Survival after treatment for carcinoma invading bladder muscle: a Dutch population-based study on the impact of hospital volume

Authors


Catharina A. Goossens-Laan, Department of Urology University Medical Centre Utrecht; HP:C04.236, PO Box 85500, 3508 GA Utrecht, The Netherlands. e-mail: C.A.Goossens-Laan@umcutrecht.nl

Abstract

Study Type – Prognosis (cohort)

Level of Evidence 2a

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

The subject of mortality and survival rates after radical cystectomies in high-volume hospitals in comparison to low-volume hospitals has been extensively studied. Postoperative mortality is known to be significantly lower with high-volume providers, but for survival rates there was only a trend forwards this finding. For this reason, we performed this Dutch population-based study on survival rates, to see if we had enough power to support this trend with significant findings. To our knowledge, this is the first study of good quality showing a significant beneficial effect for survival in high-volume hospitals.

OBJECTIVE

  • • To examine the volume–outcome relationship for carcinoma invading bladder muscle (MIBC) with respect to differences in survival rates among all hospitals in the Netherlands as a guide for regionalization initiatives.

MATERIALS AND METHODS

  • • This population-based retrospective study included all patients (n= 13 033) newly diagnosed with MIBC during the period 1999–2008 in the Netherlands, selected from the Netherlands Cancer Registry.
  • • Data were collected on demographics, morphology, stage at diagnosis and after surgery, primary treatment, vital status and date of follow-up or death.
  • • The relative survival rate (RSR) per treatment was analysed for age, stage and hospital surgical volume.

RESULTS

  • • Overall 5 and 10-year RSR for all treatments of MIBC was 32% and 25%, respectively.
  • • Although 71.7% of the patients featured stages II and III, radical cystectomy was only performed in only 42% and 44% of these patients, respectively.
  • • Relative survival for MIBC remained unchanged in the two consecutive time periods (1999–2003 and 2004–2008).
  • • In all, 34% of patients diagnosed in low-volume hospitals (<10 cystectomies/year) underwent cystectomy vs 42% of those diagnosed in high-volume hospitals (P= 0.000).
  • • In a multivariate analysis long-term survival (>30 days after surgery) was significantly lower in patients after cystectomy for stage II/III in low-volume hospitals (hazard ratio [HR] 1.17, P= 0.036). A high lymph node count (>20) was associated with a lower risk of death (HR 0.52, P= 0.000).

CONCLUSIONS

  • • The 10-year RSR for patients with MIBC in the Netherlands was modest (25%) and has remained unchanged in the last decade.
  • • The chance of undergoing cystectomy is significantly higher in high-volume hospitals. Long-term survival after cystectomy is higher in high-volume hospitals.
  • • Regionalization of bladder cancer treatments could improve overall outcomes.
Abbreviations
MIBC

carcinoma invading bladder muscle

RSR

relative survival rate

IRT

interstitial radiotherapy

EBRT

external beam radiotherapy

TURBT

transurethral resection of bladder tumour

NCR

Netherlands Cancer Registry

CCC

Comprehensive Cancer Centre

ASR

absolute survival rate.

INTRODUCTION

In the Netherlands in 2008 there were 5100 cases of newly diagnosed bladder cancer, with a lifetime cumulative risk of 2%. Of these, 30% were cases of carcinoma invading bladder muscle (MIBC) [1]. Radical cystectomy with bilateral pelvic lymphadenectomy is considered to be the optimum therapy for patients with MIBC or with progression to MIBC after local treatment for stage I disease. Other curative treatment options are interstitial radiotherapy (IRT), e.g. brachytherapy (for small solitary clinical stage II tumours), and external beam radiotherapy (EBRT) [2]. When a patient is not eligible for any of the above-mentioned therapies owing to co-morbidity or preference, a non-curative option usually follows: a transurethral resection of the bladder tumour (TURBT) or palliative radiotherapy.

In the Netherlands in recent years, 900 radical cystectomies have been performed in 88 general and nine university hospitals [3]. During the period 2000–2006 the postoperative mortality rate ranged from 1.2% in high-volume hospitals (>10 cystectomies/year) to 6.4% in low-volume hospitals (<10 cystectomies/year) [4]. In a prospective study on complications for cystectomy, the postoperative complication rate of 64% appeared to be related to this postoperative mortality rate [5]. Annually in the Netherlands, IRT is performed 30–40 times as a primary treatment, compared with EBRT which is performed 300–350 times [3].

To guide regionalization initiatives for cystectomy, information on the influence of hospital volume on long-term survival rates for hospitals in the Netherlands is needed; therefore, data from the Netherlands Cancer Registry (NCR) were used to examine long-term (5- and 10-year) relative survival rates (RSRs) among patients undergoing different types of treatment, according to stage, age and hospital volume.

MATERIALS AND METHODS

CANCER REGISTRY DATA

All patients (n= 13 033) with MIBC newly diagnosed in the Netherlands between 1 January 1999 and 31 December 2008 were selected from the population-based NCR (which has had complete national coverage since 1989). Dutch hospitals were served in nine largely coordinating and facilitating Comprehensive Cancer Centres (CCCs) that host the cancer registry. Notified by the Dutch national pathology database (PALGA), registration clerks of the CCCs extract information for the registry from the medical records in the hospitals. Various data are collected, including demographics, morphology, stage at diagnosis and after treatment, primary treatment, vital status and date of follow-up or death. Tumour site and morphology are coded according to the International Classification of Diseases for Oncology [6]. Quality of data is generally considered to be high and completeness is estimated to be >95% [7,8]. Comorbidity at diagnosis is only registered in the south of the Netherlands and was therefore not available for use in the present study. Follow-up of the vital status through the national common database of municipalities was complete until 1 February 2010. The cause of death was generally not available.

The present registration study was performed according to the privacy regulations and approved by the Privacy Commission of the NCR.

DATA ANALYSIS

Data analysis of this retrospective cohort study was done using anonymized data. Stage was grouped according to the Union Internationale Contre le Cancer TNM classification, 6th edition [9]. The clinical TNM system was used for comparison of various treatment policies concerning all patients. The pathological TNM system was used for a subset analysis of radical cystectomy cases. Excluded from the analysis were patients with invasive bladder cancers with unknown stage or without histologically proven diagnosis, neuro-endocrine carcinomas, bladder sarcomas and bladder cancer only found post mortem. MIBC as a result of progression or recurrence after a previous Ta or Tis tumour was included, while progression of T1 tumours to MIBC was excluded. This was owing to registration practices, e.g. T1 is registered as an invasive tumour, as is MIBC, and progression of T1 is not accounted for as progression in the system. Only the initial treatment for every new tumour was registered, thereby disregarding cystectomy ≥ 6 months after the first diagnosis or a salvage cystectomy after radiotherapy. When the initial treatment (e.g. neo-adjuvant chemotherapy) had taken ≥6 months to complete, the (intended) cystectomy was recorded. About 20% of the cystectomies performed annually were done on T1 tumours (6%), or as secondary treatment.

Primary treatment was grouped as cystectomy, EBRT, IRT, chemotherapy only, local (intravesical) treatment only, or no/unknown treatment. Local treatment was defined as one or multiple TURBTs often followed by bladder installations. A patient who underwent cystectomy and radiotherapy and/or chemotherapy was classified as ‘cystectomy’. A patient receiving radiotherapy as well as chemotherapy was classified as ‘radiotherapy’. Combined EBRT and IRT was classified as ‘IRT’. No information was available on the number of patients treated with palliative intent of cystectomy and/or radiotherapy.

The hospitals were classified according to the average annual number of cystectomies performed. Hospitals with <10 cystectomies per year were classified as low volume and hospitals with ≥10 cystectomies per year were classified as high volume. The number of hospitals performing ≥20 cystectomies was too small to analyse separately.

STATISTICAL ANALYSIS

All analyses were performed in stata, version 10. A P value of <0.05 was considered to indicate statistical significance. The RSR was estimated using the Hakulinen method [10]– the ratio of observed survival to the expected survival in the general population of the Netherlands of the same age and sex – using the strs-command for stata[11]. The absolute survival rate (ASR) was calculated using the Kaplan–Meier survivor estimator. Univariate and multivariate Cox regression analysis were performed for the risk of death for patients, conditional upon surviving ≥30 days after surgery, to examine the influence of age of diagnosis, gender, postoperative stage, residual disease and hospital volume on RSR.

RESULTS

GENERAL

Of the 13 033 patients with MIBC, 9441 were male (72%) and the median (range) age at diagnosis was 73 (22–100) years (Table 1) The median (interquartile range) length of follow-up was 70 (40–100) months. Table 2 shows the treatment of MIBC according to clinical stage and age group. During the whole study period, 167 patients received neoadjuvant chemotherapy before undergoing cystectomy (3.6%), and 94 cystectomy patients underwent adjuvant chemotherapy (2%); however, neoadjuvant therapy increased over time and in 2008, 7% of the patients undergoing cystectomy received neoadjuvant chemotherapy.

Table 1.  Characteristics of patients with MIBC in the Netherlands in 1999–2008
CharacteristicAge groupTotal
<75 years≥75 years
Sex, n (%)   
 Males5527 (74.5)3914 (69.7)9 441 (72.4)
 Females1890 (25.5)1702 (30.3)3 592 (27.6)
Morphological classification, n (%)   
 Urothelial cell carcinoma6891 (92.9)5304 (94.4)12 195 (93.6)
 Squamous cell carcinoma295 (4.0)212 (3.8)507 (3.9)
 Adenocarcinoma173 (2.3)58 (1.0)231 (1.8)
 Undifferentiated carcinoma58 (0.8)42 (0.7)100 (0.8)
Clinical stage, n (%)   
 II4013 (54.1)3429 (61.1)7 442 (57.1)
 III1036 (14.0)873 (15.2)1 909 (14.6)
 IV2368 (31.9)1314 (23.4)3 682 (28.3)
Treatment, n (%)   
 cystectomy3777 (50.9)845 (15.0)4 622 (35.5)
 EBRT1121 (15.1)2030 (36.1)3 151 (24.2)
 IRT301 (4.1)59 (1.1)360 (2.8)
 Chemotherapy542 (7.3)68 (1.2)610 (4.7)
 Local therapy only1091 (14.7)1721 (30.6)2 812 (21.6)
 None/unknown585 (7.9)890 (15.8)1 475 (11.3)
Total7417561613 033
Table 2.  Treatment of patients with MIBC, according to clinical stage and age group, in the Netherlands in 1999–2008
 <75 years≥75 yearsAll ages
Stage IIStage IIIStage IVStage IIStage IIIStage IVStage IIStage IIIStage IV
n= 4013n= 1036n= 2368n= 3429n= 873n= 1314n= 7442n= 1909n= 3682
Cystectomy, %64652217207424417
EBRT, %141716394224262819
IRT, %711200410
Chemotherapy, %12210140115
Local only, %13721351629231124
None/unknown, %17207223641425

During 2004–2008, in clinical stage II–III, the proportion was almost equal in low- and high-volume hospitals (2.4% and 2.0%, respectively). In clinical stage IV, those figures were 15% and 30%, respectively. Of the 3151 patients receiving EBRT, 118 patients (4%) received chemotherapy.

RSRS

The RSRs of patients with MIBC for all treatments combined were 32% and 25% at 5 and 10 years after diagnosis, respectively. Figure 1A–C show the long-term RSRs according to stage and treatment. Survival was clearly inferior for patients receiving only local treatment, only chemotherapy or no oncological therapy. In clinical stage II disease the 5-year RSR after IRT was higher than after cystectomy (70% vs 57%). When only cystectomy patients with pathologically confirmed stage II MIBC were selected, the 5-year RSR was 75%.

Figure 1.

A, The RSRs of patients with MIBC clinical stage II (A), clinical stage III (B) and clinical stage IV (C) in the Netherlands, 1999–2008, according to primary treatment. Curves stop if the number of cases at risk is ≤15.

For the age group <55 years, the 5-year ASR was 41% and the 10-year ASR was 34%; in patients aged 55–64 years the 5-year ASR was 38% and the 10-year ASR was 28%; ASRs for patients aged 65–74 years were 29% and 18% at 5 and 10 years, respectively and in patients >74 years ASRs were 14% and 5% at 5 and 10 years, respectively.

When comparing all patients diagnosed during 1999–2003 with those diagnosed during 2004–2008, no increase in survival was observed (31–32%). In the later period, more patients ≥75 years underwent cystectomy than in the earlier period (16% vs 12% ), whereas about the same numbers in both groups received radiotherapy (35% and 34%, respectively). The 5-year RSRs for cystectomy showed no significant change: 52% (95% CI 49–54%) and 48% (95% CI 45–51%) for the earlier and later period, respectively. The 5-year RSRs for EBRT also showed little change over time: 25% (95% CI 22–28%) and 26% (95% CI 23–30%) for the earlier and later period, respectively.

CYSTECTOMY AND HOSPITAL VOLUME

Of the 2168 cystectomies performed between 2004 and 2008, 781 (30%) were performed in a high-volume hospital. In multivariate Cox regression analysis (>30 days after surgery) higher age, squamous cell carcinoma, higher pT stage, presence of regional/distant metastasis, residual disease and no lymph node dissection or low lymph node count were independent predictors of a lower survival rate (Table 3). Gender was not an independent predictor of lower survival.

Table 3.  Multivariate Cox regression analysis for the risk of death after cystectomy of patients with clinical stage II or III bladder cancer, 2004–2008 (excluding patients who died <30 days after cystectomy)
CharacteristicnUnivariate analysisMultivariate analysis
HR 95% CIHR 95% CI
  1. Values in bold and italic are significant.

Sex     
 Male15931 (Ref.) 1 (Ref.) 
 Female5751.080.94–1.251.110.96–1.29
Age group, years     
 <552651 (Ref.) 1 (Ref.) 
 55–646261.150.91–1.461.230.97–1.57
 65–748441.421.13–1.771.511.191.90
 75+4331.841.44–2.331.821.412.35
Morphological classification     
 Urothelial cell carcinoma20391 (Ref.) 1 (Ref.) 
 Squamous cell carcinoma861.741.32–2.311.361.02–1.81
 Adenocarcinoma390.790.48–1.320.630.37–1.07
 Undifferentiated carcinoma42.030.65–6.292.910.93–9.14
pT stage     
 T29981 (Ref.) 1 (Ref.) 
 T39352.762.37–3.202.462.11–2.87
 T42355.14.21–6.163.693.00–4.55
pN stage     
 N018061 (Ref.) 1 (Ref.) 
 N+3622.091.80–2.421.841.56–2.17
pM stage     
 M021571 (Ref.) 1 (Ref.) 
 M1116.603.53–12.32.551.32–4.91
Residual disease     
 R017991 (Ref.) 1 (Ref.) 
 R1/R21693.202.66–3.862.021.65–2.48
 RX2001.461.20–1.781.281.05–1.57
Hospital volume     
 ≥10 cystectomies per year15291 (Ref.) 1 (Ref.) 
 <10 cystectomies per year6391.211.05–1.401.171.01–1.35
Systemic chemotherapy     
 No20781 (Ref.) 1 (Ref.) 
 Adjuvant401.300.86–1.960.760.49–1.17
 Neo-adjuvant500.760.47–1.230.670.41–1.09
Lymph node dissection     
 No3951 (Ref.) 1 (Ref.) 
 Yes, 1–9 nodes8290.890.75–1.060.780.65–0.93
 Yes, 10–19 nodes4050.730.59–0.890.620.50–0.78
 Yes, ≥20 nodes1260.550.39–0.780.520.36–0.75
 Yes, number unknown4130.840.69–1.020.840.68–1.03

Patients who underwent cystectomies in low-volume hospitals showed a significantly higher risk of death (>30 days after surgery) than in high-volume hospitals: hazard ratio (HR) 1.2 (95% CI 1.01–1.35; P= 0.036). The postoperative 30-day mortality rate was 2.9% and the 90-day mortality rate was 7.8% in low-volume hospitals.

Data on patient age, stage and therapeutic strategies, stratified according to the cystectomy volume of the hospitals, for all patients with MIBC (n= 13 033), is shown in Table 4. Patients diagnosed in hospitals with a low volume of cystectomies were slightly older, while the chance of undergoing a cystectomy was significantly smaller (34%) compared with patients diagnosed in hospitals with a high volume (42%; P= 0.000).

Table 4.  Differences in patient age, stage, distribution on therapeutic strategies, for low- vs high-volume hospitals in the Netherlands, 1999–2008
 Low-volume hospitals,n (%)High-volume hospitals,n (%)P
Age group, years   
 <45183 (2)46 (2) 
 45–54708 (7)186 (8) 
 55–641 847 (17)421 (19) 
 65–743 340 (31)686 (31) 
 75+4 711 (44)905 (40)0.002
Clinical stage   
 II6 187 (57)1255 (56) 
 III1 591 (15)318 (14) 
 IV3 011 (28)671 (30)ns
Treatment   
 Cystectomy3 684 (34)938 (42) 
 EBRT2 739 (25)412 (18) 
 IRT311 (3)49 (2) 
 Chemotherapy504 (5)106 (5) 
 Local therapy only2 284 (21)528 (24) 
 None/unknown1 267 (12)211 (9)0.000
Total10 7892244 

DISCUSSION

Using data from the population-based NCR we evaluated RSRs for patients with MIBC treated with different therapeutic methods. Seven main findings emerged. (i) Total RSRs for all treatments combined at 5 and 10 years after diagnosis of MIBC were 32% and 25%, respectively. (ii) Survival was similar for groups of patients with MIBC treated with different therapeutic methods and did not change over time. (iii) Although 71.7% of the patients presented with clinical tumour stages II or III, radical cystectomy was performed in only 42% and 44% of these patients, respectively. (iv) Age appears to be an important factor in the choice of treatment strategy: patients ≤75 years underwent a cystectomy more often, while older patients were more likely to undergo radiotherapy. (v) The chance of undergoing a cystectomy was significantly lower when diagnosed in low-volume hospitals (34% vs 42%, respectively; P= 0.000). Furthermore, RSRs after cystectomy for stage II/III disease differ between low- and high-volume hospitals, with a 17% higher risk of death for patients operated in low-volume hospitals (<10 cystectomies/year; HR 1.17). (vi) Having no lymph node dissection performed or having a low lymph node count were independent predictors of lower survival (vii).

Population-based studies on survival incorporating all treatments of bladder cancer are scarce. The overall 5-year survival rates reported for stage I–IV bladder cancer range from 58 to 82% [12,13]. We measured RSR as the ratio of observed survival to the expected survival in the general population of the Netherlands of the same age and sex. Only one other population-based study (also from the Netherlands) has reported RSRs for patients with MIBC who underwent various different treatment methods. For the period 1988–2003, 5-year RSRs for stages II, III and IV were reported as 44%, 28% and 9%, respectively, and 10-year RSRs for these stages were reported as 36%, 21% and 6%, respectively [14].

The present Dutch study shows that 42% of clinical stage II patients underwent a cystectomy. A population-based study using Surveillance, Epidemiology and End Results data showed that an even lower percentage of stage II patients (21%) underwent cystectomy [15]. As radical cystectomy is the guideline-recommended treatment for MIBC, data from both of these studies imply underuse of this procedure [15].

Our earlier study [4] also reported interesting regional and/or hospital variation in the choice of treatment methods. Between regions the proportion of patients aged ≤75 years with MIBC undergoing cystectomy ranged from 60 to 77%, the proportion undergoing EBRT from 10 to 28%, and the proportion undergoing IRT from 2 to 13%.

Interstial radiotherapy is a bladder-sparing treatment in which radioactive sources are positioned in or close to a tumour. The technique was first used in the Netherlands for bladder cancer in the 1970s [16]. As this treatment is given to a selected group of patients with small (<5 cm) solitary stage II tumours (a favourable prognostic factor for MIBC), comparison of outcomes with other treatment methods should be done with great care.

(Neo-)adjuvant chemotherapy is a treatment option for patients with MIBC undergoing cystectomy [2]. In the present study, only a small percentage received neo-adjuvant (4%) or adjuvant (2%) chemotherapy. Chemotherapy was predominantly given in patients participating in a trial, however, an increased use of chemotherapy was observed: in 2008, 7% of the cystectomy patients received neoadjuvant chemotherapy. RSRs in the Netherlands have remained unchanged in the last 10 years [17]. This suggests little improvement, despite the development of treatment strategies such as extended lymph node dissections.

Age seems to be an important factor in the choice of treatment strategies which, in the present study, differed widely between younger and elderly patients. Although the RSR after cystectomy was superior to the RSR after the other treatment methods, local therapy and EBRT (Fig. 1), only 15% of the patients aged ≥75 years underwent cystectomy. The difference in survival between treatment methods might also be attributable to selection bias. Comorbidity is negatively associated with undergoing a cystectomy and about two-thirds of the Dutch patients aged ≥75 years with MIBC also suffered from serious comorbid conditions [18]; however, multiple studies have shown that radical cystectomy is a safe option for elderly patients, even for those with comorbidity [19–21]. Good patient selection, therefore, seems to be imperative in the choice of treatment in the elderly.

In addition to the high postoperative mortality rates in low-volume hospitals reported earlier [4], survival after cystectomy seems to be lower in hospitals performing ≤10 cystectomies a year compared with higher-volume hospitals. To our knowledge, the present study is the first high-quality study showing a significant beneficial effect for survival in high-volume hospitals [22].

In multivariate analysis lymph node dissection and lymph node count were significant, showing that the process of care given to a patient, e.g. the extent of the lymph node dissection, is a prognostic factor.

With the ongoing discussion on quality of care of patients with MIBC, regionalization for cystectomy seems justified, but mortality is not the only determinant of quality of care, and procedural volume is not the only measure of quality. Underperforming high-volume providers do exist, as do low-volume providers with excellent outcomes. Minimum volume standards fall short in identifying underperforming high-volume centres and might undeservedly lead to the closure of well-performing low-volume centres. Further research is needed to look for additional quality-of-care indicators and should be the next step for the guiding of regionalization initiatives [23].

The present study has a few important limitations. The NCR provides population-based data from patients diagnosed with MIBC in the Netherlands, but data on the pre-treatment comorbidity status of patients with MIBC are incomplete. Comorbidity is only registered in the region of the CCC South. For patients with MIBC diagnosed in this region, a recent study (submitted) showed a negative association between comorbidity and the chance of undergoing cystectomy. The lack of comorbidity data and information on the cause of death might introduce bias if the comorbidity status of patients in high- and low-volume hospitals differ. On the one hand, if patients with high comorbidity are selectively referred to high-volume hospitals the survival outcome in those hospitals would even be better when corrected for comorbidity. On the other hand, patients with high socio-economic status and lower comorbidity might choose to be operated in a high-volume hospital, which would be the cause of the relatively favourable survival outcome of the high-volume hospitals. As little information was available to the general population on cystectomy volume during the study period, the latter effect is unlikely. It is also unlikely that low-volume hospitals selectively refer patients with low comorbidity to high-volume hospitals. A slight tendency towards increased referral to the high-volume hospitals emerged in the last few years of our study period, increasing further in 2009–2010 after publication of the hazards in high- vs low-volume hospitals in the Netherlands [4]. The assumptions regarding patient selection among the elderly with MIBC for EBRT, although very probably accurate, remain to be proven.

Other limitations of the study are inadequate control of additional confounding factors (e.g. no registration on follow-up treatments) and the impossibility of estimating disease-specific survival. Comorbidities and comorbidity-associated events represent very important causes of mortality in patients undergoing cystectomy. A recent study on cancer-specific and other-cause mortality showed that, after stratification according to disease stage and patient age, cancer-specific mortality was the main cause of mortality in all patient strata. Nonetheless, at 5 years after radical cystectomy, between 8.5% and 27.1% of deaths were attributable to other causes [24].

The use of cystectomy instead of all treatment methods to define volume measurements was based on the following reasons. The problem regarding patients undergoing EBRT is that all hospitals refer patients to radiotherapy centres, making this treatment method unsuitable for distinguishing between low- and high-volume centres. The same problem arises for brachytherapy/IRT as this treatment is given in a few specialized centres and the annual volume is low (≈150 cases in 2004–2006). Neo-adjuvant chemotherapy is also mostly given in high-volume centres, in trials that run mostly in oncological centres.

A major strength of the present study is that its results are based on the entire population in the Netherlands, of unselected patients with diverse treatment regimes.

In conclusion, less than one third of patients with MIBC in the Netherlands are being cured. Despite improved treatment strategies, no improvement in survival was observed in the last decade. Survival appears to be significantly higher in high-volume hospitals.

CONFLICT OF INTEREST

None declared.

Ancillary