The 2004 WHO classification system identifies several forms of non-malignant and pre-malignant urothelial tumour [26].

Urothelial hyperplasia is defined as markedly thickened mucosa without cytological atypia. It is seen in the mucosa adjacent to low-grade papillary tumours. There is no direct evidence of any premalignant potential, although it may be clonally related to nearby tumours.

Urothelial dysplasia by contrast does involve cytological atypia, but falls short of the changes seen in carcinoma-in-situ. It is challenging to diagnose and must be distinguished from reactive atypia, which is an inflammatory finding that may be seen in association with stones, infection or instrumentation. Urothelial dysplasia is a potentially premalignant finding. Clinical information on the outcome of de novo primary dysplasia is limited. In one study, 7 of 36 (19 %) patients with primary dysplasia developed carcinoma [28]. A later study showed that after a mean follow-up of 3.9 years, 4 (15 %) developed biopsy-proven cancer [29]. Another retrospective study of 15 patients found that 15 % with primary urothelial dysplasia and a mean follow-up of 4.8 years developed carcinoma in situ (CIS) [30]. The presence of urothelial dysplasia in association with papillary tumours increases the risk of tumour progression [31].

Non-malignant papillary tumours include urothelial papillomas and inverted papillomas. They are not pre-malignant and have a very low risk of local recurrence [26].

The 2009 tumour, node, metastasis (TNM) classification is currently recommended for bladder cancer staging (Table 31.2) [38].

Histological categories Ta, T1, and Tis are regarded as superficial bladder cancers.

Sub-staging of T1 bladder tumours according to the depth of lamina propria invasion, using the muscularis mucosae as a reference point, has been proposed. T1a describes disease into the lamina propria but not invading the muscularis mucosae, T1b disease invades to the muscularis mucosae, and T1c beyond it. This sub-classification does not form part of the current TNM classification system, but has been shown to correlate with likelihood of progression to muscle invasion by Orsola and colleagues [39]. There was no difference in recurrence rates between the different sub-categories. Accurate T1 sub-staging represents a challenge for the pathologist, and was only possible in 87 % of cases in the study quoted above.

Some molecular alterations are found in all grades and stages of TCC, such as chromosome 9 deletions, which appear to be an early genetic event. However, there are well-defined differences in the molecular changes seen in low-grade non-invasive tumours and high-grade tumours including those with lamina propria or muscle invasion, and carcinoma in-situ (CIS) [40].

Mutations of the fibroblast growth factor receptor gene (FGFR3) on chromosome 4p and the TP53 gene on chromosome 17p are almost mutually exclusive [41]. FGFR3 mutations are seen in PUNLMPs and low-grade non-invasive urothelial carcinoma, whereas TP53 mutations are seen in high-grade and/or invasive tumours. These differences at the molecular level underline the quite different behaviour of high-grade tumours in terms of their potential for invasion, metastasis and consequent mortality. These molecular alterations are outlined in Fig. 31.5.

By contrast, mutations of the FGFR3 and Ras genes in low-grade tumours are absolutely mutually exclusive, but Ras mutations appear to have no prognostic significance [41].

It is likely that in the future, molecular profiling using high throughput assessment of gene and protein expression will play an increasingly important role in stratifying superficial bladder tumours according to their likelihood of progression, and may be used to determine treatment or follow-up pathways.

Two principal theories have been proposed to explain the frequent multifocality of bladder tumours. The field-change, or oligoclonal, theory suggests that multifocal tumours arise due to individual transformation of areas of unstable or dysplastic urothelium into clonally unique tumours [42]. This theory is supported by the observation that tumours may recur months or years after treatment of the primary lesion.

There is some molecular evidence however, to suggest that in at least some cases recurrent or concurrent tumours may arise due to implantation or transepithelial spread of tumour cells, which are of a common clonal origin to the primary tumour [43].

The success of single instillations of intravesical chemotherapy after transurethral resection of superficial bladder tumours in preventing tumour recurrence has been attributed to prevention of implantation of tumour cells, which would lend further support to the theory of monoclonal implantation.

Cystoscopy remains the gold standard for the detection of both new and recurrent bladder cancer. Despite this, its sensitivity is limited to around 93 % for the detection of tumours during follow-up [46], and may be as low as 60 % for the detection of CIS [47].

Urine cytology was first described by Papanicolau and Marshall in 1945 [48]. When malignant cells are detected, it carries a high specificity for underlying urothelial malignancy (>95 %). However, its sensitivity is limited to 30–50 % mainly due to the fact that low-grade tumours are often not associated with malignant cytology. The accuracy of urine cytology is operator-dependent, with more accurate results from laboratories processing larger numbers of samples [49].

There are a variety of urinary biomarkers available or being investigated for use in the detection and/or monitoring of urothelial cancer. In general, they possess greater sensitivity compared to cytology, at the expense of lower specificity. None, however, have sufficient sensitivity to obviate the need to perform cystoscopy [50].

Currently available point-of-care biomarker tests include Nuclear Matrix Protein 22 (NMP22) and Bladder Tumour Antigen (BTA Stat). Laboratory-based tests include BTA TRAK, ImmunoCyt and UroVysion.

Despite the demonstrably higher sensitivity of these tests compared to urinary cytology, their usefulness is limited by their reduced specificity. The concern regarding their use as a diagnostic adjunct in patients presenting with haematuria instead of cytology is that the increased false positive rate would lead to unnecessary anxiety and further investigations in patients with a positive biomarker assay but negative imaging and cystoscopy.

Other potential uses for these tests include for population-based screening, for which there is currently no evidence base, and as part of surveillance protocols for follow-up of patients after treatment for low- or intermediate-risk superficial bladder cancer (see section “Risk stratification”). It is proposed that the use of biomarker assays may reduce the frequency with which these patients require follow-up cystoscopy. It is this last application that looks most promising, but currently large prospective trials validating such protocols are lacking.

The European Organisation for Research and Treatment of Cancer (EORTC) genitourinary (GU) group has devised a scoring system to predict the risk of recurrence and progression in the short and long term based on six clinicopathological variables (Table 31.3). These predictive variables were identified via a multivariate analysis of data from 2,596 patients with Ta or T1 bladder tumours, enrolled in 7 EORTC-GU trials [34]. The variables are weighted differently according to their influence on recurrence and progression to produce recurrence and progression scores (Table 31.4). These scores can then be used to stratify patients into low-, intermediate-, and high-risk categories for both recurrence and progression (Table 31.5). Criticisms of the EORTC tables include the possible bias inherent in using patients enrolled in trials, who may have more intensive follow-up than non-trial patients, and the fact that the cohort includes few patients with CIS, and few patients treated with BCG. Those patients that did receive BCG did not receive a maintenance regime and repeat transurethral resection (TUR) was not standard practice for high-grade T1 tumours. The Club Urológico Español de Tratamiento Oncológico (CUETO) group from Spain evaluated the EORTC tables in a cohort of 1,062 patients with superficial bladder cancer treated with BCG and found that they overestimated the risk of recurrence and progression in that group [55]. Despite these caveats, the EORTC tables have been validated in an external patient cohort with long-term follow-up [56], and are the risk stratification tool recommended by the European Association of Urology (EAU) [57].