Chromosomal Alterations Detected by Fluorescence In Situ ...

Anatomic Pathology / FISH for Bladder Cancer Variants

Chromosomal Alterations Detected by Fluorescence In Situ Hybridization in Urothelial Carcinoma and Rarer Histologic Variants of Bladder Cancer Benjamin R. Kipp, PhD,1 Harmony L. Tyner, MPH,1 Michael B. Campion,1 Jesse S. Voss, CT, MP(ASCP),1 R. Jeffrey Karnes, MD,2 Thomas J. Sebo, MD, PhD,1 Kevin C. Halling, MD, PhD,1 and Jun Zhang, MD1 Key Words: Fluorescence in situ hybridization; FISH; Urothelial carcinoma; UroVysion DOI: 10.1309/DFJUHY3WPC9GUU2W

Abstract Fluorescence in situ hybridization (FISH) with the UroVysion probe set (Abbott Molecular, Des Plaines, IL) was used to assess 31 bladder cancers for chromosomal abnormalities, including 4 adenocarcinomas, 5 urachal adenocarcinomas, 6 small cell carcinomas, 7 squamous cell carcinomas, and 9 typical urothelial carcinomas. FISH was also used to assess the benign urothelium in 4 cases. There was a significant increase (P < .001) in the mean number of chromosome 3 (2.64 vs 1.51), chromosome 7 (2.61 vs 1.48), and chromosome 17 (2.41 vs 1.41) centromeric signals observed in cells from patients with cancer compared with patients without cancer. Of the 31 tumors, 29 (94%) demonstrated polysomic signal patterns in more than 10% of cells. In the 2 remaining tumor specimens, there was a high percentage of cells (>75%) demonstrating homozygous 9p21 deletion. The data from this study suggest that chromosomal abnormalities detectable by FISH in urothelial carcinoma are also common in rarer histologic variants of bladder cancer.

552 552

Am J Clin Pathol 2008;130:552-559 DOI: 10.1309/DFJUHY3WPC9GUU2W

More than 67,000 people will be diagnosed with bladder cancer this year in the United States, with approximately 90% of cases being classified as urothelial carcinoma (UC).1,2 The remaining approximately 10% of all bladder cancers are mesenchymal tumors or epithelial tumors that seem to derive from pluripotent urothelial cells, a phenomenon known as UC with divergent differentiation.2 These less common bladder tumors include, but are not limited to, squamous cell carcinoma (SCC), small cell/neuroendocrine carcinoma, adenocarcinoma, sarcomatoid tumors, urachal tumors, and other undifferentiated carcinomas.3-6 The identification and correct classification of these variants is important because these tumors generally have a worse prognosis than typical UC, and they often necessitate a more complex therapeutic approach beyond surgery.2,5 It is well known that malignant cells generally contain various types of chromosomal alterations, including aneusomy (abnormal chromosome copy number), gene amplification or deletion, and translocations.7-9 Fluorescence in situ hybridization (FISH) is increasingly used clinically to detect bladder cancer in urine specimens. The FISH assay for the detection of bladder cancer works by detecting cells in the urine that have chromosomal abnormalities consistent with a diagnosis of neoplasia. However, the chromosomal abnormalities observed are not specific for UC and could indicate the presence of a rarer histologic variant of bladder cancer. Numerous studies have shown that FISH has good sensitivity for the detection of UC in patients with a history of superficial bladder cancer or hematuria.10-14 However, there are limited data exploring whether less common histologic variants of bladder cancer might be detected using FISH. In addition, there are no known data that compare the chromosomal abnormalities © American Society for Clinical Pathology

Anatomic Pathology / Original Article

detected by FISH in rarer histologic variants of bladder cancer compared with typical UC. The goal of this study was to assess paraffin-embedded samples of typical UC bladder cancer, rarer histologic variants of bladder cancer (SCC, small cell carcinoma [SmCC], adenocarcinoma, and urachal adenocarcinoma [UA]), and normal bladder tissue with FISH to determine if FISH can detect chromosomal abnormalities in the less common variants of bladder cancer and to see if there are differences in the chromosomal abnormalities observed in these rarer variants as compared with typical UC.

Materials and Methods Patient Population For the study, 35 paraffin-embedded bladder biopsy (n = 28) and cystectomy (n = 7) specimens from 35 patients were analyzed by FISH on areas determined by 2 board certified pathologists (T.J.S. and J.Z.) to contain cells that represented the diagnosis of adenocarcinoma (n = 4) zImage 1Az, UA (n = 5) zImage 1B, SmCC (n = 6) zImage 1Cz, SCC (n = 7) zImage 1Dz, transitional cell carcinoma (TCC; n = 9) zImage 1Ez, or benign bladder tissue (negative; n = 4). All tumors evaluated in this study represented primary bladder cancers. Specimens diagnosed as UA had a tumor localized to an area at or near the dome of the bladder in relation to the medial umbilical ligament without evidence of in situ carcinoma or glandular metaplasia.15 The 9 specimens with typical UC were TaG1 (n = 2), TaG3 (n = 1), TIS (n = 1), T1 (n = 1), T2 (n = 3), and T3 (n = 1) stage tumors. The 4 specimens diagnosed as negative represented 4 patients without diagnostic abnormalities. FISH Processing For FISH, 5-µm paraffin sections were placed in a 90°C oven for 15 minutes. The tissue sections were then placed in xylene for 30 minutes followed by dehydration in 100% ethanol for 5 minutes. After air drying for 3 minutes, the slides were immersed in 10 mmol/L of citric acid (80°C; pH 6.0) for 45 minutes and 2× saline sodium citrate (SSC) at 37°C for 5 minutes. The samples were then digested in a 0.2% pepsin solution (2,500-3,500 U/mg; Sigma, St Louis, MO) at 37°C for 48 minutes and dehydrated in 75%, 85%, and 100% alcohol solutions for 3 minutes each. Next, 10 µL of UroVysion probe (Abbott Molecular, Des Plaines, IL) was applied to the tissue area of interest as identified on the corresponding H&E-stained slide. Probe and target DNA were codenatured and hybridized on a Vysis HYBrite instrument (Abbott Molecular) set at a denaturation temperature of 80°C for 3 minutes and a hybridization temperature of 37°C for 10 to 16 hours. After hybridization, unbound probe © American Society for Clinical Pathology

was removed by washing in 2× SSC/0.1% NP-40 (Abbott Molecular) at 76°C for 2 minutes followed by 1 minute at room temperature in 2× SSC/0.1% NP-40. Then, 10 µL of DAPI I (Abbott Molecular) counterstain was applied, and the slides were coverslipped. Enumeration of FISH Signals in Tumor and Normal Tissue Specimens Using standard accepted criteria for enumeration,16 the signal copy number for each of the 4 probes (centromere enumeration probe [CEP] 3, CEP 7, CEP 17, and locus specific indicator [LSI] 9p21) was enumerated in 50 consecutive tumor or normal nuclei from each patient. The percentage of cells demonstrating 9p21 homozygous loss (loss of both gold signals; Image 1D), single chromosomal gain (Image 1A), and polysomy (gain of 2 or more of the 4 probes; Images 1A through 1E) were calculated and recorded for each specimen. Cases with cells containing signal patterns consistent with homozygous deletion and either a single chromosomal gain or polysomy were recorded as having 2 abnormalities for the individual cell. Statistics CEP 3, CEP 7, CEP 17, and LSI 9p21 signal patterns and the type of chromosomal abnormality that they exhibited (negative, homozygous 9p21 deletion, single chromosome gain, polysomy, or polysomy with homozygous 9p21 deletion) were documented for each cell and entered into a JMP spreadsheet (version 6.0, SAS Institute, Cary, NC). The mean copy number for each probe (from the 50 cells for each specimen) and absolute number of cells with specific chromosomal classifications were calculated and entered into an SPSS spreadsheet (version 11.5, SPSS, Chicago, IL) for statistical evaluation. Comparisons between the mean number of probe signals and mean number of cells with specific chromosomal abnormalities based on tumor type were calculated by using an independent t test and SPSS statistical software. P values of .05 or less were considered statistically significant.

Results We evaluated 31 cases with histopathologic evidence of adenocarcinoma (n = 4), UA (n = 5), SmCC (n = 6), SCC (n = 7), or TCC (n = 9) and 4 samples of benign urothelial tissue by FISH using the UroVysion probe set. An example of enumeration data from a tumor case (specimen 6, SCC) is shown in zTable 1z. Benign tissue from 4 cases was also evaluated by FISH to assess the probe copy number distribution and frequency of cells showing single gain, 9p21 loss, and polysomy in normal tissue. As shown in zTable 2z, the mean probe copy number for each of the 4 probes in cells from benign tissue

553


553 553

Kipp et al / FISH for Bladder Cancer Variants

B

A

P SG P

C

D

PL

P P N

L

GL

zImage 1z Representative examples of histologic findings and fluorescence in situ hybridization signal patterns from patients with adenocarcinoma (A), urachal adenocarcinoma (B), small cell carcinoma (C), squamous cell carcinoma (D), and transitional cell carcinoma (E). Arrows point to cells having polysomic (P), single chromosome gain (SG), homozygous 9p21 deletion (L), polysomic and homozygous 9p21 deletion (PL), single gain and homozygous 9p21 loss (GL), or normal (N) signal patterns.

E

P

PL

554 554


© American Society for Clinical Pathology


was fairly consistent with means of 1.51 (CEP 3), 1.48 (CEP 7), 1.41 (CEP 17), and 1.30 (LSI 9p21) copies per cell. There were relatively few cells within the benign tissue demonstrating gains of a single chromosome (mean, 2.5 cells per specimen; range, 1-5 cells per specimen) or polysomy (mean, 0.5 cells per specimen; range, 0-1 cells per specimen). Table 2 shows that there was an increase in the number of chromosomal abnormalities found in all variants of bladder cancer compared with benign tissue. Although there

was not a significant difference in the mean number of 9p21 signals from patients with and without cancer (1.61 vs 1.30; P = .191), there was a significant increase (P < .001) in the mean number of CEP 3 (2.64 vs 1.51), CEP 7 (2.61 vs 1.48), and CEP 17 (2.41 vs 1.41) signals in tumor cells from patients with cancer compared with benign cells from patients without cancer, respectively. There was also a significant increase in the absolute number of cells demonstrating gains of a single chromosome (12.3 vs 2.5; P < .001) and

zTable 1z Example of FISH Enumeration Data for Specimen 6, Squamous Cell Carcinoma

Cell No.

CEP 3

CEP 7

CEP 17

LSI 9p21

Cell Categorization

1 1 0 1 0 Homozygous 9p21 deletion only 2 2 1 1 0 Homozygous 9p21 deletion only 3 2 1 1 0 Homozygous 9p21 deletion only 4 1 3 1 0 Homozygous 9p21 deletion and single gain 5 3 2 0 1 Single gain only 6 2 0 2 0 Homozygous 9p21 deletion only 7 4 1 1 0 Homozygous 9p21 deletion and single gain 8 2 1 3 0 Homozygous 9p21 deletion and single gain 9 3 3 2 0 Homozygous 9p21 and polysomy 10 2 1 2 0 Homozygous 9p21 deletion only 11 3 3 2 0 Homozygous 9p21 and polysomy 12 5 0 2 0 Homozygous 9p21 deletion and single gain 13 1 1 1 0 Homozygous 9p21 deletion only 14 3 1 3 0 Homozygous 9p21 and polysomy 15 2 1 1 0 Homozygous 9p21 deletion only 16 4 2 4 0 Homozygous 9p21 and polysomy 17 2 0 4 0 Homozygous 9p21 deletion and single gain 18 4 2 1 0 Homozygous 9p21 deletion and single gain 19 5 2 2 0 Homozygous 9p21 deletion and single gain 20 3 2 2 2 Single gain only 21 4 1 1 0 Homozygous 9p21 deletion and single gain 22 2 0 2 0 Homozygous 9p21 deletion only 23 0 1 1 0 Homozygous 9p21 deletion only 24 1 2 1 0 Homozygous 9p21 deletion only 25 2 1 2 0 Homozygous 9p21 deletion only 26 3 3 4 0 Homozygous 9p21 and polysomy 27 2 2 2 0 Homozygous 9p21 deletion only 28 1 2 2 2 Normal 29 1 3 1 0 Homozygous 9p21 deletion and single gain 30 2 1 2 0 Homozygous 9p21 deletion only 31 3 2 2 0 Homozygous 9p21 deletion and single gain 32 2 0 2 1 Normal 33 1 1 0 0 Homozygous 9p21 deletion only 34 3 2 2 0 Homozygous 9p21 deletion and single gain 35 3 2 4 0 Homozygous 9p21 and polysomy 36 0 2 2 0 Homozygous 9p21 deletion only 37 2 0 2 0 Homozygous 9p21 deletion only 38 3 2 2 0 Homozygous 9p21 deletion and single gain 39 1 1 3 0 Homozygous 9p21 deletion and single gain 40 2 3 2 0 Homozygous 9p21 deletion and single gain 41 1 1 3 0 Homozygous 9p21 deletion and single gain 42 3 4 4 0 Homozygous 9p21 and polysomy 43 3 1 3 0 Homozygous 9p21 and polysomy 44 3 2 1 1 Single gain only 45 1 1 0 0 Homozygous 9p21 deletion only 46 3 1 3 0 Homozygous 9p21 and polysomy 47 2 1 3 0 Homozygous 9p21 deletion and single gain 48 2 4 1 1 Single gain only 49 1 3 2 0 Homozygous 9p21 deletion and single gain 50 2 2 3 0 Homozygous 9p21 deletion and single gain Average 2.26 1.56 1.96 0.16 Totals: Normal, 2; 9p21 loss only, 17; single gain only, 4; 9p21 loss and single gain, 18; 9p21 loss and polysomy, 9

CEP, centromere enumeration probe; FISH, fluorescence in situ hybridization; LSI, locus specific indicator.


555


555 555


polysomy (24.9 vs 0.5; P < .001) from patients with cancer compared with the number from patients without cancer, respectively. Table 2 and zFigure 1z summarize the FISH signal patterns of individual specimens based on the different histopathologic diagnoses. Figure 1A shows that among the different bladder cancers examined, SmCC showed the fewest chromosomal alterations (mean, 33%; median, 29%; range, 6%-60%). There was relatively little difference in

the percentages of cells with gains of a single chromosome based on cancer subtype (Figure 1B). It is interesting that a higher percentage of cells demonstrating homozygous 9p21 loss was observed in SCC (mean, 70%; median, 89%; range, 14%-96%; Figure 1C). The mean percentage of cells demonstrating homozygous 9p21 deletion in the 7 SCC specimens was significantly greater than the percentage of cells demonstrating homozygous 9p21 deletion among the remaining 24 cancers (69.7 vs 19.8, respectively; P < .001).

zTable 2z Individual Specimen FISH Results FISH Scores*

Mean No. of Signals per Cell

Histologic Diagnosis/ Specimen ID CEP 3 CEP 7 CEP 17 LSI 9p21 Normal

Homozygous 9p21 Loss Only

Absolute No. of Cells Single Gain Only

Polysomy Only

Single Gain Polysomy + 9p21 Loss + 9p21 Loss

Negative 17 1.42 1.60 1.58 1.40 40 7 2 1 0 0 19 1.68 1.64 1.44 1.52 47 2 1 0 0 0 23 1.40 1.30 1.24 0.94 31 17 2 0 0 0 24 1.54 1.38 1.38 1.32 36 8 4 1 1 0 Average 1.51 1.48 1.41 1.30 38.5 8.5 2.3 0.5 0.3 0.0 Carcinoma Adenocarcinoma 4 1.74 2.82 2.22 1.90 10 5 9 22 4 0 10 2.52 3.24 1.82 1.96 9 5 8 27 1 0 11 3.50 2.94 3.22 3.26 3 0 4 43 0 0 16 3.12 2.64 1.48 0.90 9 4 11 19 6 1 Squamous cell carcinoma 6 2.26 1.56 1.96 0.16 2 17 4 0 18 9 7 2.52 2.42 2.70 0.16 3 3 0 2 22 20 8 1.36 1.38 1.30 0.28 10 38 1 1 0 0 9 3.40 2.44 2.54 0.08 2 10 0 0 14 24 13 2.30 2.36 2.20 0.06 2 10 0 0 23 15 14 1.22 2.38 2.28 2.24 13 4 8 22 3 0 31 2.84 2.38 2.66 1.10 4 1 9 23 6 7 Small cell carcinoma 1 2.18 2.06 1.98 1.96 16 3 12 18 1 0 2 2.06 2.60 2.46 1.96 11 2 5 28 4 0 3 3.24 2.58 3.22 3.28 3 0 8 39 0 0 5 1.98 1.86 1.68 1.78 30 2 7 9 2 0 12 2.64 2.28 2.66 1.92 13 4 5 27 0 1 15 1.70 1.50 2.16 1.52 25 4 15 6 0 0 Transitional cell carcinoma 18 4.84 2.58 3.44 0.02 0 5 1 0 7 37 20 2.00 1.92 1.48 1.56 25 5 8 10 2 0 21 2.10 2.16 2.00 2.20 9 3 12 24 2 0 25 3.34 3.14 2.42 1.58 2 0 11 32 3 2 30 2.98 2.92 3.84 0.16 1 8 1 2 8 30 32 3.84 3.94 3.12 3.26 5 1 3 40 0 1 33 2.42 2.70 2.36 2.54 6 1 13 29 0 1 34 4.68 3.98 3.82 1.14 1 0 3 39 0 7 35 4.26 3.94 3.40 3.20 1 0 5 44 0 0 Urachal adenocarcinoma 22 1.66 3.38 1.70 2.14 8 1 19 21 1 0 26 1.40 3.02 1.42 0.34 6 13 4 0 26 1 27 3.10 1.84 2.82 2.64 3 0 12 32 2 1 28 2.16 2.92 2.70 1.16 5 6 4 23 10 2 29 2.48 3.14 1.58 3.42 4 1 12 32 1 0 Average 2.64 2.61 2.41 1.61 7.8 5.0 6.9 19.8 5.4 5.1

CEP, centromere enumeration probe; FISH, fluorescence in situ hybridization; LSI, locus specific indicator. * Based on 50 cell counts.

556 556




Of the 31 tumors, 29 (94%), independent of histologic subtype, demonstrated polysomic signal patterns in more than 10% of cells (Table 2). The bladder tumors that had the highest percentages of polysomic cells included TCC (mean, 68%; median, 66%; range, 20%-92%), adenocarcinoma (mean, 56%; median, 49%; range, 40%-86%), and UA (mean, 45%; median, 50%; range, 2%-66%; Figure 1D). However, these differences were not statistically significant. In the 2 specimens (specimen IDs 8 and 26, Table 2) in which fewer than

Although numerous studies have assessed the performance characteristics (sensitivity and specificity) of FISH with UroVysion in cases of “typical” UC,10-14 there is a lack

B 100

100

90

90

80

80

70 60 50

*

40 30 20 10 0

n=

4 4 ADCA Benign

7 SCC

6 SmCC

9 TCC

70 60 50 40 30 20 10 0

5 UA

C

n=

4 4 ADCA Benign

7 SCC

6 SmCC

9 TCC

5 UA

n=

4 4 ADCA Benign

7 SCC

6 SmCC

9 TCC

5 UA

D 100

100

* *

90

90

80

% of Cells With Polysomy

% of Cells With Homozygous 9p21 Loss

Discussion

% of Cells With Single Gain

% of Cells With Normal Signal Patterns

A

10% of the cells were polysomic, there was an extremely high percentage of cells demonstrating homozygous 9p21 deletion (76% and 80%).

70 60 50 40 30 20

70 60 50 40 30 20 10

10 0

80

n=

4 4 ADCA Benign

7 SCC

* 6 SmCC

9 TCC

5 UA

0

zFigure 1z Box and whisker plots showing the mean percentage of cells with normal (A), single chromosomal gain (B), homozygous 9p21 loss (C), and polysomic (D) fluorescence in situ hybridization signal patterns among individual patients with adenocarcinoma (ADCA), squamous cell carcinoma (SCC), small cell carcinoma (SmCC), transitional cell carcinoma (TCC), or urachal adenocarcinoma (UA) of the bladder. Horizontal lines within boxes indicate median values. Upper and lower ends of the boxes indicate the two quartiles (25th and 75th percentages). Upper and lower whiskers indicate 1.5 times the difference of the upper and lower ends of the boxes. Circles and asterisks indicate outliers whose values fall outside the whiskers; asterisks indicate cases that exceed the quartiles by more than 3 times the interquartile range. © American Society for Clinical Pathology

557


557 557


of data assessing the ability of this probe set to detect other histologic variants of bladder cancer. The results from the present study suggest that FISH with the UroVysion probe set should be able to detect other types of bladder cancer in urine specimens because it detected chromosomal abnormalities in paraffin-embedded adenocarcinoma, UA, TCC, SmCC, and SCC specimens (Images 1A-1E). This is important because FISH is frequently used for the detection of bladder cancer, and approximately 10% of patients with bladder cancer have one of these less common histologic variants. To determine the frequency of FISH-identified abnormalities in normal urothelial tissue, we performed FISH analysis on paraffin-embedded benign tissue from 4 patients. This is important because the paraffin sections that are used for FISH analysis are only 5 µm thick, and tissue sectioning leads to nuclear truncation.16 Table 2 shows the mean copy number for each of the 4 probes in normal tissue. As expected, normal cells showed fewer than the expected 2 signals per cell owing to nuclear truncation. The normal value data served as a baseline on which to determine if the chromosome abnormalities observed in the tumor cases were truly abnormal. There is relatively little known about the chromosomal alterations found in less common variants of bladder cancer. In the present study, in 94% of all patients with cancer, including the 91% of patients with the rarer variants of bladder cancer, polysomic signal patterns were found in more than 10% of cells within the individual specimens (Table 2). This finding was not unexpected because these rarer variants are aggressive tumors with numerous genetic abnormalities.4,6,17-26 Adenocarcinomas, for example, are rare aggressive bladder cancers (0.5%-2.0% of all bladder cancers) with reported 5-year survival rates as low as 11% to 55% for nonurachal adenocarcinomas and 27% to 61% for UA.18 The present study assessed both urachal and nonurachal adenocarcinomas because patients with UA are often younger and frequently have metastasis before therapeutic initiation.4,18 There were no significant differences in the mean percentage of cells demonstrating single chromosomal gains (36% vs 22%; P = .132), homozygous 9p21 deletion (26% vs 13%; P = .477), and polysomy (45% vs 56%; P = .507) when comparing tumor cells from patients with UA and adenocarcinoma, respectively. More important, FISH detected polysomic signal patterns in more than 20% of cells from all adenocarcinoma and UA specimens with the exception of 1 case (specimen 26, Table 2). That case showed a low percentage of cells with polysomy (only 2%) but a high percentage of cells with homozygous 9p21 loss (80%) and gain of chromosome 7 (60%), suggesting that the tumor is likely a near-diploid tumor with less marked aneuploidy than found in most of the other tumors in this study. SCC is the most common type of bladder cancer in Middle Eastern countries where schistosomiasis is endemic 558 558


and the second most common type of bladder cancer in Western countries, where it represents 2% to 5% of all bladder cancers.17 Patients with SCC of the bladder tend to have a poor prognosis, with a majority of patients dying within 3 years of diagnosis.26 An interesting finding of this study was that SCC specimens had significantly more cells (69.7%) with homozygous 9p21 deletion compared with all other cancers (19.8%; Figure 1C; P < .001). However, further analysis of the 7 SCC cases revealed that 75 (30.7%) of the 244 cells with homozygous 9p21 deletion also had polysomic signal patterns in combination with the homozygous 9p21 deletion (Table 2). This type of chromosomal pattern might suggest that 9p21 loss is an early event followed by events that lead to aneuploidy, which is consistent with previous studies that demonstrated that SCC bladder cancer has numerous chromosomal gains and a high frequency of chromosome 9p allelic loss.19-21,27,28 SmCCs are also rare (