Stage, grade and pathological characteristics of bladder cancer in the UK: British Association of Urological Surgeons (BAUS) Urological Tumour Registry

Authors


Abstract

Objective

  • To investigate Tumour-Node-Metastasis (TNM) stage and demographics at presentation in a very large, contemporary UK cohort of patients with bladder cancer and compare them with other published series, as little published data exists on the pathological characteristics of bladder cancer at presentation.

Patients and Methods

  • The British Association of Urological Surgeons (BAUS) Section of Oncology started a new urological tumour registry in 1998.
  • We performed a data analysis of all bladder cancer cases between 1999 and 2008.
  • Tumour TNM stage, grade and histopathological diagnosis were reviewed along with standard epidemiological data.

Results

  • In all, 69 712 bladder cancer registrations were recorded. Complete T, N and M stage and grade was available for 32 240 patients.
  • The male to female ratio of the study population was 3:1 and the overall median (sd, range) age at presentation was 73 (11.6, 6–108) years.
  • Final pathological T staging showed that non-muscle-invasive bladder cancer accounted for 75% of cases with the remaining 25% being muscle-invasive disease. Of these patients, 8% had nodal disease and 4% other metastatic sites at presentation.
  • The tumour grade was G1–2 in 65% and G3 in 35% of cases. Transitional cell carcinoma (TCC) accounted for 92%, squamous cell carcinoma and adenocarcinomas 1.5% each, with 5% other histological variants.

Conclusions

  • Non-muscle-invasive TCC accounted for 75% of bladder cancer cases in the UK. The 1973 World Health Organization classification remains in widespread use amongst pathologists in the UK.
  • Obtaining complete and standardised staging and pathology reporting systems in bladder cancer remains a challenge.

Introduction

Bladder cancer remains a major health burden. In most Western countries including the UK, bladder cancer ranks as the fourth most common cancer in males, and eighth in females. Globally an estimated 383 000 new cases of bladder cancer occur each year and is ranked sixth in terms of overall cancer frequency [1]. At least 75–80% of newly diagnosed bladder cancer is superficial, non-invasive papillary TCC, and at least 50–75% of these patients will recur over time. This high recurrence rate impacts heavily on prevalence rates making bladder cancer the second most common cancer behind prostate cancer in males. This necessitates long-term surveillance over decades making multiple hospital treatments common. Non-muscle-invasive bladder cancer (NMIBC) should therefore be regarded as a chronic disease state.

It has been suggested that bladder cancer is the most costly common malignancy to manage per patient from diagnosis to death [2]. In the UK the total cost of bladder cancer between 2001 and 2002 has been estimated at £55.4 million (£8349 per patient), of which £32.25 million was associated with NMIBC [1]. In the USA bladder cancer is the fifth most expensive cancer in terms of total medical care expenditures, estimated at 3.7 billion USA dollars using 2001 indices [3, 4].

On the other hand, muscle-invasive bladder cancer (MIBC) remains a lethal disease entity, with little improvement in survival over the last three decades [5]. Despite the prevalence of bladder cancer, little published data exists on the pathological characteristics of these tumours at presentation. Most reports are from relatively small cohorts of patients, with inherent selection bias [6-9]. Also, specific TNM stage and grade data collected by cancer registries is not always complete. For example in 2010, the TNM data from the United Kingdom Association of Cancer Registry (UKACR) is only 15% complete [10].

Patients and Methods

The BAUS Section of Oncology started a new urological tumour registry in 1998. These registrations were voluntarily submitted by participating surgeons from >50 Urological Departments across the UK, with anonymous data on all newly diagnosed urological cancers. Currently this database holds information on ≈270 000 cancers, of which >70 000 are bladder cancer registrations. This data complements that collected by the National Cancer Data Repository (NCDR) using unique identifiers and is shared on an annual basis. Data were collected prospectively on standardised data sheets for each organ site. The bulk of the data were submitted electronically to a secure central access database, a minority were submitted in paper format. After 2007, a secure web-based database was used.

We performed a data analysis of all bladder cancer cases between 1999 and 2008. Tumour TNM stage, grade and histopathological diagnosis were reviewed along with standard epidemiological data. T stage was usually derived from transurethral resection of bladder tumour (TURBT) and bimanual examination under anaesthesia. N and M staging was derived from cross-sectional imaging (CT or MRI). We followed the same principles as the UKACR attributing an integrated stage, which is a hybrid of pathological and clinical T, N and M values. Generally, the T and N values will come from the pathology and/or imaging, and the M value will be attributed clinically or from imaging. For example M staging in NMIBC is all attributed N0 M0 unless specifically reported as being otherwise [10]. All X category data was excluded from analysis. The histopathology was recorded in free-text boxes. Central pathology review was not feasible. Data calculations and statistical analysis was performed with StatsDirect™ and Tableau™ software.

Results

We obtained data on 69 712 bladder cancer registrations during the 10 years of data analysed. The number of new bladder cancer registrations per year and the sex distribution is outlined in Fig. 1. The median number of new cases was just >7000 per annum. This represents ≈50% of the total bladder cancer registrations recorded by the UKACR per year. The relative amounts of complete data for all the patients are summarised in Table 1.

Figure 1.

Numbers of new bladder cancer registrations by year between 1999 and 2008 (69 712 patients).

Table 1. Summary of relative amounts of complete patient data by endpoint
Data endpointNumber of patients
T stage43 899
N stage43 133
M stage42 772
Tumour grade54 476
Histological type66 873
Complete TNM, grade and histology32 240

Complete TNM staging was available for 32 240 patients. This represents 48% of all the patients in our database. The sex distribution showed a male to female ratio of 3:1. The median (sd, range) age at presentation overall was 73 (11.6, 6–108) years. The median (sd, range) age in males was 72 (11.4, 6–108) years and in females 74 (12, 7–108) years. Women were significantly older at presentation (P < 0.001, 95% CI –2.205 to –1.795).

The T stage and the relative frequency of N1–3 and M1 disease at presentation of patients with complete staging (32 240 patients) is summarised in Table 2. NMIBC (pTa; pT1; pTis) represent 76% of cases, with MIBC in the rest. In all, 2% of patients presented with nodal disease and 1.3% with metastatic disease overall. Of the patients with MIBC, 8% had nodal disease and 3.7% other metastatic sites at presentation. It should be noted that the incidence of node-positive disease outlined in Table 2 is much less than what would be expected for patients with T2–T4 tumour. This is a reflection of using integrated staging using clinical and cross-sectional imaging rather than pathological staging in attributing N stage.

Table 2. TNM stage at presentation of patients with complete staging data using an integrated TNM staging (32 240 patients)
T Stagen%N1%N2%N3%M1%
Ta15 38347.700000000
T18 62126.7320.4120.120.08340.4
T25 34516.61232.3280.560.11542.9
T31 5734.919212.2785.0100.61358.6
T46392.010115.860.9233.68813.7
Tis6792.110.110.10040.6
Total32 240 449 125 41 416 

Tumour differentiation was recorded by most pathologists across the UK using the 1973 WHO classification. The tumour grade was accurately recorded in 32 240 cases, with G1 in 26% of cases, G2 in 39% and G3 in 35%. Tumour grade stratified by stage is shown in Table 3. In all, 80% of NMIBC cases were graded as well to moderately differentiated (G1–2), while 80% of muscle-invasive tumours were high grade (G3). There were some staging discrepancies, which are difficult to account for, e.g. the small incidence of G1 muscle-invasive disease. This can only be explained by the methods of data collection and lack of centralised pathology review. It is likely this reflects reporting by pathologists without a sub-specialist interest in bladder cancer.

Table 3. Tumour grade at presentation stratified by stage (32 240 patients)
StageG1, %G2, %G3, %
Ta21.223.43.4
T13.51211.4
T20.43.613.7
T30.70.74.1
T40.20.21.3
Tis001.2
Total263935

The histopathological features of this large cohort of 66 873 patients, is outlined in Tables 4 and 5. TCC accounted for 92% of bladder cancer cases in the UK. Squamous cell cancer and adenocarcinoma accounted for 1.5% respectively, with 5% being other histological variants. These variants are clearly outlined in Table 5. TCC variants most commonly reported include TCC with squamous or sarcomatoid (spindle cell) differentiation. Of the other rare histological variants, small cell neuroendocrine bladder tumours and carcino-sarcomas were the most common. Not all variants were reported on, for example micro-papillary variant was not reported, and reflects the weakness of databases with voluntary free text fields, as opposed to mandatory fields from pull-down menus.

Table 4. Proportions of bladder cancer subtypes (66 873 patients)
Histological group% of total
TCC91.7
Adenocarcinoma1.7
Squamous cell carcinoma1.6
Other histological variants5.1
Table 5. Histopathology at diagnosis from initial TURBT and relative frequency of bladder cancer subtypes (66 873 patients)
Histological typeSubtypeNumber patientsProportion, %
Urothelial neoplasmNon-invasive38 25757.208
Infiltrating25 08237.507
Flat450.067
CIS5640.843
TCC variantsKeratinising70.010
Squamous differentiation1700.254
Nested variant20.003
Sarcomatoid/ spindle cell800.120
Undifferentiated carcinoma210.031
Signet ring180.027
Clear cell90.013
Inverted papilloma290.043
Squamous carcinoma 1 1041.651
Adenocarcinoma 1 0801.615
MesenchymalSarcoma530.079
Leiomyosarcoma70.010
Rhabdosarcoma10.001
Haematopoetic/lymphoid 150.022
Lymphoma110.016
Carcino-sarcoma 1100.164
NeuroendocrineCarcinoid240.036
Small cell1290.193
Phaeochromocytoma40.006
Metastases 510.076
Total 66 873100.000

Discussion

Data Quality and Study Limitations

Comparison of newly diagnosed bladder cancers recorded by cancer registries and those collected by the BAUS database indicates that roughly 50–60% of all new bladder cancers were recorded in the decade between 1999 and 2008. Comparing the lowest registration year of 2008 with the UKACR figures, BAUS registered 4747 of 8791(54%) new bladder cancers.

The present study provides complete TNM stage and grade data for 32 240 patients, representing 48% of patients in the database. We admit that because this is incomplete, we run the risk of introducing selection bias when analysing the data. However, this problem is common, and occurs in National Cancer Registries [11]. As referenced above, the grade data for bladder cancer from UKACR was 75% complete in 2010, while the TNM data was much worse than this at just 15% complete. This difference between BAUS and UKACR data has been recognised. While we accept that it does not give full coverage of the population, the higher quality stage and grade data allow us to share data with the UKACR in a complementary fashion. When we compared TNM stage and grade data for the whole cohort (66 873 patients) vs those with complete data (32 240), there was no significant effect on the relative frequency of different stage or grade of tumour at presentation.

Comparison with Published Data

In the present series, 75% of tumours were NMIBC (Ta 48%, T1 27% and carcinoma in situ [CIS] 2%). Zhang et al. [7] reported similar clinicopathological features in 658 patients with urothelial carcinoma of the bladder (UCB) from Shanghai, China showing pTa in 49% and pT1 in 29% of cases. While, deVere White et al. [8] reported that ≈80% of patients with newly diagnosed bladder carcinomas presented with superficial diseases (Ta/T1), ≈20% of which were stage T1. In a similar study, Matalka et al. [9] reviewed 115 patients with UCB showing 55% of the cases were stage pTa, 17% were pT1, 20% were pT2, 7% were pT3, and 0.9% were pT4. The percentages of Ta/T1 patients in the present study are very similar (Table 2). Kirkali et al. [12], quoted 70% of bladder tumours presenting as superficial disease and 30% as muscle-invasive disease, with ± 70% of the superficial tumours present as Ta lesions, 20% as T1 lesions, although 10% of their cohort were reported to have CIS (Tis lesions). Rafique and Javed [13] reporting on a series of 221 patients with bladder cancer in Pakistan, 98% showed invasion of lamina propria (pT1 at least), and >60% of the patients had ≥ pT2 disease at the time of diagnosis, possibly relating to delayed presentation. The median age at presentation in the present study was significantly older in women than in men. This fits with similar data reported by Scosyrev et al. [14].

In all, 80% of the newly diagnosed NMIBC in the present study cohort were well to moderately differentiated (G1–2) at presentation, while 80% of patients with MIBC had high-grade (G3) disease. Messing et al. [15] reported similar findings in a cohort of 509 males from Wisconsin, confirming that grade–stage category at diagnosis predicts overall and bladder cancer mortality but not mortality from other causes. Two other reports confirm that most NMIBC cases are well to moderately differentiated, while almost all MIBC cases are high grade. In a large study reporting on 1515 cases of NMIBC, Pan et al. [16] showed that 60% of the cases were low grade, while Zhang et al. [7] found 56% of the superficial tumours were of low grade, while 95% of the muscle invasive tumours were of high grade. In the present study, tumour grading at initial TURBT (32 240 patients) was G1 in 26%, G2 in 39% and G3 in 35%, where most pathologists were still using the WHO 1973 classification as opposed to the WHO/International Society for Urological Pathology (ISUP) system. The limitations of the WHO 1973 classification for Grade (particularly Grade 2 disease) has long been recognised. Despite this, uro-pathologists have not embraced newer classification systems reporting only low- and high-grade lesions.

The low-risk cases (1973 WHO Grade1 and papillary urothelial neoplasm of a low malignant potential [PUNLMP]) represent at least 50% of newly diagnosed bladder cancer, and while regarded as having low malignant potential in terms of invasiveness and risk of progression, they can often be more problematic and costly than muscle-invasive tumours, and impact significantly on patient's quality of life. These tumours also account for a significant proportion of the workload and resources in urology departments.

Pathological Reporting

The histological classifications proposed by the ISUP in 1987 and the WHO in 1998 and 2004 have not been adopted widely within the uro-pathology community across the UK for good reasons. When first introduced, both the new classifications were supported by Level 5 evidence. (scale of 1, strongest to 5, weakest, of the Oxford Centre for Evidence-Based Medicine Levels of Evidence – http://www.cebm.net). The use of Level 5 evidence (expert opinion without explicit critical appraisal) can potentially affect patient care and interpretation of epidemiological data, as stage and grade cancer registry data on bladder cancer can be misleading. Population-based studies comparing the 1973 and 2004 WHO classifications have shown conflicting results, with some studies showing an advantage of the 2004 system for inter-observer variability, whilst others have not [17]. The Royal College of Pathologists (RCPath) dataset on reporting of urinary bladder cancers recognised two inherent deficiencies in the 2004 system. The first related to the larger group of patients with tumours classified as high grade according to the 2004 WHO system, resulting in more aggressive intravesical therapy, which does not translate into net patient benefit. The second problem was the identification of the new category of PUNLMP recognised in the 2004 classification scheme. The identification of this entity has been problematic, with lack of good reproducibility for the diagnosis. In routine practice, discrepancies in the discrimination between PUNLMP and low-grade carcinomas are reportedly between 39% and 50% in a non-specialist setting [17].

The RCPath bladder cancer dataset observed that there was limited robust evidence available to support the introduction and use of the 2004 classification scheme. In contrast, the 1973 WHO classification has been repeatedly validated and it is incorporated into clinical practice. Nevertheless, the RCPath recommended that both classifications be reported in parallel and results compared through prospective audit of patient outcomes. This recommendation has seen the lack of widespread use of the 1998 and 2004 WHO/ISUP classification [18] and the present data confirms that the 1973 WHO classification remains in widespread use. Current problems in genitourinary pathology reporting were recently highlighted by Hansel et al. [19]. Discrepancies of 15–20% on pathology review are not uncommon. Other problem areas highlighted include grading differences between classification systems, and the increasing reporting of histological variants of TCC. In the present data for example, micro-papillary variant TCC did not appear, meaning only that it may not have been reported specifically due to the methods of data collection.

Issues in Cancer Registration

There are considerable differences amongst national cancer registries in collection of bladder cancer data due to issues in the classification, grading and pathological staging of bladder tumours. The rates of International Classification of Diseases (ICD) code C67 (T1–T4) disease are much more consistent than they were before 2000, so that consistency between registries has improved. Nevertheless, this has still led to unfavourable comparisons of outcomes and survival rates for bladder cancer in the UK compared with Europe and the USA, as survival for bladder cancer in the UK is based in T1–T4 only. Crow and Ritchie [20] reported results of a survey of pathological grading in 2003. They showed that after a change in coding definitions by the UKACR in 2000, that only four of the 11 regions were coding correctly following the UKACR 2000 Guidelines, and without exception the regions not following the protocol were miscoding pTa G1–3 bladder cancers as benign or of uncertain behaviour. Before 2003, none of the regions were coding pTa lesions or CIS with an invasive behaviour code. This situation has probably improved (personal communication with data managers at South West Public Health Observatory). Data from the NCDR for 2006–2010 showed no pTa G3 tumours registered as uncertain behaviour. However, roughly half of G1/G2 pTa were registered as CIS, indicating that the coding is still inconsistent. This is important because in the UK, as highlighted above, only invasive lesions are used to calculate the national incidence and survival values and hence were skewing UK national bladder cancer statistics.

Further, the incorporation of invasive carcinoma (T1 or T2) alongside PUNLMP, which represents a non-invasive lesion in some national cancer registries, leads to artificial variability in the geographic incidence of bladder cancer. This makes the interpretation of any geographical patterns and international comparisons in bladder cancers difficult. For example in 2008 estimated age-standardised bladder cancer incidence rates reported for UK men are well below the European Union (EU) average (18.5 vs 27.4 per 100 000 population respectively) and the rate for women in the UK is slightly lower (5.1 vs 5.6 per 100 000 population respectively) as shown in Fig. 2 [21]. The lower incidence in the UK in contrast to the EU average may reflect inappropriate incorporation of non-invasive (PUNLMP lesions) and invasive bladder cancers in European registries.

Figure 2.

European age-standardised incidence rates per 100 000, EU-27 countries (2008 data).

The difference in cancer registration practices between the UK and USA also affect incidence and survival statistics. The present data indicates that 48% of bladder tumours registered in the UK were in the pTa TCC or CIS categories (Table 1). Thus a significant proportion of bladder tumours may be excluded from the UK incidence, but included in the USA incidence values. This difference in the way the incidence is calculated will tend to falsely lower the UK incidence compared with that in the USA. The age-standardised incidences of bladder cancer in men and women in the UK and the USA in 1997 were respectively, 18 and 5, and 22 and 5 per 100 000 population, but the degree to which the higher value in the USA is explained by differences in registration practices is difficult to determine [20].

In conclusion, the BAUS cancer registry gives a unique insight into bladder cancer TNM stage, grade and tumour pathology at presentation. Information of this type is often not available in national cancer registries. This issue has long been recognised and recently addressed in England with the introduction of the Cancer Outcomes and Services Dataset (COSD), which will be the new national standard for reporting all cancer information in England from 2013. One of the explicit aims of the COSD is to improve TNM staging across all cancer sites. This will be achieved through cancer multi-disciplinary team meetings, where every new cancer diagnosis is attributed with a completed TNM staging.

Limitations of the present study include the incomplete data, which is to be expected in a multi-institutional study of this nature. Also central pathology review was not practically possible, which could affect outcomes of histopathological diagnosis. Nevertheless, it provides a real-life insight into uro-pathology reporting across a wide range of pathologists over a decade. Additional limitations include failure to present outcome data. Survival data in the BAUS database is poor and insufficient to draw valid conclusions. However, we can cross-reference our population with NCDR survival data using unique patient identifiers to obtain accurate survival data, and this may be a possible future study.

Conflict of Interest

None declared.

Abbreviations
CIS

carcinoma in situ

COSD

Cancer Outcomes and Services Dataset

EU

European Union

ISUP

International Society for Urological Pathology

NCDR

National Cancer Data Repository

(N)MIBC

(non-)muscle-invasive bladder cancer

PUNLMP

papillary urothelial neoplasm of a low malignant potential

RCPath

The Royal College of Pathologists

TURBT

transurethral resection of bladder tumour

UCB

urothelial carcinoma of the bladder

UKACR

United Kingdom Association of Cancer Registry

Ancillary