This article is mentioned in:

Introduction

Breast cancer (BC) is the most common malignancy (28% of all cancers in females) and the second most common cause of cancer mortality in women (15% of female cancer mortality) in the United States (1). The incidence of BC is also increasing in Taiwan. The number of new cases per 100,000 individuals increased from 34.3 in 1994 to 53.1 in 2007. The median age at diagnosis was 51 years, and stage IV disease accounted for approximately 1-5% of new cases. Although there are numerous adjuvant therapies for early stage BC, the recurrence rate remains 20–30%. Treatments for BC rely on the availability of robust clinical and pathological prognostic and predictive factors to guide decision making and the selection of treatment options. Pathological characteristics, including the size of the primary tumor, regional lymph node (LN) metastasis, estrogen receptor (ER), progesterone receptor (PR), Her-2 expression, lymphovascular invasion (LVI) and Ki-67 expression, are all established prognostic markers. In addition to histological prognostic factors, there are increasing numbers of molecular classification systems for BC, including the 21-gene RT-PCR assay (Oncotype DX), which can also be used to predict the benefits of adjuvant chemotherapy (2).

Angiogenesis, the formation of new vessels, is an essential process in the progression of malignant tumors since solid tumors cannot grow beyond 1-2 mm in diameter without angiogenesis (3). As in other malignancies, angiogenesis is also critical for the growth, invasion and metastasis of BC. In BC, extensive neovascularization and tumor thrombus in vessels have been reported to be signs of poor prognosis (4). Also, the tumor microvasculature may constitute a target for anti-angiogenic therapy. Clinical studies have demonstrated that anti-angiogenic agents such as bevacizumab are able to improve the progression-free survival (PFS) of patients with advanced BC when combined with chemotherapy. However, to date there are few predictive markers for anti-angiogenic agents.

Microvascular density (MVD) has become the morphological gold-standard for assessing the neovascularization in human tumors. Several studies have shown that the angiogenic potential of BC as assessed by MVD correlates with progression and metastasis; it therefore predicts clinical outcome (5). The evaluation of MVD requires the use of specific markers for the vascular endothelium and immunohistochemical (IHC) procedures to visualize microvessels (6). Traditionally, histological assessments of MVD in tumors have used pan-endothelial markers, including CD31, CD34 and von Willebrand factor (vWF) (7–9). Although CD34 has been reported to be more sensitive than CD31 or vWF for MVD in BC (10), these markers are specific for all endothelia and do not target only the vascular endothelium from tumor-induced neovascularization. In non-small-cell lung cancer, CD105 (endoglin) has been shown to be superior to CD34 and CD31 in the evaluation of neovascularization since it has a greater affinity for the endothelial cells in tumor-related angiogenic tissue, whereas CD34 and CD31 react nonspecifically with normal and pathological vessels (11). In solid tumors, CD105 has been revealed to be upregulated in the endothelial cells of peri- and intra-tumoral blood vessels and in the stromal components of several types of cancer (12). Studies have shown that increased MVD, as assessed by a CD105 antibody, is associated with worse overall and disease-free survival (DFS) for various types of cancer (13–15).

CD105 is expressed on the cell surface as a 180-kDa homodimeric transmembrane protein. It is a co-receptor of TGF-β1 and TGF-β2. TGF-β has a complex role in carcino-genesis, as it has tumor-suppressor and oncogenic activities. In the early stages of epithelial tumorigenesis, TGF-β has strong anti-proliferative, pro-apoptotic and tumor growth-inhibiting effects; however, it also acts as a tumor-promoting factor by stimulating the epithelial-to-mesenchymal transition and the invasiveness of cancer cells, inhibiting immune surveillance and stimulating angiogenesis by inducing vascular endothelial growth factor (VEGF) expression (16,17). It is expressed almost exclusively in the endothelial cells of peri- and intra-tumoral blood vessels and in tumor stromal components. CD105 antagonizes the inhibitory effects of TGF-β on proliferation and migration, thus promoting the growth and migration of tumor cells (18).

The use of an antisense agent to inhibit CD105 protein translation in cultured human endothelial cells has been reported to markedly enhance the ability of TGF-β1 to inhibit in vitro angiogenesis, suggesting that CD105 is a pro-angiogenic component in the endothelial cells (18). Monoclonal antibodies and their immunoconjugates (immunotoxins and radioimmunoconjugates) against CD105 have been reported to have anticancer effects (19).

In our current study, we analyzed the correlation between CD105 expression and prognostic markers of BC. We sought to determine whether CD105 has the potential to be a therapeutic target for further anti-angiogenic therapies in the future.

Materials and methods

Breast cancer tissue

This study was approved by the Ethics Committee of Kaohsiung Chang Gung Memorial Hospital. With permission from the Institutional Review Board of Chang Gung Memorial Hospital, we collected clinical data and pathological specimens from patients with BC who were diagnosed between 2000 and 2002 and treated in our hospital. Patients provided written consent for their tissue samples to be used for research purposes. In addition, all data were analyzed anonymously. Clinical data, including age at diagnosis, clinical stage and pathological stage, date of recurrence and mortality, and pathological features, including ER, PR and Her-2 status, were obtained from a combination of clinical and pathological record reviews and reports of external medical records.

Tissue microarray

Tissue microarray (TMA) blocks were constructed using the Manual Tissue Arrayer (MTA-1; Beecher Instruments, Sun Prairie, WI, USA). Targets for arraying (areas with BC) were identified by marking the areas on hematoxylin and eosin-stained sections from each paraffin-embedded block. Three tissue cores with a diameter of 2 mm were transferred from each donor block to the recipient TMA block. Liver and skeletal muscle tissue was placed in the first lane core of the three upper left and two lower left cores of the TMA block to ensure correct orientation.

Immunohistochemical staining

TMA blocks constructed from formalin-fixed paraffin-embedded human BC tissue were sectioned at 3-μm thickness on adhesive-coated glass slides and dried overnight at 37°C. The slides were deparaffinized in xylene and rehydrated through graded alcohols to water. For antigen retrieval, the slides were heated in 10 mM citrate buffer (pH 6.0) for 17 min using a pressure cooker. The slides were subsequently washed using TBS buffer with 0.1% Tween 20 for this and subsequent washes. Endogenous peroxidase activity was quenched by treatment with 3% H2O2. After washing, the slides were incubated with primary antibodies targeting CD105 (NCL-CD105; Novocastra, Leica, Newcastle, UK) for 3 h at room temperature (RT), ER (ER, clone SP1, #RM-9101, Neomarkers, Thermo Fisher Scientific, Suwanee, GA, USA) for 1 h at RT, PR (PR, NCL-L-PGR-312/2; Novocastra, Leica) for 1 h at RT and Her-2 (clone SP3, #RM-9103; Neomarkers, Thermo Fisher Scientific) for 1 h at RT. Following the incubation of the primary antibody and a wash step, the HRP polymer (87-8963; Invitrogen, Grand Island, NY, USA) was added. The slides were analyzed using the DAB substrate chromogen system K3468 (Dako, Glostrup, Denmark). The TMA slides were counterstained with hematoxylin and then coverslipped using Entellan® new (Merck KGaA, Darmstadt, Germany) mounting medium. Incubation without the primary antibody was used as a negative control.

The intensity of these markers was determined by two independent pathologists and classified into low and high. A score of low was given if fewer than 10% of the cells were stained either in the nucleus or cytosol; if more than 10% of the cells were stained, this was scored as high. The intensity of the protein expression was tested for correlation with the tumor stage and the prognosis of the disease. The Fisher's exact test was used to test the hypothesis of independence between categorical variables.

Evaluation of MVD

The MVD was evaluated by the immunohistochemical analysis of tumor vessels for CD105 expression in tissue microarrays. Any immune-positive single cell or cluster of cells clearly with lumen was considered to be an individual vessel, as previously recommended (20). Areas of fibrosis, necrosis and inflammation, as well as vessels with a muscle wall, were excluded from the count. The sections were scanned at a magnification level of x100 by two observers simultaneously to select the most vascularized (hot-spots) of the three tissue array spots for each patient. The microvessels in the hot-spots were counted at a magnification level of x200, and their density was expressed as the mean number of microvessels/mm2. Mean values of CD105 staining were calculated for each individual tumor and used for further analysis.

Statistical analysis

Survival curves in relation to predictive markers were illustrated by Kaplan-Meier survival plots and the difference in time to progression was evaluated by the log-rank test. Spearman's rho correlation was used to test the correlation between the immunohistochemical findings and conventional clinical features, including pathological stage and expression of markers. P<0.05 was considered to indicate a statistically significant result. Statistics software (SPSS, version 10.0; SPSS, Inc., Chicago, IL, USA) was used for all calculations.

The cumulative probability of DFS and overall survival (OS) was estimated by the product-limit method. The log-rank test was used to compare the homogeneity of DFS and survival functions across strata defined by categories of prognostic variables.

Results

Characteristics of the patients and pathological findings

In total, 201 patients were enrolled into this study. The median age at diagnosis was 66 years. The major histological type was infiltrating ductal carcinoma. Approximately one-half of the patients were diagnosed at stage II disease. ER was positive in 132/201 (65.7%) patients, PR was positive in 117/201 (58.2%) patients and Her-2 was positive in only 31/201 (15.4%) patients (Table I).

Table I. Patient and disease characteristics at baseline (n=201).

Table I.

Patient and disease characteristics at baseline (n=201).

Characteristics	No. of patients	%
Age at diagnosis, years median (range)	66 (24–84)
Histology
Infiltrating ductal carcinoma	186	92.5
Infiltrating lobular carcinoma	6	3.0
Medullary carcinoma	5	2.5
Other	4	2.0
Stage at diagnosis
I	34	16.9
II	106	52.7
III	52	25.9
IV	9	4.5
Estrogen receptor status
Positive	132	65.7
Negative	63	31.3
Unknown	6	3.0
Progesterone receptor status
Positive	117	58.2
Negative	66	32.8
Unknown	18	9.0
Her-2 status
Negative	145	72.1
Positivea	31	15.4
Unknown	25	12.4
Triple-negative	29	14.4

a Her-2 positive indicates a immunohistochemical staining score of 3 or fluorescence in situ hybridization-positive.

Evaluation of MVD from the intensity of CD105 staining

Positive CD105 was observed as thin, linear deposits in the membranes and cytoplasm of the endothelial cells within microvessels. Evaluation of MVD from the intensity of the CD105 staining revealed a median of 11 microvessels. The median microvessel count of 11 was selected as the cut-off value to define low and high groups (Fig. 1).

Figure 1. Scoring and illustrations of the expression of CD105. Positive CD105 was observed as thin, linear deposits in the membrane and cytoplasm of endothelial cells within the microvessels. The mean values of CD105 staining in (A–C) are 0, 13 and 43, respectively (x400).

CD105 correlated inversely with hormone receptor (HR) expression

In a bivariant correlation analysis between clinical-pathological variables and CD105, no correlation was found among age of diagnosis, stage and tumor grade. However, we found that CD105 correlated inversely with ER (p<0.001) and PR (p<0.001) expression. On the other hand, it correlated positively with Her-2 (p=0.005) expression (Fig. 2). No correlation between CD105 expression and triple-negative BCs was identified.

Figure 2. Correlation analysis between pathological variables and CD105. CD105 correlated (A) inversely with estrogen receptor expression (p<0.001), (B) inversely with progesterone receptor expression (p<0.001) and (C) positively with Her-2 expression (p=0.005). MVD, microvascular density.

CD105 is a poor prognostic factor for early stage BCs

Since angiogenesis is a predictive marker for cancer prognosis, we also analyzed the correlation between CD105 expression and survival for different stages of BC. We found that although the number of microvessels (cut-off, 11 microvessels) did not correlate with the DFS of stage I to III BC patients, it had a stronger correlation with OS in patients with early stage BC (stages I and II) than in patients with late stage BC (stages III and IV; Fig. 3).

Figure 3. The prognostic effect of CD105 expression in various stages of BC. (A) In stage I and II BCs, lower expression of CD105 correlated with higher overall survival (p=0.039). (B) In stage III and IV BCs, the expression of CD105 did not correlate with OS. MVD, microvascular density; BC, breast cancer.

Hormone receptors correlate with the OS of BC patients of all stages

We also analyzed a combination of CD105 expression and the expression of HRs and Her-2, using a microvessel count of 11 as a cut-off point. We found that at the early stages (stages I and II), both HRs and MVD influenced OS; we found no difference between CD105 and HRs in the prediction of OS. However, at stage III, only HRs had a predictive value for OS, either in patients with tumors expressing lower levels of CD105 (p=0.018) or in patients with tumors expressing higher levels of CD105 (p=0.034; Fig. 4).

Figure 4. In multivariate analysis, only hormone receptors (HRs) were of prognostic value for the overall survival of (C) stage III patients. For patients with (A and B) early stage cancers, both CD105 and HRs influenced overall survival. MVD, microvascular density; ER, estrogen receptor; PR, progesterone receptor.

Discussion

Breast cancer is at present the most common cancer in women in Taiwan and its incidence is growing. Although TNM status is the first consideration when selecting treatment for BC, in this era of high-throughput methods, there are numerous novel biomarkers which have been reported for prognostic and predictive purposes. Among these, the most commonly used bio markers are HRs and Her-2, which categorize BCs into luminal A, luminal B, Her-2 type and basal-type classifications (21); such classifications are also relevant in prognosis. The identification of new markers has led to a more definitive insight into tumor biology and substantiates the importance of existing biomarkers.

There are several drugs that target molecules in various oncogenic pathways. These include trastuzumab and lapatinib which target Her-2, RAD-001 which targets mTOR and bevacizumab which targets VEGF. The development of anti-Her-2 agents has improved the poor prognosis for Her-2-positive BC. In order to maximize the therapeutic effects of the anti-Her-2 agents, only Her-2-positive BCs are indicated for these drugs. In contrast, anti-angiogenic agents have been reported to be more effective in the treatment of triple-negative BC (22).

Angiogenesis is critical for the proliferation, growth and metastasis of cancer cells. Therefore, by inhibiting angiogenesis, these processes are also inhibited. Currently there are several agents that target the neovascularization pathway, but none of them has a predictive marker. Even elevated VEGF levels in the circulations of cancer patients may not be a relevant biomarker (23). Although several types of malignancies are hypervascular (24,25), it is not known how significant the angiogenesis pathway is to these tumors. High MVD alone is not a predictive marker for anti-angiogenic agents since anti-angiogenic agents target new vessels in tumors only. As CD105 expression is more specific for areas of neovascularization, CD105 has the potential to be a predictive marker for anti-angiogenic agents.

HR-positive BCs usually have a more favorable prognosis than HR-negative cases of BC. A primary reason for this is the development of antihormonal agents which can prevent the recurrence and delay the progression of HR-positive BC (26). This reverse correlation also explains the favorable prognosis and slow progression of HR-positive BC. However, the overexpression of Her-2 has been reported to upregulate VEGF synthesis and thus may increase angiogenesis and metastasis in BC (27). In the current study, we found that the expression of CD105 correlated negatively with HR expression but positively with Her-2 expression, which is compatible with the behavior and prognostic significance of these markers.

Tumorigenesis comprises multiple steps. When a tumor grows to more than 1 mm3 in size, angiogenesis becomes critical for the tumor to grow and metastasize. However, when the tumor grows beyond this size, the signal transduction pathways become more complicated, with a greater number of overactive pathways which are responsible for tumor growth, survival, angiogenesis, metastasis and drug resistance (28). At an advanced stage, neovascularization may not be the most crucial prognostic factor since there are already several active oncogenic pathways. In clinical practice and animal models, small tumors are more easily controlled by anti-angiogenic agents than larger tumors, which also implies that the role of angiogenesis is less significant when tumors grow larger. In our study, we found that the prognostic effect of MVD was even more considerable when tumors were at their early stages, which might correlate with the aggressiveness of primary tumors and primary resistance to adjuvant therapies.

The poor outcome for the patients with increased MVD as measured by CD105 staining is in line with previously published studies concerning BC and other malignancies (29). Higher MVD correlated with reduced DFS and OS in BC patients, and decreased survival rates in patients with increased CD105 and CD31 counts have also been reported (30).

We report for the first time that MVD measured by CD105 staining is significantly associated with Her-2 molecular subtypes, but is inversely associated with HR expression. In Her-2-overexpressing BC, hormone therapy is less effective than in Her-2-negative BCs, which may be due to cross-talk between HRs and the Her-2 pathway (31,32). Trastuzumab has been proved to be effective on Her-2-overexpressing BC as a monotherapy or as a part of combination therapy with either chemotherapy or endocrine therapy (32), but the reported response rate was less than 70%, and resistance always developed (33). Methods of improving the therapeutic effects of anti-Her-2 agents and preventing the emergence of resistance are necessary.

The targeting of TGF-β signaling leads to the inhibition of angiogenesis and reverses a stem cell phenotype to a more differentiated luminal phenotype (34). These effects are potentially inhibited by a dual blockade of the TGF-β and CD105 pathways (35). Moreover, the anti-CD105 antibody as a single agent as well as CD105 as a target for an oral DNA vaccine have also been reported to be effective agents for suppressing or preventing tumor progression and prolonging survival in in vitro models (36–38).

Since overactivity of the Her-2 pathway promotes angiogenesis and neovascularization contributes to the progression and metastasis of Her-2-overexpressing BC, a combination of anti-Her-2 and anti-angiogenic treatments may be feasible This combination may not only increase the therapeutic effect but also delay the emergence of resistance (39). The targeting of tumor neovascularization with vaccines against CD105 in a Her-2-driven BC cell line has been reported (38). In our study, we found that MVD measured by CD105 staining correlated with Her-2 expression, which indicates that combination therapy may be of benefit in the treatment of Her-2-overexpressing BC which also has high CD105 expression.

In conclusion, we found that MVD measured by CD105 staining correlated positively with Her-2 expression, but negatively with HR expression. The significance of MVD on OS is greater for early stage BCs.

Acknowledgements

We thank J.K. Hsiao for his technical support. This study was supported by a grant from the Chang Gung Memorial Hospital (CMRPD890111).

References