3305jum759-928online_Layout 1 4/22/14 8:54 AM Page 835
ORIGINAL RESEARCH
Three-Dimensional Contrast-Enhanced Sonography in the Assessment of Breast Tumor Angiogenesis Correlation With Microvessel Density and Vascular Endothelial Growth Factor Expression Man Chen, MD, PhD, Wen-ping Wang, MD, PhD, Wan-ru Jia, MD, Lei Tang, MD, Yi Wang, MD, Wei-wei Zhan, MD, PhD, Xiao-chun Fei, MD Objectives—The purpose of this study was to differentiate perfusion and vascular characteristics between benign and malignant breast lesions by 3-dimensional contrastenhanced sonography and evaluate their correlation with microvessel density and vascular endothelial growth factor (VEGF) expression for further clinical exploration. Methods—Three-dimensional contrast-enhanced sonography was performed in 183 patients with breast lesions, and sonographic characteristics were carefully observed for further analysis. The mean microvessel density and VEGF expression were measured by immunohistochemical analysis.
Received May 23, 2013, from the Departments of Diagnostic Ultrasound (M.C., W.J., L.T., Y.W., W.Z.) and Pathology (X.F.), Ruijin Hospital, School of Medicine, Shanghai Jiao Tong University, Shanghai, China; and Department of Diagnostic Ultrasound, Zhong Shan Hospital, Fudan University School of Medicine, Shanghai, China (W.W.). Revision requested July 10, 2013. Revised manuscript accepted for publication August 30, 2013. We thank Yun-yun Hu for preparing the illustrations, Claudia for assistance with manuscript preparation, and everyone who has given us help during our research. This work was supported in part by National Natural Science Foundation of China. Address correspondence to Wen-ping Wang, MD, Department of Diagnostic Ultrasound, Zhong Shan Hospital, Fudan University School of Medicine, 180 Fen Lin Rd, 200032 Shanghai, China. E-mail: [email protected] Abbreviations
MRI, magnetic resonance imaging; NPV, negative predictive value; PPV, positive predictive value; 3D, 3-dimensional; 2D, 2dimensional; VEGF, vascular endothelial growth factor doi:10.7863/ultra.33.5.835
Results—Pathologic results showed 35 benign and 148 malignant cases. Malignancy and benignity differed significantly in peripheral vessel characteristics (number, distribution, course, degree of dilatation, and penetrating vessels), rim perfusion and coarseness degree, intratumoral perfusion type, and intratumoral vessel dilatation (P < .05) but not the presence of peripheral and intratumoral vessels and intratumoral perfusion (P > .05). The specificity of penetrating vessels was 88.6% for diagnosing malignancy. The sensitivity, specificity, and accuracy of rim perfusion coarseness were 90.2%, 70.4%, and 85.3% respectively. The sensitivity of the intratumoral perfusion type was 77.8%, whereas the specificity of intratumoral vessel dilatation was 88.9%. Microvessel density and VEGF expression were significantly correlated with perfusion and vascular characteristics (P < .05), except the presence of peripheral vessels, rim perfusion, and intratumoral perfusion (P > .05). The presence of intratumoral vessels was related to VEGF (P < .05) but not microvessel density (P > .05). Conclusions—Three-dimensional contrast-enhanced sonographic characteristics were statistically different between benign and malignant breast lesions. Some of them also correlated significantly with microvessel density and VEGF expression and therefore have potential for objective evaluation of tumor angiogenesis. Key Words—angiogenesis; breast tumor; breast ultrasound; 3-dimensional contrastenhanced sonography
T
he incidence of breast cancer has been the highest among malignant tumors of female patients in China.1 Tumor angiogenesis is defined histologically as the postembryonic formation of new blood vessels by capillary sprouting.2 It plays a central role in local tumor growth, invasion, and distant metastasis, which was first confirmed in breast cancer.3 Neovascularization of malignant tumors is entirely different from blood vessels of benign tumors, both in morphologic characteristics and in hemodynamics.
©2014 by the American Institute of Ultrasound in Medicine | J Ultrasound Med 2014; 33:835–846 | 0278-4297 | www.aium.org
3305jum759-928online_Layout 1 4/22/14 8:54 AM Page 836
Chen et al—3D Contrast-Enhanced Sonography of Breast Tumors
Therefore, the characteristics of tumor angiogenesis are of vital importance in differentiating malignant lesions from benign ones. Microvessel density has been accepted as the reference standard for evaluation of tumor angiogenesis.4 Vascular endothelial growth factor (VEGF) has been found to be an important cytokine in regulating endothelial cell proliferation and function development, which may influence the expression of microvessel density.5 However, in clinical application, microvessel density and VEGF are more likely to be measured by immunohistochemistry during postoperative resection of tumor tissue samples. Therefore, it would be important to look for a simple, noninvasive, and reproducible in vivo imaging modality to evaluate tumor angiogenesis, differentiate malignant from benign breast lesions, predict the prognosis, and monitor the response to systemic treatment. Several quantitative and semiquantitative studies have been performed previously to evaluate the vascular density of breast masses by sonography. Huber et al6 found that color Doppler sonography was superior to conventional duplex sonography in helping differentiation. Carson et al7 found that visual assessment of color Doppler images may be aided by quantitative analysis of information provided in the images. Sehgal et al8 concluded that quantitative Doppler imaging, when used in combination with a nonlinear rule-based approach, has potential for differentiating between malignant and benign masses. Threedimensional (3D) sonograms are reconstructed from data obtained from a single sweep of the ultrasound beam across the lesion of interest. With the development of 3D reconstruction techniques and the application of integrated volume probes, 3D image resolution is obviously improved. These techniques can depict tumor anatomic structures and simultaneously show tortuous angiogenic vasculature associated with breast cancer by providing a more complete overview of the tumor neovascularization.9,10 Contrast-enhanced sonography of the breast has recently been studied for characterization of indeterminate breast lesions, and it can be used to visualize tumor vasculature even at the capillary level, permitting a more sensitive measure of tumor perfusion.11,12 Therefore, 3D contrastenhanced sonography, which combines the advantages of 3D-sonography and contrast-enhanced sonography, has become more sensitive for revealing breast tumor angiogenesis and depicting reconstruction of the stereoscopic vascular structure. The purpose of our study was to show the perfusion and vascular 3D contrast-enhanced sonographic characteristics of benign and malignant breast lesions and their association with microvessel density and VEGF expression, and then to evaluate the potential
836
clinical diagnostic value of 3D contrast-enhanced sonography in the assessment of tumor neovascularization.
Materials and Methods Study Population In our institution, all patients with breast lesions treated by a multidisciplinary team approach were entered prospectively into a database, which was approved by The Institutional Review Board, and patients’ informed consent was obtained. From May 2011 to February 2012, 183 consecutive patients were included. Among these patients, 182 were female, and 1 was male (age range, 20–76 years; mean age, 50.25 years). All patients underwent conventional preoperative sonography, 2-dimensional (2D) contrastenhanced sonography, and 3D contrast-enhanced sonography and subsequently underwent surgery (modified radical mastectomy, radical mastectomy, or conservative breast surgery) within 1 week. None of them received preoperative neoadjuvant chemotherapy or endocrine therapy. Histologic analysis showed 35 benign breast lesions, including fibroadenoma (n = 28), intraductal papilloma (n = 5), and phyllodes tumors (n = 2), and 148 malignant lesions, including invasive ductal carcinoma (n = 108), mixed invasive carcinoma (n = 18), ductal carcinoma in situ (n = 8), invasive lobular carcinoma (n = 4), apocrine carcinoma (n = 4), invasive micropapillary carcinoma (n = 2), metaplastic carcinoma (n = 2), and neuroendocrine carcinoma (n = 2). Sonographic Examinations All sonographic examinations, including conventional sonography, 2D contrast-enhanced sonography, and 3D contrast-enhanced sonography, were performed with the same ultrasound machine (MyLab 90; Esaote SpA, Genoa, Italy). Conventional sonography and color Doppler sonography were performed with an LA532 transducer with a frequency of 13–4 MHz, whereas 2D contrast-enhanced sonography was evaluated with an LA522 transducer with a frequency of 9–3 MHz. A BL433 volume transducer with a frequency of 15–9 MHz was used for 3D scanning. The contrast agent was SonoVue (BR1; Bracco SpA, Milan, Italy), a sulfur hexafluoride–filled microbubble contrast agent. To avoid interobserver variability, all sonographic scanning was performed by a single radiologist (M.C., with >15 years of experience in breast sonography, 5 years in breast 2D contrast-enhanced sonography, and 1 year in breast 3D contrast-enhanced sonography). All scanning was usually performed less than 1 week before surgery.
J Ultrasound Med 2014; 33:835–846
3305jum759-928online_Layout 1 4/22/14 8:54 AM Page 837
Chen et al—3D Contrast-Enhanced Sonography of Breast Tumors
Examinations were performed as follows: Before injection of the contrast agent, the breast tumors were scanned first on conventional and color Doppler sonography to obtain the best imaging in the maximum plane from which both the tumors and the normal adjacent breast tissue could be observed. Imaging parameter settings were optimized to ensure high-quality conventional sonograms after the target lesion was determined. Baseline 3D scanning was then performed using the same scanning route to define the appropriate volume angle so that the whole lesion would be included in the volume data without signal loss. Subsequently, all patients manually received injections of 2.4 mL of SonoVue as a bolus through an antecubital vein, followed by a flush with 5 mL of 0.9% saline. Breast mass 2D contrast-enhanced sonography was evaluated in real time until 2 minutes after the beginning of the injection. The following settings were used for 2D contrast-enhanced sonography: The selected plane included the lesion and its surrounding normal tissue if possible. The scanner settings for contrast-enhanced sonography were as follows: mechanical index, 0.09; single focus at the bottom of the image; dynamic range, 70 dB; image depth, 3 or 4 cm; the probe was stabilized manually; and no pressure was exerted. The timer was activated immediately at the beginning of injection. The lesion was observed for at least 2 minutes, and the whole process of contrast-enhanced sonography was stored in the hard disk of the ultrasound machine in the Digital Imaging and Communications in Medicine format for further analysis. Five to 10 minutes after contrastenhanced sonography, when the signals from the microbubbles in the large vessels such as the axillary vein disappeared, the transducer was shifted from LA522 to BL433, and 3D contrast-enhanced sonography was initiated. The contrast agent was injected in the same dose and fashion as for 2D contrast-enhanced sonography. Ten seconds later, 3D contrast-enhanced sonograms were obtained more than 5 times continuously with a total time of more than 2 minutes. The imaging settings for 3D contrast-enhanced sonography were as follows: mechanical index, 0.08 to 0.13; 1 focal zone; power output, 3% to 6%; dynamic range, 40 to 60 dB; volume angle, 30° to 50°; and the scanning route was consistent with that of 2D contrastenhanced sonography. During 3D scanning, the transducer was kept in a stable position without movement. After scanning, reformatted 2D contrast-enhanced images of 3 orthogonal planes were displayed on the screen. The data acquired were also stored in the Digital Imaging and Communications in Medicine format on the hard disk of the ultrasound system.
J Ultrasound Med 2014; 33:835–846
Reconstruction of 3D Contrast-Enhanced Sonograms The 3D contrast-enhanced sonographic measurements were analyzed by 2 investigators (L.T., with 6 years of experience in conventional breast sonography and 1 year in 3D contrast-enhanced sonography; and Y.W., with 5 years of experience in conventional breast sonography and 1 year of experience in 3D contrast-enhanced sonography), without knowledge of pathologic information. After independent interpretations by the observers, a consensus was obtained by a conference. If different assessments were assigned by the readers, then a consensus was reached after discussing the findings. There are various 3D display and rendering modes, such as triplanar imaging, tomographic mode imaging, volume rendering and analysis, thick-slice imaging, surface mode, transparency mode, and a combination of surface and transparency modes. In this study, the transparent mode and tomographic mode imaging were mainly selected to depict neovascularization and to distinguish intratumoral enhancement. The characteristics of tumor vasculature and perfusion were focused to assess the following: (1) peripheral vessels (presence or absence), which were defined as radial or parallel vessels surrounding the tumor margin with or without branches, and their number (
ORIGINAL RESEARCH
Three-Dimensional Contrast-Enhanced Sonography in the Assessment of Breast Tumor Angiogenesis Correlation With Microvessel Density and Vascular Endothelial Growth Factor Expression Man Chen, MD, PhD, Wen-ping Wang, MD, PhD, Wan-ru Jia, MD, Lei Tang, MD, Yi Wang, MD, Wei-wei Zhan, MD, PhD, Xiao-chun Fei, MD Objectives—The purpose of this study was to differentiate perfusion and vascular characteristics between benign and malignant breast lesions by 3-dimensional contrastenhanced sonography and evaluate their correlation with microvessel density and vascular endothelial growth factor (VEGF) expression for further clinical exploration. Methods—Three-dimensional contrast-enhanced sonography was performed in 183 patients with breast lesions, and sonographic characteristics were carefully observed for further analysis. The mean microvessel density and VEGF expression were measured by immunohistochemical analysis.
Received May 23, 2013, from the Departments of Diagnostic Ultrasound (M.C., W.J., L.T., Y.W., W.Z.) and Pathology (X.F.), Ruijin Hospital, School of Medicine, Shanghai Jiao Tong University, Shanghai, China; and Department of Diagnostic Ultrasound, Zhong Shan Hospital, Fudan University School of Medicine, Shanghai, China (W.W.). Revision requested July 10, 2013. Revised manuscript accepted for publication August 30, 2013. We thank Yun-yun Hu for preparing the illustrations, Claudia for assistance with manuscript preparation, and everyone who has given us help during our research. This work was supported in part by National Natural Science Foundation of China. Address correspondence to Wen-ping Wang, MD, Department of Diagnostic Ultrasound, Zhong Shan Hospital, Fudan University School of Medicine, 180 Fen Lin Rd, 200032 Shanghai, China. E-mail: [email protected] Abbreviations
MRI, magnetic resonance imaging; NPV, negative predictive value; PPV, positive predictive value; 3D, 3-dimensional; 2D, 2dimensional; VEGF, vascular endothelial growth factor doi:10.7863/ultra.33.5.835
Results—Pathologic results showed 35 benign and 148 malignant cases. Malignancy and benignity differed significantly in peripheral vessel characteristics (number, distribution, course, degree of dilatation, and penetrating vessels), rim perfusion and coarseness degree, intratumoral perfusion type, and intratumoral vessel dilatation (P < .05) but not the presence of peripheral and intratumoral vessels and intratumoral perfusion (P > .05). The specificity of penetrating vessels was 88.6% for diagnosing malignancy. The sensitivity, specificity, and accuracy of rim perfusion coarseness were 90.2%, 70.4%, and 85.3% respectively. The sensitivity of the intratumoral perfusion type was 77.8%, whereas the specificity of intratumoral vessel dilatation was 88.9%. Microvessel density and VEGF expression were significantly correlated with perfusion and vascular characteristics (P < .05), except the presence of peripheral vessels, rim perfusion, and intratumoral perfusion (P > .05). The presence of intratumoral vessels was related to VEGF (P < .05) but not microvessel density (P > .05). Conclusions—Three-dimensional contrast-enhanced sonographic characteristics were statistically different between benign and malignant breast lesions. Some of them also correlated significantly with microvessel density and VEGF expression and therefore have potential for objective evaluation of tumor angiogenesis. Key Words—angiogenesis; breast tumor; breast ultrasound; 3-dimensional contrastenhanced sonography
T
he incidence of breast cancer has been the highest among malignant tumors of female patients in China.1 Tumor angiogenesis is defined histologically as the postembryonic formation of new blood vessels by capillary sprouting.2 It plays a central role in local tumor growth, invasion, and distant metastasis, which was first confirmed in breast cancer.3 Neovascularization of malignant tumors is entirely different from blood vessels of benign tumors, both in morphologic characteristics and in hemodynamics.
©2014 by the American Institute of Ultrasound in Medicine | J Ultrasound Med 2014; 33:835–846 | 0278-4297 | www.aium.org
3305jum759-928online_Layout 1 4/22/14 8:54 AM Page 836
Chen et al—3D Contrast-Enhanced Sonography of Breast Tumors
Therefore, the characteristics of tumor angiogenesis are of vital importance in differentiating malignant lesions from benign ones. Microvessel density has been accepted as the reference standard for evaluation of tumor angiogenesis.4 Vascular endothelial growth factor (VEGF) has been found to be an important cytokine in regulating endothelial cell proliferation and function development, which may influence the expression of microvessel density.5 However, in clinical application, microvessel density and VEGF are more likely to be measured by immunohistochemistry during postoperative resection of tumor tissue samples. Therefore, it would be important to look for a simple, noninvasive, and reproducible in vivo imaging modality to evaluate tumor angiogenesis, differentiate malignant from benign breast lesions, predict the prognosis, and monitor the response to systemic treatment. Several quantitative and semiquantitative studies have been performed previously to evaluate the vascular density of breast masses by sonography. Huber et al6 found that color Doppler sonography was superior to conventional duplex sonography in helping differentiation. Carson et al7 found that visual assessment of color Doppler images may be aided by quantitative analysis of information provided in the images. Sehgal et al8 concluded that quantitative Doppler imaging, when used in combination with a nonlinear rule-based approach, has potential for differentiating between malignant and benign masses. Threedimensional (3D) sonograms are reconstructed from data obtained from a single sweep of the ultrasound beam across the lesion of interest. With the development of 3D reconstruction techniques and the application of integrated volume probes, 3D image resolution is obviously improved. These techniques can depict tumor anatomic structures and simultaneously show tortuous angiogenic vasculature associated with breast cancer by providing a more complete overview of the tumor neovascularization.9,10 Contrast-enhanced sonography of the breast has recently been studied for characterization of indeterminate breast lesions, and it can be used to visualize tumor vasculature even at the capillary level, permitting a more sensitive measure of tumor perfusion.11,12 Therefore, 3D contrastenhanced sonography, which combines the advantages of 3D-sonography and contrast-enhanced sonography, has become more sensitive for revealing breast tumor angiogenesis and depicting reconstruction of the stereoscopic vascular structure. The purpose of our study was to show the perfusion and vascular 3D contrast-enhanced sonographic characteristics of benign and malignant breast lesions and their association with microvessel density and VEGF expression, and then to evaluate the potential
836
clinical diagnostic value of 3D contrast-enhanced sonography in the assessment of tumor neovascularization.
Materials and Methods Study Population In our institution, all patients with breast lesions treated by a multidisciplinary team approach were entered prospectively into a database, which was approved by The Institutional Review Board, and patients’ informed consent was obtained. From May 2011 to February 2012, 183 consecutive patients were included. Among these patients, 182 were female, and 1 was male (age range, 20–76 years; mean age, 50.25 years). All patients underwent conventional preoperative sonography, 2-dimensional (2D) contrastenhanced sonography, and 3D contrast-enhanced sonography and subsequently underwent surgery (modified radical mastectomy, radical mastectomy, or conservative breast surgery) within 1 week. None of them received preoperative neoadjuvant chemotherapy or endocrine therapy. Histologic analysis showed 35 benign breast lesions, including fibroadenoma (n = 28), intraductal papilloma (n = 5), and phyllodes tumors (n = 2), and 148 malignant lesions, including invasive ductal carcinoma (n = 108), mixed invasive carcinoma (n = 18), ductal carcinoma in situ (n = 8), invasive lobular carcinoma (n = 4), apocrine carcinoma (n = 4), invasive micropapillary carcinoma (n = 2), metaplastic carcinoma (n = 2), and neuroendocrine carcinoma (n = 2). Sonographic Examinations All sonographic examinations, including conventional sonography, 2D contrast-enhanced sonography, and 3D contrast-enhanced sonography, were performed with the same ultrasound machine (MyLab 90; Esaote SpA, Genoa, Italy). Conventional sonography and color Doppler sonography were performed with an LA532 transducer with a frequency of 13–4 MHz, whereas 2D contrast-enhanced sonography was evaluated with an LA522 transducer with a frequency of 9–3 MHz. A BL433 volume transducer with a frequency of 15–9 MHz was used for 3D scanning. The contrast agent was SonoVue (BR1; Bracco SpA, Milan, Italy), a sulfur hexafluoride–filled microbubble contrast agent. To avoid interobserver variability, all sonographic scanning was performed by a single radiologist (M.C., with >15 years of experience in breast sonography, 5 years in breast 2D contrast-enhanced sonography, and 1 year in breast 3D contrast-enhanced sonography). All scanning was usually performed less than 1 week before surgery.
J Ultrasound Med 2014; 33:835–846
3305jum759-928online_Layout 1 4/22/14 8:54 AM Page 837
Chen et al—3D Contrast-Enhanced Sonography of Breast Tumors
Examinations were performed as follows: Before injection of the contrast agent, the breast tumors were scanned first on conventional and color Doppler sonography to obtain the best imaging in the maximum plane from which both the tumors and the normal adjacent breast tissue could be observed. Imaging parameter settings were optimized to ensure high-quality conventional sonograms after the target lesion was determined. Baseline 3D scanning was then performed using the same scanning route to define the appropriate volume angle so that the whole lesion would be included in the volume data without signal loss. Subsequently, all patients manually received injections of 2.4 mL of SonoVue as a bolus through an antecubital vein, followed by a flush with 5 mL of 0.9% saline. Breast mass 2D contrast-enhanced sonography was evaluated in real time until 2 minutes after the beginning of the injection. The following settings were used for 2D contrast-enhanced sonography: The selected plane included the lesion and its surrounding normal tissue if possible. The scanner settings for contrast-enhanced sonography were as follows: mechanical index, 0.09; single focus at the bottom of the image; dynamic range, 70 dB; image depth, 3 or 4 cm; the probe was stabilized manually; and no pressure was exerted. The timer was activated immediately at the beginning of injection. The lesion was observed for at least 2 minutes, and the whole process of contrast-enhanced sonography was stored in the hard disk of the ultrasound machine in the Digital Imaging and Communications in Medicine format for further analysis. Five to 10 minutes after contrastenhanced sonography, when the signals from the microbubbles in the large vessels such as the axillary vein disappeared, the transducer was shifted from LA522 to BL433, and 3D contrast-enhanced sonography was initiated. The contrast agent was injected in the same dose and fashion as for 2D contrast-enhanced sonography. Ten seconds later, 3D contrast-enhanced sonograms were obtained more than 5 times continuously with a total time of more than 2 minutes. The imaging settings for 3D contrast-enhanced sonography were as follows: mechanical index, 0.08 to 0.13; 1 focal zone; power output, 3% to 6%; dynamic range, 40 to 60 dB; volume angle, 30° to 50°; and the scanning route was consistent with that of 2D contrastenhanced sonography. During 3D scanning, the transducer was kept in a stable position without movement. After scanning, reformatted 2D contrast-enhanced images of 3 orthogonal planes were displayed on the screen. The data acquired were also stored in the Digital Imaging and Communications in Medicine format on the hard disk of the ultrasound system.
J Ultrasound Med 2014; 33:835–846
Reconstruction of 3D Contrast-Enhanced Sonograms The 3D contrast-enhanced sonographic measurements were analyzed by 2 investigators (L.T., with 6 years of experience in conventional breast sonography and 1 year in 3D contrast-enhanced sonography; and Y.W., with 5 years of experience in conventional breast sonography and 1 year of experience in 3D contrast-enhanced sonography), without knowledge of pathologic information. After independent interpretations by the observers, a consensus was obtained by a conference. If different assessments were assigned by the readers, then a consensus was reached after discussing the findings. There are various 3D display and rendering modes, such as triplanar imaging, tomographic mode imaging, volume rendering and analysis, thick-slice imaging, surface mode, transparency mode, and a combination of surface and transparency modes. In this study, the transparent mode and tomographic mode imaging were mainly selected to depict neovascularization and to distinguish intratumoral enhancement. The characteristics of tumor vasculature and perfusion were focused to assess the following: (1) peripheral vessels (presence or absence), which were defined as radial or parallel vessels surrounding the tumor margin with or without branches, and their number (
