Main

Breast cancer is one of the most frequent malignancies in women and ranks second among cancer deaths in adult females.1 At diagnosis, approximately 70–80% of breast carcinoma are positive for estrogen receptors (ERs) and/or progesterone receptors (PgRs), and ∼20% overexpress Her-2/neu.2, 3, 4 These findings could be related with gene expression-profiling studies,5, 6, 7, 8, 9, 10, 11, 12, 13, 14 which identify diverse varieties of breast carcinoma, mainly the ‘luminal’ one associated with hormone receptors, the ‘Her-2/neu-positive’ and the ‘basal’ or ‘triple negative’ breast carcinoma (negative for ERs, PgRs and Her-2/neu).

Basal breast carcinoma includes 15–25% of breast cancers, and it has been correlated with aggressive behavior, poor prognosis and better response to chemotherapy than conventional breast carcinoma. This variety of breast carcinoma was characterized by genetic studies and by immunohistochemical expression of basal cytokeratins, mainly types 5, 14 and 17, epidermal growth factor receptor (EGFR) and p53,15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30 but results from diverse studies are not always concordant. Recently, ‘pure’ and ‘myoepithelial’ variants of basal breast carcinoma were described, based on S-100, actin or p63 expression. Although these subtypes share some concordances in gene profiles and cytokeratin expressions, only the pure variant was associated with shorter survival in the multivariate analyses.17 Based on the above-reported heterogeneity of basal breast carcinomas, further studies seem necessary to better describe the characteristics of this neoplasm, which in turn may improve the knowledge of its growth and dissemination pathways. In this study, we analyze the role and relationship of several factors related with the cell cycle (Ki67, p53, p27, cyclin D1, p50, p52 and p65), membrane growth factor receptors other than Her2 (EGFR, insulin-like growth factor-I receptor (IGF-IR), platelet-derived growth factor receptor-α (PDGFRα) and CD117) and proteins involved in the phosphoinositide-3 kinase (PI3K)/protein tyrosine phosphatase (PTEN)/Akt signaling pathway (PTEN, Akt, and Bad and caspase-3) in a series of basal breast carcinomas.

Materials and methods

Sixty-four breast carcinomas that were triple negative for ERs, PgRs and Her-2/neu were diagnosed between 2000 and 2006. All hematoxylin- and eosin-stained sections of these tumors were reviewed to confirm conventional morphological parameters. Histological grades were assigned using modified Elston & Ellis criteria.31 Three representative areas of each tumor were then selected to build three tissue microarrays. Serial, 5-μm- thick tissue sections were used for immunohistochemical staining with the Envision method (Dako). Antibodies, dilutions, antigen retrieval methods and suppliers are listed in Table 1. Then, we studied the expression of S-100 and smooth muscle actin (SMA) to identify myoepithelial differentiation, basal cytokeratins (types 5/6, 14, 17), luminal cytokeratins (types 8/18) and vimentin. Ki67, p53, p27 and cyclin D1 were studied among the cell cycle-related proteins. Analyses of the nuclear factor-κB (NF-κB) pathway included the study of p50, p52, p65 and IKKα. In addition, diverse membrane receptors other than Her-2 were studied: Her-1(EGFR), platelet-derived growth factor receptor-α (PDGFRα), IGF-IR and c-kit (CD117). Investigation of the PI3K/PTEN/Akt pathway included the analysis of PTEN, phosphorylated Akt (pAkt), phosphorylated Bad (pBad) and activated caspase-3 expression.

Table 1 Immunohistochemistry
Full size table

Two pathologists (EL and SF) evaluated the immunohistochemical stainings blinded to patient outcome. Discordances were reviewed to achieve agreement.

The percentage of stained nuclei was evaluated, independent of the intensity, for Ki67, p53, p27, cyclin D1 and p50, p52 and p65. The percentage of cells with cytoplasmic staining was determined for S-100, actin, CKs 5/6, 14, 17 and 8/18, and vimentin. Distinct membrane staining in >10% of tumor cells for EGFR, IGF-IR and CD117 was counted as positive for expression. PDGFRα was positive when at least 10% of tumor cells showed cytoplasmic and/or membrane stainings. Immunostaining cytoplasmic scores for PTEN, pAkt and pBad were calculated by multiplying the percentage of labeled cells by the intensity of the staining (range 0–300). Scores >150 were considered positive. Cytoplasmic active caspase-3 was rarely detected and was always in less than 5% of cells, and the cutoff was established at 2%.

Clinical followup was reviewed by one of the authors (TR). Mean followup was 21 months (range, 3–70 months), but overall survival and disease-free survival were not conclusive due to the good initial response to treatment and short followup. Finally, the pathological and immunohistochemical findings were related with clinical data and outcome.

Statistical analyses were performed with the SPSS version-4 statistical software package. Continuous and ordinal variables between the two groups were compared using the Student's t-test. Dichotomous variables were compared with the Fisher exact test. The Kaplan–Meier method was used to estimate survival and the log-rank test was performed to assess survival differences between groups. A two-sided P-value of 0.05 was considered significant.

Results

The mean age of the 64 patients was 53 years (range, 25–86 years). Five patients had a history of familial breast carcinoma, and a BRCA1 mutation was detected in another case. Two patients had a previous history of breast cancer, which could not be reviewed.

The most frequent location of tumors was the upper-external quadrant of the breast (36%) as detailed in Table 2. Thirty-eight patients (59%) underwent a tumorectomy and 26 (41%) a modified radical mastectomy. Sentinel lymph node dissection was studied in 17 cases and was positive in 7. Dissection of axillary lymph nodes at levels I and II was performed in 33 patients with a mean of 18 nodes per patient (median: 15; range 5–45), and metastases were identified in 18 (55%) cases.

Table 2 Clinicopathological findings
Full size table

The pathological stage at diagnosis and followup was determined in 58 patients: 21 cases were in stage I (36%), 13 in IIA (22%), 14 in IIB (24%), 4 in IIIA (7%), 3 in IIIB (5%) and 3 in IV (5%).

Most carcinomas were between 2 and 5 cm (58%) and most were of the ductal type (NOS), although 2 were medullary carcinomas, 14 atypical medullary carcinomas and 1 had extensive papillar features (2%). A ductal ‘in situ’ (DCIS) component >10% was present in 36% of tumors and >25% in 6%. Histological grading was II in 25% of tumors and III in 75%. Other microscopic features are detailed in Table 2.

Immunohistochemical data are shown in Table 3 and Figure 1. Briefly, S-100 was positive in 14 (22%) cases and SMA was positive in 5 (8%), but 2 cases co-expressed both Ab. Therefore, 17 (27%) of our basal breast carcinomas had myoepithelial differentiation and were considered of the myoepithelial subtype.

Table 3 Immunohistochemical findings
Full size table
Figure 1
figure 1

Expression of diverse Ab in basal carcinomas of breast. (a) S-100 (×40), (b) vimentin (×20), (c) CD117 (×40), (d) caspase-3 (×40), (e) EGFR (×20), (f) PTEN (×20), (g) pAkt (×20) and (h) p50 (×20).

Full size image

Expressions of CK 5/6, 14, 17 and 8/18 were identified in 50, 56, 58 and 80% of tumors, respectively. Fifty-two (81%) tumors presented one of the basal cytokeratins studied (types 5/6, 14, 17). Expression of cytokeratins 5/6 was often associated with cytokeratin 17 (P=0.002) but not with cytokeratin 14. Vimentin was positive in 55% of tumors.

The mean percentage of Ki67-positive nuclei was 46% (median of 35 and range 10–90%) and only 20% of tumors had a Ki67<20%. Aberrant p53 expression (>10% of nuclei) was present in 53% of cases, cyclin D1 overexpression (>10% nuclei) was detected in 51% of tumors and loss of p27 was identified in 56%. Nuclear positivity for p50, p52 and p65 was detected in 19, 14 and 13%, respectively. Eight (13%) cases had coincident p50 and p65 expressions, suggesting activation of this NF-κB heterodimer.

Several membrane receptors were also studied: EGFR was identified in 34% of tumors, PDGFRα in 31%, IGF-IR in 36% and CD117 in 11%. Overlapping in the staining of IGF-IR, EGFR and CD117 was very rare (P=0.078), except in one myoepithelial basal breast carcinoma, which stained for EGFR and CD117. Overexpression of at least one of these three membrane receptors was associated with tumor growth in 44 (77%) cases. No relationship was found between PDGFRα and the other growth factor receptors.

Loss of PTEN expression (score <150) was seen in 14% of tumors, whereas pAkt was positive in 28% and pBad in 14%. Activated caspase-3 was detected in 58% and was associated with the presence of pBad (P=0.04).

Some correlation was detected between the expressions of IGF-IR, pAkt and pBad, although this was not statistically significant. Interestingly, loss of p27 appeared only in tumors lacking IGF-IR (P=0.062).

No significant differences were detected between pure and myoepithelial variants of basal breast carcinomas in most clinicopathological features or in the immunohistochemical studies, with the exception that pure basal breast carcinomas were more frequently associated with DCIS (45 vs 12%; P=0.019) and pBad overexpression (19 vs 0%; P=0.098), and myoepithelial basal breast tumors more frequently showed necrosis in >10% of the tumor cells (76 vs 47%; P=0.048), marked inflammatory reaction (59 vs 40%, P=0.19), expression of vimentin (>50% tumor cells) (82 vs 19%; P=0.0001), CD117 (35 vs 2%; P=0.001) and activated caspase-3 (76 vs 51%; P=0.089).

Two patients died from tumor progression: one 4 months after diagnosis of a stage IV pure variant and the second at 15 months after diagnosis of a stage IV myoepithelial variant. Five patients are presently alive with disease, three with pure variants (followup of 32, 30 and 27 months) and two with myoepithelial variants (followup of 27 and 38 months). Another patient died of other causes.

Disease-free survival was calculated according to the main characteristics. No differences were observed between the two groups and survival curves lacked significance due to the low mortality.

Discussion

Triple negative (ER, PgR and Her-2/neu negatives) or basal cell carcinomas are a variety of highly proliferative breast neoplasms that could also be identified by the immunohistochemical expression of basal cytokeratins (type 5, 14 or 17) and by DNA microarray studies.5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30 Table 4 summarizes relevant contributions in the characterization of these neoplasms, which are associated with a high histological grade, metaplastic features, wide tumor necrosis, pushing borders and marked lymphoid reaction.19 Furthermore, these tumors are related with medullary features32 and BRCA1 mutations.18, 28 Similar clinical and pathological findings were found in the present series.

Table 4 Historical contributions for characterization of basal breast carcinomas
Full size table

Rakha et al17 have recently subdivided the basal breast carcinomas into two variants: one, ‘pure’, associated with early recurrence and reduced survival, and another with ‘myoepithelial’ differentiation. Approximately, one-fourth of our series presented myoepithelial differentiation, confirmed by immunohistochemical stains for S-100 and SMA, the former being the more sensitive marker. Although most pathological and immunohistochemical results were markedly similar in both subtypes of basal breast carcinomas, we identified several differences: DCIS was more frequent in tumors of the pure variant, whereas necrosis, lymphoid reaction, CD117, activated caspase-3 and vimentin were more common in those with myoepithelial differentiation. Vimentin expression was previously related with some breast carcinomas that today could have been included in the myoepithelial group,33, 35 whereas the remaining breast carcinomas rarely express vimentin.33 Indeed, these tumors can be related to a previously described variety of mammary neoplasms with frequent vascular dissemination, characterized by central acellular areas and expression of cytokeratin 14 and S-100.36

The main objective of this study was the exploration of alternative growth metabolic pathways to hormone receptors and Her2/neu in basal breast carcinomas. Rapid growth of these neoplasms could be related with a high Ki67 index and frequent p53 overexpression. The mean Ki67 index of 46% and the overexpression of p53 in 53% of our cases is much higher than the mean values described in sporadic and familial breast carcinomas.37

Cyclin D1 overexpression has been reported in 60% of DCIS and 70% of invasive cancers,37, 38 in association with a favorable clinical outcome and response to tamoxifen.38 In the present series, cyclin D1 overexpression was present in 52% of tumors in association with loss of p27 nuclear expression, but unrelated to pAkt or ER. In conclusion, cyclin D1 overexpression seems less relevant in the basal subtype than in conventional breast carcinomas.

NF-κB has an important role in cell cycle regulation39 and may participate in the invasion/metastasis of breast cancer by upregulation of vascular endothelial growth factor mRNA40 and matrix metalloproteinases.41 Coincident expression of p50 and p65 in 13% of our tumors suggests the activation of this NF-κB heterodimer in basal breast carcinomas. The p52–p65 heterodimer was not identified in our series.

Several growth factor receptors were studied in the present series. Epidermal growth factor receptor-1 (EGFR or Her-1) and c-erbB-2 (neu or Her-2) expression had been previously detected in 20 and 32% of breast neoplasms, respectively.15 In basal breast carcinoma, overexpression of EGFR is described between 30 and 54%,15, 17, 21, 24 as in this study. EGFR may relate with intracellular signaling kinases, notably mitogen-activated protein kinase and Akt, and promote ER transcriptional activity in a ligand-independent manner,42 and this cross-linking could exist in basal breast carcinoma. IGF-IR is a member of the type II receptor tyrosine kinase family that regulates cell growth in a variety of neoplasms, including breast cancers. Recently, IGF-IR was identified in ∼50% of breast carcinomas,43 including ER-negative tumors,44 although its presence was unrelated with prognosis or survival.45, 46 IGF-IR was identified in 36% of our series and rarely coincided with other growth factor receptors.

CD117 has been detected in 19–31% of basal breast carcinoma, whereas in conventional breast carcinomas its expression varies from 4 to 14%.21, 24, 34 In this paper, as in others,34 CD117 was mainly present in tumors with myoepithelial differentiation.

Coexpression of EGFR, IGF-IR and CD117 was very rare in our study, and 77% of all basal breast carcinomas showed at least one of these receptors. Our results suggest that IGF-IR may be relevant in the development of basal breast carcinoma without EGFR.

PTEN is a tumor suppressor gene that inhibits PI3K-dependent activation of Akt.47 Lack of PTEN induces Akt phosphorilization (activation), which subsequently promotes cell cycle arrest by decreasing the level and nuclear localization of cyclin D1,48, 49 inactivation of p-Bad, expression of Bcl-2 and inhibition of the Forkhead family of transcription factors.50 Lack of PTEN is found in 8–48% of breast carcinomas and is generally associated with a high grade, necrosis, node metastasis and absence of ER.51, 52, 53, 54, 55, 56 Generally, PTEN absence in breast carcinomas is due to gene promoter hypermethylation51, 52, 53, 54, 55 or loss of heterozygosity,53 whereas PTEN mutations are exceptional in sporadic neoplasms.57 Loss of PTEN is rare in basal breast carcinoma, and its relationship with Akt expression could not always be confirmed,51, 56 as in our series where Akt, present in 28% of tumors, was more frequently associated with growth factor receptors than with the PTEN pathway.52, 57 Recently, pAkt overexpression was described in 33% of in situ and 38% of ductal invasive breast carcinomas, although without any significant clinicopathological relationship,56 as in our cases.

In summary, Ki67 and p53 are higher in basal than in conventional breast carcinomas. The expression of EGFR, IGF-IR and CD117 may have a significant role in basal breast carcinomas, as well the heterodimer p50–p65 (NF-κB) in some cases. PTEN, pAkt, p27 and cyclin D1 alterations occur as in conventional mammary neoplasms. In addition to myoepithelial markers such as S-100 and actin, among others, myoepithelial variant could be differentiated from the pure variant of basal breast carcinomas for more frequent expression of vimentin and CD117.