotal RCTs of NACT (49% and 57%).[1, 2] Likewise, OS at 5 years among our ACT patients was 29% but was closer to 50% in the updated ABC meta-analysis.[4] Several potential factors may explain the inferior outcomes observed in our population-based series compared with the results of clinical trial cohorts, including selection and referral biases, differences in surgical technique, and the known association between greater cystectomy case volumes and improved outcomes.[27] Ongoing analyses will explore the quality of lymph node dissection in this study population and how it might explain the observed outcomes.

Although our study provides detailed data regarding the use of perioperative chemotherapy for bladder cancer in a contemporary population-based cohort, several methodologic limitations merit comment. Although the electronic data sources used in this study describe general aspects of disease, treatment, and outcome for all patients in the province, detailed information related to chemotherapy dose/cycles, treatment toxicity, performance status, and renal function is not available and limits our ability to evaluate the appropriateness of case selection for NACT/ACT. It is possible that other unmeasured confounders may have contributed to the observed survival benefit with ACT. When there is marked change in practice over a short period, it is possible to measure effectiveness by looking at population-level outcomes among all patients before and after the change in practice.[28, 29] However, this was not the case in the current study, as there was no pivotal shift in practice at a specific point. Accordingly, in addition to traditional Cox analyses, we used a propensity score analysis to adjust for potential confounders. A potential example of residual confounding relates to the possibility of misclassification and stage migration of NX patients. However, our results remained robust in Cox analyses stratified for surgical hospital and surgical case volume (data not shown). Despite these potential limitations our results are internally consistent (ie, Cox analysis and propensity score analysis) and are also consistent with the effect size reported in 2 studies pooling RCT data[4, 5] and an institution-based retrospective cohort study.[26] Finally, because our study population was identified on the basis of the presence of MIBC in cystectomy pathology reports, we would not have captured outcomes for the small proportion of patients whose only evidence of MIBC was at the time of transurethral resection.

In summary, our study demonstrates low utilization of perioperative chemotherapy for bladder cancer in the general population. Our results suggest that ACT is associated with a substantial survival benefit in the general population. Given the potential for perioperative chemotherapy to improve patient outcomes, further efforts are needed to understand reasons for underutilization. Efforts in knowledge translation are required to better understand possible barriers to care. Finally, the comparatively poor outcomes of patients in routine practice compared with survival reported in clinical trials suggest a substantial efficacy–effectiveness gap that requires further attention to improve the outcomes of patients with MIBC in the general population.

FUNDING SUPPORT

This research is supported by Cancer Care Ontario and the Canadian Foundation for Innovation. This study was supported by the Institute for Clinical Evaluative Sciences (ICES), which is funded by an annual grant from the Ontario Ministry of Health and Long-Term Care (MOHLTC). The opinions, results and conclusions reported in this paper are those of the authors and are independent from the funding sources. No endorsement by ICES or the Ontario MOHLTC is intended or should be inferred.

CONFLICT OF INTEREST DISCLOSURES

The authors made no disclosures.

Ancillary