Parameters of dynamic contrast-enhanced MRI as imaging markers for angiogenesis and proliferation in human breast cancer.

BACKGROUND: Breast cancer is the most common malignancy and the leading cause of cancer death in women worldwide; however, early diagnosis has been difficult due to its complex pathological structure. This study evaluated the value of morphological examination in conjunction with dynamic contrast-enhanced MRI (DCE-MRI) for more precise diagnosis of breast cancer, as well as their correlation with angiogenesis and proliferation biomarkers. MATERIAL/
METHODS: DCE-MRI parameters (including Ktrans: volume transfer coefficient reflecting vascular permeability, Kep: flux rate constant, Ve: extracellular volume ratio reflecting vascular permeability, and ADC: apparent diffusion coefficient) were obtained from 124 patients with breast cancer (124 lesions). Microvessel density (MVD) was evaluated by the immunohistochemical analysis of tumor vessels for CD31 and CD105 expression. The proliferation was assessed by analyzing Ki67.
RESULTS: Ktrans values were in the order of: malignant lesions>benign lesions>normal glands. Similar results were observed for Kep. The opposite changes were seen with Ve. Ktrans and Kep values were significantly higher in invasive ductal carcinoma (IDC) and ductal carcinoma in situ (DCIS) than in mammary ductal dysplasia (MDD; ANOVA followed by Dunnett's test). In sharp contrast, ADC values were lower in IDC and DCIS than in MDD, and Ve was not significantly different among the three groups. The data from MIP (maximum intensity projection) showed that benign breast lesions had no or only one blood vessel, whereas malignant lesions had two or more blood vessels. In addition, expression of CD105 and Ki67, the commonly recognized markers for angiogenesis and proliferation, respectively, were closely correlated with MRI parameters as revealed by Pearson analysis.
CONCLUSIONS: Determination of Ktrans, Kep and ADC values permits estimation of tumor angiogenesis and proliferation in breast cancer and DCE-MRI parameters can be used as imaging biomarkers to predict patient prognosis and the biologic aggressiveness of the tumor.

Background

Tumor progression and metastasis are angiogenesis-dependent and quantification of tumor angiogenesis has prognostic value in breast cancer. Magnetic resonance imaging (MRI) is considered a highly sensitive, noninvasive technique for the detection of breast cancer [1]. Of all MRI techniques used for assessment of breast cancer, dynamic contrast enhanced MRI (DCE-MRI) is the most established and widely used. DCE-MRI has been reported to have sensitivity of 88–100% and specificity of 68–96% for cancer detection [2,3]. Recently, MRI has been used to analyze combined morphologic characteristics and kinetic enhancement patterns of breast lesions, and integration of these parameters offers an enhanced predictive power [4,5].

Angiogenesis is the term describing the complex process leading to the formation of new blood vessels from the preexisting vascular network [6]. The growth, invasion, and metastasis of cancers depend on angiogenesis, and quantification of tumor microvasculature is thus considered an approach for anti-angiogenic therapy of cancers [7]. The association of CD105 with angiogenesis was initially identified by Wang et al. [8], who observed that monoclonal antibodies to CD105 reacted strongly with the endothelium in various tumor tissues but only weakly in normal tissues [9]. On the other hand, unlimited proliferation is one of the most important characteristics of cancer. Studies revealed that Ki67 is highly expressed in proliferative cells in many kinds of cancers, but rarely in normal cells [10].

In this study, we investigated that there is any correlation between angiogenesis and proliferation of tumors and DCM-MRI parameters including Ktrans (volume transfer coefficient reflecting vascular permeability), Kep (flux rate constant), Ve (extracellular volume ratio reflecting vascular permeability) and ADC (apparent diffusion coefficient) so as to assess the diagnostic value of DCE-MRI for breast lesions. To gain insight into the mechanisms associated with tumor angiogenesis, we also investigated the relationship between the DCE-MRI parameters and expression levels of vascular markers CD31, CD105 and Ki67.

Material and Methods

Subjects

The study enrolled 124 patients with breast carcinoma ranging from 23 to 76 years old (mean ±SD=49.1±11.9 years) under the procedures approved by the Ethnic Committee for Use of Human Samples of Harbin Medical University. These patients all underwent surgery (lumpectomy or mastectomy) from September 2011 to October 2013, and they did not receive chemotherapy or hormone therapy prior to surgery. The inclusion criteria were: (1) Gender: women; (2) Age: 18 years or older; and (3) Therapy: not receiving chemotherapy or hormone therapy prior to surgery. And the exclusion criteria were: (1) currently lactating and (2) currently being pregnant. Conventional MRI scan, DWI and DCE-MRI dynamic contrast-enhanced scan were performed on the patients, within two weeks post surgery. All measurements were examined by three senior pathologists experienced in breast carcinoma diagnosis and screening from the First, the Third and the Fourth Hospitals affiliated to the Harbin Medical University, respectively, and they worked independently in a double-blinded manner.

MRI study protocol

MRI was performed using the 3.0 T Achieva (Philips, Netherland). Patients were scanned in prone position, eupnea, with the bilateral breast naturally suspended in the breast coil. Conventional sequences and scan parameters were: (1) T2-SPAIR fat suppression sequences, TR60, ms; TE, 4000 ms; FOV, 24×34×15 cm; thickness, 3 mm; gap, 0 mm; matrix, 300×325; NSA, 2; (2) Coronal 3De-THRIVE, TR, 3 ms; TE, 1.4 ms; FOV, 37.5×40×7 cm; thickness, 4 mm; gap, 0 mm; matrix, 512×512; NSA, 2; (3) The diffusion-weighted images (DWI) were acquired with the following parameters: TR, 2900 ms; TE, 64 ms; FOV, 35×24×15 cm; thickness, 3 mm; gap, 0 mm; matrix, 256×256; NSA; and (4) Dynamic contrast-enhanced axial scans using 3D fast low-angle shot dynamic imaging sequences (FLASH-3D). Scan parameters were as follows: TR, 4.5 ms; TE, 2.2 ms; FOV, 25×34×9 cm; matrix, 256×256; scan layers, 75. For dynamic MRI, gadopentetate dimeglumine (Gd-DTPA) was intravenously injection at 2 mL/s, 0.15 mmol/kg using a power injector (Sonic Shot 50; Nemoto Kyorindo, Tokyo, Japan), followed by 20 mL saline flush. After scanning, the images were transmitted to strengthen the image and make maximum signal projection (maximum intensity projection, MIP) images.

Interpretation of MRI Measurements

DCE-MRI parameters (including Ktrans: volume transfer coefficient reflecting vascular permeability, Kep: flux rate constant, Ve: extracellular volume ratio reflecting vascular permeability, and ADC: apparent diffusion coefficient) were measured.

Post-processing of the measurements was performed off-line on a Pentium IV personal computer with Philips Achieva. Breast MRI was retrospectively reviewed by two experienced radiologists. The breast MRI was analyzed according to the criteria set by the American College of Radiology Breast Imaging Reporting and Data System, including tumor shape (round, oval, lobular, or irregular), margin (smooth, irregular, or spiculated), and internal enhancement (homogeneous, heterogeneous, rim enhancement, dark internal septa, enhancing internal septa, or central enhancement). In addition, we analyzed and calculated tumor size, pattern of time–signal intensity (TSI) curve, and the ADC value.

For the signal-intensity analysis, the areas with the highest visual enhancement were chosen as regions of interest (ROIs) for evaluating the enhancement pattern and TSI curves were constructed. Kinetic analysis was performed based on the Breast Imaging Reporting and Data System MRI guidelines. Depend on the images obtained during the last four phases of contrast-enhanced dynamic imaging, TSI curve patterns were classified into three types: (1) the persistent pattern, with a continuous increase in the signal intensity over time; (2) the plateau pattern, in which the signal intensity does not change after its initial increase; and (3) the washout pattern, with an initial increase and reaching the highest point, then followed by a decrease of the signal intensity.

ROIs were used for the ADC measurement. The size of ROI was ≥15 mm2 and the area ranged from 30 to 100 mm2. Three ROIs within the area were used for kinetic analysis. Obvious cystic areas were avoided for ROIs, which was for time-signal intensity curve construction and ADC measurement. The lowest ADC of three ROIs was selected.

Immunohistochemical analyses for microvessel density (MVD) and Ki67

Formalin-fixed, paraffin-embedded sections of breast specimens were placed on sterile glass cover slips, penetrated by Triton X-100 for 1 h, and blocked by normal goat serum for 1 h at 37°C. Then, the slides were incubated with CD31 (Abcam, Cambridge, MA, USA; Cat#: ab28364), CD105 (Abcam; Cat#: ab107595) or Ki67 antibody (Abcam; Cat#: ab16667) overnight at 4°C and subsequently incubated with goat anti-mouse or goat anti-rabbit antibody (LI-COR/Rockland) for 1 h. The slides were washed with PBS, stained nuclei with DAPI (Roche Molecular Biochemicals) for 5 min at room temperature. Immunohistochemistry was analyzed under a fluorescence microscope (Nikon, Japan).

The MVD was evaluated by the immunohistochemical analysis of tumor vessels for CD31 and CD105 expression. An individual vessel was defined as an immune-positive single cell or a cluster of cells with clear lumen. Areas with fibrosis, necrosis, and inflammation, as well as vessels with a muscle wall, were excluded from the count. The sections were scanned at a magnification of 40× by two observers simultaneously to select the most vascularized areas (hot-spots) of the three tissue array spots for each patient. The microvessels in the hot-spots were counted at a magnification of 100× and their density was expressed as the mean number of microvessels/mm2. Mean values of CD31 and CD105 staining were calculated for each individual tumor.

Ki67 expression was assessed with Image Pro Plus 6.0. Five areas were analyzed in each immunostaining slide and each area was divided into 100 squares. The percentage of Ki67-positive staining was classified into weakly positive (+ for 10–25%), positive (++ for 25–50%) and strong positive (+++ for >50%). Results are shown as the percentage of area occupied by Ki67 positive staining over the total area examined.

Statistical analysis

All data are presented as mean ±SEM. The chi-square test was used to analyze significance between malignant and benign lesions. One-way analysis of variance (ANOVA) followed by Dunnett’s test was used for multiple group comparisons. Pearson analysis was used to evaluate the correlation between MRI measurements and pathological indicators. A two-tailed value of p<0.05 was considered as a statistically significant difference. All computations were done with SPSS 18.0.

Results

Clinical characteristics of breast lesions

Breast lesions were classified as malignant lesions in 59/124 patients (including invasive ductal carcinoma (IDC), invasive lobular carcinoma, ductal carcinoma in situ (DCIS), medullary carcinoma, and mucinous carcinoma) and benign lesions in 65/124 patients (including fibrocystic breast disease, fibroadenoma, intraductal papilloma, and mammary ductal dysplasia MDD). Clinical manifestations included breast lumps, breast tingling, inverted nipple, and nipple discharge. Table 1 shows the histopathological characterizations of varying types of lesions.

Table 1

Histopathological characteristics of benign and malignant breast lesions.

Tumor group	Number	Percentage
Malignant lesions	59	47.58
Invasive ductal carcinoma	14	11.29
Invasive lobular carcinoma	35	28.23
Ductal carcinoma in situ	7	5.65
Medullary carcinoma	2	1.61
Mucinous carcinoma	1	0.81
Benign lesions	65	52.42
Adenosis	24	19.35
Fibroadenoma	20	16.13
Intraductal papilloma	5	4.03
Mammary ductal dysplasia	9	7.26
Inflammation	7	5.65

Characteristics of MRI in patients with breast cancer

The DCE-MRI measurements are presented in Table 2. MRI demonstrated that benign lesions were characterized primarily by morphological alterations: round or oval-based shape with clear and sharp boundaries. However, mastitis, mammary ductal hyperplasia, and atypical form of adenovirus had irregular shapes with rough fuzzy boundaries. The dynamic contrast-enhanced features revealed by DCE-MRI showed that benign breast lesions mostly had no enhancement or had early low-ratio, homogeneous enhancement and progressive TIC. In contrast, malignant breast lesions had multiple patterns, such as lobulated, round, or irregular shape, and common glitches or spines surrounded by rough blurred “crab feet”-like edges. Most of the malignant lesions had heterogeneous enhancement or annular enhancement with DCE-MRI. In addition, apparent early enhancement was observed, and TIC was mainly of multi-platform and dissection types.

Table 2

DCE-MRI Performance of benign and malignant breast lesions.

	DCE-MRI performance	Malignant lesions	Benign lesions	χ2	p value
		n=59	n=65
Shape	Regular	13 (22.03)	38 (58.46)	16.95	<0.01
	Irrrgular	46 (77.97)	27 (41.54)
Edge	Clear	13 (22.03)	41 (63.08)	21.19	<0.01
	Rough fuzzy	46 (77.97)	24 (36.92)
Enhanced Features	Homogeneous or no enhancement	7 (11.86)	45 (69.23)	42.39	<0.01
	Heterogeneous enhancement	28 (47.46)	13 (20.00)
	Ring enhancement	24 (40.68)	7 (10.77)
TIC	subtype I	1 (1.69)	50 (76.92)	78.33	<0.01
	subtype II	21 (35.59)	12 (18.46)
	subtype III	37 (62.71)	3 (4.62)
MIP	0	2 (3.39)	51 (78.46)	81.15	<0.01
	1	2 (3.39)	7 (10.77)
	≥2	55 (93.22)	7 (10.77)

The data from MIP (maximum intensity projection) showed that benign breast lesions had no or only one blood vessel, whereas malignant lesions had two or more blood vessels. There was visible local skin thickening and sagging, and occasionally orange peel-like skin change when malignant lesions occurred superficially in the skin and Cooper’s ligament. Nipple retraction or bridge levy was observed when tumors invaded into the nipple and milk ducts. The fat line between pathological lesions and muscle disappeared if the lesions were located deep into the pectoral muscle and fascia. In addition, 11/59 malignant lesions had axillary lymph node metastasis, 7/59 demonstrated lateral skin thickening, and 5/59 developed nipple and skin sagging.

By comparison, among the benign breast lesions, 27 had lobulated and irregular forms of nidus; 20 had annular or irregular enhancement; 17 had early high-ratio enhancement with a platform or clearance type of TIC; and seven had more than two blood vessels. Three DCIS cases were misdiagnosed as breast adenosis, which had irregular flake-like shapes and even enhancement with progressive TIC and without blood vessels, but did not have any defined lump. Another four IDC were misdiagnosed as fibro adenomas, which were rounded in shape with distinct contour, even enhancement, and a platform type of TIC with only one blood vessel.

Comparisons of MRI parameters among normal glands, benign lesions, and malignant lesions

As shown in Table 3, mean value of Ktrans in the malignant lesion group was significantly higher than that of the benign lesion and normal gland groups (p<0.001), despite (in rare cases) the values overlapping between benign lesions and malignant lesions. On the other hand, Ktrans in the benign lesion group was higher than the normal gland group (0.280±0.193 vs. 0.049±0.021, p<0.001). Similarly, the averaged Kep was significantly greater in malignant lesions than in benign lesions and normal glands (p<0.001), and greater in benign lesions relative to normal glands (0.483±0.259 vs. 0.121±0.079, p<0.001). Ve values were slightly lower in malignant lesions than in benign lesions (0.620±0.160 vs. 0.633±0.293, p=0.760), but significantly higher than in normal glands (p=0.008). Ve in benign lesions was also significantly greater compared with normal glands (p=0.020).

Table 3

DCE-MRI parameters of normal glands, benign lesions and malignant lesions.

DCE-MRIParameters	Normaln=59	Benignn=65	Malignantn=59	p value benign vs. normal	p value malignant vs. normal	p value malignant vs. benign
Ktrans (min−1)	0.049±0.021	0.280±0.193	0.783±0.209	<0.001	<0.001	<0.001
Kep (min−1)	0.121±0.079	0.483±0.259	1.304±0.335	<0.001	<0.001	<0.001
Ve	0.523±0.225	0.633±0.293	0.620±0.160	0.020	0.008	0.760

MRI parameters in MDD, IDC, and DCIS

DCE-MRI parameters were further compared among MDD, IDC, and DCIS. As shown in Table 4, Ktrans and Kep values were significantly higher in IDC and DCIS than in MDD. In sharp contrast, ADC values were lower in IDC and DCIS than in MDD. Ve was not significantly different among the three groups; for instance, the averaged Ktrans value was 0.713±0.169/min DCIS and 0.803±0.224 for IDC, whereas it was 0.313±0.238 for MDD. The averaged Kep value was 1.282±0.375 and 1.338±0.326 in DCIS and IDC, respectively, and was 0.449±0.296 in MDD.

Table 4

DCE-MRI parameters of mammary ductal dysplasia, ductal carcinoma in situ and invasive ductal carcinoma.

DCE-MRIParameters	Ductal dysplasia n=9	Ductal carcinoma in situ n=14	Invasive ductal carcinoma n=41	p valueDCIS vs. MDD	p valueIDC vs. MDD
Ktrans (min−1)	0.313±0.238	0.713±0.169	0.803±0.224	<0.001	<0.001
Kep (min−1)	0.449±0.296	1.282±0.375	1.338±0.326	<0.001	<0.001
Ve	0.729±0.318	0.601±0.172	0.617±0.159	0.292	0.329
ADC	1.221±0.113	1.008±0.097	0.947±0.236	<0.001	<0.001

Correlation between MRI parameters and pathological indicators

To further explore the association between MRI parameters and pathological indicators in malignant breast cancer, we examined the status of angiogenesis and proliferation in breast cancer by applying immunohistochemistry to assess the expression of CD31, CD105, and Ki67 in breast cancer samples. We found that the expression of CD31, CD105, and Ki67 was significantly increased in patients with DCIS and IDC compared with those with MDD (Figure 1), indicating that there were more neovascularization and proliferative cells in DCIS and IDC than in MDD. The data from DCE-MRI showed that Ktrans and Kep were markedly higher in DCIS and IDC than in MDD, whereas the ADC value showed the opposite alterations (Table 4). Table 5 shows that the mean Ktrans or Kep values were positively correlated with CD105 level (p=0.024, r=0.563, or p=0.010, r=0.723, respectively). The mean ADC value showed a significant inverse correlation with Ki67 (p=0.025, r=−0.566).

Figure 1

Comparison of expression of CD31, CD105, and Ki67 in the patients with MDD, DCIS and IDC, as assessed by immunohistochemistry.

Table 5

Correlation between DCE-MRI parameters and pathological indicators.

	CD31		CD105		Ki67
	r value	p value	r value	p value	r value	p value
Ktrans (min−1)	0.187	0.393	0.563	0.024	0.329	0.092
Kep (min−1)	0.295	0.164	0.723	0.010	0.418	0.061
Ve	−0.105	0.635	−0.075	0.888	−0.540	0.268
ADC	−0.079	0.721	−0.293	0.271	−0.566	0.025

Discussion

Breast cancer is the most common malignancy in women, and the incidence is still increasing [11]. There is an urgent need for early diagnosis of breast cancer. However, due to the complex pathological processes, it is difficult to diagnose breast cancer at an early stage [12]. In recent years, MRI has been developed as a highly sensitive and specific tool for breast cancer diagnosis [13]. In the present study, we evaluated the value of conventional morphological examination in conjunction with DCE-MRI for more precise diagnosis of breast cancer. We found that CD105 and Ki67, the commonly recognized markers for angiogenesis and proliferation, respectively, were closely correlated with MRI parameters. These results provide further evidence for the feasibility and accuracy of MRI in diagnosis of breast cancer. Moreover, our findings suggest that determination of Ktrans, Kep, and ADC values permits estimation of tumor angiogenesis and proliferation in breast cancer, and DCE-MRI parameters can be used as imaging biomarkers to predict the biologic aggressiveness of the tumor and patient prognosis.

Positive correlation between CD105 and Ktrans or Kep

Compared to normal tissues, tumors require more rigorous supplies of nutrients and oxygen, as well as an enhanced capacity to eliminate metabolic wastes and carbon dioxide [14]. In meeting these needs, tumors develop a more active process of angiogenesis to promote the tumor-associated neovasculature. Indeed, angiogenesis is not only a basis for solid tumor growth and metastasis, but is also an important determinant for prognosis [15,16]. Recent studies found that MVD is closely related to metastasis and tumor recurrence [17,18]. Although a number of molecules expressed in vascular endothelial cells, including CD31, CD34, and FVIII, have been used as essential markers for angiogenesis, their instability in proliferative microvascular cells of tumor tissues has hampered their applications. Thus, whether MVD count can be used to determine the prognosis of malignant tumors remains controversial. In contrast, CD105, belonging to the transformation growth factor-β (TGF-β) family, is highly expressed in endothelial cells of tumor vessels, but only minimally expressed in normal vessels [19,20]. Zhang et al. identified CD105 as a marker of endothelial cells involved in the process of tumor angiogenesis [21]. CD105 was found to be inversely proportional to the survival of cancer patients [22]. These findings suggest CD105 as a novel indicator for predicting cancer development and prognosis.

In this study, we found that the expression of CD31 and CD105 in breast cancer was significantly higher than that in normal breast tissue. These results indicate that there is increased neovascularization in breast cancer, which may play a key role in supplying nutrition for the development and metastasis of breast cancer. Moreover, our results showed that MRI parameters Ktrans and Kep values were significantly higher in malignant than in benign lesions. Furthermore, CD105, but not CD31, was closely correlated with MRI parameters Ktrans or Kep values, indicating that Ktrans and/or Kep in combination with CD105 might serve as a better diagnostic marker for breast cancer.

Ki67 in combination with MRI parameters for early diagnosis of breast cancer

In contrast to Ktrans and Kep values, ADC values were significantly lower in malignant than in benign lesions, and the mean ADC values were inversely correlated with Ki67. These results indicate that in addition to Ktrans and Kep values, ADC might also offer a valuable MRI parameter for breast cancer diagnosis, particularly when combined with the histological parameter Ki67.

Sustained proliferation is a hallmark of cancer [14]. Ki67, a nuclear antigen expressed in the nucleus of cells in active proliferation, is considered a valid nuclear marker of cell proliferation. Studies have shown that expression of Ki67 is upregulated in a variety of human tumors, such as breast cancer, cervical cancer, non-Hodgkin’s malignant tumor, and glioma [10,23]. Evidence also exists showing that increased expression of Ki67 reflects proliferation of malignant tumor cells in breast cancer, lung cancer, colorectal cancer, and other cancers [24]. The positive staining of Ki67 also provided evidence for the use of Ki67 in combination with MRI for early diagnosis of breast cancer.

Conclusions

We have demonstrated that MRI parameters Ktrans, Kep, and ADC allow an estimation of tumor angiogenesis and proliferation in breast cancer and, when combined with the histological features such as CD105 or Ki67, these parameters can be used as imaging biomarkers for more precise and earlier prediction of the biologic aggressiveness and prognosis of breast tumors.