Flat urothelial hyperplasia (FUH), which is defined as markedly thickened mucosa without cytological atypia, is frequently found adjacent to papillary urothelial tumours. For many years, this lesion has been regarded non-neoplastic without malignant potential, although clinicopathological studies documented an association between hyperplasia and papillary neoplasia.1 Genetic analysis on FUH and concomitant papillary tumours in bladder cancer patients revealed that many hyperplasias showed the same genetic aberrations as the papillary tumour, suggesting that FUH could be the first neoplastic step in the development of urothelial carcinoma.2, 3 Chromosome 9 alterations, especially loss on 9q, are the most frequent genetic events in bladder tumours, and can be found in histologically normal urothelium.2, 4, 5, 6, 7 It has therefore been suggested that loss of 9q may be the earliest event in the initiation of bladder cancer.8, 9 Deletions at chromosome arm 8p are also frequent but predominantly associated with advanced papillary bladder tumours.10
Mutations in the Fibroblast Growth Factor Receptor 3 (FGFR3) gene, which occur in the germ line of patients with skeletal dysplasias, are present in about 50% of bladder tumours. The majority of these tumours are of low stage and grade, and the presence of an FGFR3 mutation correlates with a favourable clinical outcome; that is, significantly fewer patients show progression and disease-specific mortality.11FGFR3 mutations were also detected in 75% of urothelial papillomas and are the only genetic events reported to date in these benign lesions.12, 13 Since, mostly tumours of low stage and grade present with FGFR3 mutations, FGFR3 mutations are considered early genetic events in bladder tumorigenesis.14, 15
The aim of the present study was to analyse FUH for FGFR3 mutations and loss of heterozygosity (LOH) of chromosomes 9 and 8 to investigate these early events in preneoplastic lesions. Since hyperplasia might be the first morphologically identifiable lesion in the development of papillary bladder tumours, we hypothesised that molecular analyses of these lesions could identify early genetic alterations in bladder cancer development.
Patient material was obtained from a clinical trial assessing the photodynamic diagnosis of neoplastic urothelial lesions.16 All patients gave written informed consent for this study. Biopsies were taken during cystoscopy after instillation of 5-ALA into the bladder from both fluorescence-positive, but otherwise inconspicuous areas, and fluorescence-negative papillary tumours. The biopsies were immediately snap-frozen in the operating room. Serial frozen sections stained with haematoxylin and eosin were used for histological diagnosis. Grading was performed according to the WHO classification17 and staging according to the Union Internationale Contre le Cancer (UICC).18 FUH was diagnosed if there was a markedly thickened mucosa (at least 10 layers of cells) without any cytological atypia in serial sections. Surface umbrella cells were preserved. Cases with significant inflammatory infiltrate and edema in the adjacent stroma were excluded as well as lesions with urothelial atypia.2, 3
Microdissection and DNA isolation of the urothelial lesions was performed as described previously.3 The microdissected samples contained at least 90% urothelial cells. Normal DNA as a reference for LOH analyses was isolated from EDTA blood, or from microdissected stromal or muscle tissue from the urinary bladder.
Whole genome amplification using I-PEP-PCR was performed for all samples to obtain sufficient amounts of DNA.3, 19 For all lesions, at least 500 cells were microdissected to avoid preferential amplification of one allele in LOH analysis because of insufficient amounts of DNA.
After preamplification, exons 5–9 of the TP53 gene were directly sequenced using single exon amplification and subsequent cycle sequencing on an ABI 373 sequencer as described previously.20
Analysis of the FGFR3 gene for mutations was based on the ABI PRISM SNaPshot Multiplex Kit (Applied Biosystems, Foster City, CA), and performed as described before.21 In short, 3 regions of interest comprising 9 mutations were amplified in one multiplex polymerase chain reaction, followed by extension of primers for each mutation with a labelled dideoxynucleotide. Extended primers were separated by capillary electrophoresis, and the presence or absence of a mutation was indicated by the incorporated nucleotide.
LOH analysis was performed using 2 microsatellite markers at chromosome arm 8p (D8S1817, D8S1145), 2 markers at chromosome arm 8q (D8S587, D8S591), 2 markers at chromosome arm 9p (D9S304, pky11) and 3 at chromosome arm 9q (D9S303, D9S747, D9S1113) as described previously.22 Informative cases were scored as allelic losses when the intensity of a signal for a tumour allele had decreased to at least 50% relative to the matched normal allele. All cases of allelic loss (LOH) were confirmed at least once.
Thirty lesions diagnosed as FUH and 9 simultaneous or consecutive papillary carcinomas from 24 patients were investigated in this study (additional clinicopathological data are detailed in Table I). Good results were obtained for all lesions, however repeated LOH analyses failed in some hyperplasias (8% of the samples could not be analysed (NA), see Fig. 1).
|Time point||Localisation||Stage/grade||Time point||Localisation|
|2||M37||S||LW right||pTaHG||S||LW left|
|3||M68||S||LW right||pTaLG||S||LW right|
|20||M60||S||LW left||pT2HG||S||LW left|
|21||M62||S||LW left||pT1HG||S||LW left|
Microsatellite analysis showed LOH in 12 of 30 hyperplasias (40%) and in 6 of 9 tumours (67%, Fig. 1). Most of the loss was observed at chromosome arm 9q (6/30 hyperplasias and 4/9 tumours), but LOH was also found on 9p (5/30 hyperplasias and 1/9 tumours) and 8p (3/30 hyperplasias and 2/9 tumours). Loss on 8q was not observed.
FGFR3 mutation analysis showed that 23% (7/30) of hyperplasias had an FGFR3 mutation (Fig. 1). Of the 9 papillary carcinomas, 2 had an FGFR3 mutation (22%, cases 3 and 7), of which one corresponded with the mutation in the concurrent hyperplasia. In the other case, the tumour was mutant and the concomitant hyperplasia wild type for FGFR3. Both tumours were of low stage and grade (Table I). None of the high grade or invasive tumours carried an FGFR3 mutation. On the other hand, 2 high-grade tumours (cases 14 and 20) showed a mutation in TP53 after analysis of all tumours (Fig. 1). This is in accordance to previous studies that demonstrated the mutual exclusiveness of FGFR3 and p53 alterations, and their role in the different molecular pathways of disease pathogenesis in bladder cancer.23, 24
When the results of LOH and FGFR3 mutation analysis for the hyperplasias were combined, 2 hyperplasia samples of 2 patients (cases 9 and 16a) displayed an FGFR3 mutation, without having deletions at chromosomes 9 or 8. These deletions, however, may be missed because of the limited set of markers we used. Alternatively, samples could be contaminated with normal tissue in spite of microdissection, and this may influence the results of LOH analysis, whereas FGFR3 mutation analysis is very robust in a large background of non-tumour DNA. Conversely, 7 hyperplasia samples of 6 patients with deletions at chromosomes 9 or 8 (cases 2, 3, 12, 14, 17, 19a and 19b) were FGFR3 wild type. Six of the 7 samples had deletions at chromosome 9. This means that 55% (6/11) of hyperplasias with LOH at chromosome 9 lacked FGFR3 mutations, while 71% (5/7) of mutated hyperplasias already had LOH (Table II, p = 0.03). This is in concordance with a previous study by Van Tilborg et al. in which we showed that FGFR3 mutations occur after loss of chromosome 9 in evolutionary genetic trees of patients with multiple tumour recurrences.25 Urothelial papillomas may also be regarded as precursors of low-grade urothelial cancer12 and the percentage of FGFR3 mutations in these lesions is high. Interestingly, so far no LOH in papillomas has been observed, although only a limited number of samples have been investigated.13 This suggests that mutations in FGFR3 may play a leading role in the formation of papillomas, whereas in this study we show that chromosome 9 deletions are more frequent in FUH. An explanation for this observation could be that FGFR3 mutations are associated with neoplasias with a papillary growth pattern, while chromosome 9 deletions might be important for all urothelial neoplasias. We previously showed that chromosome 9 deletions occur both in hyperplasias3 and in dysplasias and CIS.20
|LOH 9 (yes)||5||6||11|
|LOH 9 (no)||2||17||19|
In summary, FGFR3 mutations were detected in FUHs of patients with bladder tumours, further supporting the role of this lesion as a precursor of papillary bladder carcinoma. Chromosome 9 deletions may occur earlier than FGFR3 mutations in most cases and still remain the earliest known genetic alterations in bladder cancer.