Runx2 in human breast carcinoma: its potential roles in cancer progression

Abstract

Runx2 has been proposed as one of the pivotal factors in the process of osteogenesis and metastasis in human malignancies including breast cancer, but its details have not been evaluated. Therefore, in this study, we evaluated its expression in human breast cancer using immunohistochemistry. One hundred and thirty‐seven formalin‐fixed and paraffin‐embedded breast cancer specimens were used in this analysis of immunohistochemical study. Immunoreactivity was evaluated using the labeling index (LI). Runx2 immunoreactivity was detected in both carcinoma and stromal cells, as well as non‐pathological ductal cells. The nuclear LI of Runx2 in carcinoma cells was associated with the clinical stage, histological grade and HER2 status of the patients examined. In addition, among the patients not associated with distant metastasis, those with high Runx2 LI demonstrated a significantly worse clinical outcome than those with a low LI. This was more pronounced in the group of estrogen receptor (ER)‐negative cases. In addition, both univariate and multivariate analyses demonstrated that the Runx2 LI in breast carcinoma cells turned out an independent prognostic factor. Results of our present study demonstrated that Runx2 plays very important roles in the progression of breast cancer, especially in those of ER‐negative cases. (Cancer Sci 2010; 101: 2670–2675)

Breast cancer is one of the most common malignancies in women worldwide. Recently, the potential association of breast cancer with its bone metastasis has been evaluated from different perspectives and, in particular, the process of osteolysis itself in its metastatic sites has been proposed to facilitate breast cancer progression.^{( 1 )} It is also well known that breast carcinoma cells themselves secrete parathyroid‐hormone‐related peptide (PTHrP), which stimulates osteoblasts in the microenvironment of bone metastatis.^{( 2 )} Osteoblasts at the sites of metastasis are also considered to secrete a receptor activator of NFκB ligand (RANKL) to facilitate the process of transition from mesenchymal cells into functional osteoclasts, which subsequently resorb bone.^{( 3} , ⁴ , ⁵ , ⁶ , ^{7 )} In normal human adult skeleton, bone is constantly renewed or maintained through the coordinated activities of both osteoclasts and osteoblasts.^{( 8 )} Metastatic breast carcinoma cells are seeded into the bone microenvironment, which results in the maturation of osteoclasts.^{( 9 )} These subsequently formed osteolytic foci are associated with bone resorption, which eventually leads to the release of growth factors including transforming growth factor‐ β (TGF‐β) and several insulin‐like growth factors (IGF) from the collapsed bone matrix.^{( 10} , ^{11 )} These factors are considered to subsequently mediate tumor cell proliferation at the sites of bone metastasis.

The Runt‐related transcription factors 1‐3 (Runxs1‐3) have been shown to be required for the process of organogenesis, and mutations in these genes have been reported to be linked to several types of cancer development.^{( 12 )} For instance, Runx1 and Runx3 mutations were reported to promote leukemia^{( 13} , ^{14 )} and gastric cancers,^{( 15 )} respectively. Among these Runx families, Runx2 plays a pivotal role in the process of bone formation or osteogenesis^{( 16} , ¹⁷ , ¹⁸ , ^{19 )} and deregulation of Runx2 itself is associated with the development of osteosarcoma.^{( 20} , ^{21 )} Runx2 was also reported to be highly expressed in both prostate and breast carcinoma cell lines, which can metastasize to bone in various transplanted models.^{( 22} , ²³ , ^{24 )} Loss of function of the Runx2 gene in the mouse was also reported to result in increased cell proliferation of ex vivo skeletal lineage cells.^{( 25} , ^{26 )} Expression of Runx2 was also reported in mammary epithelial cells of the mouse.^{( 27} , ^{28 )} In addition, aberrant Runx2 expression has been reported in breast and prostate primary tumors.^{( 22} , ^{25 )} Runx2 was reported to be involved in the regulation of a mammary‐gland‐specific β‐casein gene and osteopontin.^{( 22} , ²⁸ , ^{29 )} In regard to its potential roles at the sites of breast carcinoma metastasis to the bone, Runx2 was reported to regulate PTHrP expression of metastatic breast carcinoma cells in the microenvironment of bone metastasis and the cell cycle of carcinoma cells themselves.^{( 30 )} Runx2 was also shown to modulate several factors, which can contribute to facilitating the process of metastasis including vascular endothelial growth factor (VEGF),^{( 31 )} several matrix metalloproteinases (MMP)^{( 24} , ^{32 )} and bone sialoprotein.^{( 33 )} However, to the best of our knowledge, its roles in the early stage of breast cancer patients have not been studied at all. In addition, the correlation of Runx2 nuclear immunoreactivity in breast carcinoma cells and histopathological features of breast cancer were reported,^{( 34 )} but the correlation between Runx2 expression and prognosis has still remained unknown.

Among the anti‐estrogen therapies available in cases with estrogen receptor (ER)‐positive breast carcinoma, the administration of selective estrogen receptor modulator (SERM) or aromatase inhibitor (AI) has been considered the gold standard.^{( 35} , ^{36 )} However, it is well known that ovarian suppression and administration of AI frequently results in osteoporosis.^{( 36} , ³⁷ , ³⁸ , ^{39 )} The suppression of estrogenic actions in osteoclasts results in inhibition of their apoptosis and enhancement of their maturation.^{( 36} , ³⁷ , ³⁸ , ^{39 )} Therefore, both suppression of estrogenic actions and elevated Runx2 expression in metastatic breast carcinoma cells might enhance the development of osteoporosis in these patients.

Therefore, in the present study, we evaluated the status of nuclear Runx2 immunoreactivity in breast carcinoma cells and correlated the findings with stage, histological grade, ER status and HER2 expression of the patients in order to study its clinicopathological significance.

Materials and Methods

Breast carcinoma cases.  One hundred and thirty‐seven cases of invasive ductal carcinoma of the breast were retrieved from the surgical pathology files of the Department of Pathology, Tohoku University Hospital, Sendai, Japan. Breast tissue specimens were obtained from Japanese female patients who underwent a mastectomy during 1988–1999 in the Department of Surgery, Tohoku University Hospital, Sendai city, Japan. The mean age was 52.9 years (range, 22–81 years). None of the patients examined in the present study received chemotherapy, administration of trastuzumab or irradiation prior to surgery. The mean follow‐up time was 81 months (range, 1–151 months). All of the specimens had been fixed in 10% formalin at room temperature and embedded in paraffin wax. Research protocols for this study were approved by the Ethics Committee at Tohoku University School of Medicine (approval number 2005‐178).

Antibodies.  Mouse monoclonal antibody for human Runx2 was purchased from Abnova Corporation (Taipei, Taiwan). The characterization of this antibody has been previously reported using both immunoblotting and immunohistochemistry.^{( 40 )} Monoclonal antibodies for estrogen receptor α (ER1D5), progesterone receptor (PR; MAB429) and Ki‐67 (MIB1) were purchased from Immunotech (Marseille, France), Chemicon (Temecula, CA, USA) and DAKO (Carpinteria, CA, USA), respectively. We used a standardized immunohistochemistry kit (HercepTest for Immunoenzymatic Staining; DAKO).

Immunohistochemistry.  A Histofine kit (Nichirei, Tokyo, Japan), which uses the streptavidin–biotin amplification method, was used in this study. Antigen retrieval was carried out by heating the slides in an autoclave at 120°C for 5 min in citric acid buffer (2 mM citric acid and 9 mM trisodium citrate dehydrate, pH 6.0) for Runx2, ER, PR, HER2 and Ki‐67 immunostaining. The dilutions of the primary antibodies used in this study are as follows: Runx2, 1/1000; ER, 1/50; PR, 1/30; HER2, 1/200; and Ki‐67, 1/50. The antigen–antibody complex was visualized with 3.3′‐diaminobenzidine (DAB) solution (1 mM DAB, 50 mM Tris‐HCl buffer [pH 7.6] and 0.006% H₂O₂), and counterstained with haematoxylin. As a negative control, normal mouse, rabbit or goat IgG was used instead of the primary antibodies, and no immunoreactivity was detected in these sections (data not shown).

Statistical analysis.  Immunoreactivity of Runx2 was detected in the nuclei and the labeling index (LI) was subsequently obtained. Briefly, Runx2 immunoreactivity was evaluated in the nuclei of more than 1000 carcinoma cells for each case, and the percentage of immunoreactivity (i.e. the LI) was subsequently determined. In breast carcinoma cells, Runx2, ER, PR and Ki‐67 immunoreactivity was detected in the nucleus, and the immunoreactivity was evaluated as a LI in the same was as described above. Cases with ER, PR or Ki‐67 of more than 10% were considered positive in this study, according to a report on ER.^{( 41 )} HER2 immunoreactivity was evaluated according to a grading system proposed in HercepTest (DAKO), and moderately or strongly circumscribed membrane staining of HER2 in more than 10% of carcinoma cells was considered positive.^{( 42 )} An association between Runx2 immunoreactivity and clinicopathological factors of breast carcinoma patients was statistically evaluated using a correlation coefficient (r) and regression equation, Student’s t‐test, or a one‐way anova and Bonferroni test. Overall and disease‐free survival curves were generated according to the Kaplan–Meier method, and the statistical significance was calculated using the log‐rank test evaluating 64 cases with the Runx2 LI ≧37% (median value) as Runx2 positive and 56 cases with the Runx2 LI <37% as Runx2 negative in a group of non‐distal metastatic breast cancer. Both univariate and multivariate analyses were evaluated by a proportional hazard model (COX) using PROC PHREG in SAS software (SAS Institute Inc., Cary, NC, USA).

Results

Immunohistochemistry of Runx2.  Runx2 immunoreactivity was detected in the nuclei of breast carcinoma cells (Fig. 1a,b), and the mean value of the Runx2 LI in 137 breast carcinoma tissues examined was 43.1% (range, 0–99%) in total. Runx2 immunoreactivity was also detected in non‐pathological myoepithelial and ductal cells (Fig. 1c). Ninety‐five cases are ER positive (LI ≧ 10%) and 42 cases are ER negative (LI < 10%). Seventy‐eight cases are PR positive (LI ≧ 10%) and 59 are PR negative (LI < 10%). The mean value ± SD of the Ki‐67 LI in 137 breast carcinoma tissues examined was 21.3 ± 17.9% (range, 0–82%) in total.

Figure 1.

 Immunohistochemistry for Runx2. Runx2 immunoreactivity was detected in the nuclei of both carcinoma and stroma cells. (a) Case I with Stage III showed a labeling index (LI) = 97.0. (b) Case II with Stage II showed a LI = 37.0. (c) Normal mammary epithelial cells also include nucleus Runx2 positive. Bar, 100 μm.

Correlation of the Nucleus Runx2 LI with the clinico‐pathological factors of the cases examined.  Table 1 summarizes the correlation of the Runx2 LI in breast carcinoma cells with the clinicopathological parameters in the breast carcinoma cases. Significant association between the Runx2 LI and stage (P = 0.0004), histological grade (P = 0.046) and HER2 status (P = 0.002) of the patients was demonstrated, but there were no significant correlation between the Runx2 LI and age (P = 0.78), menopausal status (P = 0.69) and lymph node status (P = 0.66) of the cases examined. The Runx2 LI tended to be correlated with ER (P = 0.13) and PR status (P = 0.06), but the correlation did not reach statistical significance. The Runx2 LI also correlated with both clinical stages and histological grades of the patients.

Table 1.

 Summary of an association between the nuclear Runx2 LI of carcinoma cells and the clinicopathological parameters in 137 breast cancers

	n	Runx2 LI	P value
Age (22–81 years)	137		0.78 (r = −0.024)
Menopausal status
Premenopausal	50	43.1 ± 4.1	0.69
Postmenopausal	87	45.1 ± 2.9
Stage
I	34	35.3 ± 4.6	0.0004
II	67	42.0 ± 3.4
III	19	47.2 ± 5.1
IV	17	68.8 ± 5.7
Tumor size
<2.0 cm	18	49.8 ± 6.3	0.68
≥2.0 cm	94	46.5 ± 3.0
Lymph node status
Positive	17	45.5 ± 3.6	0.66
Negative	120	43.4 ± 3.2
Distant metastasis
Positive	60	68.8 ± 5.7	<0.0001
Negative	77	40.9 ± 2.5
Histological grade
1 (well)	27	34.7 ± 5.6	0.046
2 (moderate)	63	43.4 ± 3.6
3 (poor)	47	51.2 ± 3.6
ER status
Positive	95	41.6 ± 2.9	0.13
Negative	42	49.3 ± 3.8
PR status
Positive	78	40.3 ± 3.2	0.06
Negative	59	49.2 ± 3.4
HER2 status
Positive	30	58.3 ± 5.4	0.002
Negative	107	40.5 ± 2.5
Ki‐67 LI (0–82%)			0.004 (r = 0.25)

Data considered significant (P < 0.05) in the univariate analysis are shown in bold. Significant values were examined in the multivariate analysis in the present study. ER, estrogen receptor; LI, labeling index; PR, progesterone receptor.

Correlation between the Runx2 LI and clinical outcome in 120 non‐distal metastatic breast carcinoma patients (stage I–III).  A significant association was detected between the Runx2 LI and recurrence (P = 0.01) or overall survival (P = 0.003) of the patients in 120 breast carcinoma patients at stage I, II and III (Fig. 2a,b). In particular, among these 120 patients, ER negative cases (23 Runx2 positive and 20 Runx2 negative), the Runx2 LI was markedly associated with an increased risk of clinical recurrence (P = 0.03) (Fig. 2c) and overall survival of the patients (P not calculated because no patients died in the ER negative/Runx2 positive group) (Fig. 2d). However, among the ER‐positive cases (41 Runx2 positive and 36 Runx2 negative), no significant association was detected between the Runx2 LI and an increased risk of recurrence (P = 0.55) (Fig. 2e) and overall survival (P = 0.39) (Fig. 2f).

Figure 2.

 Disease‐free (a) and overall (b) survival of 120 cases associated with non‐distal metastatic breast carcinoma according to nucleus Runx2 immunoreactivity (Kaplan–Meier method), respectively. Disease‐free (c) and overall survival (d) of 77 estrogen receptor (ER)‐positive breast carcinoma cases associated with non‐distal metastatic breast carcinoma according to nucleus Runx2 immunoreactivity, respectively. Disease‐free (e) and overall survival (f) of 43 ER negative breast carcinoma cases associated with non‐distal metastatic breast carcinoma according to nucleus Runx2 immunoreactivity, respectively.

In a Univariate analysis, the Runx2 LI evaluated as a continuous variable also turned out to be a significant prognostic factor (P = 0.049 in disease‐free survival and P = 0.004 in overall survival), and an independent prognostic factor when it was included in a multivariate analysis instead of the dichotomized variable (P = 0.01 and P = 0.04, respectively) (Table 2). Because no cases had received administration of trastuzumab agent in this study, HER2 positive was a remarkably poor prognostic factor.^{( 43 )}

Table 2.

 Summary of univariate and multivariate analyses of overall survival in 137 breast carcinoma patients

Variable	Univariate	Multivariate
	P‐value	P‐value	Relative risk (95% CI)
Disease‐free survival
Lymph node status (positive/negative)	0.001	0.001	7.1 (2.4–21.5)
HER2 status (positive/negative)	0.002	0.002	4.0 (1.6–10.1)
Runx2 LI (99–0%)	0.049	0.01	1.02 (1.01–1.1)
Ki‐67 LI (82–0%)	0.49
Histological grade (3/1, 2)	0.56
Tumor size (≥2.0 cm/< 2.0 cm)	0.96
Overall survival
HER2 status (positive/negative)	0.001	0.02	4.0 (1.2–13.2)
Runx2 LI (99–0%)	0.004	0.04	1.03 (1.001–1.1)
Histological grade (3/1, 2)	0.01	0.5
Lymph node status (positive/negative)	0.03	0.02	4.2 (1.1–1.68)
Ki‐67 LI (82–0%)	0.048	0.29
Tumor size (≥2.0 cm/<2.0 cm)	0.51

Data considered significant (P < 0.05) in the univariate analysis are shown in bold. CI, confidence interval; LI, labeling index.

Discussion

In the present study, the Runx2 LI in breast carcinoma cells was significantly associated with stage (P = 0.0004) and histological grade (P = 0.046) of the patients examined. These findings indicate the possible roles of Runx2 in the biological behavior of breast carcinoma patients, including those without metastasis. We also demonstrated that the prognosis or clinical outcome of cases associated with a high Runx2 LI is generally poor. In particular, in 120 cases not associated with distant metastasis, a significant positive association was detected between the Runx2 LI and both the risk of recurrence and overall survival of patients. Furthermore, this association was more pronounced in the group of 43 ER‐negative cases (36%). This group of ER‐negative carcinoma included HER2 positive and basal‐like subgroups of breast carcinoma. These findings also suggested that Runx2 could serve as a marker of aggressive biological behavior and its inhibition might open a new strategy of therapy for these cases.

Breast cancer development consists of many sequential steps, including primary tumor growth, neovascularization around the tumor, invasion, extravasation and subsequently formation of bone metastasis.^{( 24 )} Many in vitro studies demonstrated that Runx2 might participate in these steps in multiple fashions. Regulation or modification of VEGF secretion by Runx2 was reported in neovascularization.^{( 44 )} Regulation of several MMP secretion by Runx2 was also postulated to be linked with subsequent invasion of carcinoma cells.^{( 45} , ^{46 )} Runx2 was proposed to subsequently mediate PTHrP expression of metastatic breast carcinoma cells in the microenvironment of bone and might be involved in the formation of a vicious cycle.^{( 2 )} All of the above might be related to an adverse clinical outcome for patients but little has actually been demonstrated in clinical cases of human breast carcinoma. In the present study, we demonstrated a significant correlation of the status of Runx2 expression in carcinoma cells with the histological grade and stage of patients. In addition, the possibility of potential involvement of Runx2 in earlier phases of breast cancer development was also raised in the present study.

In the present study, the groups of patients with elevated Runx2 expression were significantly associated with a poor prognosis in the ER‐negative group of patients, while this association was not detected in the ER‐positive carcinoma patients. In our cohort of ER‐positive breast cancer patients, 34 of 95 cases (35.8%) received administration of tamoxifen, while in that of the ER‐negative breast cancer patients, six of 42 cases (14.3%) did. Estrogenic depletion affects not only breast carcinoma cells but also the entire body of patients. It is true that SERM can prevent the systemic effects of estrogen depletion to some extent, but ovarian suppression and administration of AI result in marked side‐effects, especially in the skeletal system causing development of clinically significant osteoporosis in some patients.^{( 36} , ³⁷ , ³⁸ , ^{39 )}

Results of various in vitro studies have shown that suppression of estrogenic activation caused maturation of osteoclasts in a direct manner.^{( 47 )} Estrogen, acting via ER, causes upregulation of Fas ligand (FasL) in osteoclast progenitors (pOC) and/or OC themselves.^{( 47 )} The increased FasL levels cause apoptosis because these OC lineage cells also express Fas. Fas ligand expression is also reported to be diminished or even suppressed without estrogens in these systems, and the life span of OC was actually reported to be elongated in the absence of estrogen in a culture medium. These OC might stimulate osteoblasts to form bone via poorly defined factors but the resorptive effects of these OC are usually considered dominant.^{( 47 )}

Results of several reported studies suggest that anti‐estrogen therapy did amplify the maturation of OC, resulting in the development of osteoporosis. Osteoporosis is a disease of increased bone turnover, in which the bone‐resorbing activity of OC outpaces the bone‐forming activity of osteoblasts, resulting in the loss of predominantly trabecular bone.^{( 47 )} Both of these cell types are reported to respond to estrogen, but results of previous studies suggest that the response of human bone to estrogen withdrawal is at least in part mediated by a network of inflammatory and osteoclastogenic cytokines, including tumor necrosis factor α (TNFα) and interleukin‐1 (IL‐1), released by stromal/osteoblast lineage cells and T cells.^{( 48} , ^{49 )}

Decreased levels of estrogens usually result in increased production of the cytokine IL‐7 by osteoblasts, which stimulates proliferation of T cells and their secretion of both TNFα and receptor activator of NF‐κB Ligand (RANKL).^{( 49 )} Tumor necrosis factor α stimulates osteoblasts to increase their synthesis of RANKL, which results in the differentiation and activation of OC. Tumor necrosis factor α also acts directly on pOC, synergizing with RANKL for OC differentiation.^{( 49 )} Additional pro‐osteoclastogenic cytokines and growth factors are also expressed in T cells and other peripheral blood mononuclear cells.^{( 49} , ^{50 )}

It is practically very difficult to confirm that bone‐metastatic breast cancer cells are also associated with elevated Runx2 expression in the cases with high Runx2 expression in primary breast carcinoma cells because clinically the availability of specimens for both primary and bone metastasis are in general rare. However, metastatic breast carcinoma cells associated with high Runx 2 expression may facilitate the process of osteoporosis in the bone microenvironment by RANKL secretion of osteoblasts via PTHrP secretion,^{( 30 )} but further investigations are required for clarification.

HER2 status of patients turned out to be a strong independent factor because administration of Herceptin had not been used in any of the patients examined in this retrospective study. Runx2 LI in carcinoma cells was also markedly correlated with the HER2 status of patients (P = 0.002). The correlation of HER2 over‐expression with Runx2 expression has not been demonstrated in any human carcinoma and it awaits further investigation for clarification.

In the present study, we showed that the Runx2 LI of breast carcinoma cells associated with clinical stage, histological grade and HER2 status and High Runx2 LI is a poor prognostic factor. Runx2 is currently considered one of the essential regulators of both skeletal development and progression of several tumors including breast and prostate tumors. These tumor cells have been known to frequently metastasize to the bone.^{( 51 )} However, the potential roles on primary carcinoma and bone metastasis still remain relatively unknown. Further investigation into the potential roles of Runx2 in these interactions should provide paths toward the establishment of effective management or control of bone metastasis with breast carcinoma patients.

Disclosure Statement

The authors have no conflict of interest.