Toward the Validation of Aneusomy Detection by Fluorescence in Situ Hybridization in Bladder Cancer: Comparative Analysis with Cytology, Cytogenetics, and Clinical Features Predicts Recurrence and Defines Clinical Testing Limitations1

Fluorescence in situ hybridization (FISH) is regarded as a potential new tool for the clinical management of bladder cancer that works by detecting cytogenetic aberrations in noncycling, exfoliated cells from bladder irrigations. However, clinical validation steps must be addressed to define the true predictive potential in a clinical setting. Toward the validation of FISH with the use of bladder washings and prior to incorporation into a large, prospective clinical trial, a pilot study was designed to determine its clinical potential, define testing limitations, optimize a panel of probes specific for bladder cancer detection, and outline protocol/data collection parameters. Correlations with standard cytogenetics and clinicopathological features of bladder cancer were investigated. Exfoliated cells obtained from benign bladder washings served as normal controls. The results of this pilot study suggest the following: (a) FISH and cytology are complementary testing procedures; however, the FISH data provided valuable pboidy and specific genotypic information for recurrent tumors in “suspicious” cases; (b) chromosomal aberrations defined by FISH are associated with tumor grade and stage (i.e., simple numerical aberrations were associated with low-grade tumors, and high-grade and invasive tumors exhibited multiple, nonrandom chromosomal aberrations and vast intratumor heterogeneity); (c) somatic pairing or homologous centromeric association can give a Received 7/7/97: revised 9/2/97: accepted 9/10/97. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1This project was funded in part by a Southwest Oncology Group Genitourinary Tumor Biology seed grant. National Cancer Institute Grant CA32102. and a private donation by Arthur Paul Berg. M. L. S. is a member of the City of Hope Cancer Research Center, which is supported by USPHS Grant CA-33572. 2 To whom requests for reprints should be addressed, at City of Hope National Medical Center, Department of Cytogenetics, Northwest Building. Room 2255. 1500 East Duane, Duarte. CA 91010. Phone: (626) 359-81 1 1, ext. 2313: Fax: (626) 301-8877: E-mail: mslovak@ smtplink.coh.org. false-positive result and appears to be linked to prior therapy; (d) dual hybridization with reference gene-specific probes must be used to control for somatic pairing; and (e) focal, deep muscle invasive lesions, with no surface exposure, may yield false-negative results. The data suggest that FISH analysis, with the use of cells isolated from bladder washings, is a powerful technique holding promise for early cancer detection, monitoring treatment outcome, and predicting recurrence of disease. INTRODUCTION Approximately 50,000 new cases of bladder cancer are diagnosed annually in the United States, making it the fifth most common cancer in Western society ( 1 ). Ninety percent of bladder cancer cases are categorized as TCC.3 whereas the remaining 10% are classified as either squamous cell carcinoma or adenocarcinoma (2). Three subgroups of TCC have been defined by histopathobogy: papillary (invasive and noninvasive), nonpapillary. and Tis, each with dramatically differing clinical histories and outcomes (2-4). Histopathological classification reflects tumor grade and stage: however, it fails to reliably predict pathobiobogical characteristics. such as drug resistance, recurrence, and invasive or metastatic potential. A major prob1cm in the management of superficial bladder cancer is accurate identification of those TCC cases with the highest risk of relapse and progression. Genetic aberrations in cancer have become a primary focus to study the pathogenesis of urothelial tumors (2, 5). In bladder cancer, assessment of chromosome content by FCM or SC has been correlated with several clinical parameters, including diagnosis. histopathology. biological behavior, and prognosis. Tumor progression and recurrence appear to correlate with aneuploidy and a high proliferative rate. However, ascertainment of ploidy by FCM has limited value in low-grade tumors, because small variations in DNA content resulting from loss or gain of partial or one to two chromosomes cannot be detected. Additionally, FCM provides no data regarding specific genetic events that may occur during tumor development or progression. and most invasive tumors are aneuploid (3). Sensitivity of DNA FCM for grade 1-3 tumors ranges from 12 to 89%. with the widest range for grade I tumors (6, 7). Alternatively, SC provides an overview of specific chromosomal aberrations and 3The abbreviations used are: TCC. transitional cell carcinoma: Tis. transitional carcinoma in sitz ; FCM, flow cytometry; SC, standard cytogenetics; FISH, fluorescence in siti4 hybridization; FISH-BW, FISH on bladder washings: FISH-PE, FISH on paraffin-embedded tissue section; Ta, papillary noninvasive tumor. Research. on October 16, 2017. © 1997 American Association for Cancer clincancerres.aacrjournals.org Downloaded from 2318 Aneusomy Detection in Bladder Cancer ploidy levels; however, analysis is limited by sample availability, mitotic yield of fresh tumor material, and in vitro cell processing complications (8, 9). Accordingly, genetic tests that will enhance the diagnostic accuracy of bladder cancer detection and surveillance and provide valuable prognostic information to the practicing urologist are warranted. FISH is a rapid, powerful and sensitive technique for studying molecular cytogenetic abnormalities in malignant cells independent of cell cycling status. Furthermore, FISH permits direct correlation of selected karyotypic anomalies with cell morphology. Recent reports indicate that specific numerical alterations of chromosomes 1, 3, 5, 7, 8, 9, 10, 1 1, 17, and Y in bladder cancer detected by SC and FISH may be useful to screen urine or bladder washings for early tumor detection or recurrence (8, 10-21); however, clinical validation steps must be addressed to define the true predictive potential of aneusomy screening in a clinical setting. The objectives of this clinical validation study were (a) to define a panel of six chromosomespecific probes and validate its usefulness as a screen to detect aneusomy in exfoliated cells from bladder washings (FISHBW), (b) to investigate the ability of FISH-BW to reflect true nonrandom numerical chromosomal aberrations in bladder cancer, and (c) to correlate the FISH-BW data with cytology, clinical cystoscopy findings, and histopathobogy. MATERIALS AND METHODS Samples. This study was a 1-year Institutional Review Board-approved single-institution pilot study for the Southwest Oncology Group Genitourinary Tumor Biology Program. SC were performed on 11 fresh tumors and 1 bladder washing; FISH analyses were performed on 23 paraffin-embedded tissues (18 tumors and 5 normal bladder tissue controls) and 37 bladder washings or urine. Tissue availability was the only criterion for analysis; no sample was excluded from the study on the basis of clinical profile. Clinicopathobogical and cytogenetic features of the cases successfully analyzed are listed in Tables 1-3. Table 1 compares SC to FISH-PE from 11 fresh bladder tumor samples, obtained at time of cystectomy or transurethral resection, and 1 bladder washing (case B21). Concurrent paraffin blocks containing the same portion of the tumor used for SC were studied by FISH-PE. Eight tumors were collected for FISH-BW and FISH-PE comparisons (Table 2). In five cases, tumor tissue was collected on the same day as was the bladder washing specimens. Tumor tissue corresponding to cases B12 and B14 was collected and embedded 9 and 8 months, respectively, before collection of the bladder washing; tumor tissue corresponding to case B20 was collected 6 months after bladder irrigation. For the FISH-BW comparison study, 35 bladder washings (150-200 ml) and two urine samples were obtained at the time of transurethral resections or at follow-up cystoscopy. FISH analysis was determined successful when a 200-cell count for each of the six probes was possible. Concurrent cytology for the 25 successful samples are listed in Table 3: benign disease (n = 4), reactive cytology (n = 7), suspicious for bladder carcinoma (n 7), and bladder carcinoma (n = 7). SC. Twelve bladder tumors, including one highly mitotic bladder washing (case B2 1), were cultured and harvested according to standard technique. Giemsa-trypsin chromosome banding (GTG banding) was used to identify nonrandom chromosomal aberrations. International System for Human Cytogenetic Nomenclature 1995 guidelines were followed for clonal definition and description of individual structural and numerical karyotypic anomalies. FISH-PE. Four-pm sections were pretreated and hybridized with chromosome-specific enumeration probes for chromosomes 1 (a satellite), 7 (a satellite), 8 (a satellite), 9 (classical satellite), 17 (a satellite), and Y (cocktail; Oncor, Inc.) according to the manufacturer’s instructions, with slight modifications. Tissue sections were pretreated with 30% sodium bisulfite in 2X SSC (lX SSC = 0.15 M NaC1 and 0.015 M Na Citrate, pH 7.0) at 45#{176}Cfor 15 mm and digested in proteinase K (0.25 mg/mi in 2X SSC; Oncor, Inc.) at 45#{176}C for 40-90 mm. FISH-BW. Exfoliated cells from bladder washings or urine samples were collected by centrifugation (400 X g for 8 mm), washed with I X PBS and exposed to 0.075 M KC1 hypotonic solution at 37#{176}Cfor 20 mm. These cells were fixed with Carnoy’s fixative before being dropped onto silanized slides. The panel of FISH probes described above for FISH-PE was used. Hybridization and signal detection were carried out according to standard technique (Oncor, Inc.). Two probes were applied to two adjacent areas (22 X 22 mm in size) on one slide. To rule out the possibility of somatic pairing, slides from three cases (B9, BlO, and Bli) previously hybridized with enumeration probes were rehybridized with either the ABL (9q34) or the HER2/neu (l7ql 1.2-12) DNA probe (Oncor, Inc.). Coverslips were removed prior to destaining in 1 X phosphate-buffered detergent for 10 mm at room temperature and dehydration in ethanol (70, 80, and 95%). The initial enumeration probes were washed off, and the target DNA was redenatured by incubation at 72#{176}C in 70% formamide/2X SSC for 1-1.5 mm. Hybridization and posthybridization washes were carried out as described above. Scoring Criteria for FISH and Statistics. FISH slides were scored using a Nikon microscope equipped with FITC/ Texas Red and FITC filters. Scoring criteria for FISH-BW were defined to ensure that the data obtained from the two independent scorers were comparable: (a) 200 nuclei were scored per probe per case; (b) only intact, nonoverlapping nuclei were scored; (c) split adjacent signals less than the diameter of a signal apart were counted as one signal; (d nuclei without hybridization signals and inflammatory cells and squamous cells with intact cell membrane and cytoplasm were recorded but not counted; and (e) nuclei with obvious cross-hybridization and/or high background were not scored. For chromosome Y signals, cells without signal were scored as chromosome loss. The following scoring criteria were used for FISH-PE: (a) a minimum of 200 nuclei from at least four different pathologically confirmed tumor areas; (b) only nonoverlapping nuclei with a size no smaller than one-third of average size and good propidium iodide staining were scored; (c) split signals were counted as a single signal; and (6) nuclei without probe signal were scored to evaluate the effects of truncation due to sectioning. The enumeration probes for the autosomes tested were chromosome-specific repetitive DNA probes for human satellite DNAs located near the centromere. Leukemia and bladder tissue Research. on October 16, 2017. © 1997 American Association for Cancer clincancerres.aacrjournals.org Downloaded from Clinical Cancer Research 2319 4 Unpublished data. Table I A comparison of numerical chromosomal abnormalities of bladder tumors detected by FISH-PE and SC for chromosomes 1. 7. 8. 9. 17, and Y” Chromosomal abnormalities listed in Tables 1-3 represent the subpopulation containing the highest number of cbonal chromosomal gains. assuming a clone equaled or exceeded 5.0% of the total cells scored in bladder washings and 4.0% in paraffin-embedded sections. Numerical abnormalies by FISH-PE Abnormalies by SC Case Grade/Stage chr 1 chr 7 chr 8 chr 9 chr I 7 chr Y chr 1 chr 7 chr 8 chr 9 chr 17 chr Y Al II Ta N N N -“ N N N N N der(9) N N A2 II Ta N N N N N N N N del(9) N N A3 II Ta N N N N N N N N N N N N A4 II Ta N N N N N N N + N N N N AS 111111 Ta +,+,+ +,+,+ +,+,+ +,+ +,+ +,+ +.+ +,+,+der +,+ +,+der +,+,+ + A6 111111 Tib +,+ +.+,+ + +,+ +,+ +,+ +der +.+,+ +der + +der + A7 III TaTF, + +,+ + +,+ +,+ / +der +,+ + +,+der + / A8 III T3,, N + + N N / N + +der N N / A9 ?III T3 + +,+,+,+ +,+ +“,+ +,+ + N +der +der der(9) N N B21 III T3 +,+ +,+ +,+,+.+ +,+ + +,+,+ +,-+-‘ +,+ +,+ +,+ +,+ + AlO IIIIIV T-,b + + +,+ + + + +,+ +,+ +,+ +,+ +,+ +,+h All IV T3 +,+ +,+ +,+ + +,+ + N’ N N N N N a chr, chromosomal: N, normal: der, derivative chromosome; del, deleted chromosome: +, chromosome gain; the number of +s indicates the chromosomal gain in comparison with a normal diploid complement; -. chromosome loss. the number of -s indicates chromosomal loss in relation to normal diploid; N, normal diploid number; I, female. F, Faint second signal observed due to partial loss of classical satellite DNA (structural aberration) in one chromosome 9. ‘. Tetraploidization was found in one small locus. whereas 1 of 15 cells was near tetraploid by SC. .1 Disomy Y was observed in the mainline. e SC revealed 3 N in 15 of 20 cells, and 6 N, idemx2 in 5 of 20 cells. f Many unidentifiable chromosome markers. g Diploid cells were detected as the main population by both FISH and SC: 4N cells were also observed by FISH-PE besides 3 N (truncated 4 N cells). I, der(Y)t(Y:l), der(12)t(Y;l2). I 2 NE 16]/4N [2], 4 N cells had 92 chromosomes, and the actual number of gains for each chromosome was uncertain due to poor banding. data4 from our laboratory confirmed an earlier observation, reporting no significant differences among repetitive DNA autosome probes to detect chromosomal gains and losses (16). Thus, a common cutoff value for FISH-BW and FISH-PE was used to determine autosomal chromosome loss and gain. Cutoff values for FISH-BW were determined using four cytologically benign samples. The percentage of nuclei with only one autosome signal was 4.3 ± 1 .8% (mean ± SD: P = 0.95; n = 20); the percentage of nuclei with more than two signals was 2.6 ± 1.94% (mean ± SD; P = 0.95; n = 20). To distinguish monosomy and polysomy from normal background, conservative cutoff values, as reported by others, were set at 3 X SD above the mean percentages from samples of benign cells with one and more than two signals, respectively (22-24). Cutoff values for monosomy and polysomy in FISH-BW were set at 10% and 9%, respectively. Because only one benign bladder washing specimen was collected from a male, the 20% cutoff for chromosome Y loss, as described by Sauter et a!. ( I 3), was used; a value of greater than 10% implied gain of chromosome Y. Urothelium from five normal bladder samples obtained at autopsy, with no history of bladder cancer, served as controls to define cutoff values for FISH-PE. The percentage of nuclei with no or only one autosome signal was 44.9 ± 3.3% (mean ± SD; P = 0.95: n = 25); the percentage of nuclei with more than two signals was 1.9 ± 0.9% (mean ± SD; P 0.95; n 25). A higher percentage of cells with no or only one autosome signal is due to the truncation effect of paraffin-embedded tissue sectioning. A similar percentage of cells with fewer than two autosome signals was reported previously for normal tissue (25). Applying the same criteria used for FISH-BW, cutoff values for monosomy and polysomy in FISH-PE were 55% and 5%, respectively. For chromosome Y, the percentage of nuclei with no signal was 17.2 ± 2.0% (mean ± SD. P = 0.95. n = 5); the percentage of nuclei with more than one Y signal was 2.0 ± 0.6% (mean ± SD, P = 0.95, n = 5). Thus, cutoff values of 24% and 4%, respectively, were set for gain and loss of chromosome Y. Intratumor genetic heterogeneity is frequently observed in bladder tumors with higher pboidy levels. Chromosomal abnormalities listed in Tables 1-3 represent the subpopulation contaming the highest number of clonal chromosomal gains, assuming a clone equaled or exceeded 5.0% of the total cells scored in bladder washings and 4.0% in paraffin-embedded sections. These values corresponded to clonal findings defined by SC. RESULTS Correlation of SC and FISH-PE. Table 1 summarizes the histopathobogy and numerical abnormalities identified by SC and FISH-PE for the 12 bladder cancer cases successfully analyzed by both methods. Ten cases (83.3%) were concordant. Abnormalities of chromosome 9 [monosomy 9 and del(9q)] were observed as the sole anomalies in low-grade, noninvasive TCCs; marked aneupboidy was common in the high-grade tuResearch. on October 16, 2017. © 1997 American Association for Cancer clincancerres.aacrjournals.org Downloaded from