Precise microdissection of human bladder carcinomas reveals divergent tumor subclones in the same tumor

Authors


Abstract

BACKGROUND

Human bladder carcinoma is thought to arise from a field change that affects the entire urothelium. Whether independently transformed urothelial cell populations exist in the same patient is uncertain.

METHODS

We studied the clonality of urinary bladder carcinoma in 18 female patients who underwent cystectomy for urothelial carcinoma. None had multiple tumors. Tumor samples were obtained from different areas of the same tumor. Sixty-seven tumor samples were analyzed. Tumor genomic DNA was microdissected and extracted from formalin-fixed, paraffin-embedded slides. The clonality of urothelial tumors was evaluated on the basis of a polymorphism of the X chromosome-linked human androgen receptor gene (HUMARA) locus. The technique is dependent on digestion of DNA with the methylation-sensitive restriction enzyme HhaI, polymerase chain reaction (PCR) amplification of HUMARA locus, and detection of methylation of this locus. With this method, only the methylated HUMARA allele is selectively amplified by PCR.

RESULTS

Eleven of 18 patients were informative. Nonrandom inactivation of the X chromosome was found in 9 of the 11 informative patients (82%). Seven patients showed different patterns of nonrandom X chromosome inactivation for tumor samples obtained from different regions of the same tumor. Two patients showed the same pattern of nonrandom X chromosome inactivation in all samples.

CONCLUSIONS

Some muscle-invasive urothelial carcinomas may arise from independently transformed progenitor urothelial cells, supporting the “field effect” theory for bladder carcinogenesis. Cancer 2002;94:104–10. © 2002 American Cancer Society.

Urothelial carcinomas are the most common carcinomas of the urinary bladder in the western world, and affect both men and women.1 Approximately one third of bladder carcinomas present as multifocal lesions at the time of diagnosis.2 More than 60% of urothelial carcinomas are superficial lesions confined to the mucosa or submucosa of the bladder at presentation. However, even after complete transurethral resection of all visible lesions, patients often develop new tumors, and 10-25% of patients later develop muscle-invasive urothelial carcinomas.3–5 This high frequency of synchronous and metachronous multifocality has provided support for the concept of a neoplastic “field effect” in the bladder that promotes independent transformation of multiple urothelial cells.6–12

Recent molecular and clinical data suggest that invasive bladder carcinomas may develop by different molecular pathways, arising de novo, or originating from either superficial papillary tumors or carcinoma in situ.13–15 Using tissue microdissection and X chromosome inactivation analysis, Tsai et al. identified macroscopic urothelial patches of monoclonality.16 Alternating patterns of X chromosome inactivation were observed in the same patient, thus providing molecular evidence for the “field effect” in bladder carcinogenesis.16 The authors hypothesized that multiple tumors may arise independently from these predisposed areas (“patches”).16, 17 Muto et al. showed that chromosomal deletions and methylation of the promoter regions of putative tumor suppressor genes occur in morphologically normal urothelium from patients with bladder carcinoma, supporting the “field effect” theory during bladder carcinogenesis. Using X chromosome inactivation and cytogenetic techniques, several studies showed that some human cancers may harbor separate distinct subclones from the same tumor.18, 19 Independently transformed cells may participate in the formation of grossly apparent macroscopic tumors.20

Whether independently transformed urothelial cell populations exist in the same patient is uncertain. Sidransky et al. studied X chromosome inactivation pattern in four female patients with multiple tumors.21 Of three informative patients, all showed the same pattern of X chromosome inactivation, supporting the monoclonal origin of bladder carcinoma.21 Other molecular data suggested the occurrence of independent clones and oligoclonality in some bladder carcinoma patients.15, 22–27 These studies were limited by the small sample size, lack of precise microdissection, and selection of clonal markers (p53 point mutations and/or loss of heterozygosity). In the current report, we sought to address the clonal process of carcinogenesis by analyzing the pattern of X chromosome inactivation in tumor components microdissected from different areas of muscle-invasive urothelial carcinoma from 18 female patients.

MATERIALS AND METHODS

Patients

Eighteen women, diagnosed with urothelial carcinoma of the bladder, underwent radical cystectomy at Indiana University Hospitals and University Hospitals of Cleveland from 1995 to 1998. All samples were procured after obtaining signed informed consent in accordance with the Institutional Committee for the Protection of Human Subjects. Muscle-invasive tumors were identified in all cases (pathologic stage T2 or greater). Histologic grading was performed according to the newly proposed World Health Organization and International Society of Urological Pathology grading system,28, 29 and the tumors were high-grade in all cases. None of the patients had multiple tumors.

Tumor Samples and Microdissection

Histologic sections were prepared from formalin-fixed, paraffin-embedded blocks and stained with hematoxylin and eosin for histopathologic review and the X chromosome inactivation analysis. Sixty-seven tumor samples were obtained from 18 female patients. Tumors in different compartments were microdissected respectively from surrounding normal tissue using a sterile 28-gauge needle (Fig. 1). Approximately 400-600 cells were microdissected from 5 μm histologic sections under direct light microscopic visualization (BH2 Olympus, Tokyo, Japan) as previously described.30, 31 Each tumor was separated by at least 5 mm from any others or sampled from different blocks. Normal urothelial tissues microdissected from the cystectomy specimens were used as control samples for analysis of X chromosome inactivation.

Figure 1.

Precise microdissection of bladder carcinoma from different areas of the same tumor (Patient 6). A) tumor before microdissection; B) after microdissection.

Detection of X Chromosome Inactivation

The dissected tissue was placed in a 15 μL aqueous solution of 10 mM Tris, 1 mM EDTA, 1% Tween 20, and 0.2 mg/mL proteinase K (pH 8.3) and incubated overnight at 37 °C for DNA extraction.30, 31 Eight microliter aliquots of the DNA extract were digested overnight with 1U of HhaI restriction endonuclease (New England Biolabs Inc., Beverly, MA) in a total volume of 10 μL. Control reactions for each sample were incubated in the digestion buffer without HhaI endonuclease.

Primers used in this reaction were: AR-sense: 5′TCC AGA ATC TGT TCC AGA GCG TGC3′ and AR-antisense: 5′GCT GTG AAG GTT GCT GTT CCT CAT3′.32 Three μL of digested or non-digested DNA were amplified in a 25-μL polymerase chain reaction volume containing 0.1μL 32[P]α-labeled deoxyadenosine triphosphate (dATP) (3000 ci/mmol), 4 μM AR-sense primer, 4 μM AR-antisense primer, 4% dimethyl sulfoxide, 2.5 mM MgCl2, 300 μM deoxycytidine triphosphate, 300μM deoxythymidine triphosphate, 300 μM deoxyguanosine triphosphate, 300 μMdATP, and 0.13U AmpliTaq Gold DNA polymerase (Perkins-Elmer Cetus, Foster City, CA). Polymerase chain reaction (PCR) amplification was performed with an initial denaturation step of 95 °C for 8 minutes followed by 32 cycles as follows: 95 °C for 40 seconds, 63 °C for 40 seconds, 72 °C for 60 seconds, and the final extension step at 72 °C for 10 minutes. The PCR products were then diluted with 4 μL loading buffer containing 95% de-ionized formamide, 20 mM EDTA, 0.05% bromophenol, and 0.05% xylene cyanol FF (Sigma Chemical Co., St. Louis, MO). The samples were heated to 95 °C for 5 minutes, then placed on ice. Three microliters of this mixture were loaded onto 6.5% polyacrylamide denaturing gels without formamide. Electrophoresis was performed at 1600V for 4-7 hours, followed by autoradiography with Kodak X-OMAT AR film for 8-16 hours.

Analysis of X Chromosome Inactivation

The clonality of the samples was evaluated on the basis of a polymorphism of the X-linked human androgen receptor gene (HUMARA) locus.32, 33 The technique is dependent on digestion of DNA with the methylation-sensitive restriction enzyme HhaI, PCR amplification of the HUMARA locus, and detection of methylation of this locus. With this method, only the methylated HUMARA allele is selectively amplified by PCR. The random inactive status of X chromosomes is established in all female somatic cells early in embryogenesis.34–36 Normal female tissues should be a cellular mosaic, with an equal distribution of cells containing maternal or paternal-derived inactivated X chromosomes. Tumors arising from a single transformed progenitor cell should contain the same inactive X chromosome in each tumor cell. Different patterns of X chromosome inactivation are consistent with independent origin of the cells.

The cases were considered informative if the control sample displayed two alleles after PCR amplification without HhaI digestion. Nonrandom inactivation of X chromosomes was defined as a complete or nearly complete absence of one or the other allele after HhaI digestion, indicating predominance of one androgen receptor (AR) allele.30, 33 Definition of multiclonality, as described previously,30, 33 was presumed as alternate predominance of AR alleles after HhaI digestion.

RESULTS

Inactivation of the X chromosome at the AR locus was examined in 67 tumor samples microdissected from muscle-invasive urothelial carcinoma from 18 female patients. In 7 of 18 patients (Numbers 1, 2, 3, 4, 5, 7, and 18), the patterns of X chromosome inactivation were non-informative. Of the 11 informative patients, two (Numbers 10 and 17) showed random inactivation of X chromosomes and the other nine demonstrated nonrandom inactivation patterns of X chromosomes. Tumor samples from two cases (Numbers 13 and 16) showed the same pattern of nonrandom inactivation of X chromosomes; whereas tumors from seven cases (Numbers 6, 8, 9, 11, 12, 14, and 15) showed different patterns of nonrandom inactivation of X chromosomes in samples microdissected from different regions of the same tumor. Table 1 summarizes the findings.

Table 1. Results of X-Chromosome Inactivation Analysis in 18 Female Patients who Underwent Cystectomy for Urothelial Carcinoma of the Bladder
Case no.NT1T2T3T4T5
  1. N: normal tissue; T1, T2, T3, T4, and T5: tumor samples microdissected from different areas of the same tumor; NI: noninformative; ▴: upper allele inactivated; ▾: lower allele inactivated; •••: heterozygous.

1NININININI
2NINININI
3NININININI
4NINININI
5NININININI
6▴ ▾▴ ▾
7NININININI
8▴ ▾
9▴ ▾
10▴ ▾▴ ▾▴ ▾▴ ▾
11▴ ▾
12▴ ▾
13▴ ▾
14▴ ▾
15▴ ▾
16▴ ▾
17▴ ▾▴ ▾▴ ▾▴ ▾▴ ▾
18NININININI
Figure 2.

Analysis of X chromosome inactivation in urothelial carcinoma of the bladder at the androgen receptor locus. Tumors with the same allelic inactivation pattern were considered compatible with monoclonal origin; tumors with different allelic inactivation patterns were consistent with polyclonal origin. An alternative nonrandom pattern of X chromosome inactivation was shown in four tumor samples obtained from different areas of the same tumor (Patient 8). N is DNA from normal tissue as control sample; T1, T2, T3, T4 are tumors from different compartments of the same urothelial carcinoma; DNA was digested (+) or not digested (−) with Hha I prior to polymerase chain reaction. The amplified DNA fragments harbor HhaI sites and CAG polymorphic repeats.

DISCUSSION

The current study provides molecular evidence of intratumoral heterogeneity of invasive urinary bladder carcinoma. We analyze the patterns of X chromosome inactivation in 67 tumor samples obtained from 18 cystectomy specimens. In seven of 11 informative cases, tumors from different areas had alternating patterns of nonrandom inactivation of X chromosomes, suggesting that they arose independently from different transformed progenitor cells. Some bladder carcinomas may harbor separate tumor clones derived from different progenitor cells. Independent transformation of urothelial cells of the bladder may be explained by the “field effect” during bladder carcinogenesis.

Two cases in this study showed the same pattern of nonrandom inactivation of X chromosome in all tumor samples analyzed. Although this observation is compatible with monoclonal origin, it also is possible that they arose independently from two or more progenitor cells which, by chance, shared the same randomly inactivated X-allele, either paternal- or maternal-derived after lyonization.

Histologic, cytogenetic, and biologic heterogeneities of bladder carcinomas have been well documented.37–39, 40 Urothelial carcinoma of the bladder often presents as multifocal lesions, and excised tumors frequently recur, findings which cannot be easily explained by a monoclonal origin theory. “Field effect” theory, independent transformation of epithelial cells under the effect of epigenetic stimuli, has been proposed for the tumorigenesis of urinary bladder carcinoma.6–9 Miyao reported that the primary tumor showed allelic loss of chromosome 9, whereas a recurrent tumor excised 10 months later retained both chromosome 9 alleles.26 Since losses of chromosomal material are irreversible, it was possible that these two tumors evolved from different transformed progenitor cells.26 Spruck et al. reported allelic loss of chromosome 9 in invasive bladder tumor, which was absent in dysplastic lesions from the same patient.15 On the contrary, a mutation at codon 245 of the p53 gene was seen in dysplasia, but not in the invasive tumor.15 The authors suggested that urothelial carcinoma of the bladder arose either from a precursor neoplastic cell, which spread in the bladder to a new location prior to the occurrence of any of the tested alterations, or resulted from independent transforming events, which may explain the current data. Using fluorescence in situ hybridization and loss of heterozygosity analysis, Hartmann et al. showed the occurrence of two independent clones in some patients.27 In 5 of 10 patients with multiple bladder tumors, two or more clones were found.27

Sidransky et al. reported that three of four female patients with multiple bladder carcinomas had the same pattern of nonrandom inactivation of X chromosome.21 In contrast to the current study, each sample of tumor DNA was obtained from geographically separated tumors of the bladder, instead of different areas of a single tumor, and the assessment of X chromosome inactivation was conducted at the hypoxanthine phosphoribosyltransferase gene or phosphoglycerate kinase gene locus, instead of AR locus. It is possible that predominance of one neoplastic clone in a given tumor may yield a pattern of nonrandom X chromosome inactivation consistent with monoclonal origin. Although the current data support independent origin of tumor clones in the same patient, it is likely that some bladder carcinomas may represent a single clonal growth. Two cases in the current study showed the same pattern of X chromosome inactivation in multiple specimens sampled from the same tumor. Thus, the current data do not rule out the possibility that some bladder tumors arise from a single clone.

Recent p53 mutation studies of large patient series suggest that multifocal bladder tumors can be either mono- or polyclonal.22–25, 41 Yasunaga et al studied p53 gene mutations in 26 bladder carcinoma patients who were exposed to aromatic amines, and found that the patterns of p53 gene mutations differed in concurrent and metachronous lesions, supporting the multiclonal origin of the tumors.23 Different p53 mutations in separate urothelial lesions as well as metachronous samples obtained from patients in the Ukraine after the Chernobyl accident suggest multiple transformation events may occur after a strong carcinogenic exposure.24 All these patients were male and X chromosome inactivation analysis was not performed. In a series of 42 patients with multiple tumors, Goto et al. found discordant p53 mutations in multiple tumors from 11 of 22 patients who had p53 mutations.25 In an experimental animal model, Yamamoto showed that most geographically separated bladder tumors had different patterns of p53 mutations, and only in a single case were identical p53 mutations found in separate urinary bladder carcinomas, implicating divergent or independent pathways for the clonal development of urothelial carcinoma.22 Our findings that some bladder carcinomas may arise from independent transformation of multiple urothelial cells are consistent with the “field effect” during bladder carcinogenesis.

In summary, we have used an inactivation of X chromosome technique to determine the clonal origin of muscle-invasive urothelial carcinoma of the urinary bladder. The current data provide molecular evidence of intratumoral heterogeneity of bladder carcinoma. Some muscle-invasive urothelial carcinomas may arise from independently transformed progenitor urothelial cells, supporting the “field effect” theory for bladder carcinogenesis.

Ancillary