Oncology–Original Article

Veterinary Pathology
2015, Vol. 52(1) 74-82
Differences in Expression of Uroplakin III, ª The Author(s) 2014
Reprints and permission:
[Link]/[Link]
Cytokeratin 7, and Cyclooxygenase-2 DOI: 10.1177/0300985814522819
[Link]
in Canine Proliferative Urothelial Lesions
of the Urinary Bladder

D. G. Sledge1, D. J. Patrick2, S. D. Fitzgerald1, Y. Xie3, and M. Kiupel1

Abstract
The expression of immunohistochemical markers that have been used in diagnosis and/or prognostication of urothelial tumors in
humans (uroplakin III [UPIII], cytokeratin 7 [CK7], cyclooxygenase-2 [COX-2], and activated caspase 3) was evaluated in a series
of 99 canine proliferative urothelial lesions of the urinary bladder and compared to the lesion classification and grade as defined by
the World Health Organization / International Society of Urologic Pathology consensus system. There were significant asso-
ciations between tumor classification and overall UPIII pattern (P ¼ 1.49  10–18), loss of UPIII (P ¼ 1.27  10–4), overall CK7
pattern (P ¼ 4.34  10–18), and COX-2 pattern (P ¼ 8.12  10–25). In addition, there were significant associations between depth
of neoplastic cell infiltration into the urinary bladder wall and overall UPIII pattern (P ¼ 1.54  10–14), loss of UPIII (P ¼ 2.07  10–4),
overall CK7 pattern (P ¼ 1.17  10–13), loss of CK7 expression (P ¼ .0485), and COX-2 pattern (P ¼ 8.23  10–21). There were no
significant associations between tumor classification or infiltration and caspase 3 expression pattern.

Keywords
dog, urinary bladder, oncology, immunohistochemistry, transitional cell carcinoma, uroplakin III, neoplasia, cytokeratin

Proliferative lesions of the urothelium of the canine urinary is the WHO / International Society of Urologic Pathology
bladder range from benign polyps and papillomas to carcino- (ISUP) consensus classification system, published in 1998 and
mas with varying metastatic potential. Also, inflammatory updated in 2004.6,7 In multiple studies, the WHO/ISUP consen-
conditions such as polypoid cystitis can form tumor-like sus classification system has been demonstrated to be signifi-
masses within the urinary bladder that can be confused with cantly associated with clinical outcome.2,32–34,37,42,43,51,59 It
urothelial neoplasms. Classification of urothelial proliferative has long been known that canine urothelial neoplasms are
lesions and histologic grading of urothelial carcinomas is key similar to urothelial neoplasms in humans in terms of
to accurate prognostication and treatment selection. A classifi- morphology, biological behavior, and response to chemother-
cation and grading scheme based on the 1986 World Health apy.14,18,19,27,28,59 Patrick et al recently examined the potential
Organization (WHO) classification scheme for human urinary use of the WHO/ISUP consensus classification system in clas-
bladder and urethral cancer was proposed in 1995 for use in sifying canine proliferative urothelial lesions.36 In that study,
evaluating urothelial neoplasms in dogs based on pattern of the authors demonstrated that the histomorphology of prolifera-
growth, nuclear atypia, and degree of infiltration into the urin- tive urothelial lesions was homologous between dogs and
ary bladder wall.50 In the initial description of this system’s use humans and that canine lesions could easily be classified
in dogs, significant correlations were found between tumor according to the system; however, data regarding the
grade and depth of infiltration, the presence of metastases, and
survival time. However, a subsequent smaller study failed to
find a significant correlation between grade and prognosis.41 1
Department of Pathobiology and Diagnostic Investigation, College of
To our knowledge, this scheme is not widely used in dogs and Veterinary Medicine, Michigan State University, East Lansing, MI, USA
has been replaced in human medicine by more modern classi- 2
MPI Research, Mattawan, MI, USA
3
fication schemes. There remains a need for a clear, reproduci- Pharmanet/i3, Haslett, MI, USA
ble, and well-accepted classification and grading scheme for
Corresponding Author:
urothelial proliferative lesions in dogs. Dodd G. Sledge, Michigan State University, Diagnostic Center for Population
Currently, the most widely accepted scheme for classifica- and Animal Health, 4125 Beaumont Road, Lansing, MI 48911, USA.
tion and grading of proliferative urothelial lesions in humans Email: sledged@[Link]
Sledge et al 75

prognostic relevance of classification system remain lacking. diagnostic cases were selected from the tissue archives of the
Additional study is needed of the biological differences Diagnostic Center for Population and Animal Health of
between classifications and grades of proliferative urothelial Michigan State University. Of these, 93 had previously been
lesions in dogs. classified and, where applicable, graded with the WHO/ISUP
Uroplakins compose a group of membrane-associated consensus classification system.36 Six additional diagnostic
proteins expressed by urothelial cells that are important for cases that presented to the center were included in the study set,
cell-to-cell adhesion and maintenance of water impermeabil- including additional low-grade neoplasms. For grading of these
ity.57,58 These proteins form a plaque-like complex along the additional samples, 5-mm sections of all samples were routinely
apical membrane of the umbrella cells that form the most processed and stained with hematoxylin and eosin for
superficial layer of the urothelium.57 Cytokeratin 7 (CK7) is microscopic examination. Proliferative urothelial lesions were
a cytokeratin expressed by simple epithelium as well as differ- categorized according to the WHO/ISUP consensus classifica-
entiated urothelial cells and a variety of carcinomas in humans, tion system as previously described and summarized in
including those of urothelial, pancreatic, cholangiolar, and Table 1.36 For urothelial carcinomas, the degree of infiltration
ovarian origin.49,52 Uroplakin III (UPIII) and CK7 are used into the urinary bladder wall was scored as no infiltration, infil-
in dogs and humans as diagnostic markers of urothelial differ- tration into the substantia propria, or infiltration into the tunica
entiation in primary tumors and metastases.8,13,31,35,39,47 In muscularis.
humans, loss of UPIII has been associated with several Of the 99 proliferative urothelial lesions examined, 44 were
prognostic features.13,25 nonneoplastic and categorized as either urothelial polyps or
The inducible enzyme cyclooxygenase-2 (COX-2) and the polypoid cystitis. Of the 55 neoplasms, there were 2 urothelial
resulting production of prostaglandin E2 have been ascribed sig- papillomas and 1 papillary urothelial neoplasm of low malig-
nificant roles in carcinogenesis, including immunosuppression, nant potential (PUNLMP). The remainders of the urothelial
inhibition of apoptosis, increased metastatic potential of neoplas- neoplasms were papillary urothelial carcinomas of varying
tic epithelial cells, promotion of drug resistance, and stimulation grade. Low-grade (grade 1) papillary urothelial carcinomas
of angiogenesis.9,11,23,54–56 Numerous studies have shown were rare, with only 2 being included in the set. Both these had
significant correlations between COX-2 expression and tumor some degree of infiltration into the urinary bladder wall, with
grade, infiltration, metastasis, and survival.21,24,30,40,44–46,53 one sample having infiltrative clusters of neoplastic urothelial
Concordantly, a decreased risk for urinary bladder cancer devel- cells within the muscularis. Papillary carcinomas infiltrated
opment has been seen in humans undergoing long-term nonster- at least into the substantia propria in 49 of 52 cases (94%) and
oidal anti-inflammatory therapy, and in vitro and in vivo studies into the muscularis in 21 of 52 (40%).
have suggested potential use of COX-2 inhibitors in treat-
ment.3,5,10,29,53 In humans and dogs, COX-2 is not expressed
by normal urothelium of the urinary bladder.17,24 Substantial
Immunohistochemistry
expression of COX-2, however, has been observed in transitional Sections (5 mm) of all samples were processed for immuno-
cell carcinomas.17,20,22 histochemistry and labeled with a mouse monoclonal anti-
Caspase 3 is an effector or executioner caspase that is UPIII antibody (1:5, RDI, Fitzgerald Industries Intl, Concord,
activated by intrinsic and extrinsic apoptosis signaling path- MA), a mouse monoclonal anti-CK7 antibody (1:75, Dako
ways to cleave multiple cellular structural and repair pro- Cytomation, Carpentaria, CA), a rabbit polyclonal anti-COX-
teins.4,12,38 Due to the fact that this protein is activated late 2 antibody (1:100, Cayman Chemical Company, Ann Arbor,
in the apoptotic pathway, immunohistochemical detection of MI), or rabbit polyclonal anti-activated caspase 3 antibody
activated caspase 3 has been used to evaluate apoptotic rate.1,48 (1:5000, RDI). Deparaffinization, antigen retrieval, immuno-
In human urinary bladder cancers, expression of caspase 3 has histochemical labeling with 3,30 -diaminobenzidine chromogen,
been suggested to have prognostic significance.15,16,26 and counterstaining with hematoxylin were performed on the
The goals of the current study were twofold: first, to Bond Max Automated Staining System (Vision BioSystems,
evaluate the expression of UPIII, CK7, COX-2, and caspase Leica, Bannockburn, IL) with the Bond Polymer Detection
3 in nonneoplastic and neoplastic proliferative lesions of System (Vision BioSystems). Sections of normal canine
the canine urothelium of the urinary bladder; second, to urothelium were similarly labeled as positive controls for UPIII
correlate the observed patterns of expression of each of these and CK7. A canine squamous cell carcinoma known to express
markers with specific lesion classification and grade as defined COX-2 and a lymph node with large numbers of cells positive
by the WHO/ISUP consensus classification system. for activated caspase 3 were respectively used as positive
controls for these antibodies. For negative controls, homolo-
gous nonimmune sera or buffer replaced primary antibodies.
Materials and Methods Immunoreactivity for UPIII and CK7 was scored according
to overall pattern within the urothelium and partial loss of
Selection of Cases and Histologic Classification immunoreactivity (Figs. 1–6). Immunoreactivity for UPIII and
A series of 99 formalin-fixed, paraffin-embedded proliferative CK7 in positively labeled cells was variably perimembranous,
urothelial lesions from 99 dogs that had been submitted as predominately cytoplasmic without distinct perimembrane
76 Veterinary Pathology 52(1)

Table 1. Histologic Features of Proliferative Urothelial Lesions According to the World Health Organization / International Society of Urologic
Pathology Consensus Classification System.a

Classification Mitoses Histologic Characteristics

Nonneoplastic lesions
Polyp/polypoid cystitis Rare and confined to the basal cell layers Exophytic protrusions of mucosa and supporting fibrovascular
stroma lacking true papillary fronds; often associated with
stromal edema and inflammation; may occur as single
fibroepithelial polyp or in multiples as polypoid cystitis
Neoplastic lesions
Urothelial papilloma Rare and confined to the basal cell layer <6 cell layers lining papillary fronds; orderly arrangement of
cells; nuclei are uniform in size, shape, and chromatin staining
PUNLMPb Rare and confined to the basal cell layer >6 cell layers lining papillary fronds; orderly arrangement of
cells; nuclei are uniform in size, shape, and chromatin staining
Papillary carcinoma
Grade 1 Infrequent and limited to the basal half of the Orderly appearance with recognizable variation in architectural
epithelium or cytologic features at low magnification; mild anisokaryosis
with variable chromatin staining
Grade 2 Low to moderate numbers throughout all levels Overall disorderly appearance with retainment of some degree
of the urothelium with possible atypia of polarity; irregular clustering and disorganization of cells;
moderate anaplasia; moderate anisokaryosis with prominent
nucleoli and clumped chromatin
Grade 3 High numbers throughout all levels with Complete loss of polarity; irregular clustering and
common atypia disorganization of cells; marked pleomorphism, anisocytosis,
and anisokaryosis; prominent nucleoli and clumped
chromatin
a
Adapted from descriptions made by Patrick et al.36
b
Papillary urothelial neoplasm of low malignant potential.

labeling, or associated with both the membrane the cell mem- Statistical Analysis
brane and the cytoplasm. Specifically, for overall pattern of
Statistical Analysis Software version 9.1.3 (2002, SAS Institute
UPIII and CK7 immunoreactivity within the urothelium, pat-
Inc, Cary, NC) was used for the data analysis. Fisher exact test
tern 1 was defined by labeling limited to the most superficial
was used to test the association between grade or degree of
layer of cells (Figs. 1, 4); pattern 2 was defined by labeling
infiltration and the pattern of immunoreactivity for all
extending to the middle layer of the urothelium (Fig. 2); pattern
evaluated markers and loss of immunoreactivity for UPIII and
3 was defined by cell labeling throughout the full thickness of
CK7. For all statistical analyses, lesions were categorized as
the urothelium (Fig. 5); and pattern 4 was defined by immunor-
follows: nonneoplastic lesions (urothelial polyps and polypoid
eactivity that was patchy and randomly distributed (Figs. 3, 6).
cystitis), low-grade neoplasms (urothelial papillomas,
Partial loss of immunoreactivity was noted when greater than
PUNLMPs, and grade 1 urothelial carcinomas), grade 2 urothe-
50% of epithelial cells were immunonegative and there were
lial carcinomas, and grade 3 urothelial carcinomas.
areas within the proliferative urothelium with no immunoposi-
Significance was set at P ¼ .05.
tive cells in at least 2 contiguous high-power (400) fields.
Immunoreactivity for COX-2 and activated caspase 3 was
classified according to overall pattern within the urothelium. Results
Cells immunoreactive for COX-2 had diffuse cytoplasmic to
perinuclear labeling. Cells immunoreactive for activated Immunohistochemistry
caspase 3 had diffuse or, more often, finely granular labeling UPIII and CK7. The urothelium of control urinary bladders and
within the cytoplasm and/or nucleus. In pattern 1 for these mar- areas of normal urothelium in tumor samples demonstrated
kers, immunoreactivity was limited to the superficial (1–3) cell strong expression of UPIII and CK7 diffusely throughout the
layers (Fig. 7). In pattern 2, cells were labeled throughout the superficial layers (umbrella cells). This distribution of
full thickness of the urothelium (Fig. 8). In pattern 3, immunor- immunolabeling typified UPIII and CK7 pattern 1.
eactivity was patchy and randomly distributed throughout the In the hyperplastic urothelium of polyps and polypoid
proliferative urothelium, but greater than 15% of neoplastic cystitis, expression of UPIII and CK7 was generally limited
cells were positively labeled (Fig. 9). to the superficial (umbrella) cell layers (UPIII pattern 1; CK7
For each marker, any section that completely lacked immu- pattern 1) or extended to only the midportion of the urothelium
noreactivity was excluded from analysis, as it could not be (UPIII pattern 2). Only 1 urothelial polyp had UPIII pattern 4,
determined whether this was true loss of expression or was and none had CK7 pattern 4. The PUNLMP and urothelial
artifactual. papillomas demonstrated UPIII pattern 3 and CK7 pattern 2
Sledge et al 77

Figure 1. Urinary bladder; dog. Hyperplastic urothelium with uroplakin III (UPIII) pattern 1: Immunolabeling (brown) is limited to the superficial (1–2)
cell layers. 3,30 -Diaminobenzidine (DAB) chromogen, hematoxylin counterstain. Figure 2. Urinary bladder; dog. Hyperplastic urothelium with UPIII
pattern 2: UPIII is expressed strongly by all but the most-basal cell layers. DAB chromogen, hematoxylin counterstain. Figure 3. Urinary bladder; dog.
Papillary urothelial carcinoma grade 2 with UPIII pattern 4: UPIII expression is randomly distributed throughout the neoplasm. Individual or small
clusters of neoplastic cells strongly express UPIII, while numerous islands of neoplastic cells lack expression. DAB chromogen, hematoxylin counter-
stain. Figure 4. Urinary bladder; dog. Hyperplastic urothelium with cytokeratin 7 (CK7) pattern 1: Strong expression of CK7 is limited to the super-
ficial (1–2) cell layers. DAB chromogen, hematoxylin counterstain. Figure 5. Urinary bladder; dog. Papillary urothelial carcinoma grade 2 with CK7
pattern 3: CK7 is strongly expressed by all cell layers. DAB chromogen, hematoxylin counterstain. Figure 6. Urinary bladder; dog. Papillary urothelial
carcinoma grade 2 with CK7 pattern 4: CK7 expression is patchy. While most cells have strong expression of CK7, individual or small groups of
neoplastic cells randomly distributed throughout the mass lack expression. DAB chromogen, hematoxylin counterstain. Figure 7. Urinary bladder;
dog. Hyperplastic urothelium with cyclooxygenase-2 (COX-2) pattern 1: Immunolabeling for COX-2 is limited to the superficial (1–2) cell layers. DAB
chromogen, hematoxylin counterstain. Figure 8. Urinary bladder; dog. Hyperplastic urothelium with COX-2 pattern 2: COX-2 is strongly expressed
by cells throughout all cell layers. DAB chromogen, hematoxylin counterstain. Figure 9. Urinary bladder; dog. Papillary urothelial carcinoma grade 2
with COX-2 pattern 3: Cells that strongly express COX-2 are randomly distributed throughout the neoplastic cell population. DAB chromogen,
hematoxylin counterstain.
78 Veterinary Pathology 52(1)

Table 2. Immunohistochemical Scoring of Uroplakin III Expression in Canine Proliferative Urothelial Lesions by Overall Pattern, n (%).

Pattern

Classification 1 2 3 4 Loss

Polyp/polypoid cystitis, n ¼ 44 19 (43) 20 (46) 3 (7) 1 (2) 1 (2)

Urothelial papilloma, n ¼ 1 0 0 1 (100) 0 0
PUNLMP,a n ¼ 1 0 0 1 (100) 0 0
Papillary carcinoma
Grade 1, n ¼ 2 0 0 0 2 (100) 0
Grade 2, n ¼ 17 1 (6) 0 1 (6) 15 (88) 1 (6)
Grade 3, n ¼ 33 2 (6) 0 2 (6) 27 (82) 14 (42)
a
Papillary urothelial neoplasm of low malignant potential.

Table 3. Immunohistochemical Scoring of Cytokeratin 7 Expression in Canine Proliferative Urothelial Lesions by Overall Pattern, n (%).

Pattern

Classification 1 2 3 4 Loss

Polyp/polypoid cystitis, n ¼ 44 16 (36) 25 (57) 3 (7) 0 0

Urothelial papilloma, n ¼ 2 0 1 (50) 1 (50) 0 0
PUNLMP,a n ¼ 1 0 0 1 (100) 0 0
Papillary carcinoma
Grade 1, n ¼ 2 0 1 (50) 0 1 (50) 0
Grade 2, n ¼ 17 2 (12) 0 5 (29) 10 (59) 2 (12)
Grade 3, n ¼ 33 0 1 (3) 6 (18) 25 (76) 1 (3)
a
Papillary urothelial neoplasm of low malignant potential.

or 3. With few exceptions, papillary carcinomas of all grades The PUNLMP and the 2 papillomas demonstrated COX-2
demonstrated patchy randomly distributed UPIII expression pattern 1 while the grade 1 papillary carcinomas exhibited
consistent with pattern 4. Partial loss of UPIII expression was either pattern 2 or 3. All high-grade (grades 2 and 3) papillary
not detected in any grade 1 papillary carcinoma and in only 1 urothelial carcinomas in which COX-2 expression was detected
grade 2 papillary carcinoma. In contrast, 14 of 33 (42%) grade exhibited COX-2 pattern 3 (Table 4).
3 papillary carcinomas had partial loss of expression of UPIII. Activated caspase 3. Approximately 10% of cells in the
More grade 2 and 3 carcinomas had a CK7 pattern 3 than had a lymph node used as a positive control for activated caspase 3 had
UPIII pattern 3 (11 of 50 CK7 pattern 3 cases compared to 3 of positive finely granular labeling of the cytoplasm and/or nucleus.
50 UPIII pattern 3 cases). The majority (71%) of grade 2 and 3 In the normal urothelium of control urinary bladders and adja-
urothelial carcinomas, however, had a CK7 pattern 4 similar to cent to proliferative lesions, immunoreactive cells composed
that observed for UPIII. Only grade 2 and 3 urothelial carcino- less than 10% of the total urothelium and were largely limited
mas had significant loss of CK7 expression (Tables 2 and 3). to the superficial (1–2) cell layers. The total percentage of cas-
COX-2. The squamous cell carcinoma from the digit of a pase 3–immunoreactive cells ranged from 10% to 40% in all
dog used as a positive control demonstrated strong positive other proliferative lesions; however, there was no appreciable
cytoplasmic and mainly perinuclear immunoreactivity for variation in the percentage of positive cells among tumor types.
COX-2 in 30% of neoplastic cells. There was no positive Immunoreactivity for activated caspase 3 was most commonly
immunoreactivity for COX-2 in any part of the normal urothe- noted in the superficial-most cell layers of all proliferative
lium from control urinary bladders; however, in the superficial lesions (caspase 3 pattern 1) but was also seen diffusely through-
layers of nonproliferative urothelium adjacent to proliferative out the urothelium (caspase 3 pattern 2) and in a randomly
lesions, there were often a small percentage (< 10%) of distributed patchy distribution (caspase 3 pattern 3) (Table 5).
COX-2 positive cells. A summary of the predominant patterns of immunoreactivity for
In all proliferative lesions, at least 10% of proliferative UPIII, CK7, and COX-2 observed in each proliferative urothelial
urothelial cells exhibited positive COX-2 immunoreactivity. lesion classification is presented in Table 6.
In urothelial polyps and polypoid cystitis, COX-2 expression
in the proliferative urothelium was usually restricted to the
superficial layers (COX-2 pattern 1) or was less commonly full Statistical Analysis
thickness (COX-2 pattern 2). Only 1 polyp had randomly dis- Based on Fisher exact test, there were significant associations
tributed and patchy expression of COX-2 (COX-2 pattern 3). between tumor classification and overall UPIII pattern
Sledge et al 79

Table 4. Immunohistochemical Scoring of Cyclooxygenase-2 Table 5. Immunohistochemical Scoring of Activated Caspase 3 in

Expression in Canine Proliferative Urothelial Lesions by Overall Canine Proliferative Urothelial Lesions by Overall Pattern, n (%).
Pattern, n (%).
Pattern
Pattern
Classification 1 2 3
Classification 1 2 3
Polyp/polypoid cystitis, n ¼ 44 27 (61) 13 (30) 4 (9)
Polyp/polypoid cystitis, n ¼ 44 32 (73) 11 (25) 1 (2) Urothelial papilloma, n ¼ 2 1 (50) 0 1 (50)
Urothelial papilloma, n ¼ 2 1 (50) 1 (50) 0 PUNLMP,a n ¼ 1 1 (100) 0 0
PUNLMP,a n ¼ 1 1 (100) 0 0 Papillary carcinoma
Papillary carcinoma Grade 1, n ¼ 2 1 (50) 0 1 (50)
Grade 1, n ¼ 2 1 (50) 0 1 (50) Grade 2, n ¼ 16 10 (63) 3 (19) 3 (19)
Grade 2, n ¼ 17 0 0 17 (100) Grade 3, n ¼ 30 22 (73) 4 (13) 4 (13)
Grade 3, n ¼ 33 0 0 33 (100) a
Papillary urothelial neoplasm of low malignant potential.
a
Papillary urothelial neoplasm of low malignant potential.

Table 6. Summary of Patterns of Immunoreactivity of Uroplakin III, Cytokeratin 7, and Cyclooxygenase-2 in Proliferative Urothelial Lesions.

Classification Uroplakin III Cytokeratin 7 COX-2

Polyp, polypoid cystitis 89% had expression limited to cells 93% had expression limited to cells 98% had expression limited to the
in the superficial-most cell layer in the superficial-most cell layer superficial one-third of cell layers
(pattern 1) or extending to the (pattern 1) or extending to the (pattern 1) or throughout all cell
middle-most layer (pattern 2) middle-most layer (pattern 2) layers (pattern 2)
Urothelial papilloma, All cases had expression throughout All cases had expression that All cases had expression limited to
PUNLMPa all cell layers (pattern 3) extended to the middle-most layer the superficial one-third of cell
of the urothelium (pattern 2) or layers (pattern 1) or throughout
throughout all cell layers (pattern 3) all cell layers (pattern 2)
Papillary carcinoma 83% had randomly distributed, 21% had expression throughout all 98% had randomly distributed and
patchy expression (pattern 4) cell layers (pattern 3); 68% had patchy expression (pattern 3)
randomly distributed, patchy
expression (pattern 4)
Grade 1 No significant loss of expression was No significant loss of expression was 50% had expression limited to the
observed observed superficial-most one-third of cell
layers (pattern 1); 50% had ran-
domly distributed, patchy expres-
sion (pattern 3)
Grade 2 6% had significant areas of 12% had significant areas of All cases had randomly distributed,
expression loss expression loss patchy expression (pattern 3)
Grade 3 42% had significant areas of 3% had significant areas of All cases had randomly distributed,
expression loss expression loss patchy expression (pattern 3)
a
Papillary urothelial neoplasm of low malignant potential.

(P ¼ 1.49  10–18), loss of UPIII (P ¼ 1.27  10–4), overall CK7 cases of polypoid cystitis, urothelial papillomas and
pattern (P ¼ 4.34  10–18), and COX-2 pattern (P ¼ 8.12  10–25). PUNLMPs, and papillary carcinomas. However, no association
Also by Fisher exact test, there were significant associations was observed between the expression of activated caspase 3
between depth of neoplastic cell infiltration and overall UPIII pat- and tumor classification, grade, or depth of infiltration into the
tern (P ¼ 1.54  10–14), loss of UPIII (P ¼ 2.07  10–4), overall urinary bladder wall.
CK7 pattern (P ¼ 1.17  10–13), loss of CK7 expression Overall, urothelial polyps and cases of polypoid cystitis
(P ¼ .0485), and COX-2 pattern (P ¼ 8.23  10–21). There were predominately had expression of UPIII and CK7 that either was
no significant associations between tumor classification and loss limited to the superficial-most cell layers (UPIII/CK7 pattern
of CK7 or caspase 3 pattern. There were also no significant asso- 1) or extended to the middle-most cell layer (UPIII/CK7 pat-
ciations between depth of infiltration and caspase 3 expression. tern 2). Urothelial papillomas and a PUNLMP had UPIII and
CK7 expression throughout the full thickness of the urothelium
or extending at least up to the middle-most cell layer. The
Discussion majority of papillary carcinomas, in contrast, had randomly
In the current study, strong differences in expression of UPIII, distributed patchy immunoreactivity for UPIII and CK7
CK7, and COX-2 were observed between urothelial polyps and (UPIII/CK7 pattern 4). In addition, there was often partial loss
80 Veterinary Pathology 52(1)

of UPIII and CK7 expression in high-grade carcinomas that lesions, as observed in the current study—might represent a
invaded into the urinary bladder wall. consequence of such surface irritation.
Based on the association with infiltration into the urinary blad- Overall, there were distinct differences in the patterns of
der wall, loss of UPIII and CK7 in urothelial carcinomas may UPIII, CK7, and COX-2 expression in canine urothelial prolif-
suggest a lack of differentiation or epithelial-mesenchymal erative lesions of the urinary bladder as defined and categor-
transition in that favors infiltration. In humans, loss of UPIII ized by the WHO/ISUP consensus classification system. As
expression is frequently seen in metastases of high-grade the cases included in the study set comprised diagnostic sam-
tumors and has been associated with lymphovascular infiltra- ples, reliable follow-up data regarding clinical outcome were
tion, stage, and grade.13,25 Loss of cell membrane adhesion not available. Definitive prospective studies of the clinical out-
molecules may also result in an increased propensity for metas- come based on these markers and the WHO/ISUP consensus
tasis in dogs; however, information on clinical outcome includ- classification system in dogs remain lacking, but this study
ing presence of metastasis and survival was not available for encourages the continued prognostic evaluation of the
the examined canine tumor set. The observed significant loss WHO/ISUP consensus classification system and these immuno-
of UPIII and CK7 in many high-grade carcinomas suggests that histochemical markers in canine proliferative urothelial lesions.
when used as diagnostic markers on small specimens, some
urothelial carcinomas might not be positively labeled. Acknowledgements
There was a clear difference in the location of COX-2- The work described in this article composed a portion of Dr Sledge’s
positive cells within the urothelium between nonneoplastic and PhD graduate program, for which a fellowship was provided from
neoplastic lesions as well as among papillary urothelial carci- Bristol-Meyers Squibb through the American College of Veterinary
nomas with different degrees of infiltration into the urinary Pathologists / Society of Toxicological Pathologists Coalition.
bladder wall. Urothelial polyps, cases of polypoid cystitis,
urothelial papillomas, and the one PUNLMP exhibited expres- Declaration of Conflicting Interests
sion of COX-2 in the superficial layers of the urothelium The author(s) declared no potential conflicts of interest with respect to
(COX-2 pattern 1) or, rarely, diffusely throughout the urothe- the research, authorship, and/or publication of this article.
lium (COX-2 pattern 2). In contrast, papillary carcinomas had
a randomly distributed, patchy pattern of COX-2 expression Funding
(COX-2 pattern 3). The author(s) received no financial support for the research, author-
Previous studies of COX-2 in canine urinary bladder urothe- ship, and/or publication of this article.
lium have focused on the difference in COX-2 expression
between normal urinary bladder and urothelial carcinomas and References
reported differences as a percentage of COX-2 positive cells. In 1. Abu-Qare AW, Abou-Donia MB. Biomarkers of apoptosis:
the current study, COX-2 immunoreactivity was observed release of cytochrome c, activation of caspase-3, induction of 8-
in the majority of nonneoplastic and neoplastic lesions, often hydroxy-20 -deoxyguanosine, increased 3-nitrotyrosine, and
in a relatively high percentage of cells, precluding the use of alteration of p53 gene. J Toxicol Environ Health B Crit Rev.
percentage of positive cells as a differentiating criterion. In one 2001;4:313–332.
previous immunohistochemical study, only 30 of 52 (58%) 2. Alsheikh A, Mohamedali Z, Jones E, et al. Comparison of the
canine transitional cell carcinomas demonstrated COX-2 WHO/ISUP classification and cytokeratin 20 expression in pre-
expression by IHC.22 In a study of 18 canine transitional cell dicting the behavior of low-grade papillary urothelial tumors:
carcinomas, all tumors expressed COX-2, and 3 showed more World/Health Organization/International Society of Urologic
than 30% of cells to be positive for COX-2.20 The reasons for Pathology. Mod Pathol. 2001;14:267–272.
discrepancy between these studies in terms of the numbers of 3. Castelao JE, Yuan JM, Gago-Dominguez M, et al. Non-steroidal
cases and the number of cells within a given case that are posi- anti-inflammatory drugs and bladder cancer prevention. Br J Can-
tively labeled for COX-2 may be due to the low number of cer. 2000;82:1364–1369.
cases examined in each study, differing methods of fixation and 4. Cohen GM. Caspases: the executioners of apoptosis. Biochem J.
processing of the tissues, differences in immunohistochemical 1997;326(pt 1):1–16.
methods including the anti-COX-2 antibody used, or interob- 5. Dhawan D, Jeffreys AB, Zheng R, et al. Cyclooxygenase-2
server variation. dependent and independent antitumor effects induced by cele-
To our knowledge, COX-2 expression has not been previ- coxib in urinary bladder cancer cells. Mol Cancer Ther. 2008;7:
ously evaluated in nonneoplastic proliferative urothelial lesions 897–904.
of the urinary bladder in dogs. In contrast to normal urothelium 6. Eble J, Sauter G, Epstein JI, et al. World Health Organization
from urinary bladders used as controls, all urothelial polyps and Classification of Tumours: Pathology and Genetics of Tumours
cases of polypoid cystitis had expression of COX-2. Urothelial of the Urinary System and Male Genital Organs. Lyon, France:
polyps and polypoid cystitis are often associated with mucosal IARC Press; 2004.
irritation such as what might occur with cystoliths. Expression 7. Epstein JI, Amin MB, Reuter VR, et al. The World Health Orga-
of COX-2—often throughout the superficial portion or the full nization/International Society of Urological Pathology consensus
thickness of the proliferative urothelium of nonneoplastic classification of urothelial (transitional cell) neoplasms of the
Sledge et al 81

urinary bladder: Bladder Consensus Conference Committee. Am J 24. Margulis V, Shariat SF, Ashfaq R, et al. Expression of
Surg Pathol. 1998;22:1435–1448. cyclooxygenase-2 in normal urothelium, and superficial and
8. Espinosa de los Monteros A, Fernandez A, Millan MY, et al. advanced transitional cell carcinoma of bladder. J Urol. 2007;
Coordinate expression of cytokeratins 7 and 20 in feline and 177:1163–1168.
canine carcinomas. Vet Pathol. 1999;36:179–190. 25. Matsumoto K, Satoh T, Irie A, et al. Loss expression of uroplakin
9. Gately S, Li WW. Multiple roles of COX-2 in tumor angiogen- III is associated with clinicopathologic features of aggressive
esis: a target for antiangiogenic therapy. Semin Oncol. 2004; bladder cancer. Urology. 2008;72:444–449.
31(2)(suppl 7):2–11. 26. Mitra AP, Castelao JE, Hawes D, et al. Combination of molecular
10. Gee J, Lee IL, Jendiroba D, et al. Selective cyclooxygenase-2 alterations and smoking intensity predicts bladder cancer out-
inhibitors inhibit growth and induce apoptosis of bladder cancer. come: a report from the Los Angeles Cancer Surveillance Pro-
Oncol Rep. 2006;15:471–477. gram. Cancer. 2013;119:756–765.
11. Greenhough A, Smartt HJ, Moore AE, et al. The COX-2/PGE2 27. Mohammed SI, Bennett PF, Craig BA, et al. Effects of the
pathway: key roles in the hallmarks of cancer and adaptation to cyclooxygenase inhibitor, piroxicam, on tumor response, apopto-
the tumour microenvironment. Carcinogenesis. 2009;30: sis, and angiogenesis in a canine model of human invasive urinary
377–386. bladder cancer. Cancer Res. 2002;62:356–358.
12. Grutter MG. Caspases: key players in programmed cell death. 28. Mohammed SI, Craig BA, Mutsaers AJ, et al. Effects of the
Curr Opin Struct Biol. 2000;10:649–655. cyclooxygenase inhibitor, piroxicam, in combination with che-
13. Gruver AM, Amin MB, Luthringer DJ, et al. Selective immuno- motherapy on tumor response, apoptosis, and angiogenesis in a
histochemical markers to distinguish between metastatic high- canine model of human invasive urinary bladder cancer. Mol Can-
grade urothelial carcinoma and primary poorly differentiated cer Ther. 2003;2:183–188.
invasive squamous cell carcinoma of the lung. Arch Pathol Lab 29. Mohammed SI, Dhawan D, Abraham S, et al. Cyclooxygenase
Med. 2012;136:1339–1346. inhibitors in urinary bladder cancer: in vitro and in vivo effects.
14. Henry CJ, McCaw DL, Turnquist SE, et al. Clinical evaluation of Mol Cancer Ther. 2006;5:329–336.
mitoxantrone and piroxicam in a canine model of human invasive 30. Mohammed SI, Knapp DW, Bostwick DG, et al. Expression of
urinary bladder carcinoma. Clin Cancer Res. 2003;9:906–911. cyclooxygenase-2 (COX-2) in human invasive transitional cell carci-
15. Karam JA, Lotan Y, Karakiewicz PI, et al. Use of combined apop- noma (TCC) of the urinary bladder. Cancer Res. 1999;59:5647–5650.
tosis biomarkers for prediction of bladder cancer recurrence and 31. Moll R, Laufer J, Wu XR, et al. [Uroplakin III, a specific mem-
mortality after radical cystectomy. Lancet Oncol. 2007;8: brane protein of urothelial umbrella cells, as a histological mar-
128–136. kers for metastatic transitional cell carcinomas]. Verh Dtsch
16. Karamitopoulou E, Rentsch CA, Markwalder R, et al. Prognostic Ges Pathol. 1993;77:260–265.
significance of apoptotic cell death in bladder cancer: a tissue 32. Oosterhuis JW, Schapers RF, Janssen-Heijnen ML, et al. Histolo-
microarray study on 179 urothelial carcinomas from cystectomy gical grading of papillary urothelial carcinoma of the bladder:
specimens. Pathology. 2010;42:37–42. prognostic value of the 1998 WHO/ISUP classification system
17. Khan KN, Knapp DW, Denicola DB, et al. Expression of and comparison with conventional grading systems. J Clin
cyclooxygenase-2 in transitional cell carcinoma of the urinary Pathol. 2002;55:900–905.
bladder in dogs. Am J Vet Res. 2000;61:478–481. 33. Pan CC, Chang YH, Chen KK, et al. Constructing prognostic
18. Knapp DW, Glickman NW, Denicola DB, et al. Naturally- model incorporating the 2004 WHO/ISUP classification for
occurring canine transitional cell carcinoma of the urinary bladder patients with non-muscle-invasive urothelial tumours of the urin-
A relevant model of human invasive bladder cancer. Urol Oncol. ary bladder. J Clin Pathol. 2010;63:910–915.
2000;5:47–59. 34. Pan CC, Chang YH, Chen KK, et al. Prognostic significance of
19. Knapp DW, Glickman NW, Widmer WR, et al. Cisplatin versus the 2004 WHO/ISUP classification for prediction of recurrence,
cisplatin combined with piroxicam in a canine model of human progression, and cancer-specific mortality of non-muscle-
invasive urinary bladder cancer. Cancer Chemother Pharmacol. invasive urothelial tumors of the urinary bladder: a clinicopatho-
2000;46:221–226. logic study of 1,515 cases. Am J Clin Pathol. 2010;133:788–795.
20. Knottenbelt C, Mellor D, Nixon C, et al. Cohort study of COX-1 35. Parker DC, Folpe AL, Bell J, et al. Potential utility of uroplakin
and COX-2 expression in canine rectal and bladder tumours. J III, thrombomodulin, high molecular weight cytokeratin, and
Small Anim Pract. 2006;47:196–200. cytokeratin 20 in noninvasive, invasive, and metastatic urothelial
21. Komhoff M, Guan Y, Shappell HW, et al. Enhanced expression of (transitional cell) carcinomas. Am J Surg Pathol. 2003;27:1–10.
cyclooxygenase-2 in high grade human transitional cell bladder 36. Patrick DJ, Fitzgerald SD, Sesterhenn IA, et al. Classification of
carcinomas. Am J Pathol. 2000;157:29–35. canine urinary bladder urothelial tumours based on the World
22. Lee JY, Tanabe S, Shimohira H, et al. Expression of Health Organization/International Society of Urological Pathol-
cyclooxygenase-2, P-glycoprotein and multi-drug resistance- ogy consensus classification. J Comp Pathol. 2006;135:190–199.
associated protein in canine transitional cell carcinoma. Res Vet 37. Pich A, Chiusa L, Formiconi A, et al. Biologic differences
Sci. 2007;83:210–216. between noninvasive papillary urothelial neoplasms of low malig-
23. Liu B, Qu L, Tao H. Cyclo-oxygenase 2 up-regulates the effect of nant potential and low-grade (grade 1) papillary carcinomas of the
multidrug resistance. Cell Biol Int. 2010;34:21–25. bladder. Am J Surg Pathol. 2001;25:1528–1533.
82 Veterinary Pathology 52(1)

38. Porter AG, Janicke RU. Emerging roles of caspase-3 in apoptosis. 49. Tot T. Cytokeratins 20 and 7 as biomarkers: usefulness in discri-
Cell Death Differ. 1999;6:99–104. minating primary from metastatic adenocarcinoma. Eur J Cancer.
39. Ramos-Vara JA, Miller MA, Boucher M, et al. Immunohisto- 2002;38:758–763.
chemical detection of uroplakin III, cytokeratin 7, and cytokeratin 50. Valli VE, Norris A, Jacobs RM, et al. Pathology of canine bladder
20 in canine urothelial tumors. Vet Pathol. 2003;40:55–62. and urethral cancer and correlation with tumour progression and
40. Ristimaki A, Nieminen O, Saukkonen K, et al. Expression of survival. J Comp Pathol. 1995;113:113–130.
cyclooxygenase-2 in human transitional cell carcinoma of the 51. Vardar E, Gunlusoy B, Minareci S, et al. Evaluation of p53
urinary bladder. Am J Pathol. 2001;158:849–853. nuclear accumulation in low- and high-grade (WHO/ISUP classi-
41. Rocha TA, Mauldin GN, Patnaik AK, et al. Prognostic factors in fication) transitional papillary carcinomas of the bladder for
dogs with urinary bladder carcinoma. J Vet Intern Med. 2000;14: tumor recurrence and progression. Urol Int. 2006;77:27–33.
486–490. 52. Vojtesek B, Staskova Z, Nenutil R, et al. A panel of monoclonal
42. Samaratunga H, Makarov DV, Epstein JI. Comparison of WHO/ antibodies to keratin no. 7: characterization and value in tumor
ISUP and WHO classification of noninvasive papillary urothelial diagnosis. Neoplasma. 1990;37:333–342.
neoplasms for risk of progression. Urology. 2002;60:315–319. 53. Wadhwa P, Goswami AK, Joshi K, et al. Cyclooxygenase-2
43. Schned AR, Andrew AS, Marsit CJ, et al. Survival following the expression increases with the stage and grade in transitional cell
diagnosis of noninvasive bladder cancer: WHO/International carcinoma of the urinary bladder. Int Urol Nephrol. 2005;37:
Society of Urological Pathology versus WHO classification sys- 47–53.
tems. J Urol. 2007;178(4)(pt 1):1196–1200. 54. Wang D, Dubois RN. Prostaglandins and cancer. Gut. 2006;55:
44. Shariat SF, Matsumoto K, Kim J, et al. Correlation of 115–122.
cyclooxygenase-2 expression with molecular markers, pathologi- 55. Wang MT, Honn KV, Nie D. Cyclooxygenases, prostanoids,
cal features and clinical outcome of transitional cell carcinoma of and tumor progression. Cancer Metastasis Rev. 2007;26:
the bladder. J Urol. 2003;170:985–989. 525–534.
45. Shirahama T. Cyclooxygenase-2 expression is up-regulated in 56. Wendum D, Masliah J, Trugnan G, et al. Cyclooxygenase-2 and
transitional cell carcinoma and its preneoplastic lesions in the its role in colorectal cancer development. Virchows Arch. 2004;
human urinary bladder. Clin Cancer Res. 2000;6:2424–2430. 445:327–333.
46. Shirahama T, Arima J, Akiba S, et al. Relation between 57. Wu XR, Kong XP, Pellicer A, et al. Uroplakins in urothelial biol-
cyclooxygenase-2 expression and tumor invasiveness and patient ogy, function, and disease. Kidney Int. 2009;75:1153–1165.
survival in transitional cell carcinoma of the urinary bladder. 58. Wu XR, Lin JH, Walz T, et al. Mammalian uroplakins: a group of
Cancer. 2001;92:188–193. highly conserved urothelial differentiation-related membrane pro-
47. Soslow RA, Rouse RV, Hendrickson MR, et al. Transitional cell teins. J Biol Chem. 1994;269:13716–13724.
neoplasms of the ovary and urinary bladder: a comparative immuno- 59. Yin H, Leong AS. Histologic grading of noninvasive papillary
histochemical analysis. Int J Gynecol Pathol. 1996;15:257–265. urothelial tumors: validation of the 1998 WHO/ISUP system by
48. Stadelmann C, Lassmann H. Detection of apoptosis in tissue immunophenotyping and follow-up. Am J Clin Pathol. 2004;
sections. Cell Tissue Res. 2000;301:19–31. 121:679–687.