Cases
The samples used originated from 48 cases: six normal mammary glands, ten benign tumours (tubulo-papillary adenoma) and 32 malignant (six in situ carcinomas, 12 tubulo-papillary carcinomas and 14 solid carcinomas) of the mammary gland of female cats. Ethical approval for the study was not required because cases were selected among those routinely collected at the laboratory of histopathology for diagnostic purposes after mastectomy.
Each specimen had been formalin-fixed and paraffin-wax embedded and histological diagnoses were performed on Haematoxylin-Eosin stained sections according to the WHO criteria by Misdorp et al. [15]. Malignancies were then graded according to the histological staging method developed by Gilbertson et al. [16] in the dog and adapted to the cat by Mandelli et al. [17] as follows: stage 0 (St0) or malignant non-infiltrating tumour, when the neoplastic proliferation was limited by the basement membrane (all these cases belong to the in situ carcinomas of the WHO [15] classification); stage I (StI) or infiltrating malignant tumour with stromal invasion (i.e. the tumour proliferated beyond the basement membrane and invaded the surrounding stroma); stage II (StII) or infiltrating malignant tumour with lymphatic or blood emboli and/or regional lymph node metastases. On the basis of this gradation system, 6 malignancies resulted stage 0 (in situ), 14 stage I and 12 stage II.
Immunohistochemistry
Four μm-thick sections from all the samples were immunohistochemically stained with a laminin/VEGFR-3 double stain, VEGF-C and CD44 stains.
Laminin/VEGFR-3 double stain reaction
Sections of formalin-fixed and paraffin-embedded samples were dewaxed and rehydrated. Antigen was unmasked with 0.05% Pronase E (1.07433, Merck, Darmstadt, Germany) in Tris buffer pH 7.4 for 10 min at room temperature followed by citrate buffer (2.1 g citric acid monohydrate/liter distilled water) pH 6.0 in a microwave oven at 750 W (two cycles of 5 min each); then the sections were allowed to cool at room temperature for 20 min. The first immunohistochemical step was the block of peroxidase reagent with DAKO EnVision™ Doublestain System (Dako, Amsterdam, The Netherlands) for 5 min followed by primary polyclonal rabbit anti-mouse laminin antibody (Sigma, St. Louis, Missouri, USA), diluted 1:200 in PBS containing 1% bovine serum albumin applied overnight at 4°C. After a wash with TRIS buffer the peroxidase labelled components (diaminobenzidine as chromogen) of the DAKO EnVision™ Doublestain System were used to reveal this part of the double immunohistochemical stain. After a step with double staining block solution of the DAKO EnVision™ Doublestain System, the primary polyclonal rabbit anti-mouse FLT-4/VEGFR-3 antibody (Alpha Diagnostic International, San Antonio, Texas, USA) diluted 1:350 in PBS containing 1% of bovine serum albumin was applied for 60 min at room temperature. After a wash with TRIS buffer the alkaline phosphatase labelled components (fast red as chromogen) of the DAKO EnVision™ Doublestain System were used to reveal this second part of the double immunohistochemical stain. Apart from primary antibodies, all other steps were conducted following the kit manufacturer's instructions. Sections were counterstained with Mayer's hematoxylin and mounted with DAKO Glycergel Mounting Medium (Dako, Amsterdam, The Netherlands). Negative controls were achieved by incubating the slides with a non-specific antibody of the same isotype. As positive controls to assess the cross-reactivity with feline tissues and the specificity of the immunohistochemical stain of the anti-VEGFR-3 antibody a section of feline intestine was used with the same protocol, employing one of the two immunohistochemical DAKO EnVision™ Doublestain System steps. The specificity and cross-reactivity of the anti-laminin stain were evaluated as expected in basement membranes surrounding glandular structures of NMG, BT and MT, and around blood containing vessels.
VEGF-C and CD44 stains
Four μm-thick sections were dewaxed in toluene and rehydrated. Endogenous peroxidase was blocked by immersion in 0.3% hydrogen peroxide for 20 min. Sections were then rinsed in Tris Buffer and antigen retrieval was obtained only for anti-CD44 stain with citrate buffer (2.1 g citric acid monohydrate/liter distilled water), pH 6.0, and heating for two 5-min periods in a microwave oven at 750 W, followed by cooling at room temperature for 20 min. Sections were then preincubated with serum free protein block (Dako, Amsterdam, The Netherlands) for 10 min at room temperature. The primary antibodies anti-VEGF-C, clone Z-CVC7, diluted 1:12.5 (Zymed laboratories inc., San Francisco, California, USA) and anti-CD44var (v5), clone VFF-8, diluted 1:12.5 (Bender MedSystems, Vienna, Austria) were applied overnight at 4°C and were followed by a commercial streptoavidin-biotin-peroxidase technique (LSAB Kit, Dako, Amsterdam, The Netherlands). Diaminobenzidine (0.05% for 10 min at room temperature) was used as chromogen. Slides were counterstained with Papanicolaou's hematoxylin. Negative controls were obtained substituting the primary antibody with an unrelated monoclonal antibody of the same isotype. As positive controls to assess the cross-reactivity with feline tissues and the specificity of the immunohistochemical stain of the anti-VEGF-C and anti-CD44 antibodies, a section of granulation tissue from a feline healing skin lesion and one from a feline anaplastic mammary carcinoma respectively were used following the same protocols.
Scoring methods
Immunohistochemical expression was evaluated in intratumoral (fields randomly chosen between lobules inside the tumour) and extratumoral areas (between the external border of the neoplasm and the subcutis or the muscle/stromal plan of the abdomen or thorax). Intramammary and extramammary fields were selected in the normal mammary gland considering any area containing normal lobuli and/or interlobular connective tissue in the first case, and the subcutis or the muscle/stroma plan of the abdomen or thorax in the second.
Lymphatic vessels expressing and not expressing VEGFR-3 were counted in ten intratumoral/intramammary fields and 20 extratumoral/extramammary fields. Each field had an area of 0.789 square mm and the microscope objective used was 20x. Counts are expressed as the number of lymphatic vessels in ten square millimitres.
VEGF-C and CD44 immunohistochemical expressions were analyzed with a 40x objective (corresponding to 0.196 mm2) on five intratumoral (intramammary) and ten extratumoral (extramammary) fields, selecting the areas with major expression (Hot Spots) at low magnification. In each immunostained area, VEGF-C and CD44 were expressed as score index of the epithelial, adenomatous and carcinomatous components as follows: absent = 0, i.e. no positive cells per field; low = 10, up to ten positive cells per field; intermediate = 100, fewer than 50% positive cells per field; high = 1000, more than 50% positive cells per field. Only for VEGF-C were the number of positive vessels in the same field also counted.
Statistical analysis
The variables (number of lymphatics expressing and not expressing VEGFR-3, VEGF-C and CD44 scores, and number of VEGF-C expressing vessels) were tested for normality with Shapiro Wilk's W Test and the data did not show a normal distribution. Comparison between cases grouped as for diagnosis (normal mammary gland, benign and malignant tumours), site of evaluation (intratumoral/intramammary, extratumoral/extramammary) and stage of invasion (non infiltrating, locally invasive (StI) and with intravascular emboli (StII)) were evaluated with the Spearman rank test. Pearson's correlation test was employed to examine the variables relationship. Analyses were performed by CSS software (Statsoft, Tulsa, OK, USA) statistics, and a conventional 5% level was used to define statistical significance.