Clonal origin of lymph node metastases in bladder carcinoma

Authors

  • Timothy D. Jones M.D.,

    1. Department of Pathology and Laboratory Medicine, Indiana University School of Medicine, Indianapolis, Indiana
    Search for more papers by this author
  • Matthew D. Carr M.D.,

    1. Department of Pathology and Laboratory Medicine, Indiana University School of Medicine, Indianapolis, Indiana
    Search for more papers by this author
  • John N. Eble M.D.,

    1. Department of Pathology and Laboratory Medicine, Indiana University School of Medicine, Indianapolis, Indiana
    Search for more papers by this author
  • Mingsheng Wang M.D.,

    1. Department of Pathology and Laboratory Medicine, Indiana University School of Medicine, Indianapolis, Indiana
    Search for more papers by this author
  • Antonio Lopez-Beltran M.D.,

    1. Department of Pathology, Cordoba University, Cordoba, Spain
    Search for more papers by this author
  • Liang Cheng M.D.

    Corresponding author
    1. Department of Pathology and Laboratory Medicine, Indiana University School of Medicine, Indianapolis, Indiana
    2. Department of Urology, Indiana University School of Medicine, Indianapolis, Indiana
    • Department of Pathology and Laboratory Medicine, Indiana University Medical Center, University Hospital 3465, 550 North University Boulevard, Indianapolis, IN 46202
    Search for more papers by this author
    • Fax: (317) 274-5346


Abstract

BACKGROUND

Evidence of genetic heterogeneity within urothelial carcinomas of the bladder has raised questions about the clonal origin of urothelial carcinoma and its metastases. High-grade urothelial carcinoma of the bladder frequently metastasizes to multiple regional lymph nodes in the pelvis. Whether or not these multiple lymph node metastases originate from the same tumor clone is uncertain. Molecular analysis of microsatellite alterations and X-chromosome inactivation status of distinct tumor cell populations from the same patient may further our understanding of the genetic basis of carcinoma progression and metastasis.

METHODS

The authors examined 24 patients who underwent radical cystectomy for urothelial carcinoma. All patients had multiple (from two to four) lymph node metastases. Genomic DNA samples were prepared from formalin fixed, paraffin embedded tissue sections using laser-assisted microdissection. Loss of heterozygosity (LOH) assays for 3 microsatellite polymorphic markers on chromosome 9p21 (D9S171, region of putative tumor suppressor gene p16), 9q32 (D9S177, putative tumor suppressor gene involved in urothelial carcinoma tumorigenesis), and 17p13 (TP53, the p53 locus) were performed. In addition, X-chromosome inactivation analysis was performed in primary tumors and metastases from 10 female patients.

RESULTS

In total, 79 tumors were analyzed. The overall frequency of allelic loss was 67% (16 of 24 tumors) in the primary urothelial carcinomas and 79% (19 of 24 tumors) in the metastatic carcinomas. The primary urothelial carcinoma showed LOH at the D9S171, D9S177, and TP53 loci in 39% (9 of 23 tumors), 30% (6 of 20 tumors), and 30% (7 of 23 tumors) of informative samples, respectively. LOH in ≥ 1 lymph node metastases was seen at the D9S171, D9S177, and TP53 loci in 35% (8 of 23 tumors), 45% (9 of 20 tumors), and 48% (11 of 23 tumors) of informative samples, respectively. Eleven tumors demonstrated identical allelic loss patterns at all DNA loci both in the primary carcinoma and in all corresponding lymph node metastases. Three tumors showed allelic loss in the metastatic carcinoma but not in its matched primary carcinoma. Six tumors demonstrated a different LOH pattern in each of its lymph node metastases. Clonality analysis showed the same pattern of nonrandom X-chromosome inactivation both in the primary urothelial carcinoma and in all of the lymph node metastases in five of nine informative tumors studied. Four tumors showed a random pattern of X-chromosome inactivation in both the primary carcinoma and in the metastases.

CONCLUSIONS

LOH and X-chromosome inactivation assays showed that multiple lymph node metastases and matched primary urothelial carcinomas of the bladder had the same clonal origin, suggesting that the capability for metastasis often arises in only a single clonal population in the primary tumor. The variable LOH patterns observed in some of the tumors likely reflect genetic divergence during the clonal evolution of urothelial carcinoma. Cancer 2005. © 2005 American Cancer Society.

An understanding of metastasis at the cellular and molecular levels is an important objective in cancer research, because the major barrier to adequate treatment of metastatic lesions is the biologic heterogeneity of carcinoma cells in both the primary neoplasm and its metastases.1 Urothelial carcinoma of the urinary bladder is responsible for approximately 13,000 cancer deaths in the U.S. each year.2 Patients with urothelial carcinomas that have invaded the bladder wall have a poor prognosis, with approximately 50% of such patients developing metastases within 2 years.3 Initially, urothelial carcinomas metastasize to regional lymph nodes in the pelvis, with tumor frequently deposited in more than one lymph node. These tumors have been shown to be genetically and phenotypically heterogeneous neoplasms,4, 5 which has raised questions about the clonal origin of urothelial carcinoma and its metastases.

To our knowledge, the precise genetic relations between primary tumors and each metastatic focus have not been characterized well in patients with urothelial carcinoma, and the biologic and molecular mechanisms responsible for tumor metastasis are not understood entirely. For example, it is unknown whether multiple lymph node metastases arise from different neoplastic subclones within the same tumor that have a different metastatic potential or whether these metastatic lesions arise from an identical cell of origin. In this study, molecular analysis of microsatellite alterations and X-chromosome inactivation status in multiple lymph node metastases of urothelial carcinoma and matched primary tumors from the same patient were used to assess clonality and the molecular genetic relations among these lesions.

MATERIALS AND METHODS

Patients

Fourteen men and 10 women with urothelial carcinoma of the bladder underwent surgical excision of their tumors and regional lymph node dissection between 1991–2003. The patients had a mean age of 67 years (range, 44–86 yrs). All patients had two or more lymph node metastases. The 2004 World Health Organization bladder tumor classification criteria were used for grading.6 The primary tumor was classified as high grade in all 24 patients. Pathologic staging was performed according to the 2002 tumor, lymph node, metastasis (TNM) classification system.7 Four patients had pathologic T2 (pT2) tumors, 11 patients had pT3 tumors, and 9 patients had pT4 tumors. The mean greatest dimension of the primary tumor was 4.5 cm (range, 1.0–9.5 cm). The lymph node stage for all 24 patients was N2.

Tissue Samples and Microdissection

Archival surgical materials from 24 patients with urothelial carcinoma of the urinary bladder (14 male patients and 10 female patients) who had ≥ 2 regional lymph node metastases accessioned from 1991–2003 were retrieved from the surgical pathology files of the Department of Pathology and Laboratory Medicine of the Indiana University School of Medicine (Indianapolis, IN) and the Department of Pathology of Corboda University (Corboda, Spain). This study included a total of 24 primary urothelial carcinomas and 55 lymph node metastases.

Histologic sections were prepared from formalin fixed, paraffin embedded tissue and were stained with hematoxylin and eosin for microscopic evaluation. From these slides, the urothelial carcinomas and lymph node metastases were identified (Fig. 1). Laser-assisted microdissection of the neoplastic cells was performed (Fig. 1) on unstained sections using a PixCell II Laser-Capture Microdissection (LCM) system (Arcturus Engineering, Mountain View, CA), as described previously.8–10 Approximately 400–1000 cells from each tumor were microdissected from the 5-μm histologic sections. Normal tissue from each patient was microdissected as a control.

Figure 1.

These photomicrographs, obtained after the laser microdissection of tumors from patients with urothelial carcinoma metastatic to multiple pelvic lymph nodes, reveal (A) metastatic tumor before microdissection, (B) metastatic tumor after microdissection, and (C) laser-captured metastatic cells (C).

Detection of LOH

The dissected cells were deparaffinized with xylene and ethyl alcohol. Polymerase chain reaction (PCR) was used to amplify genomic DNA at 3 specific loci on 2 different chromosomes: 9p21 (D9S171), 9q32 (D9S177), and 17p13 (TP53). Previous studies have demonstrated that loss of heterozygosity (LOH) at these loci occurs frequently in urothelial carcinoma.10–12 D9S171 includes regions of the putative tumor suppressor gene p16. D9S177 is located within a putative tumor suppressor gene that is involved in the carcinogenesis of squamous cell carcinomas and urothelial carcinomas. Alterations of chromosome 9 are some of the earliest and most frequent events in papillary urothelial carcinogenesis.13, 14 The TP53 locus corresponds to the gene that encodes the p53 protein. Mutations of the p53 gene are the most common genetic abnormalities in carcinoma.15 PCR amplification and gel electrophoresis were performed as described previously.8, 9, 16 The criterion for allelic loss was the complete or nearly complete absence of one allele in tumor DNA.9, 16 PCR analyses for each polymorphic microsatellite marker were repeated at least twice from the same DNA preparations, and the same results were obtained.

Analysis of Allelic Loss Pattern

When the genetic material from a patient was identified as homozygous for the polymorphic markers (i.e., it showed only one allele in the normal control tissue), then the material was considered noninformative. Patients who had genetic material that was informative (i.e., it showed two alleles in the normal control tissue) were divided into two categories.17 Their DNA either showed no allelic deletions in the tumor, retaining two different alleles of similar intensity on autoradiographs, or it showed the absence of one allele. DNA that is sampled from the cells of a primary urothelial carcinoma and from lymph node metastases that demonstrate identical allelic loss patterns is compatible with a common clonal origin, whereas different patterns of allelic deletions are compatible with independent clonal origins.

Detection of X-Chromosome Inactivation

X-chromosome inactivation analysis was performed on the primary urothelial carcinoma and matching lymph node metastases from the 10 female patients, as described previously.4, 18 DNA samples were prepared from the tumor and from each metastatic focus from the same patient. The dissected cells were placed in 50 μL of buffer (i.e., 10 mM Tris-HCl, 1 mM ethylenediamine tetraacetic acid [EDTA], 1% Tween-20, and 3 mg/mL proteinase K, pH 8.3) and incubated overnight at 37 °C. The solution was boiled for 10 minutes to inactivate the proteinase K and was used directly for subsequent clonal analysis without further purification. Aliquots (8 μL) of the DNA extract were digested overnight at 37 °C with 10 U of HhaI restriction endonuclease (New England Biolabs Inc., Beverly, MA) in a total volume of 10 μL. Equivalent aliquots of the DNA extracts also were incubated in the digestion buffer without HhaI endonuclease as control reactions for each sample. After the incubation, 3 μL of digested or nondigested DNA were amplified in a 25-μL PCR volume containing 0.1 μL 32[P]α-labeled deoxyadenosine triphosphate (dATP) (3000 Ci/mmol), 4 μM androgen receptor (AR)-sense primer (5′TCC AGA ATC TGT TCC AGA GCG TGC3′), 4 μM AR-antisense primer (5′TGCT GTG AAG GTT GCT GTT CCT CAT3′), 4% dimethyl sulfoxide, 2.5 mM MgCl2, 300 μM deoxycytidine triphosphate, 300 μM deoxythymidine triphoshpate, 300 μM deoxyguanosine triphosphate, 300 μM dATP, and 0.65 U Taq DNA polymerase (Perkin-Elmer Corp., Norwalk, CT). Each PCR amplification had an initial denaturation step at 95 °C for 8 minutes followed by 38 cycles at 95 °C for 40 seconds, at 58 °C for 40 seconds, and at 72 °C for 60 seconds and then followed by a single, final extension step at 72 °C for 10 minutes. The PCR products were then diluted with 4 μL of loading buffer containing 95% formamide, 20 mM EDTA, 0.05% bromphenol blue, and 0.05% xylene cyanole FF (Sigma Chemical Company, St. Louis, MO). The samples were heated to 95 °C for 8 minutes and then placed on ice. Three microliters of the reaction mixture were loaded onto 6.5% polyacrylamide-denaturing gels without formamide, and the PCR products were separated by electrophoresis at 80 Watts for 2 hours. The bands were observed after autoradiography with Kodak X-OMAT film (Eastman Kodak Company, Rochester, NY) for 8–16 hours.

Analysis of X-Chromosome Inactivation

The materials were considered informative if two AR allelic bands were detected after PCR amplification in normal control samples that had not been treated with HhaI. Only informative materials (i.e., tissues without a skewed pattern of X-chromosome inactivation after treatment with HhaI in normal control samples) were included in the analysis. In tumor samples, nonrandom X-chromosome inactivation was defined as a complete or a nearly complete absence of an AR allele after HhaI digestion, which indicated a predominance of one allele. Tumors were considered to have the same clonal origin if the same AR allelic inactivation pattern was detected in both the primary tumor and in each lymph node metastasis.18, 19 Tumors were considered to have independent origins if alternate predominance of AR alleles after HhaI digestion (different allelic inactivation patterns) was detected in the primary tumor and the lymph node metastases (Fig. 2).

Figure 2.

This schematic illustrates the determination of X-chromosome inactivation patterns. Nonrandom methylation of X chromosomes may occur in women by the hypermethylation of exon 1 of the androgen receptor (AR) gene on the X chromosome. These genetic alterations (both the hypermethylation and the X-chromosome inactivation) will be maintained in subsequent cell doublings. Clonal analysis was performed by a polymerase chain reaction (PCR)-based, endonuclease restriction enzyme-dependent method, which took advantage of an X-chromosome-linked polymorphic marker, a CAG-nucleotide repeat of the AR gene (HUMARA), and several methylation-sensitive HhaI (GCGC) endonuclease sites to detect the nonrandom inactivation of X-chromosomes. Tumors with identical allelic inactivation patterns (Pattern B) are of monoclonal origin. Tumors with different allelic loss inactivation patterns (Pattern A) are of independent origin. LN: lymph node metastasis; +: positive; −: negative.

RESULTS

The overall frequency of allelic loss was 67% (16 of 24 patients) in primary urothelial carcinomas and 79% (19 of 24 patients) in metastases (Table 1). The number of specific loci lost in a single primary urothelial carcinoma ranged from 1 to 2 with none of the 24 primary tumors showing LOH at all 3 loci. The number of specific loci lost in a single metastatic lesion ranged from one to three. The frequency of allelic loss in all of the informative samples of primary urothelial carcinoma was 39% (9 of 23 tumors) for D9S171, 30% (6 of 20 tumors) for D9S177, and 30% (7 of 23 tumors) for TP53 (Fig. 3). The frequency of allelic loss in ≥ 1 lymph node metastases in the informative samples was 35% (8 of 23 metastases) for D9S171, 45% (9 of 20 metastases) for D9S177, and 48% (11 of 23 metastases) for TP53.

Table 1. Loss of Heterozygosity Analysis and X-Chromosome Inactivation Analysis of Urothelial Carcinomas of the Bladder and Multiple Lymph Node Metastases
PatientMicrosatellite markers (allelic loss)X-chromosome inactivation
D9S171D9S177TP53
  1. UC: urothelial carcinoma; LN: lymph node metastasis; NI: noninformative; ↕ both alleles present; ▴: loss of lower allele; ▾: loss of upper allele.

1    
 UC NI
 LN 1NI
 LN 2NI
2    
 UC
 LN 1
 LN 2
3    
 UCNINI
 LN 1NINI
 LN 2NINI
4    
 UCNI
 LN 1NI
 LN 2NI
5    
 UC
 LN 1
 LN 2
6    
 UC
 LN 1
 LN 2
7    
 UC
 LN 1
 LN 2
 LN 3
8    
 UC
 LN 1
 LN 2
 LN 3
9    
 UC
 LN 1
 LN 2
 LN 3
10    
 UC
 LN 1
 LN 2
 LN 3
 LN 4
11    
 UC 
 LN 1 
 LN 2 
12    
 UC 
 LN 1 
 LN 2 
13    
 UCNI 
 LN 1NI 
 LN 2NI 
 LN 3NI 
14    
 UCNI 
 LN 1NI 
 LN 2NI 
15    
 UC  
 LN 1 
 LN 2 
16    
 UC 
 LN 1 
 LN 2 
17    
 UC 
 LN 1 
 LN 2 
18    
 UC 
 LN 1 
 LN 2 
19    
 UC 
 LN 1 
 LN 2 
20    
 UC 
 LN 1 
 LN 2 
21    
 UCNI 
 LN 1NI 
 LN 2NI 
22    
 UC 
 LN 1 
 LN 2 
23    
 UC 
 LN 1 
 LN 2 
24    
 UC 
 LN 1 
 LN 2 
 LN 3 
Figure 3.

These are representative results from the loss of heterozygosity analysis and X-chromosome inactivation analysis in patients with urothelial carcinoma. N: normal control tissue specimen from the same patient; P: primary urothelial carcinoma; LN: lymph node metastasis. The arrowheads indicate loss of either the upper or lower allele. Note the bottom three images, which highlight the concordant nonrandom pattern of X-chromosome inactivation in the primary tumor and multiple lymph node metastases and the one noninformation case (Case 4) (+: after HhaI endonuclease digestion; −: without HhaI endonuclease digestion).

Eleven patients showed identical allelic loss patterns at all DNA loci in both the primary tumor and in all corresponding lymph node metastases. Three patients showed allelic loss in the metastatic tumor but not in its matched primary tumor. Six patients showed a different LOH pattern in each lymph node metastasis. Eight patients showed an LOH pattern in the lymph node metastases that differed from the LOH pattern of the primary tumor. However, in four of those eight patients, the primary tumor and the lymph node metastases showed LOH at some of the same loci, suggesting the possibility of a common clonal origin with subsequent genetic divergence. In fact, X-chromosome inactivation analysis was performed on two of those eight patients, and the same pattern of nonrandom X-chromosome inactivation was seen in the primary tumors and in each lymph node metastasis, consistent with a common clonal origin. Five patients did not show LOH at any of the three loci examined in either the primary tumor or the metastases. Among these five patients, X-chromosome inactivation analysis was performed on three patients and showed the same nonrandom pattern of X-chromosome inactivation in the primary tumor and in each lymph node metastasis in one patient and showed a random pattern in the other two patients.

Overall, X-chromosome inactivation analysis was performed on the tumor and lymph node metastases from 10 female patients. Clonality analysis showed the same pattern of nonrandom X-chromosome inactivation both in the primary urothelial carcinoma and in all of the lymph node metastases in five of nine informative samples, consistent with a common clonal origin. Four patients showed a random pattern of X-chromosome inactivation both in the primary tumor and in the metastases. No patients showed a discordant pattern of nonrandom X-chromosome inactivation between the primary tumor and multiple lymph node metastases.

DISCUSSION

The majority of cancer deaths are caused by metastases that are resistant to conventional therapies. An understanding of the pathogenesis of tumor metastasis has been an important objective of cancer research. It is believed that tumors are composed of numerous subpopulations of neoplastic cells, with only rare cells that have the capacity to metastasize.20–24 It long has been hypothesized that metastases are clonal and that they arise from rare cells in genetically heterogeneous primary tumors. However, recent studies using DNA array analysis suggest that the majority of cells in a primary tumor are capable of metastasizing and that the important determinant of metastatic potential is the genetic background of the host, suggesting that metastases may not necessarily be clonal.25, 26 Urothelial carcinoma of the urinary bladder is a common malignancy. It has been shown that urothelial carcinoma exhibits significant genetic and phenotypic heterogeneity5, 27, 28 and frequently metastasizes to regional pelvic lymph nodes. Thus, urothelial carcinoma serves as an ideal model for the study of the molecular genetic correlations between multiple lymph node metastases and the matched primary tumor. In the current study, we examined 24 patients with urothelial carcinoma of the urinary bladder who had ≥ 2 regional lymph node metastases using LOH and X-chromosome inactivation analyses to assess tumor clonality. We found evidence for a common clonal origin in primary urothelial carcinomas and their multiple lymph node metastases in the majority of patients. Our data suggest that the capability for metastasis often arises in only a single clonal population in the primary tumor and then spreads to regional lymph nodes. The relative constancy of genetic changes at the loci studied in primary bladder carcinoma and multiple, paired, metastatic lesions may aid in the diagnosis and identification of tumor origins in difficult cases.

At the molecular and cellular levels, carcinoma metastasis is a complicated process that is not understood entirely. The process of acquiring metastatic potential involves a series of genotypic and phenotypic cellular alterations, which depend both on the intrinsic properties of the tumor cells and on physiologic and immunologic responses of the host.29–33 Many tumors, including urothelial carcinomas, are composed of heterogeneous subpopulations of neoplastic cells with different biologic properties and variable potentials for metastasis.20–24, 27, 34 This diversity may arise as a result of a multicellular origin or as a consequence of genetic divergence during clonal expansion of a single transformed cell. Metastatic lesions themselves can display intralesional or interlesional genetic heterogeneity, which is likely because of the increased rate of mutation observed in tumor cell populations with high metastatic potential.35, 36 Multifocal metastases, such as the multiple regional lymph node metastases of urothelial carcinoma observed in the current study, can have a unicellular or multicellular origin.1, 37

Clonality studies have been performed in several malignancies and metastatic lesions. In a study by Talmadge et al.,29 cells from a melanoma cell line were treated with γ-irradiation to induce specific chromosomal alterations and then were implanted subcutaneously in mice. Spontaneous lung metastases were isolated from different animals, established in culture as individual lines, and then karyotyped. Metastases exhibited an identical pattern of chromosomal alterations in each cell, indicating that each metastasis was clonal but that each probably originated from a different progenitor cell. Similarly, when a heterogeneous mixture of two distinct melanoma cell lines was injected intravenously, all subsequent lung metastases were of unicellular origin.38 LOH analysis of in-transit melanomas in humans has shown that these in-transit metastases are clonal in origin.39 These in-transit melanomas are considered to be a form of regional metastasis resulting from the intralymphatic trapping of melanoma cells between the primary tumor and regional lymph nodes. Similar findings have been seen in comparative genomic hybridization studies of squamous cell carcinomas40 of the head and neck and in esthesioneuroblastomas,41 in which the majority of metastases shared a common clonal origin with matched primary tumors. DNA ploidy analysis in breast carcinomas and matched regional lymph node metastases showed that, in 90% of metastatic DNA clones, the corresponding clone was identified in a primary tumor sample representing ≥ 25% of the tumor cell population.42 In a comparative genomic hybridization study of primary renal cell carcinoma and both regional and distant metastases, a high probability of a common clonal progenitor was found in 58% of patients.43 Whether or not regional metastases and matched primaries of most or all human malignancies are clonal in origin is not known; however, our finding of a common clonal origin between primary urothelial carcinomas of the urinary bladder and multiple pelvic lymph nodes suggests that the process of early metastasis in these tumors is similar to that of melanoma, breast carcinoma, and squamous cell carcinoma of the head and neck.

The pathogenesis of tumor metastasis has been studied for > 100 years. Paget first proposed his “seed and soil” hypothesis in 1889, because he observed that certain malignancies had a predilection for certain sites of distant metastasis.44 He proposed that specific tumor cell types (the seed) had an affinity for the microenvironment of other organs (the soil) and that metastases formed when “seed” and “soil” were well matched. Although others have suggested that the distribution of metastatic lesions is solely a reflection of the vascular system and patterns of blood flow,45, 46 studies in syngeneic mice have shown that, whereas tumor cells could be trapped in the capillary beds of distant organs, the formation of a metastatic lesion was influenced by interactions with specific organ cells, thus supporting Paget's original hypothesis.47 Others have concluded that, although distant metastases from many types of malignancy occur in a site-specific distribution, regional metastases, such as that seen in our study, are distributed mainly in patterns related to blood flow and lymphatic drainage.48, 49 Our finding that multiple regional lymph node metastases and matched primary urothelial carcinomas arise from a common cell of origin is compatible with this notion. However, other metastatic mechanisms cannot be ruled out entirely. In a study of patients with metastatic colorectal carcinoma, a much greater median value of fractional allelic imbalance was reported in distant metastases than in either primary tumors or in local recurrences,50 suggesting that there are molecular genetic differences between regional and distant metastases. Similarly, Bonsing et al. found differences in DNA ploidy status between lymphatic and hematogenous metastases of breast carcinomas.51 Whether or not distant metastases of urothelial carcinoma exhibit very different genetic properties from the primary tumor or from the regional lymph node metastases is an area for future research. It is unknown whether each lymph node metastasis in our study resulted from an independent, metastasizing cell population in the primary tumor or whether some metastatic lesions are able to metastasize to other lymph nodes. In patients with breast carcinoma, preliminary data suggest that some metastases do metastasize52; however, this concept has not been studied in urothelial carcinoma. Our current data are not incompatible with this phenomenon, but neither do they prove that it is a legitimate mechanism of tumor spread in bladder carcinoma.

The idea that metastatic lesions arise from rare subpopulations of neoplastic cells in primary tumors that have acquired the capability for distant spread recently has been challenged. One hypothesis is that the metastatic potential of a primary tumor is determined intrinsically by the host. In support of this notion, significant differences in metastatic efficiency were seen among the hosts in a transgenic tumor model without alteration of tumor initiation or growth kinetics.53, 54 Ramaswamy et al.25 compared the gene expression profiles of adenocarcinoma metastases from multiple tumor types with unmatched primary adenocarcinomas and found a specific gene expression signature that distinguished primary from metastatic adenocarcinomas. Moreover, those investigators found that some primary tumors carried the same gene expression signature that was found frequently in metastatic tumors. This subset of primary tumors was the most likely to be associated with metastasis and a poor clinical outcome. Thus, Ramaswamy et al. concluded that the capacity for metastasis is encoded in the bulk of the primary tumor, challenging the notion that metastases arise from rare cells in a heterogeneous primary tumor that have acquired the capacity to metastasize.

Likewise, Bernards and Weinberg26 concluded that the idea of rare tumor cells within a primary tumor acquiring the potential for metastasis has a major conceptual inconsistency. Those authors pointed out that the acquisition of genes permitting metastasis would not provide a proliferative advantage at the primary site. Hence, rare subpopulations of tumor cells that have acquired the ability to metastasize should remain rare in the primary tumor. Thus, Bernards and Weinberg reasoned that the tendency for metastasis is determined by mutations that are acquired early in the process of multistep tumorigenesis. This line of reasoning is compatible with both the findings of Ramaswamy et al. and with the findings in the current study. The allelic loss patterns and X-chromosome inactivation patterns were almost identical in the urothelial carcinomas and multiple lymph node metastases in the majority of our patients. This would be expected if the ability to metastasize were conferred on tumor cells at a very early stage of tumorigenesis. It also suggests that analysis of the gene-expression profiles of the dominant population of urothelial carcinoma cells within the primary tumor may predict the clinical behavior of the malignancy. It is unlikely that DNA alterations at chromosomes 9p21 (D9S171), 9q32 (D9S177), and 17p13 (TP53) play a critical role in tumor metastasis, because LOH at any of these loci was not observed in all or the majority of lymph node metastases. It remains possible that allelic loss at these loci may contribute to a neoplastic clone's acquisition of metastatic potential; however, if this is true, then our data suggest that alternate genetic pathways to metastatic capability also must exist.

Understanding the genetic correlations between urothelial carcinoma and its multiple metastases not only is noteworthy biologically but also is relevant clinically, in that the majority of disease-specific deaths in patients with urothelial carcinoma occur as a result of treatment-resistant metastases. Genetic and phenotypic heterogeneity of both primary and metastatic tumors makes eradication difficult. Indeed, metastatic cells in several malignancies exhibit a greater survival ability and resistance to treatment than cells with limited metastatic potential. One hypothesis for this therapeutic resistance is the disruption of the apoptotic pathway in the metastatic cells of some carcinoma types55–58; however, many other possible mechanisms for chemoresistance have been proposed.59 In breast carcinoma models, different genetic changes were observed among metastases from lung, bone, and lymph nodes with variable levels of organ-dependent chemoresistance to docetaxal.60 The majority of lymph node metastases in the current study demonstrated a similar or identical pattern of allelic loss and/or X-chromosome inactivation to that demonstrated in matched primary urothelial carcinomas, consistent with a common clonal origin. The close genetic correlation between the primary tumor and its regional metastases may suggest that both are sensitive equally to adjuvant therapy. However, additional studies need to be performed to confirm this hypothesis and to compare the incidence and mechanisms of treatment resistance in patients who have distant metastases with those in patients who have only regional lymph node metastases. It is expected that elucidating the mechanism of metastasis in bladder carcinoma will help both in identifying which tumors are at risk of or are capable of metastasizing and in developing new therapies.

In conclusion, the current data indicate that the pattern of allelic loss in primary bladder carcinoma generally was maintained during disease progression to multifocal, regional metastasis. LOH and X-chromosome inactivation assays showed that multiple lymph node metastases and matched primary urothelial carcinomas of the bladder had the same clonal origin, suggesting that the capability for metastasis often arises in only a single clonal population in the primary tumor that subsequently spreads to regional lymph nodes. The current study data also suggest the possibility that LOH and X-chromosome inactivation analysis may be used as an adjunct in the diagnosis of a metastasis of unknown primary origin.

Ancillary