Abstract
Co-expression of cytokeratin and vimentin has been traditionally associated with a few select tumors. However, this phenomenon is being recognized in a wider range of tumors. Twenty-one canine primary pulmonary epithelial neoplasms were evaluated for the co-expression of cytokeratin and vimentin. The histologic pattern and grade, and an immunohistochemical grade for cytokeratin and vimentin staining, were determined for each neoplasm. Adenocarcinomas predominated, and histologically, most tumors were grade II. All of the neoplasms stained positive for cytokeratin, while only 8 (38%) stained positive for both vimentin and cytokeratin. Papillary adenocarcinomas were consistently vimentin negative. The anaplastic histologic pattern had significantly more vimentin staining than the other histologic patterns. There was no significant difference in histologic grade or grading criteria between those tumors that stained with vimentin and those that did not. The present study established that cytokeratin and vimentin co-expression occurs in canine primary pulmonary epithelial tumors at a similar frequency to human pulmonary neoplasms. Further investigation will be needed to characterize the significance of this finding, particularly with respect to prognosis.
Introduction
Cytokeratin and vimentin co-expression has been reported in several tumors of domestic animals, including mesotheliomas, anaplastic carcinomas, amelanotic melanomas, renal carcinomas, canine prostatic carcinoma, a canine hepatocellular carcinoma, and feline bronchogenic adenocarcinomas. 4,8,10,12,14,20,21 Although intermediate filament co-expression has rarely been reported in canine pulmonary epithelial neoplasms, 3 to the authors' knowledge, there have been no published observations on this phenomenon in dogs, nor has the prevalence been established.
Expression of intermediate filaments in primary pulmonary neoplasia has been studied and well characterized in human diagnostic pathology. Co-expression of 2–4 different intermediate filaments, including cytokeratin and vimentin, has been identified in 40% of human pulmonary carcinomas. 7 Co-expression was not unique to a group of highly unusual tumors but rather was found throughout the spectrum of tumors examined and did not correlate with proliferative grade. 7
Primary pulmonary neoplasms are uncommon to rare in domestic animals and are most frequently encountered in dogs and cats. 2,5 Epithelial neoplasms predominate 2 ; however, there is significant overlap in the phenotype within this group of tumors despite arising from a variety of cellular origins. 22 Therefore, classification is generally based on histologic pattern rather than histogenesis. 5 As metastatic rate and survival appear to vary with classification, 19,22 pattern distinction is of prognostic significance rather than being solely academic. Classification is not always clear cut, which can result in uncertainty and, in human medicine, has led to a lack of consensus in the best clinical approach to individual cases. 7 The purpose of the current study was to evaluate canine primary pulmonary epithelial neoplasms to ascertain whether co-expression of cytokeratin and vimentin occurs and, if so, if this finding is limited to a specific group of tumors and/or correlated with histologic grade.
Materials and methods
A search of laboratory records from the Western College of Veterinary Medicine (Saskatoon, Saskatchewan, Canada) and Prairie Diagnostic Services (Saskatoon, Saskatchewan, Canada) yielded 26 cases of canine primary epithelial pulmonary neoplasia out of 27,654 canine biopsy and necropsy submissions spanning an 11-year period (1996–2007). Only 21 cases had available, archived, formalin-fixed, paraffin-embedded pulmonary tissue in which the pulmonary origin of the tumor could be confirmed and sufficient, nonautolyzed tissue was available for histologic evaluation and grading. Patient signalment was noted for all cases. Hematoxylin and eosin-stained sections of each pulmonary tumor were examined by the authors using light microscopy, and the histologic type was determined primarily using criteria outlined in the Armed Forces Institute of Pathology, World Health Organization Fascicle for Tumors of the Respiratory System. 5 As per the fascicle, both the mixed-pattern adenocarcinomas and adenosquamous carcinomas contained more than 1 histologic pattern. The mixed-pattern adenocarcinomas were composed of 2 or more of the following growth patterns: papillary, acinar, solid, bronchioloalveolar, or anaplastic, whereas the adenosquamous carcinomas had an intermixed adenocarcinomatous growth pattern and a malignant squamous cell component. Although not defined within the fascicle, for the purpose of the current study, anaplastic patterns and tumors were defined as in the study by Brodey and Craig 1 : varying size and less regularly shaped cells that had no specific pattern of growth. Subsequently, a histologic grade was assigned to each tumor using a grading scheme developed in 1997. 13 Briefly, a score was assigned for overall differentiation (1–3), degree of nuclear pleomorphism (1–3), mitotic figures in ten 400× fields (1–4), nucleolar size (0.5–1.5), amount of tumor necrosis (0–3), degree of tumor fibrosis (0–1.5), and demarcation of the tumor (1–3). Scores for each feature were added. A sum ≤8.5 correlated with a grade I tumor, 9–14 with a grade II, and ≥14.5 with grade III. Immunohistochemical (IHC) analysis was performed on each case using monoclonal, murine, anti-vimentin (clone V9) primary antibodies a and monoclonal, murine, anti-cytokeratin (AE1/AE3) primary antibodies. b The antibodies were diluted 1:8,000 and 1:100, respectively. The staining method was adapted from that described in a previous study. 6 Briefly, staining was performed using a commercial staining platform c and a streptavidin-biotin amplification system. Slides were counterstained with hematoxylin and bluing reagent. Slides were stained in 1 of 3 batches to minimize between-run variation. Nonneoplastic pulmonary tissue served as a positive control for adequate staining of intermediate filaments, and primary antibody omission controls were also used to ensure specificity of staining. The IHC staining was scored using a semiquantitative method (Table 1) based on both intensity and proportion of cells stained, which was modified from an image analysis protocol (http://www.ihcworld.com/ihc_scoring.htm). Scores were assigned for each tumor, and for those tumors comprising multiple patterns, individual scores were assigned for each pattern in addition to providing a total score for the tumor. SPSS 16.0 d was used for statistical analysis. Because of the low sample number, nonparametric analysis was pursued. The Mann-Whitney test was used to compare medians and to test the null hypothesis that there was no difference in grading criteria or IHC staining between histologic types and between different histologic patterns within tumors. For the purposes of analysis, to compensate for low numbers, each histologic type and histologic pattern was compared with all other types and patterns combined into a single group.
Immunohistochemical semiquantitative scoring method adapted from an image analysis protocol (http://www.ihcworld.com/ihc_scoring.htm).*
Total score = positive cells × intensity; maximum = 12.
Results
Fifteen pulmonary tumor samples were submitted as biopsy specimens, while the remaining 6 samples were acquired during necropsy. The sample population included 14 male and 7 female dogs with an average age of 12 years (range: 8–15 years) comprising a variety of breeds. Case numbers were insufficient to evaluate for breed or sex predilections. The signalment, histologic type and grade, and immuno-histochemical grade for each case are summarized in Table 2.
The most common histologic types were adenosquamous carcinoma and papillary adenocarcinoma (6 cases each), followed by mixed-pattern adenocarcinoma (5 cases), bronchioloalveolar carcinoma (3 cases), and anaplastic carcinoma (1 case). As previously mentioned, both mixed-pattern adenocarcinomas and adenosquamous tumors contained more than 1 histologic pattern. When histologic patterns, rather than overall histologic type, were evaluated, the bronchioloalveolar pattern was most frequently noted (11 tumors), followed by papillary (10), squamous (6), anaplastic (3), and acinar (2) patterns. Most tumors were classified as a histologic grade II (16/21), while 3 grade I tumors and 2 grade III tumors were identified. Eight tumors (38%) were found to co-express cytokeratin and vimentin. The IHC grade for vimentin in these cases varied from 1 to 9. Positive vimentin staining was identified at least once in each tumor type with the exception of the papillary adenocarcinomas. Maximal cytokeratin staining was maintained in each of the tumors co-expressing vimentin, barring a single mixed-pattern adenocarcinoma in which there was a notable decrease (score of 10) in cytokeratin staining intensity, as seen in Figure 1B. Cytokeratin staining was diffusely cytoplasmic, and vimentin staining, while limited to the cytoplasm, could vary in pattern between tumors. Vimentin staining was often diffuse but could also take on a predominantly basilar distribution. This latter distribution appeared most frequently in the squamous pattern; however, case numbers were too low to evaluate for statistical significance. Figure 1 shows the co-expression of cytokeratin and vimentin in a mixed-pattern primary pulmonary adenocarcinoma.
Summary of the signalment, histologic type and grade, and immunohistochemical grade for 21 cases of canine primary pulmonary epithelial neoplasia.
M = male; F = female; N = neutered; S = spayed.
Histologic grade assigned based on the scheme developed in a previous study. 13
Immunohistochemical grade assigned based on both intensity and proportion of cells stained (see Table 1).
Statistical analysis to evaluate for differences in histologic grading criteria and IHC staining between histologic types and histologic patterns was performed. A P value of <0.05 was considered significant. The Mann-Whitney test revealed that the degree of differentiation was significantly higher (P = 0.011) and mitotic index significantly lower (P = 0.018) in papillary adenocarcinomas when compared with all other histologic types. Similarly, adenosquamous carcinomas had a significantly higher mitotic index (P = 0.023) when compared with the other histologic types. The papillary pattern was found to have significantly less vimentin staining than other histologic patterns (P = 0.035), while anaplastic patterns were found to have significantly more vimentin staining than other patterns (P < 0.001). When overall grade and grading criteria were evaluated with respect to the presence or absence of vimentin staining, no significant difference was found.
Discussion
The current study evaluated the IHC staining characteristics of canine primary pulmonary epithelial neoplasia collected over an 11-year period. Only 21 tumors out of 27,654 canine biopsy and necropsy submissions could be confirmed as primary pulmonary origin, reinforcing that this is an uncommon neoplasm in dogs. 1,16 Although a sex predilection has not been identified in previous studies, 1,16 there was a higher representation of males in the present study population. As a group, adenocarcinomas were the most commonly diagnosed tumor, and anaplastic neoplasms were rare, which is similar to previous findings. 17 Most of the tumors in the current study were classified as a grade II, with grade I and grade III tumors being less common, results that are similar to that of a 1997 study. 13 However, grade II neoplasms were more prevalent in the current study population (76%) compared with the aforementioned study (48%). The significance of this finding is unknown. It may reflect the lower sample size in the present study or simply reflect the subjectivity inherent to morphologic evaluation and grading of tumors. 13 As this was a retrospective study, complete medical records were not always available to evaluate outcome based on the variety of variables examined. Therefore, correlations could not be made between the assigned grade and the expected outcome for comparison with the 1997 study. In the 1997 study, papillary adenocarcinomas had a better prognosis than other lung tumors. 13 Interestingly, the papillary adenocarcinomas in the current study tended to be well differentiated and had a significantly lower mitotic index compared with the other histologic types. In contrast, the adenosquamous carcinomas in the current study had a significantly higher mitotic index than the other neoplasms. Although this particular histologic type was not specifically identified as influencing patient outcome in the 1997 study, historically, adenosquamous carcinomas have been considered highly aggressive malignancies in domestic animals, with an increased likelihood to invade interstitium or lymphatic vessels. 5
Mixed-pattern primary pulmonary adenocarcinoma exhibiting co-expression of cytokeratin and vimentin.
Historically, cytokeratin and vimentin staining have been used to differentiate epithelial from mesenchymal cell origins, with only a few tissues recognized to express both intermediate filaments in health. Characteristic cytokeratin-vimentin co-expression has been used to support a diagnosis of mesothelioma over neoplasms of epithelial origin that are expected to express only cytokeratin. However, the expanding list of tissues that can be found to co-express these intermediate filaments has decreased the diagnostic utility of vimentin for differentiating such neoplasms. Co-expression of cytokeratin and vimentin in canine primary pulmonary epithelial neoplasms has rarely been reported in the veterinary literature. 3 A prevalence of 38% was found in the present study, using clone V9 anti-vimentin antibodies and anti-cytokeratin (AE1/AE3) antibodies, and this is similar to the 38–50% prevalence documented in humans. 7,15,18 In human diagnostic pathology, such a high prevalence of vimentin expression has resulted in the recommendation of alternate markers, such as calretinin, specific molecular weight cytokeratins, and Wilm's tumor gene 1, to differentiate mesothelial tumors and pulmonary adenocarcinomas, 18 and such action may be warranted in veterinary pathology. Thyroid transcription factor-1 appears to be a potential marker for differentiating canine pulmonary epithelial tumors from mesotheliomas and other mesenchymal tumors. 20 Similar to human studies on pulmonary neoplasia, 7 co-expression of cytokeratin and vimentin in the current study was not limited to a group of highly unusual tumors but, with the exception of papillary patterns, was found throughout the spectrum of tumors examined. Thus, to date, the vimentin staining pattern does not appear useful in assisting with histologic classification of primary pulmonary tumors. However, there appeared to be some differential staining characteristics (basilar, rather than diffuse cytoplasmic staining) predominantly, but not exclusively, in the adenosquamous pattern. Because of the small sample population, it is difficult to infer the significance of this finding, but it would be interesting to evaluate staining characteristics in relation to clinical outcome.
The epithelial to mesenchymal transition theory is a potential mechanism to explain the dissemination and metastasis of carcinomas and may explain the aberrant expression of vimentin in epithelial tumors. 9 To disseminate and metastasize, cohesive epithelial cells have to lose their stable, polarized, nonmigratory properties and transdifferentiate into migratory cells acquiring mesenchymal characteristics. 9,11 Elevated and aberrant expressions of vimentin have been found to correlate well with these features, and as these cells lose their epithelial features, there is an associated decrease in cytokeratin expression that is inversely proportional to vimentin expression. 9,11 In human breast carcinoma, this theory is supported by the significant correlation between vimentin expression, high tumor grade, mitotic index, and invasiveness. 9,11 In the present study, while there was a significant difference between some grading criteria (mitotic index, differentiation) and different histologic types, there was no significant difference between tumor grade, grading criteria, or cytokeratin staining and the presence or absence of vimentin expression. In the current study, there was a statistically significant increase in vimentin staining within the anaplastic histologic pattern compared with other histologic patterns, but the lack of significance between vimentin staining and tumor grade or grading criteria was somewhat surprising. Two of the 3 anaplastic patterns were incorporated into mixed-pattern tumors, so the overall tumor grade in these cases may have misrepresented this pattern and masked an association, or perhaps an association merely required higher case numbers. Interestingly, a lack of correlation between vimentin staining and grade was also found in a study of human pulmonary carcinomas, 7 suggesting that vimentin expression may not provide an adaptive advantage in pulmonary neoplasia. Ideally, to evaluate this relationship in dogs, a larger number of pulmonary neoplasms with known clinical outcomes would need to be evaluated.
The low sample number and the absence of clinical outcome data for most of the patients were limitations of the current study and hindered the ability to evaluate the relationship between tumor behavior and vimentin staining. While the present study has initiated characterization of co-expression in canine primary pulmonary tumors and established that co-expression alone is not an effective way to differentiate between tumor types, the authors have also established that further investigation of this phenomenon is necessary, particularly with respect to prognosis. To overcome the obstacles of the current study, a larger, possibly multicenter, prospective study would likely be best suited to characterize the significance of co-expression in canine primary pulmonary tumors.
Acknowledgements
The authors thank Dr. D. Godson for his assistance and expertise with immunohistochemical analysis and Dr. S. Hendrick for assistance with statistical analysis. Dr. Erin K. Zachar is acknowledged for submitting the case that opened Pandora's box. Funding for this study was provided by the Western College of Veterinary Medicine, University of Saskatchewan, through The Gavina Maggie Reekie Fund for Canine Research.
Footnotes
a.
Sigma-Aldrich, Oakville, Ontario, Canada.
b.
BioGenex, San Ramon, CA.
c.
Benchmark®, Ventana Medical Systems Inc., Tucson, AZ.
d.
SPSS Inc., Chicago, IL.