Cytogenetic Analysis of Multifocal Bladder Cancer Supports a Monoclonal Origin and Intraepithelial Spread of Tumor Cells¹

About one-third of all bladder cancers occur as a multifocal disease forming several tumors simultaneously at different sites of the bladder wall. During the last years, different concepts have been assumed to explain this phenomenon. One is the field defect hypothesis,according to which individual cells of the bladder urothelium are primed to undergo transformation because of environmental mutagens,consequently leading to the development of independent multiclonal tumors. This theory is mainly based on morphological and immunohistochemical mapping studies demonstrating areas of modified cells adjacent to the sites of the tumors (1, 2). In contrast, genetic studies have given evidence for a monoclonal origin. Sidransky et al. (3) showed that all tumors of a single patient with multifocal disease revealed the same X chromosome inactivation pattern. Additionally, mostly the same patterns of loss of heterozygosity (4) as well as identical TP53mutations (5) were detected. Most investigators,therefore, concluded that an intraluminal seeding or intraepithelial migration of cells originating from a single primary tumor might be responsible for multifocal tumor occurrence.

Most of the previous studies on multiple bladder cancer have been focused on few specific genetic changes (4, 6, 7, 8, 9, 10). However, bladder cancer is characterized by highly complex chromosomal changes affecting numerous chromosomal loci. Recently, a cytogenetic study provided a more complex overview about chromosomal changes in six cases of multifocal bladder cancer, showing a highly similar pattern of genetic changes in distinct tumors (11). However, a systematic analysis of both multifocal tumors and the surrounding urothelium to find clues for the mechanism of multifocal tumor development has not been performed yet.

We analyzed a set of 32 multifocal bladder carcinomas and 52 tissue samples from macroscopically uninvolved urothelium originating from six cystectomy specimens. In our study, we used CGH³that detects all DNA gains and losses present in a tissue sample to search for typical patterns of chromosomal aberrations in multiple bladder cancer. Our data not only provide further evidence for a monoclonal origin of multifocal bladder cancer but also point at intraepithelial migration of tumor cells carrying specific chromosomal aberrations.

Cystectomy specimens of six bladder cancer patients (five males and one female) with multifocal disease were investigated. Samples were taken from 32 different tumors of the bladder specimens containing three,four, five, five, six, and nine tumors, respectively. Additionally, a total of 52 tissue samples were taken from sites of macroscopically uninvolved urothelium surrounding the tumors as well as from distant sites. The histological classification of each sample is given in Table 1.

Genomic DNA of the 32 unfixed tumor samples was prepared following standard DNA extraction procedures. If necessary, laser microdissection(PALM) was carried out on 10-μm sections to assure a tumor cell content of at least 80%.

All 32 tumors were investigated by CGH. CGH analysis and the criteria for the evaluation of copy number changes have been described elsewhere(12, 13). CGH profile shifts were rated as gains and losses if they at least reached the 1.25 and 0.75 thresholds. The Cytovision 3.1 software package (Applied Imaging International, Ltd.)was used for digital image analysis and subsequent karyotyping.

Sections (4 μm) of all formalin-fixed tissue samples of the 32 tumors and the 52 mucosa samples were investigated for TP53accumulation by immunostaining. The procedure and the antibodies used have been described elsewhere (14). Scoring was performed to the following criteria: − (negative), cells completely lacking nuclear staining; (+) (very weak), weak nuclear staining in widely scattered and <10% of cells; + (weak), weak nuclear staining or strong immunostaining in widely scattered and <10% of cells; ++ (moderate), strong immunostaining in 10–50% of cells or moderate staining in >50% of cells; +++ (strong), strong immunostaining in >50% of cells.

All tumors of cases 3227, 3253, and 3312 (n = 15) were analyzed for TP53 gene mutations because they showed at least moderate TP53 immunostaining (see “Results”). In addition, the TP53 IHC-positive mucosa sample IX of case 3253 was analyzed. Sequencing of the remaining IHC-positive mucosa samples was not performed because the content of IHC-positive cells was too small to allow the detection of a potential mutation. Mutation analysis was performed by solid phase sequencing of single-stranded PCR products from exons 5 to 8 of the TP53 gene, which are known to harbor ∼80% of all mutations (15). The procedure was performed as described before (16). Gel electrophoresis,data collection, and analysis was performed on an automated laser fluorescence sequencer (A.L.F.; Pharmacia).

CGH ratio profiles were generated from DNA samples of each macroscopically identified tumor. All tumors derived from the same patient shared a set of 7–13 identical alterations (hereafter referred to as “basic changes”). Additional genetic changes were found in individual tumors or subsets of the tumors. On average, all 32 tumors investigated showed 20.4 chromosomal changes (range, 6–33).

The most frequent change was a loss of chromosome 17p material that was present as a basic change in 32 of 32 tumors (100%) of the six bladder specimens. Gains involving 20p affected all tumors of four cases. Interestingly, particular aberrations appeared frequently in combination as basic changes, whereas others were seen in the course of progression only (Table 2).

Following the theory of an accumulation of genetic changes during tumor development and progression (17), we developed cytogenetic pedigrees reflecting the accumulation of chromosomal aberrations for each case of multifocal bladder cancer (Fig. 1). Starting from a hypothetical precursor cell population characterized by the set of basic changes (X₀ in Fig. 1),additional clones (X₁–X_n)were identified based upon the highest number of identical chromosomal aberrations in addition to the precursor clone (Fig. 2).

All 32 tumors and 39 of 52 mucosa samples were successfully analyzed by IHC. The results are given in detail in Table 1. A moderate or strong positive immunostaining [++ or +++ (positive)] was detected in individual samples of cases 3227, 3253, and 3312. In case 3227, only tumors A and B showed positive staining but none of the remaining samples. In cases 3253 and 3227, the positive samples were located in a contiguous area. In case 3253, this area included tumors A, B, and C as well as mucosa samples IX (dysplasia), XI (pT₁tumor), and XIV (carcinoma in situ; Fig. 3). In case 3312, positive staining was found in tumors B, C, D, E, and F and mucosa samples X (dysplasia), XI (normal urothelium), and XII(dysplasia; Fig. 4).

The findings of 17p deletions in all 32 tumors and positive TP53 immunostaining in 15 tumors investigated suggested an involvement of the TP53 tumor suppressor gene. Sequencing of TP53 of the IHC-positive tumors revealed point mutations in individual tumors of two bladder specimens. In case 3253, a transition from guanine to adenine at exon 8 codon 269 was found in tumors A, B,and C that resulted in the change of GAG (glutamic acid) to AAG(lysine). Tumor D showed the wild-type sequence. The TP53 IHC-positive mucosa sample (sample IX) revealed a mixed population of tumor cells carrying the known mutation and TP53 wild-type mucosa cells(Fig. 3). In case 3227, also a transition from guanine to adenine occurred but affected exon 5 codon 179. It was present in tumors A and B but not in tumors C, D, and E (data not shown). In both cases, the mutations resulted in a moderate or strong TP53 immunostaining (++ or+++). None of the tumors without mutations was IHC positive. The tumors of case 3227 did not show mutations within the analyzed exons.

In our present study, we demonstrated a close genetic relationship between all tumors of a particular cystectomy specimen in six cases of multifocal bladder cancer. Comparison of the CGH ratio profiles obtained from the tumors belonging to the same case enabled us to elaborate individual cytogenetic pedigrees portraying the accumulation of chromosomal aberrations during multiple bladder cancer progression.

In our sets of tumors, between 7 and 13 identical chromosomal aberrations were found in each case as well as identical TP53 mutations in two cases, indicating a clonal origin of the tumors. This is in concordance with previous studies that showed clonality by X chromosome inactivation patterns, genetically closely related or identical tumors, or TP53 mutations in patients with multiple or recurrent tumors (3, 7, 10, 11).

The sequence of genomic changes acquired during progression of the tumors was highly individual and complex in each case. It can be assumed that early aberrations frequently affect all tumors of a case. Potentially early changes in this study include alterations such as−17p, +20p, −9p, −9q, +2q34–qter, +12q14–q21, +1q22–q25,−8p22–pter, −5q31–qter, and +17q. Subsequent tumor progression may be characterized by the accumulation of changes like +11q14, −21q,−5q13–q14, +8q22, +10p, −10q22–qter, and −11p, which occurred only rarely in all tumors of a case. Alterations appearing in particular tumors only, such as +3q24–q26, +2p21–cen, +3p26, +5p13–p14,+6pter–p22, and −22q, are likely to be very late changes. Under this assumption, the order of chromosomal aberrations outlined in Table 2(from top to bottom) represents a hypothetical, unified sequence of frequent genetic changes acquired during evolution of multiple bladder cancer.

The majority of chromosomal changes detected in our study are commonly found in singular bladder cancer as well (12, 18, 19, 20, 21, 22). Some of them have been linked to special tumor properties, e.g., tumor initiation (−9p, −9q; Refs. 12, 20, and 23), papillary growth (−9q; Ref.21), high-grade or invasive phenotype (−8p, −11p, −17p;Refs. 12, 21, 22, and 24), or are suggested to represent late events (+5p, −5q; Refs. 21 and22). Numerous genes that are known to play a role for tumor development may be affected by these alterations, e.g., TP53 (17p13), MDM2 (12q13–q15),and PMS1 (2q34; apoptosis and maintenance of genomic stability), CDK4 (12q13–q15), CDK7 (2p15–cen), CDKN2A and CDKN2B (9p21; cell cycle regulation),or DBCCR1 at 9q34 (tumor initiation). However, the findings of 17p losses in all tumors of all bladder specimens, 20p gains in all tumors of four bladder specimens, and the high number of aberrations per tumor are strikingly different from previous reports. In this regard, it is important to note that the majority of tumors investigated were late-stage, advanced carcinomas, because otherwise no cystectomy specimens would have been available. This means that our set of tumors is not entirely representative for typical multiple bladder carcinomas that are usually noninvasive.

The average number of aberrations per tumor (20.4) was remarkably higher as reported for pT_2–4 (7.9 aberrations)or grade 3 carcinomas (7.8 aberrations) by means of CGH(21), suggesting an exceptional high degree of genomic instability in multifocal bladder cancer. However, it cannot be ruled out that the high number of chromosomal aberrations might be attributable to the preselection of extremely advanced tumors. Another point is that most of the previous CGH analyses have been performed on DNA obtained from formalin-fixed tissue samples. The direct comparison with CGH results from fresh tissue and may be problematic because shifts of the CGH profile are usually stronger in this case and might account for the larger number of profile shifts passing the thresholds. Nevertheless, it has been reported that mere reduction in TP53 levels may be sufficient to promote tumorigenesis (25). Loss of 17p that was present in all of our tumors may consequently account for a reduced activity of TP53 downstream mechanisms like induction of apoptosis, cell cycle arrest, and prevention of chromosomal instability, providing an alternative explanation for the high number of chromosomal defects.

However, the 100% rate of 17p losses must be interpreted with care because of the small number of cases. Nevertheless, it is remarkably higher than reported in non-multifocal bladder cancer by means of allelic loss (30–70%; Refs. 3 and 26, 27, 28, 29)or CGH (up to 28%; Refs. 12, 20, and 21) and suggests a possible involvement of the TP53 tumor suppressor gene. Yet, in our set of tumors, TP53 mutations were detected in two of six cases only. Consequently, loss of 17p might be attributable to advanced tumor stage and grade rather than reflecting a characteristic alteration in multiple bladder cancer. On the other hand, it can be hypothesized that TP53 may not be the target gene responsible for multiple tumor development. There might be one or more currently unknown genes located on 17p that may exert an influence on multifocality, e.g., by inhibiting cell migration capabilities in healthy urothelium. A similar effect may be induced by gains involving chromosome 20p that appeared as a basic change in four of six cases (66%). This finding was surprising, because in non-multifocal bladder cancer, this alteration has been described rather rarely (9%; Ref. 21), arguing for a certain role of genes located on 20p for multiple bladder cancer.

Interestingly, cytogenetically closely related tumors revealed a close spatial relationship in this study. This raises the question on the mechanism of tumor cell spread leading to multifocal bladder cancer. To address this aspect we applied IHC for the detection of cells accumulating the TP53 protein in tumors and normal urothelium located between the tumors.

Notably, weak immunostaining (+) was found in most tumor samples of all cases but did not indicate a TP53 gene mutation. Nuclear accumulation of TP53 can occur for several reasons besides intronic mutations, e.g., overexpression of wild-type TP53(30), interactions of TP53 protein with other intracellular or viral proteins (31, 32), or extensive posttranslational modification of wild-type TP53 (33). TP53 gene mutations were solely detected in samples showing a moderate or a strong (++ or +++) immunostaining in our study. For this reason, analysis of IHC data to search for TP53 mutated cells in the urothelium was restricted to those cases showing moderate or strong (++ or +++) immunostaining.

In both cases where TP53 IHC-positive (++ or +++) urothelium samples were found, these were located in contiguous areas as shown in Figs. 3 and 4. DNA sequencing of samples containing TP53 IHC-positive cells showed the same TP53 mutation in three tumors and in an adjacent mucosa sample of one of these cases. Together with the CGH findings indicating a common progenitor for the TP53 mutant tumors, these results give strong evidence for lateral migration of neoplastic cells throughout the urothelium.

Other studies had suggested intraluminal seeding of cells shed from a primary tumor as a probable mechanism for multiple clonal tumors. Shedding of tumor cells is a frequent phenomenon and can be used for the analysis of tumor cells in urine samples (34, 35). However, intraluminal seeding is likely to require complex regulation mechanisms; cell adhesion must be lost to allow cells to shed from the primary tumor but must be retained for the process of attachment and implantation into the mucosa at a different site(36). For subsequent invasion of the urothelium, cell adhesion has to be disabled again to allow cell migration. The spread of a neoplastic clone throughout the epithelium by mucosal migration may be less complex. Although it requires loss of cell adhesion in the onset of migration mechanisms as well, there is no need for switching between cell-cell adhesion and dispersion. In addition, cells are not forced to leave the epithelial environment and to survive in the urine. However, probably most important, the spread of a neoplastic clone by intraepithelial migration can be expected to lead to the formation of a single circumscribed area of transformed cells, whereas intraluminal seeding would cause a number of randomly distributed tumors. Consequently, the observation that TP53 mutant cells were restricted to a particular area adjacent to the tumors gives strong evidence for lateral migration as the underlying mechanism for multifocal cancer development.

Taking together the cytogenetic and immunohistochemical data, it can be hypothesized that the decision whether a neoplastic cell population becomes multifocal might depend on the order in which particular genetic defects are acquired. In non-multifocal cancer, a growth advantage with tumor formation may be the initial step, followed by genetic instability, invasion capability by lysis of the lamina propria, migration into the muscularis mucosae and blood vessels attributable to loss of cell adhesion, and finally, metastatic settlement. In contrast, multiple tumors might be characterized by early genetic instability and loss of cell adhesion, leading to the migration of neoplastic cells through wide areas of the urothelium. It can be expected that this process is driven by specific genetic changes, e.g., loss of 17p that might inactivate genes preventing a lateral spread of cells throughout the urothelium or maintain genetic stability. The close spatial relationship of tumors revealing identical genetic features, e.g., TP53mutations and patterns of chromosomal aberrations, as well as the detection of tumor cells within continuous areas of the urothelium reflect the migration of tumor cells of clonal origin throughout the bladder epithelium.

We thank Ulrike Neubert, Lydia Grote, Petra Meyer, Frauke Schmidt, and Petra Fischer for excellent technical support. We are grateful to Dr. Guido Sauter (Institute of Pathology, University of Basel, Basel, Switzerland) for helpful discussions.