Atypical lesions (ADH; DCIS) are predictors of invasive breast cancer [3
]. However, monitoring the progression and invasion of these lesions in humans is not practical because the current standard therapy for DCIS is complete excision [35
]. Thus, establishing an animal model for IELs that correlates with invasive mammary carcinoma is important to develop preventive measures and effective treatments as well as for understanding the pathogenesis of the breast cancer.
Mammary IELs have not been well characterized in genetically engineered mouse models [36
], principally because they are not spontaneous, but rather induced by chemicals, radiation, or genetic modification. As in humans, and in contrast to mice and rats, spontaneous mammary tumors are quite common in cats [9
]. Even though cats may not develop mammary neoplasia as frequently as dogs, their tumors more closely resemble those in women. For example, the benign mixed tumor that is so common in dogs almost never develops in cats or women [37
Although feline IELs (ductal hyperplasia and carcinoma in situ) have been reported, these lesions were not described in detail or compared with human IELs. Consequently, we evaluated mammary IELs and expression of ER, PR, and HER-2/neu in feline mastectomy specimens. Ki67 proliferation index was also estimated.
IELs were observed in 28% of mastectomy specimens from female cats with clinical mammary disease; 79% were associated with malignant neoplasms. DCIS was the most common lesion, as in human mammary biopsy specimens [26
]; 89% of DCIS lesions were adjacent to malignant tumors. ADH was detected less commonly than DCIS; nevertheless, 95% of these lesions were adjacent to malignant tumors. In contrast, about 50% of UH lesions were adjacent to benign tumors, duct ectasias or fibroadenomatous change, consistent with its only slightly elevated cancer risk in women [38
]. In our study, the prevalence of IELs in feline mammary gland may have been underestimated because only minimal peritumoral tissue was available for histologic evaluation.
Estrogen receptor expression in benign mammary epithelium could be a risk factor for malignancy by rendering cells susceptible to the proliferative stimulus of estrogens [39
]. In this study, ER was expressed in 62.5% of UH and in 7% of ADH, whereas all DCIS and 93% of tumors were negative. These data confirm that some feline mammary dysplasias and most neoplasms are estrogen receptor-negative as reported by Martin de las Mulas [20
] and Millanta [40
]. In cats, ER expression dramatically decreased as the IELs increased in grade; almost all neoplasms were negative for this marker. Most preinvasive lesions were ER-negative. Allred suggested that human ER-negative IELs could be involved in the development of ER-negative DCIS and its evolution into the 30% ER-negative human breast cancers [19
PR immunoreactivity was low in non-lesional mammary gland, IELs, and tumors in contrast to the findings of Millanta and de las Mulas [40
]. This disparity may be due to a different proportion of ovariectomized cats, different stages of the estrus cycle, administration of exogenous progestins, or different PR immunohistochemical technique. Positivity was observed in only 6% of UH, in 7% of ADH, and in 17% of low-grade DCIS. No immunoreactivity was detected in intermediate-grade or high-grade DCIS, or in any of the 44 tumors. As for ER, PR expression decreased with increasing grade of IEL.
In agreement with Millanta and Dias Pereira [42
], the Ki67 proliferative index increased from normal mammary tissue through IELs to malignant tumors. In our study, the expression of Ki67 correlated with the grade of malignant lesions and was inversely associated with ER expression. In fact, highly proliferative lesions tended to lose ER expression. In humans, Ki67 expression increased with increasing tumor grade and correlated with decreased overall survival rates and poor response to hormonal therapy. In cats, use of Ki67 as a prognostic factor for survival with mammary carcinoma has produced conflicting results. Studies by Castagnaro et al
revealed an association between Ki-67 index and biological behavior [44
]; however, Millanta et al
reported no significant prognostic importance in feline mammary carcinomas [42
]. In a recent investigation by Dias Pereira, the Ki67 index correlated positively for different histologic lesions and tumor types with grade [43
IHC results were surprising and differed from those of De Maria, Ordas, and Millanta [23
]. Those authors reported no immunoreactivity [24
], or a faint, barely perceptible signal in part of the cell membrane [23
] in normal mammary ducts and acini. A number of normal tissues, including breast, express this receptor, which probably has a role in normal cell function, regulating growth and proliferation [45
]. However, we found HER-2/neu
protein expression in normal mammary epithelium with strong, complete membrane staining (3+), contrary to what is observed in humans [46
]. Immunohistochemical HER-2 protein overexpression was found in 27% of IELs and in 27% of tumors. HER-2/neu
expression was confirmed by Western Blot, in which both normal and neoplastic tissue showed a 185 kDa band, corresponding to human HER-2/neu
. Differences in signal intensity, however, were observed at comparable total protein loads. This result could reflect a higher expression of HER-2/neu
in some neoplastic tissues, as in the case of samples corresponding to lanes 3-4 of Figure , characterized by a stronger signal compared to neoplastic samples in lanes 1-2, and to healthy tissues (lanes 5 to 8). However, the presence of HER-2/neu
signal in adjacent histologically normal tissues, although constantly observed throughout this study, is unexpected, and needs explanation. If the DAKO antibody cross-reacts with a physiological epidermal growth factor normally expressed in the feline mammary gland, an increase in antibody specificity could overcome this issue. If the polyclonal antibody reacts with both HER-2/neu
and another epidermal growth factor receptor (EGFR) normally expressed in ducts and acini of non neoplastic mammary feline tissue, that would explain the differences in signal intensity both within neoplastic samples, and between neoplastic and healthy tissue samples. Furthermore, the HER-2/neu
protein could be present at higher levels in the normal feline mammary gland compared to the normal human mammary gland. Further investigations will be necessary to clarify the exact nature of this unexpected reactivity.
Similar to what was reported by Antuofermo in dogs, about half the feline IELs without atypia (UH) were associated with benign disease, whereas atypical IELs (ADH and DCIS) were generally associated with mammary cancer [47
]. The histologic grades of feline mammary carcinomas in our study were similar to those reported by Castagnaro [31
]. Like Seixas [34
], we recognized cases of micropapillary carcinoma.