- To investigate the incidence of carcinoma in situ (CIS) in Australia and examine implications for its diagnosis and management, as CIS of the urinary bladder is a non-reportable disease in Australia.
Bladder carcinoma (BC) globally is the sixth most common cancer with increasing incidences and treatment costs seen in the USA, Japan and European countries during recent decades [1-3]. While the incidence of BC in Australia appears to have declined over the same timeframe, a paucity of data exists regarding carcinoma in situ (CIS), including its incidence. This is primarily due to the fact that bladder CIS is not a reportable malignancy in many countries including Australia, due to its classification. Furthermore, despite recent efforts, bladder CIS is not universally categorised together with other malignancies in all cancer registries, further limiting available data .
Bladder CIS (hereafter CIS) is a distinct entity, which is defined as a flat (e.g. non-papillary) high-grade non-invasive urothelial carcinoma (TCC) . Therefore unlike testicular and prostatic ‘in situ’ disease, CIS is not a precursor to malignancy but an actual malignancy . Untreated, its natural history has a >50% 5-year progression rate, with even higher recurrence rates [5, 6]. As such, CIS remains a relatively poorly understood entity, yet it has a tendency to progress to invasive malignancy and should be treated aggressively with early intervention and strict follow-up.
Against this background, our aim was to investigate the incidence of CIS in Australia and examine implications for its diagnosis and management.
As CIS is not a reportable condition in Australia, there is no record of its incidence in cancer registries. To ascertain the incidence of CIS, annual hospital admissions data (for the whole of Australia) from the AIHW in the form of data cubes from 1993 to 2006 were examined, which showed an increase in the rates of CIS from 2001 onwards. The International Statistical Classification of Diseases and Related Health Problems, 10th Revision (ICD 10) codes sought included: CIS (D09.0); benign bladder neoplasms (D30.3) and neoplasms of uncertain or unknown behaviour of the bladder (D41.4) as the principal diagnosis code. The coding using ICD-10 has a lag period of at least 2 weeks from discharge of a patient, allowing the final diagnosis assigned for the admission to take into account cytological, histological, staging of disease and imaging results; coding is performed by trained administrative staff.
The AIHW hospital admissions data cubes consist of separations (patient counts) where one individual with multiple admissions is counted multiple times. Therefore to exclude any possible bias from multiple counting of the same patient (over-estimation), we investigated linked patient level data from Australia's most populous state, New South Wales (NSW) accounting for about one third of Australia's population . This was possible by accessing the NSW CHeReL from 2003 to 2008.
CHeREL is the entity responsible for health record linkage and is administered through the NSW Department of Health; the centre receives data for all admissions to public and private hospitals in NSW, including day surgery. CHeReL datasets have been widely used as a population-based linkage database and has been validated by several studies . As both the AIHW and CHeReL cover all eligible citizens (Australia and NSW, respectively) both are representative and can be generalised to the Australian population. Additionally, any diagnoses in the office setting would also probably be included as most interventions, e.g. cystoscopy, are performed as day surgery procedures. Ethics approval was obtained from the NSW Population and Health Services Research Ethics Committee.
The statistical analyses were carried out by the Clinical Research Design, IT and Statistical Support unit (CReDITSS) within the School of Medicine and Public Health at the University of Newcastle, Australia.
Surveying all types of bladder tumours in Australia, all the first admissions for CIS, benign neoplasm, neoplasm of unknown behaviour and BC to public and private hospitals in NSW were calculated for each financial year from 2003 to 2008.
The first diagnosis of bladder neoplasm was defined as the first record of a bladder neoplasm diagnosed, where principal diagnosis for the admission was taken as the diagnosis and when this was not cancer, the most severe cancer diagnosis was recorded according to the hierarchy of: malignant, CIS, neoplasm of unknown behaviour and benign bladder neoplasm. The latter conditions were included to observe for progression rates from one type of bladder neoplasm to another.
Age-standardised incidence rates of each bladder neoplasm type were calculated for the financial years 2003–2008, in 5-year age groups for both sexes (aged > 50 years) using the direct method of standardisation and the Australian population in 2006 (obtained from the Australian Bureau of Statistics). Patients having urothelial carcinoma with concomitant CIS were excluded. This was to ensure that the first presentation of CIS was identified and to observe its natural progression.
Although we obtained records from 2000/2001, we excluded a window period of the first 3 years to ensure that we accurately obtained the first recorded diagnosis of a bladder neoplasm. We acknowledge the asymmetry in exclusion that occurs over time using this approach. Therefore as a sensitivity analysis we included cancer diagnoses that occurred 3 years after the first recorded diagnosis. Results from the sensitivity analysis were similar to those reported in this study.
A linear regression of the natural logarithm of the rates vs year was used to test for trends over time. The estimate of the co-efficient for year in the model was used to determine the annual percentage change (APC) using the formula:
To ensure further reliability, we also compared the hospital admission data for ‘BC incidence ICD 10 (C67)’ with the Cancer registry data (Fig. 1a).
To identify if more neoplasms were being detected due to an increase in the rate of cystoscopy, we observed if any cystoscopic procedure was performed in the 12 months before the diagnosis of an individual's first neoplasm. To classify if a cystoscopy was used for the identification of a neoplasm or treatment, cystoscopies were separated into ‘diagnostic’ and ‘therapeutic’ groups. Hence any endoscopic procedure of the bladder (without destruction/resection/laser destruction), cystoscopy, biopsy and TURP were included as ‘diagnostic procedures’. Therapeutic cystoscopy was defined as endoscopic destruction/resection/laser destruction of a single or multiple lesions of the bladder or tissue of bladder. Any transurethral resection of bladder tumour conducted was counted as a treatment rather than a diagnostic procedure.
The proportion of individuals who had a procedure leading to the diagnosis of a neoplasm was calculated by the number who had the event during the12 months before detection of first neoplasm, divided by the total number of first neoplasms identified for that financial year. For male specific procedures/diseases (TURP) the proportions refer to males only. Some individuals may have had more than one of the procedures 12 months before detection hence the proportions do not add to one for any given financial year and subtype of cancer.
The time to therapeutic treatment from the highest priority cancer was determined (according to hierarchy as previously described). For example: if a patient had been first diagnosed as benign then progressed to CIS and BC, only the time from the diagnosis of ‘BC’ to the procedure was recorded for that patient. This approach was used to remove variation resulting from different classes of neoplasms that were prone to progression being considered the same as those that do not progress.
The cancer registry reported incidence for BC in Australia was 17.47 per 100 000 in males and 5.89 per 100 000 in females in the early 1980s and steadily declined in both sexes over time (Figure 1a). Although the absolute incidence rates differed using the CHeReL data linkage dataset, the trend in declining incidence rates of BC was similar in both males (31 per 100 000, APC – 6.56, P = 0.02) and females (11 per 100 000, APC – 5.99, P = 0.05) (Figure 1b).
Similarly, the trends in benign bladder neoplasm incidence rates of the AIHW separations database mirrored that of the CHeReL linkage database. Males averaged 1.4 incident cases per 100 000 and females 0.5 per 100 000 (Figure 2), showing concordance between the two databases.
Of the CHeReL linked data of 13 790 males and 5902 females analysed over the 5-year period, there were 1094 new diagnoses of primary CIS. There was an increasing trend for the incidence of CIS in both males (7 per 100 000) and females (2.3 per 100 000), which increased to 20.9 per 100 000 in males aged > 50 years (APC 10.79, P = 0.04) and 6.52 per 100 000 in females aged > 50 years (APC 14.15, P = 0.02) (Figure 3).
There was also evidence of an increasing trend for neoplasms of unknown behaviour diagnosis in males (23.3 per 100 000, APC 9.20, P = 0.09) but not in females (13.3 per 100 000, APC 0.76, P = 0.91) aged > 50 years (Figure 4).
The rates of all cancer types (BC, CIS, and neoplasms of unknown behaviour) increased with age in both males and females with a ratio of ≈3:1 (Figure 5).
Potential confounders to explain the increased incidence of CIS are increased instrumentation and/or an increase in associated conditions leading to incidental identification. Table 1 lists the diagnostic procedures and potential indications that might lead to incidental discovery of bladder neoplasms up to 1 year before its coding. The data indicate that there was no overall increase of any diagnostic procedures to suggest an increase in surveillance accounting for the increase in the CIS bladder rates. Most CIS (95.6%) were diagnosed cystoscopically, having either a biopsy (36.4%) and/or a more extensive therapeutic transurethral resection (79%) performed at the initial suspicion of the neoplasm. Few cases of CIS (8%) were diagnosed incidentally at the time of doing another endoscopic procedure, e.g. TURP.
|Category||% diagnosis of each neoplasm type by each method|
|Malignant neoplasm of bladder||19||19||19||18||19|
|Benign bladder neoplasm||40||43||33||44||46|
|Neoplasms of unknown behaviour||22||24||21||28||21|
|Malignant neoplasm of bladder||87||84||86||85||85|
|Benign bladder neoplasm||89||88||75||87||90|
|Neoplasms of unknown behaviour||89||88||94||90||89|
|Malignant neoplasm of bladder||9||8||6||6||6|
|Benign bladder neoplasm||20||12||27||16||23|
|Neoplasms of unknown behaviour||9||9||7||7||9|
In all, 79% of patients with CIS and 76% of BCs had therapeutic endoscopic procedures, usually conducted at the same time as discovery (median 0 days, mean 17.9 days and median 0 days, mean 24 days, respectively).
In all, 29% of the neoplasms initially diagnosed as CIS transformed to invasive disease at a median time of 160 days, while 15% of the neoplasms coded as neoplasm of unknown behaviour also became invasive (Table 2). A similar percentage of those with BC (12%) were also found to have CIS, at a median follow-up of 208 days.
|Initial||Subsequent||N (%)||Median time, days|
|Benign bladder neoplasm||Malignant neoplasm of bladder||11 (3.5)||91|
|Neoplasms of unknown behaviour||8 (2.6)||196|
|Malignant neoplasm of bladder||CIS||755 (12)||208|
|CIS||Malignant neoplasm of bladder||417 (29)||160|
|Neoplasms of unknown behaviour||Malignant neoplasm of bladder||358 (15)||111|
The main finding of the present study is that the incidence of CIS is increasing steadily in Australia, particularly since 2001 (Figure 3a). While it is possible that the CHeReL could overestimate the rates of CIS in NSW, it is likely that CHeRel rates are probably a better reflection of the burden of the disease in Australia, as AIHW data includes only CIS when coding for the principal diagnosis leading to the current admission, while CHeReL data accounts for the first presentation of CIS in all positions of coding in the current admission. In keeping with the present results Luke et al.  also reported an increasing trend of combined CIS and invasive BC rates in a South Australian population-based study. Presentation of isolated CIS can be very variable and diagnosis can be challenging. Most of these patients present with irritative bladder symptoms only [10-12], with 26% of patients being asymptomatic at diagnosis  and less than half have macroscopic or microscopic haematuria . Hence an increase in the incidence of CIS, as seen in the present study, could go unnoticed, particularly as it is not a reportable condition due to its current classification as an ‘in situ’ lesion.
Furthermore, the increase in CIS is unlikely to be due to increased awareness with more cystoscopic procedures being performed, as the proportion of lesions detected after bladder instrumentation has remained constant over the last few years. However, the changes could represent a change in the pattern of disease.
Although there is a decrease in the rates of BC in Australia (Fig. 1a), NSW Cancer Registry data show a sharp increase in the trend of locally advanced (T2/T3a) and aggressive (T3b/T4) BC at diagnosis, predominantly since 2002 . These rates almost doubled and between 2004 and 2008, with muscle invasive (T2) or higher stage tumours accounting for 39% of all known new bladder diagnoses , corresponding to the present findings in CIS rates. While speculative, one possible explanation is that CIS may progress to higher stage tumours directly and since T1 tumours form a preponderance of the tumour bulk, the effects of the rising trend in CIS may not be noticed until a dramatic increase in the numbers of higher stage tumours occurs. This would ultimately translate to high morbidity and mortality rates and increases in healthcare costs .
Intravesical instillation of BCG is the mainstay treatment for CIS therapy [15-17]. National Australian healthcare data (Medicare Australia) indicate an increase of 148% in the average rates of intravesical BCG (item 1131M) use from 2001–2005 when compared with the previous 5-year average, supporting the present study findings in relation to CIS that could also be due to a change in the practice advocating more BCG use . The untreated natural history of CIS typically shows a 50% progression rate to invasive disease at 5 years [5, 6], while the present results showed that even by 160 days, 29% of CIS had transformed into BC, despite high rates of initial treatment. These increasing trends in CIS should be identified, highlighted and treated early to avoid, where possible, progression to invasive disease.
Cystoscopy and biopsy is currently the method of choice for diagnosis of CIS, where 96 % of CIS diagnoses in the present study had been detected by cystoscopy and 79% had a ‘therapeutic’ procedure at the time of diagnosis (Table 1). As the diagnosis of CIS by cystoscopy can be challenging, sampling of even the normal mucosa is advocated in the high-risk patient or in the presence of positive cytology [12, 16]. While additional tools, e.g. nuclear matrix protein (NMP)-22 test and fluorescent in situ hybridization (FISH) [12, 19, 20], are available to aid diagnosis of CIS, they are not widely used in Australia. Despite the European Urology Association's recommendations for the use of fluorescent-light cystoscopy in detecting CIS, when equipment is available [16, 21], fluorescent-light cystoscopy is not currently approved by authorities in Australia.
A limitation of the present study is inter-observer variability when coding neoplasms and the procedures, as well as reporting the pathology. In an attempt to overcome this bias, we examined and compared different databases and also individual patient level linkage data. Although, the breakdown of grading and staging data (i.e. muscle involvement) would have added further validity to the present study, these data were unobtainable due to the non-reportable nature of CIS. Detection bias due to improvements in technology leading to better visualisation could also partially account for increased rates of CIS over time. Under-sampling of the lesion and re-staging at re-resection may also contribute to the results. Furthermore, we assumed that the NSW population being the largest state in the CHeReL database was representative of data for the whole of Australia. The lack of a code for cytology in the ICD-10 coding may have also been a limitation. However, despite such limitations this is the highest order available on CIS at present and will hopefully provoke further research and debate about this important topic.
In conclusion, national data (AIHW) appears to underestimate the incidence of CIS in the Australian population. Estimates derived from patient level data suggest the rates are increasing in Australia in both sexes and progression of disease occurring, despite apparently high treatment rates with BCG at the time of diagnosis.
CIS transformation into higher stage BCs may explain the parallel increase in such tumours, which has associated morbidity and economic implications. Closer surveillance and awareness of these high rates of CIS warrants further study and we recommend that CIS be considered a reportable disease.
We would like to thank Patrick McElduff of the School of Medicine and Public Health, University of Newcastle, Newcastle, NSW, Australia, Mr Leyshon Griffiths, Senior Lecturer/Honorary Consultant Urological Surgeon of the Leicester General Hospital, UK and Mr Hugh Mostafid, Consultant Urologist of the North Hampshire Hospital, UK, for all their support in the initial design of this project and A/Prof. Rod O'Connor, Research Project Manager CHeReL, NSW and Mr Allan Went, Acting Director, Demand and Performance Evaluation Branch, NSW Department of Health for their support in data acquisition.
Australian Institute of Health and Welfare
annual percentage change
Centre for Health record linkage
carcinoma in situ