European Journal of Radiology 83 (2014) 2129–2136
Contents lists available at ScienceDirect
European Journal of Radiology journal homepage: www.elsevier.com/locate/ejrad
Correlation between enhancement characteristics of MR mammography and capillary density of breast lesions Alexander Poellinger a,∗ , Sahra El-Ghannam a , Susanne Diekmann a , Thomas Fischer a , Glen Kristiansen b , Florian Fritzsche c , Eva Fallenberg a , Lars Morawietz d , Felix Diekmann a a
Charité, Universitätsmedizin Berlin, Department of Radiology, Augustenburger Platz 1, 13353 Berlin, Germany Universitätsklinikum Bonn, Department of Pathology, Sigmund-Freud-Str. 25, D-53127 Bonn, Germany c Institut für Histologie und Zytologie, Bahnhofplatz 11, Postfach, 9101 Herisau, Switzerland d Diagnostik Ernst von Bergmann GmbH, Charlottenstr. 72, 14467 Potsdam, Germany b
a r t i c l e
i n f o
Article history: Received 11 June 2014 Received in revised form 13 September 2014 Accepted 15 September 2014 Keywords: Magnetic resonance imaging Breast neoplasms Pathology Blood supply
a b s t r a c t Objective: To correlate capillary density of breast lesions using the markers D2-40, CD31, and CD34 with early and late enhancement of magnetic resonance mammography (MRM). Materials and methods: The local ethics committee approved this study, and informed consent was available from all patients. The study included 64 women with 66 histologically proven breast lesions (41 malignant, 25 benign). MR-enhancement 1 min after contrast medium administration was determined in the tumor (It1 /It0 ratio) and in comparison to the surrounding tissue (It1 /It1-fat ratio). Capillary density was quantified based on immunohistological staining with D2-40, CD31, and CD34 in breast tumors and surrounding breast tissue. Mean capillary densities were correlated with contrast enhancement in the tumor and surrounding breast tissue. The Kruskal–Wallis test was used to test whether lesions with different MR enhancement patterns differed in terms of capillary density. Results: For CD34, there was statistically significant correlation between capillary density and tumor enhancement (r = 0.329, p = 0.012), however not for the malignant or benign groups separately. Mean vessel number identified by staining with D2-40 and CD31 did not correlate significantly with tumor enhancement (D2-40: r = −0.188, p = 0.130; CD31: r = 0.095, p = 0.448). There were no statistically significant differences in capillary density between breast lesions with delayed enhancement or a plateau and lesions showing washout (Kruskal–Wallis test. D2-40: p = 0.173; CD31: p = 0.647; CD34: p = 0.515). Conclusion: Of the three markers tested, CD34 showed best correlation between early contrast enhancement on MRM and capillary density. Further studies are necessary to clearly demonstrate an association between capillary density and contrast enhancement in breast tumors and surrounding tissue. © 2014 Elsevier Ireland Ltd. All rights reserved.
1. Introduction In recent years, magnetic resonance mammography (MRM) has been increasingly used to supplement conventional mammography and other diagnostic breast imaging modalities [1]. When a suspicious breast lesion is detected by MRM that is not visible by any other imaging modality, it is recommended to obtain an MR-guided transcutaneous biopsy for histological confirmation [2]. To ensure correct lesion sampling it would be desirable to prove
∗ Corresponding author at: Charité, Universitätsmedizin Berlin, Department of Radiology, Charité Campus Virchow Klinikum, Augustenburger Platz 1, 13353 Berlin, Germany. Tel.: +49 30 450 627269; fax: +49 30 450 550914. E-mail address: [email protected] (A. Poellinger). http://dx.doi.org/10.1016/j.ejrad.2014.09.007 0720-048X/© 2014 Elsevier Ireland Ltd. All rights reserved.
that the samples origin indeed from the suspicious region. While specimen mammography or specimen ultrasound should be performed for breast lesions identified by these imaging modalities, no such recommendation exists for breast specimens obtained by MR-guided biopsy. When a biopsy has been obtained from a breast lesion detected by MRM only, it is often not very helpful to perform specimen radiography and it is sometimes difficult to prove that histological findings match the imaging appearance. In a study by Lee et al. [3], histology was discordant with imaging in 7% of lesions that were examined by MR-guided vacuum-assisted biopsy. Among discordant lesions, i.e. lesions that showed suspicious features on MRI but were evaluated as benign by histology following MR-guided biopsy, surgical excision revealed cancer in 30%. Liberman et al. [4] encountered histology-imaging discordance in nine of 98 MR-guided vacuum-assisted breast biopsies (9%).
2130
A. Poellinger et al. / European Journal of Radiology 83 (2014) 2129–2136
Four of the nine discordant lesions were subsequently found to be cancers. Thus, in all patients with discordant imaging findings and histology, repeated percutaneous biopsy or open biopsy should be contemplated. To minimize the rate of repeated biopsies, it is desirable to have a method for reliable correlation of imaging findings and histopathological parameters. A technique for verifying a suspected lesion sampled by MR-guided biopsy could be developed if one demonstrated a direct numerical correlation between contrast enhancement on MRI and histopathological microvessel density (MVD) in breast lesions. Several study groups have used immunohistological staining to determine the capillary density of breast tumors in order to identify possible correlations with signal enhancement on dynamic contrast-enhanced MRM [5–9]. The results are very heterogeneous, one reason being the diversity of histopathological markers used, including markers that are not very common. As a result, further studies are necessary to provide definitive evidence regarding a possible association between tumor angiogenesis and signal enhancement of breast lesions at MRM. In the present study, we investigated three widely used immunohistological markers, D2-40, CD31 and CD34, to quantify vessel density within breast tumors and directly adjacent breast tissue. These results were compared with contrast enhancement of breast tumors and surrounding breast tissue in dynamic contrastenhanced MR mammography. 2. Material and methods The study was approved by the local ethics committee, and all patients gave written informed consent. Data of 64 consecutive patients who underwent MRM with subsequent histological workup of suspicious breast lesions was analyzed. The interval between MRM and histological verification was no longer than 90 days (median 9 days). 2.1. MR mammography protocol All MR mammographies were performed on a 1.5-Tesla MR imager (Siemens Magnetom, SymphonyVision® , Erlangen, Germany) using a dedicated breast double coil (Siemens, NORAS® ). In premenopausal patients, MRM was performed on days 7–15 of the menstrual cycle. The MR protocol included a coronal T1-weighted turbo spin echo (TSE) sequence and an axial T2weighted TSE sequence before contrast medium administration, and a dynamic coronal 3D T1-weighted fast low-angle shot (FLASH) sequence (TR = 8.1 ms, TE = 4.48 ms, Flip angle = 25◦ , matrix: 256 × 256, Field of View: 320 mm, slice thickness: 2 mm) acquired before and at five time points after automated intravenous bolus injection of 0.2 mmol/kg body weight of Gd-DTPA (Magnevist® , Schering, Berlin, Germany), followed by a 20 ml saline flush, both at an injection rate of 2 ml/sec. The dynamic sequence was followed by an axial T1-weighted fat-suppressed spin echo (FS SE) sequence and an axial STIR sequence. 2.2. Data analysis Dynamic breast MR images were analyzed by quantifying signal intensities over time in regions of interest (ROI) set by a diagnostic radiologist with more than 10 years of experience in MRM. The ROI was placed in an area in which signal intensity was highest within the first minute after contrast medium administration. The ROI comprised at least 5 pixels to avoid local fluctuation and thus misinterpretation of contrast enhancement. The FuncTool software package (Advantage Windows Workstation, GE Healthcare, Milwaukee, WI) was used to measure signal intensities in the
ROI. Contrast enhancement was analyzed on dynamic T1-weighted images in the lesion before contrast administration (It0 ) and one minute after administration (It1 ) as well as in surrounding tissue at timepoint t1 (It1-fat ). Relative signal enhancement at timepoint t1 was calculated as the ratio of signal intensity at t1 to the baseline intensity in the tumor at t1 (It1 /It0 ). A second ratio, It1 /It1-fat , was calculated to express signal enhancement in the lesion in relation to the enhancement in surrounding fat. In addition, we documented the timepoint of maximum signal enhancement (peak) on dynamic MR images as well as the course of signal enhancement. “Washout” was defined as at least 10% decrease in signal enhancement, “prolongated enhancement” as 10% or more increase in signal enhancement relative to the intensity at t = t1. 2.3. Histopathology Magnetic resonance-guided biopsies were performed using a vacuum-assisted biopsy system with an 11-gauge needle (Mammotome® Breast Biopsy System, Cincinnati, USA). The tissue sampled from the 66 breast lesions was fixed in formalin and grossly evaluated before paraffin embedding. The paraffinembedded specimens were cut into 2–3 m sections, which were mounted on glass slides. Routine hematoxylin and eosin (H&E) was used for visualization of tissue structures. Following gross and microscopic evaluation of the tissue samples, routine immunohistochemical examinations were supplemented by staining for D2-40, CD31, and CD34 for quantification of capillary densities in tumors. D2-40 is a commercially available antibody for visualization of both normal and reactive or neoplastic lymphatic vessels. The lymphatic vessels are visualized by labeling of endothelium [10,11]. In this study, we used the D2-40 clone from Dako (Dako Deutschland GmbH, Hamburg, Germany) at a 1:100 dilution. The antibody also labels lymphatic channels of lymph nodes, though less strongly [12]. Nonendothelial cells reacting with D2-40 include the myoepithelium of the breast. Blood vessels are always negative in immunohistological specimens incubated with D2-40, both in normal tissue and in reactive tissue such as scar tissue. The D2-40 antibody thus ensures specific visualization of lymphatic vessels. CD31, also known as PECAM-1 (platelet endothelial cell adhesion molecule 1), is a transmembrane glycoprotein with a molecular weight of 130 kDa. It is mainly expressed by endothelial cells but also by some leukocytes [13]. The JC70A antibody clone (Dako Deutschland GmbH, Hamburg, used at a concentration of 1:25) serves to detect endothelial cells in formalin-fixed tissue, and thus has been used for evaluating angiogenesis in proliferating breast tumors. The CD31 antibody selectively labels the endothelial cells of blood vessels but not of lymphatic vessels. CD34 is a membrane-bound glycoprotein with a molecular weight of 116 kDa. CD34 antigen is expressed by hematopoietic stem cells. In stem cell transplantation, the CD34 antigen can be used to enrich hematopoietic stem cells from peripheral blood; this is possible as blood stem cells can be identified by means of the CD34 antigen and separated from other blood cells [13]. The CD34 antibody is especially useful for labeling the endothelial cells of blood vessels [9,11,14]. However, unlike the CD31 antibody, CD34 also reacts with lymphatic endothelia cells to a small degree [13]. In the present study we used the QBEnd1 antibody clone (Dako Deutschland GmbH, Hamburg) at 1:100 dilution. The specimens were stained automatically using the Ventana BenchMark® XT staining system (Ventana, Tucson, AZ, USA). For correlation of immunohistological capillary densities with MR contrast enhancement, five high-power fields in the tumor and five high-power fields in surrounding tissue stained for D2-40, CD31, and CD34 were counted per lesion to calculate arithmetic means.
A. Poellinger et al. / European Journal of Radiology 83 (2014) 2129–2136 Table 1 Distribution of benign breast lesions. Benign breast lesions Mastitis Mastopathy Common ductal hyperplasia Lobular hyperplasia Fibrosis/sclerosis Fibroadenoma Papilloma Adenosis
Table 3 Histological composition of the tissue surrounding breast lesions. Number (percentage) 5 (20%) 2 (8%) 2 (8%) 1 (4%) 8 (32%) 3 (12%) 2 (8%) 2 (8%) 25 (100%)
Histology
Number (percentage)
Fat Fat and glandular tissue Fat and connective tissue Glandular tissue Fibrotic glandular tissue Fibrosis Fat and isolated aggregated glands Fat and lymphatic follicles
40 (60.6%) 12 (18.3%) 3 (4.5%) 3 (4.5%) 3 (4.5%) 3 (4.5%) 1 (1.5%) 1 (1.5%)
66 (100%)
of the surrounding tissue of all 66 breast lesions is summarized in Table 3.
Table 2 Distribution of malignant breast lesions. Malignant breast lesions
Number (percentage)
DCIS LCIS IDC ILC
5 (12.2%) 1 (2.4%) 23 (56.1%) 12 (29.3%)
2131
41 (100%)
The diameter of a high-power field was 0.5 mm. Counts were performed at 400× magnification and five representative areas with high capillary densities in the tumor and in five areas of surrounding tissue for each lesion were evaluated. Histological capillary densities were correlated with contrast enhancement in the tumor (It1 /It0 ratio) and the surrounding tissue (It1 /It1-fat ratio). 2.4. Statistical analysis Statistical analysis was performed using SPSS, version 14.0 (Chicago, IL, USA). Correlations between imaging and histopathological capillary density were assessed by calculating two-tailed Spearman’s rho. The zero hypothesis (H0 ) postulates that no correlation exists between immunohistochemical vessel counts in breast tumors and surrounding tissue and MR contrast enhancement. The alternative hypothesis (H1 ) states that there is a significant correlation between imaging and histopathology. Statistical significance was assumed at a p-value of 0.05. 3. Results 3.1. Pathology Benign lesions. There were 24 patients with a total of 25 benign lesions. Mean patients’ age was 49.4 ± 11.53 years (ranged between 25 and 74 years). Fibroadenoma was the most common benign breast tumor. The distribution of benign breast lesions is summarized in Table 1. Malignant lesions. The 40 patients with a total of 41 malignant lesions ranged in age between 29 and 79 years (mean age, 53.8 ± 11.54 years). Twenty-two patients with a total of 23 malignant lesions had invasive ductal carcinoma (IDC), accounting for 56.1% of all malignant tumors. There were 12 invasive lobular carcinomas (ILC). The high percentage of ILCs is due to the criteria for performing MRM at the breast center of our hospital (all patients with ILC and with extremely dense breast parenchyma get a recommendation for MRM). Table 2 lists the distribution of malignant breast tumors in the study population. Tissue surrounding breast lesions. In 26 of all 66 cases, the perilesional tissue consisted not only of fat but also other tissue such as glandular and connective tissue. The histological composition
3.2. Contrast enhancement at time t1 and course of signal intensities For benign lesions the average relative signal increase (It1 /It0 ) in breast lesions was 2.53 ± 0.97 versus 0.92 ± 0.27 in perilesional fat. In malignant lesions this ratio was 3.14 ± 1.25 with a ratio of 1.07 ± 0.44 in perilesional fatty tissue. 3.3. Immunohistochemical vessel counts Mean values of capillary density in lesions and surrounding tissue were calculated from the mean values of five fields of view of the individual lesions for all breast lesions and for each of the three markers used (Table 4). For D2-40, the average number of vessels was less than one per field of view in malignant and benign lesions as well as in the surrounding tissue, for benign and malignant tumors. With the other two markers, CD31 and CD34, the mean vessel count was markedly higher in both malignant (number of vessels more than five) and benign breast lesions and surrounding tissue (Table 4). Fig. 1 shows a tissue sample of an invasive ductal carcinoma and the surrounding tissue with different histopathological markers. Photomicrographs of lesion samples from another patient with mastitis and corresponding dynamic MR-images are illustrated in Fig. 2. 3.4. Correlations 3.4.1. Tumor enhancement and capillary density For benign and malignant lesions taken together (n = 66), twotailed Spearman’s rho showed a significant correlation between tumor enhancement (It1 /It0 ratio) at the time of the first postcontrast measurement and the tumor vessel count using CD34 (r = 0.259, p = 0.035) (Table 5). The scatter diagram in Fig. 3 illustrates the statistical association between the CD34 vessel counts and tumor enhancement (t1/t0 ratio). There is marked residual scatter with individual residues very widely deviating from the regression line. There was no significant association between vessel counts after staining with D2-40 and CD31 and tumor signal intensities (Fig. 3). Analysis for correlations between tumor enhancement (It1 /It0 ratio) and mean immunohistochmemical vessel counts using twotailed Spearman’s rho yielded no statistically significant correlation for the 25 benign lesions alone, neither for D2-40 nor for CD31 or CD34. The same was found for the 41 malignant lesions in our study if tested alone. 3.4.2. Perilesional enhancement and capillary density Correlations between enhancement of surrounding tissue and capillary density were only calculated for a subset of specimens where the surrounding tissue was exclusively composed of fat, which was the case for 40 lesions – 16 benign and 24 malignant.
2132
A. Poellinger et al. / European Journal of Radiology 83 (2014) 2129–2136
Table 4 Immunohistochemical staining and average microvessel densities (MVD) with standard deviations. MVD was measured as microvessels per surface area (0.25 mm2 ) and is expressed here as microvessels/mm2 . Marker D2-40 CD31 CD34
Tissue
All lesions (n = 66)
Tumor Surrounding Tumor Surrounding Tumor Surrounding
3.36 1.72 31.52 12.08 29.88 13.76
± ± ± ± ± ±
Benign lesions
6.16 2.76 25.96 8.44 15.20 10.40
3.24 0.88 22.88 10.76 25.92 12.8
± ± ± ± ± ±
3.92 1.24 9.88 7.52 12.04 9.28
Malignant lesions 3.44 2.24 36.80 12.92 32.32 14.32
± ± ± ± ± ±
7.24 3.32 31.00 8.96 16.56 11.12
Table 5 Correlation of D2-40, CD31, and CD34 expression with contrast enhancement at dynamic MRM for benign tumors (a), malignant tumors (b), all tumors (c), and all lesions detected by diagnostic MRM (d); two-tailed Spearman’s rho. It1 /It0 ratio
N
Benign
25
Malignant
41
All lesions
66
*
Correlation coefficient Significance Correlation coefficient Significance Correlation coefficient Significance
D2-40-tumor
CD31-tumor
CD34-tumor
r = −0.127 p = 0.545 r = −0.185 p = 0.248 r = −0.188 p = 0.130
r = −0.018 p = 0.932 r = 0.001 p = 0.996 r = 0.095 p = 0.448
r = 0.146 p = 0.485 r = 0.236 p = 0.137 r = 0.259* p = 0.035
Correlation is assumed to be significant at p = 0.05.
For quantitative comparison, ratios of the numbers of vessels in lesions to surrounding tissue were calculated for all three stains investigated (D2-40-tumor to D2-40-normal ratio, CD31-tumor to CD31-normal ratio, and CD34-tumor to CD34-normal ratio). Testing using two-tailed Spearman’s rho yielded no significant correlation between vascular density in surrounding tissue and tumor enhancement (It1 /It1-fat ratio) at the first measurement after contrast medium administration. For all immunohistochemical markers used, the Kruskal–Wallis test revealed no statistically significant differences in capillary density between breast lesions with delayed enhancement or a plateau and lesions showing washout (D2-40: p = 0.173; CD31: p = 0.647; CD34: p = 0.515). 4. Discussion Three immunohistological stains, CD34, CD31, and D2-40, were used to determine capillary density in breast lesions and surrounding tissue. The average numbers of vessels in tumors and surrounding tissue found in our study are in line with formerly published results obtained using the same immunohistochemical stains for detecting tumor neoangiogenesis [7,9,15,16]. Teifke at al. found the microvessel density (MVD) of malignant lesions (stained with CD34) to be between 19 and 50 microvessels/mm2 , which is in the range of the MVD in our study (32–37 microvessels/mm2 ). The rather lower MVD found in our study might be attributable to the relative high frequency of lobular cancers in our study that show variable degree of neoangiogenesis. Slight differences may be due to the use of different methods for determining vascular density. For instance, Buadu et al. [15] divided breast lesions into four groups based on vascular density. A similar approach was used by Su et al. [9], who distinguished three groups of tumors based on microvessel density (MVD) – tumors with high, intermediate, and low MVD. Müller-Schimpfle et al. [7] determined capillary density in the tumor center, in the tumor margin, and in the surrounding tissue; in contrast to our approach, they counted three high-power fields at 200× magnification (in our study: five high-power fields at 400× magnification). Because of the different methods used for quantifying neoangiogenesis, our results cannot easily be compared with the reported results. Correlation between MR-imaging and histopathology yielded a statistically significant association between vessel counts using
CD34 and tumor enhancement for all 66 breast lesions. Several investigators tried to identify possible effects of tumor capillary density on tumor enhancement on Gd-DTPA-enhanced MRM [6–9,16–20]. Teifke et al. [16] also observed a significant association between tumor capillary density determined with CD34 and early enhancement of the tumor center for both benign and malignant breast lesions. The correlation coefficient was higher for breast cancer compared with benign lesions. In our study, separate analysis of benign and malignant lesions failed to reveal a significant association between tumor enhancement and capillary density. Our results are in agreement with the study of Buadu et al. [15]. They also found the intensity of lesion enhancement one minute after contrast administration to correlate with the number of tumor vessels stained with CD34. In addition, they showed tumor capillary density as well as contrast enhancement to be heterogeneously distributed in tumorous lesions. Supporting the experience of many investigators [15,21–23] that the degree of contrast enhancement alone is insufficient to differentiate benign from malignant breast lesions, they stated that breast lesion characterization should also include the course of enhancement and the MR morphology of focal lesions. We found no statistically significant correlation between tumor capillary density after staining with CD31 and tumor enhancement. Knopp et al. and Müller-Schimpfle et al. [6,7] also used CD31 for immunohistological staining to investigate the effect of capillary density of breast tumors on enhancement characteristics in dynamic MRM. In contrast to our results, Knopp et al. observed a significantly faster signal increase in malignant breast lesions compared to benign lesions, but no significant association between capillary density and tumor signal intensity one minute after contrast medium administration. Müller-Schimpfle et al., on the other hand, found a slightly significant correlation between peripheral enhancement of breast lesions and capillary density in the lesion margin, while the association between enhancement of the tumor center and capillary density did not reach significance. Su et al. [9], using CD31 staining in a total of 71 breast cancers, observed no significant correlation in the tumor periphery or center, which is in agreement with our findings. It should be noted that a direct comparison of these results is not possible as different image postprocessing techniques were used by Su et al. and in the present study. Although no definite answer can be given, it may be assumed that some of the discrepancy between the study results is attributable to the use of different methods for determining vessel
A. Poellinger et al. / European Journal of Radiology 83 (2014) 2129–2136
2133
Fig. 1. Photomicrographs (original magnification, ×200) show invasive ductal carcinoma (left column) and surrounding tissue (right column) with different immunohistochemical markers: H&E staining showing atypical glandular tumor infiltrates with numerous mitotic figures and a lymphocytic inflammatory infiltrate (a) and connective tissue with low cellularity in the surrounding tissue (b). Immunohistochemical staining with D2-40 displaying some lymph vessels in the tumor (c) and the surrounding tissue (d). CD31 staining of the tumor with multiple microvessels (e) and only few vessels in the surrounding tissue (f). CD34 staining of the ductal carcinoma shows several microvessels (the dark blue structure in the middle of the image is an artifact) (g). The surrounding tissue also exhibits some microvessels (h).
density and MR signal intensities as well as to the use of different statistical tests. In our study, staining with the endothelial marker D2-40 for identifying and counting lymphatic vessels revealed no significant
correlation with MR contrast enhancement. To date, published studies investigating lymphatic capillary density in breast lesions have not correlated this parameter with MR contrast enhancement. Rather this parameter was used to investigate the possible
2134
A. Poellinger et al. / European Journal of Radiology 83 (2014) 2129–2136
Fig. 2. Photomicrographs (A) of lesion samples of the left breast of a 25-years old woman and dynamic MRI images (B). (A) Photomicrographs (original magnification, 200×) show mastitis (left column) and surrounding tissue (right column) with different immunohistochemical markers: (a) H&E staining exhibits a cell-rich inflammatory response. Microscopically the milk ducts can be hardly discerned due to massive inflammation and cell proliferation. (b) Connective tissue surrounding the lesion. (c) Immunohistochemical staining with D2-40 displays a lymph vessel in the inflamed area (upper right corner) and (d) a rather large lymph vessel the surrounding tissue. (e) CD31 staining of the mastitis area with multiple microvessels and (f) only few vessels in the surrounding tissue. (g) CD34 staining of the sample shows several microvessels. (h) The surrounding tissue also exhibits some microvessels. (B): Corresponding dynamic MRI: from top to bottom: non-contrast enhanced T1-weighted coronal scan followed by scans at one, three and five minutes after contrast administration. The bottom image displays subtraction between the images at 3 min and the non-contrast enhanced scan.
diagnostic relevance of lymphatic vessel density in breast tumors in terms of metastatic spread, recurrence-free survival, and overall survival [11,12,24]. El-Gohary et al. and Mohammed et al. [11,12] found a significant association between intratumoral lymphatic vessel density and the presence of distant metastasis, which in turn correlated with a shorter overall survival. In our study, we identified lymphatic vessels using D2-40 in 19 of 41 breast cancers (46%) and in 17 of 25 benign lesions (68%). As expected, lymphatic capillary density was much smaller than the number of intratumoral blood capillaries. This finding is in agreement with the results of El-Gohary and Mohammed. We identified on average less than one lymphatic vessel per tumor, for both benign and malignant tumors. In addition, we also investigated capillary density in fat tissue surrounding breast lesions (but clearly outside the lesion) and correlated this histopathological parameter with early contrast
enhancement of peritumoral fat. These investigations revealed no statistically significant correlations for any of the three stains used. Only one of the research groups investigating capillary density and contrast enhancement also investigated capillary density in the tissue surrounding breast lesions [7]. In contrast to our results, they found a statistically significant correlation between capillary density around breast lesions and MR signal intensity. Note that, when interpreting these results, our analysis only included specimens (n = 40) in which the surrounding tissue was exclusively composed of fat. This approach was necessary as the relative contrast enhancement was based on the surrounding fat. Therefore, 26 lesions with heterogeneous surrounding tissue could not be included in this analysis. A comparison between our results and those reported is difficult because of the use of different histopathological methods and the use of different formulas for calculating relative signal enhancement.
A. Poellinger et al. / European Journal of Radiology 83 (2014) 2129–2136
2135
4.2. Conclusion Of the immunohistochemical stains used in the present study, CD34 appears to be most suitable to identify a possible correlation between capillary density and MR contrast enhancement of breast lesions. To obtain more reliable results, further studies using a standardized methodology are necessary to clearly demonstrate an association between capillary density and contrast enhancement in breast tumors and surrounding tissue. Once such data are available, a strong correlation between capillary density and enhancement might be interpreted to confirm lesion sampling. Conversely, a weak correlation, along with discordant imaging and histopathology findings, might be one more criterion in support of repeating the biopsy. Nevertheless, an interdisciplinary tumor conference continues to be indispensible for a comprehensive evaluation and correlation of imaging findings and histological results and timely identification of false-negative findings. Funding Fig. 3. Statistical association between the CD34 vessel counts and tumor enhancement (t1/t0 ratio). For benign and malignant lesions taken together (n = 66), two-tailed Spearman’s rho showed a significant correlation between tumor enhancement (It1 /It0 ratio) at the time of the first postcontrast measurement and the tumor vessel count using CD34 (r = 0.259, p = 0.035).
4.1. Limitations of the study Our study has some methodological limitations: 1. The fact that only lesions suspected for malignancy were submitted to histopathological verification results in a selection bias. This bias mainly affects the signal intensity time course in benign breast lesions and explains the high proportion of intensity-time curves with washout or a plateau phase we observed in benign lesions. This bias has no direct effect on the correlation between capillary density and contrast enhancement in tumors and surrounding tissue. Hence, the effect is negligible for the purpose of the present study. 2. The maximum delay between the detection of a suspicious breast lesion by MRM and histological verification was 90 days. This is critical in view of the average growth rate of breast cancers of 0.6 mm per month, depending on the patient’s age and the initial tumor size [25]. However, the median delay between MRM and histological verification was 9 days. Only for 5 of the 66 lesions the delay was longer than 30 days. Therefore, the risk of a possible statistical distortion resulting from differences in tumor size between the time of mammography and the time of histology again appears to be small. 3. Another limitation of our study is that the signal intensities in tumors and surrounding tissue were only assessed visually in subjectively placed ROIs, which limits reproducibility. However, experienced researchers as Kuhl et al. [26] consider visual assessment of ROIs and signal intensity-time curves to be sufficiently reproducible. Mammograms and pathologic specimens were assessed by only one radiologist and one pathologist. However, both readers were blinded to the results of the other. 4. Our study is also limited by the fact that only 40 of the 66 cases were included in the analysis of capillary density and contrast enhancement of surrounding fatty tissue as the remaining 26 cases did not meet the criteria (lesion surrounded by fat). 5. Finally, the composition of the benign lesion group was very heterogeneous consisting e.g. of five cases of mastitis and two cases of mastopathy. On the other hand there was a high percentage of invasive lobular carcinomas in the malignant group that exhibited a lesser degree of MVD. This heterogeneity might have weakened the correlation between MVD and signal intensity.
No funding. References [1] Akita A, Tanimoto A, Jinno H, Kameyama K, Kuribayashi S. The clinical value of bilateral breast MR imaging: is it worth performing on patients showing suspicious microcalcifications on mammography? Eur Radiol 2009;19(9):2089–96. [2] Peters NH, Meeuwis C, Bakker CJ, et al. Feasibility of MRI-guided largecore-needle biopsy of suspicious breast lesions at 3 T. Eur Radiol 2009;19(7):1639–44. [3] Lee JM, Kaplan JB, Murray MP, et al. Imaging histologic discordance at MRI-guided 9-gauge vacuum-assisted breast biopsy. AJR Am J Roentgenol 2007;189(4):852–9. [4] Liberman L, Bracero N, Morris E, Thornton C, Dershaw DD. MRI-guided 9-gauge vacuum-assisted breast biopsy: initial clinical experience. AJR Am J Roentgenol 2005;185(1):183–93. [5] Ikeda O, Nishimura R, Miyayama H, et al. Evaluation of tumor angiogenesis using dynamic enhanced magnetic resonance imaging: comparison of plasma vascular endothelial growth factor, hemodynamic, and pharmacokinetic parameters. Acta Radiol 2004;45(4):446–52. [6] Knopp MV, Weiss E, Sinn HP, et al. Pathophysiologic basis of contrast enhancement in breast tumors. J Magn Reson Imaging 1999;10(3):260–6. [7] Muller-Schimpfle M, Noack F, Oettling G, et al. Influence of histopathological factors on dynamic MR mammography. Rofo 2000;172(11):894–900. [8] Stomper PC, Winston JS, Herman S, Klippenstein DL, Arredondo MA, Blumenson LE. Angiogenesis and dynamic MR imaging gadolinium enhancement of malignant and benign breast lesions. Breast Cancer Res Treat 1997;45(1):39–46. [9] Su MY, Cheung YC, Fruehauf JP, et al. Correlation of dynamic contrast enhancement MRI parameters with microvessel density and VEGF for assessment of angiogenesis in breast cancer. J Magn Reson Imaging 2003;18(4):467–77. [10] Marinho VF, Metze K, Sanches FS, Rocha GF, Gobbi H. Lymph vascular invasion in invasive mammary carcinomas identified by the endothelial lymphatic marker D2-40 is associated with other indicators of poor prognosis. BMC Cancer 2008;8:64. [11] Mohammed RA, Ellis IO, Elsheikh S, Paish EC, Martin SG. Lymphatic and angiogenic characteristics in breast cancer: morphometric analysis and prognostic implications. Breast Cancer Res Treat 2009;113(2):261–73. [12] El-Gohary YM, Metwally G, Saad RS, Robinson MJ, Mesko T, Poppiti RJ. Prognostic significance of intratumoral and peritumoral lymphatic density and blood vessel density in invasive breast carcinomas. Am J Clin Pathol 2008;129(4):578–86. [13] Pusztaszeri MP, Seelentag W, Bosman FT. Immunohistochemical expression of endothelial markers CD31, CD34, von Willebrand factor, and Fli-1 in normal human tissues. J Histochem Cytochem 2006;54(4):385–95. [14] Mohammed RA, Green A, El-Shikh S, Paish EC, Ellis IO, Martin SG. Prognostic significance of vascular endothelial cell growth factors -A, -C and -D in breast cancer and their relationship with angio- and lymphangiogenesis. Br J Cancer 2007;96(7):1092–100. [15] Buadu LD, Murakami J, Murayama S, et al. Breast lesions: correlation of contrast medium enhancement patterns on MR images with histopathologic findings and tumor angiogenesis. Radiology 1996;200(3):639–49. [16] Teifke A, Behr O, Schmidt M, et al. Dynamic MR imaging of breast lesions: correlation with microvessel distribution pattern and histologic characteristics of prognosis. Radiology 2006;239(2):351–60. [17] Buadu LD, Murakami J, Murayama S, et al. Patterns of peripheral enhancement in breast masses: correlation of findings on contrast medium enhanced MRI with histologic features and tumor angiogenesis. J Comput Assist Tomogr 1997;21(3):421–30.
2136
A. Poellinger et al. / European Journal of Radiology 83 (2014) 2129–2136
[18] Ikeda O, Nishimura R, Miyayama H, et al. Magnetic resonance evaluation of the presence of an extensive intraductal component in breast cancer. Acta Radiol 2004;45(7):721–5. [19] Oshida K, Nagashima T, Ueda T, et al. Pharmacokinetic analysis of ductal carcinoma in situ of the breast using dynamic MR mammography. Eur Radiol 2005;15(7):1353–60. [20] Van den Eynden GG, Van der Auwera I, Van Laere SJ, et al. Comparison of molecular determinants of angiogenesis and lymphangiogenesis in lymph node metastases and in primary tumours of patients with breast cancer. J Pathol 2007;213(1):56–64. [21] Bick U. Typical and unusual findings in MR mammography. Rofo 2000;172(5): 415–28. [22] Siegmann KC, Muller-Schimpfle M, Schick F, et al. MR imaging-detected breast lesions: histopathologic correlation of lesion characteristics and signal intensity data. AJR Am J Roentgenol 2002;178(6):1403–9.
[23] Wiener JI, Schilling KJ, Adami C, Obuchowski NA. Assessment of suspected breast cancer by MRI: a prospective clinical trial using a combined kinetic and morphologic analysis. AJR Am J Roentgenol 2005;184(3):878–86. [24] Van der Auwera I, Van Laere SJ, Van den Eynden GG, et al. Increased angiogenesis and lymphangiogenesis in inflammatory versus noninflammatory breast cancer by real-time reverse transcriptase-PCR gene expression quantification. Clin Cancer Res 2004;10(23):7965–71. [25] Weedon-Fekjaer H, Lindqvist BH, Vatten LJ, Aalen OO, Tretli S. Breast cancer tumor growth estimated through mammography screening data. Breast Cancer Res 2008;10(3):R41. [26] Kuhl CK, Mielcareck P, Klaschik S, et al. Dynamic breast MR imaging: are signal intensity time course data useful for differential diagnosis of enhancing lesions? Radiology 1999;211(1):101–10.
Contents lists available at ScienceDirect
European Journal of Radiology journal homepage: www.elsevier.com/locate/ejrad
Correlation between enhancement characteristics of MR mammography and capillary density of breast lesions Alexander Poellinger a,∗ , Sahra El-Ghannam a , Susanne Diekmann a , Thomas Fischer a , Glen Kristiansen b , Florian Fritzsche c , Eva Fallenberg a , Lars Morawietz d , Felix Diekmann a a
Charité, Universitätsmedizin Berlin, Department of Radiology, Augustenburger Platz 1, 13353 Berlin, Germany Universitätsklinikum Bonn, Department of Pathology, Sigmund-Freud-Str. 25, D-53127 Bonn, Germany c Institut für Histologie und Zytologie, Bahnhofplatz 11, Postfach, 9101 Herisau, Switzerland d Diagnostik Ernst von Bergmann GmbH, Charlottenstr. 72, 14467 Potsdam, Germany b
a r t i c l e
i n f o
Article history: Received 11 June 2014 Received in revised form 13 September 2014 Accepted 15 September 2014 Keywords: Magnetic resonance imaging Breast neoplasms Pathology Blood supply
a b s t r a c t Objective: To correlate capillary density of breast lesions using the markers D2-40, CD31, and CD34 with early and late enhancement of magnetic resonance mammography (MRM). Materials and methods: The local ethics committee approved this study, and informed consent was available from all patients. The study included 64 women with 66 histologically proven breast lesions (41 malignant, 25 benign). MR-enhancement 1 min after contrast medium administration was determined in the tumor (It1 /It0 ratio) and in comparison to the surrounding tissue (It1 /It1-fat ratio). Capillary density was quantified based on immunohistological staining with D2-40, CD31, and CD34 in breast tumors and surrounding breast tissue. Mean capillary densities were correlated with contrast enhancement in the tumor and surrounding breast tissue. The Kruskal–Wallis test was used to test whether lesions with different MR enhancement patterns differed in terms of capillary density. Results: For CD34, there was statistically significant correlation between capillary density and tumor enhancement (r = 0.329, p = 0.012), however not for the malignant or benign groups separately. Mean vessel number identified by staining with D2-40 and CD31 did not correlate significantly with tumor enhancement (D2-40: r = −0.188, p = 0.130; CD31: r = 0.095, p = 0.448). There were no statistically significant differences in capillary density between breast lesions with delayed enhancement or a plateau and lesions showing washout (Kruskal–Wallis test. D2-40: p = 0.173; CD31: p = 0.647; CD34: p = 0.515). Conclusion: Of the three markers tested, CD34 showed best correlation between early contrast enhancement on MRM and capillary density. Further studies are necessary to clearly demonstrate an association between capillary density and contrast enhancement in breast tumors and surrounding tissue. © 2014 Elsevier Ireland Ltd. All rights reserved.
1. Introduction In recent years, magnetic resonance mammography (MRM) has been increasingly used to supplement conventional mammography and other diagnostic breast imaging modalities [1]. When a suspicious breast lesion is detected by MRM that is not visible by any other imaging modality, it is recommended to obtain an MR-guided transcutaneous biopsy for histological confirmation [2]. To ensure correct lesion sampling it would be desirable to prove
∗ Corresponding author at: Charité, Universitätsmedizin Berlin, Department of Radiology, Charité Campus Virchow Klinikum, Augustenburger Platz 1, 13353 Berlin, Germany. Tel.: +49 30 450 627269; fax: +49 30 450 550914. E-mail address: [email protected] (A. Poellinger). http://dx.doi.org/10.1016/j.ejrad.2014.09.007 0720-048X/© 2014 Elsevier Ireland Ltd. All rights reserved.
that the samples origin indeed from the suspicious region. While specimen mammography or specimen ultrasound should be performed for breast lesions identified by these imaging modalities, no such recommendation exists for breast specimens obtained by MR-guided biopsy. When a biopsy has been obtained from a breast lesion detected by MRM only, it is often not very helpful to perform specimen radiography and it is sometimes difficult to prove that histological findings match the imaging appearance. In a study by Lee et al. [3], histology was discordant with imaging in 7% of lesions that were examined by MR-guided vacuum-assisted biopsy. Among discordant lesions, i.e. lesions that showed suspicious features on MRI but were evaluated as benign by histology following MR-guided biopsy, surgical excision revealed cancer in 30%. Liberman et al. [4] encountered histology-imaging discordance in nine of 98 MR-guided vacuum-assisted breast biopsies (9%).
2130
A. Poellinger et al. / European Journal of Radiology 83 (2014) 2129–2136
Four of the nine discordant lesions were subsequently found to be cancers. Thus, in all patients with discordant imaging findings and histology, repeated percutaneous biopsy or open biopsy should be contemplated. To minimize the rate of repeated biopsies, it is desirable to have a method for reliable correlation of imaging findings and histopathological parameters. A technique for verifying a suspected lesion sampled by MR-guided biopsy could be developed if one demonstrated a direct numerical correlation between contrast enhancement on MRI and histopathological microvessel density (MVD) in breast lesions. Several study groups have used immunohistological staining to determine the capillary density of breast tumors in order to identify possible correlations with signal enhancement on dynamic contrast-enhanced MRM [5–9]. The results are very heterogeneous, one reason being the diversity of histopathological markers used, including markers that are not very common. As a result, further studies are necessary to provide definitive evidence regarding a possible association between tumor angiogenesis and signal enhancement of breast lesions at MRM. In the present study, we investigated three widely used immunohistological markers, D2-40, CD31 and CD34, to quantify vessel density within breast tumors and directly adjacent breast tissue. These results were compared with contrast enhancement of breast tumors and surrounding breast tissue in dynamic contrastenhanced MR mammography. 2. Material and methods The study was approved by the local ethics committee, and all patients gave written informed consent. Data of 64 consecutive patients who underwent MRM with subsequent histological workup of suspicious breast lesions was analyzed. The interval between MRM and histological verification was no longer than 90 days (median 9 days). 2.1. MR mammography protocol All MR mammographies were performed on a 1.5-Tesla MR imager (Siemens Magnetom, SymphonyVision® , Erlangen, Germany) using a dedicated breast double coil (Siemens, NORAS® ). In premenopausal patients, MRM was performed on days 7–15 of the menstrual cycle. The MR protocol included a coronal T1-weighted turbo spin echo (TSE) sequence and an axial T2weighted TSE sequence before contrast medium administration, and a dynamic coronal 3D T1-weighted fast low-angle shot (FLASH) sequence (TR = 8.1 ms, TE = 4.48 ms, Flip angle = 25◦ , matrix: 256 × 256, Field of View: 320 mm, slice thickness: 2 mm) acquired before and at five time points after automated intravenous bolus injection of 0.2 mmol/kg body weight of Gd-DTPA (Magnevist® , Schering, Berlin, Germany), followed by a 20 ml saline flush, both at an injection rate of 2 ml/sec. The dynamic sequence was followed by an axial T1-weighted fat-suppressed spin echo (FS SE) sequence and an axial STIR sequence. 2.2. Data analysis Dynamic breast MR images were analyzed by quantifying signal intensities over time in regions of interest (ROI) set by a diagnostic radiologist with more than 10 years of experience in MRM. The ROI was placed in an area in which signal intensity was highest within the first minute after contrast medium administration. The ROI comprised at least 5 pixels to avoid local fluctuation and thus misinterpretation of contrast enhancement. The FuncTool software package (Advantage Windows Workstation, GE Healthcare, Milwaukee, WI) was used to measure signal intensities in the
ROI. Contrast enhancement was analyzed on dynamic T1-weighted images in the lesion before contrast administration (It0 ) and one minute after administration (It1 ) as well as in surrounding tissue at timepoint t1 (It1-fat ). Relative signal enhancement at timepoint t1 was calculated as the ratio of signal intensity at t1 to the baseline intensity in the tumor at t1 (It1 /It0 ). A second ratio, It1 /It1-fat , was calculated to express signal enhancement in the lesion in relation to the enhancement in surrounding fat. In addition, we documented the timepoint of maximum signal enhancement (peak) on dynamic MR images as well as the course of signal enhancement. “Washout” was defined as at least 10% decrease in signal enhancement, “prolongated enhancement” as 10% or more increase in signal enhancement relative to the intensity at t = t1. 2.3. Histopathology Magnetic resonance-guided biopsies were performed using a vacuum-assisted biopsy system with an 11-gauge needle (Mammotome® Breast Biopsy System, Cincinnati, USA). The tissue sampled from the 66 breast lesions was fixed in formalin and grossly evaluated before paraffin embedding. The paraffinembedded specimens were cut into 2–3 m sections, which were mounted on glass slides. Routine hematoxylin and eosin (H&E) was used for visualization of tissue structures. Following gross and microscopic evaluation of the tissue samples, routine immunohistochemical examinations were supplemented by staining for D2-40, CD31, and CD34 for quantification of capillary densities in tumors. D2-40 is a commercially available antibody for visualization of both normal and reactive or neoplastic lymphatic vessels. The lymphatic vessels are visualized by labeling of endothelium [10,11]. In this study, we used the D2-40 clone from Dako (Dako Deutschland GmbH, Hamburg, Germany) at a 1:100 dilution. The antibody also labels lymphatic channels of lymph nodes, though less strongly [12]. Nonendothelial cells reacting with D2-40 include the myoepithelium of the breast. Blood vessels are always negative in immunohistological specimens incubated with D2-40, both in normal tissue and in reactive tissue such as scar tissue. The D2-40 antibody thus ensures specific visualization of lymphatic vessels. CD31, also known as PECAM-1 (platelet endothelial cell adhesion molecule 1), is a transmembrane glycoprotein with a molecular weight of 130 kDa. It is mainly expressed by endothelial cells but also by some leukocytes [13]. The JC70A antibody clone (Dako Deutschland GmbH, Hamburg, used at a concentration of 1:25) serves to detect endothelial cells in formalin-fixed tissue, and thus has been used for evaluating angiogenesis in proliferating breast tumors. The CD31 antibody selectively labels the endothelial cells of blood vessels but not of lymphatic vessels. CD34 is a membrane-bound glycoprotein with a molecular weight of 116 kDa. CD34 antigen is expressed by hematopoietic stem cells. In stem cell transplantation, the CD34 antigen can be used to enrich hematopoietic stem cells from peripheral blood; this is possible as blood stem cells can be identified by means of the CD34 antigen and separated from other blood cells [13]. The CD34 antibody is especially useful for labeling the endothelial cells of blood vessels [9,11,14]. However, unlike the CD31 antibody, CD34 also reacts with lymphatic endothelia cells to a small degree [13]. In the present study we used the QBEnd1 antibody clone (Dako Deutschland GmbH, Hamburg) at 1:100 dilution. The specimens were stained automatically using the Ventana BenchMark® XT staining system (Ventana, Tucson, AZ, USA). For correlation of immunohistological capillary densities with MR contrast enhancement, five high-power fields in the tumor and five high-power fields in surrounding tissue stained for D2-40, CD31, and CD34 were counted per lesion to calculate arithmetic means.
A. Poellinger et al. / European Journal of Radiology 83 (2014) 2129–2136 Table 1 Distribution of benign breast lesions. Benign breast lesions Mastitis Mastopathy Common ductal hyperplasia Lobular hyperplasia Fibrosis/sclerosis Fibroadenoma Papilloma Adenosis
Table 3 Histological composition of the tissue surrounding breast lesions. Number (percentage) 5 (20%) 2 (8%) 2 (8%) 1 (4%) 8 (32%) 3 (12%) 2 (8%) 2 (8%) 25 (100%)
Histology
Number (percentage)
Fat Fat and glandular tissue Fat and connective tissue Glandular tissue Fibrotic glandular tissue Fibrosis Fat and isolated aggregated glands Fat and lymphatic follicles
40 (60.6%) 12 (18.3%) 3 (4.5%) 3 (4.5%) 3 (4.5%) 3 (4.5%) 1 (1.5%) 1 (1.5%)
66 (100%)
of the surrounding tissue of all 66 breast lesions is summarized in Table 3.
Table 2 Distribution of malignant breast lesions. Malignant breast lesions
Number (percentage)
DCIS LCIS IDC ILC
5 (12.2%) 1 (2.4%) 23 (56.1%) 12 (29.3%)
2131
41 (100%)
The diameter of a high-power field was 0.5 mm. Counts were performed at 400× magnification and five representative areas with high capillary densities in the tumor and in five areas of surrounding tissue for each lesion were evaluated. Histological capillary densities were correlated with contrast enhancement in the tumor (It1 /It0 ratio) and the surrounding tissue (It1 /It1-fat ratio). 2.4. Statistical analysis Statistical analysis was performed using SPSS, version 14.0 (Chicago, IL, USA). Correlations between imaging and histopathological capillary density were assessed by calculating two-tailed Spearman’s rho. The zero hypothesis (H0 ) postulates that no correlation exists between immunohistochemical vessel counts in breast tumors and surrounding tissue and MR contrast enhancement. The alternative hypothesis (H1 ) states that there is a significant correlation between imaging and histopathology. Statistical significance was assumed at a p-value of 0.05. 3. Results 3.1. Pathology Benign lesions. There were 24 patients with a total of 25 benign lesions. Mean patients’ age was 49.4 ± 11.53 years (ranged between 25 and 74 years). Fibroadenoma was the most common benign breast tumor. The distribution of benign breast lesions is summarized in Table 1. Malignant lesions. The 40 patients with a total of 41 malignant lesions ranged in age between 29 and 79 years (mean age, 53.8 ± 11.54 years). Twenty-two patients with a total of 23 malignant lesions had invasive ductal carcinoma (IDC), accounting for 56.1% of all malignant tumors. There were 12 invasive lobular carcinomas (ILC). The high percentage of ILCs is due to the criteria for performing MRM at the breast center of our hospital (all patients with ILC and with extremely dense breast parenchyma get a recommendation for MRM). Table 2 lists the distribution of malignant breast tumors in the study population. Tissue surrounding breast lesions. In 26 of all 66 cases, the perilesional tissue consisted not only of fat but also other tissue such as glandular and connective tissue. The histological composition
3.2. Contrast enhancement at time t1 and course of signal intensities For benign lesions the average relative signal increase (It1 /It0 ) in breast lesions was 2.53 ± 0.97 versus 0.92 ± 0.27 in perilesional fat. In malignant lesions this ratio was 3.14 ± 1.25 with a ratio of 1.07 ± 0.44 in perilesional fatty tissue. 3.3. Immunohistochemical vessel counts Mean values of capillary density in lesions and surrounding tissue were calculated from the mean values of five fields of view of the individual lesions for all breast lesions and for each of the three markers used (Table 4). For D2-40, the average number of vessels was less than one per field of view in malignant and benign lesions as well as in the surrounding tissue, for benign and malignant tumors. With the other two markers, CD31 and CD34, the mean vessel count was markedly higher in both malignant (number of vessels more than five) and benign breast lesions and surrounding tissue (Table 4). Fig. 1 shows a tissue sample of an invasive ductal carcinoma and the surrounding tissue with different histopathological markers. Photomicrographs of lesion samples from another patient with mastitis and corresponding dynamic MR-images are illustrated in Fig. 2. 3.4. Correlations 3.4.1. Tumor enhancement and capillary density For benign and malignant lesions taken together (n = 66), twotailed Spearman’s rho showed a significant correlation between tumor enhancement (It1 /It0 ratio) at the time of the first postcontrast measurement and the tumor vessel count using CD34 (r = 0.259, p = 0.035) (Table 5). The scatter diagram in Fig. 3 illustrates the statistical association between the CD34 vessel counts and tumor enhancement (t1/t0 ratio). There is marked residual scatter with individual residues very widely deviating from the regression line. There was no significant association between vessel counts after staining with D2-40 and CD31 and tumor signal intensities (Fig. 3). Analysis for correlations between tumor enhancement (It1 /It0 ratio) and mean immunohistochmemical vessel counts using twotailed Spearman’s rho yielded no statistically significant correlation for the 25 benign lesions alone, neither for D2-40 nor for CD31 or CD34. The same was found for the 41 malignant lesions in our study if tested alone. 3.4.2. Perilesional enhancement and capillary density Correlations between enhancement of surrounding tissue and capillary density were only calculated for a subset of specimens where the surrounding tissue was exclusively composed of fat, which was the case for 40 lesions – 16 benign and 24 malignant.
2132
A. Poellinger et al. / European Journal of Radiology 83 (2014) 2129–2136
Table 4 Immunohistochemical staining and average microvessel densities (MVD) with standard deviations. MVD was measured as microvessels per surface area (0.25 mm2 ) and is expressed here as microvessels/mm2 . Marker D2-40 CD31 CD34
Tissue
All lesions (n = 66)
Tumor Surrounding Tumor Surrounding Tumor Surrounding
3.36 1.72 31.52 12.08 29.88 13.76
± ± ± ± ± ±
Benign lesions
6.16 2.76 25.96 8.44 15.20 10.40
3.24 0.88 22.88 10.76 25.92 12.8
± ± ± ± ± ±
3.92 1.24 9.88 7.52 12.04 9.28
Malignant lesions 3.44 2.24 36.80 12.92 32.32 14.32
± ± ± ± ± ±
7.24 3.32 31.00 8.96 16.56 11.12
Table 5 Correlation of D2-40, CD31, and CD34 expression with contrast enhancement at dynamic MRM for benign tumors (a), malignant tumors (b), all tumors (c), and all lesions detected by diagnostic MRM (d); two-tailed Spearman’s rho. It1 /It0 ratio
N
Benign
25
Malignant
41
All lesions
66
*
Correlation coefficient Significance Correlation coefficient Significance Correlation coefficient Significance
D2-40-tumor
CD31-tumor
CD34-tumor
r = −0.127 p = 0.545 r = −0.185 p = 0.248 r = −0.188 p = 0.130
r = −0.018 p = 0.932 r = 0.001 p = 0.996 r = 0.095 p = 0.448
r = 0.146 p = 0.485 r = 0.236 p = 0.137 r = 0.259* p = 0.035
Correlation is assumed to be significant at p = 0.05.
For quantitative comparison, ratios of the numbers of vessels in lesions to surrounding tissue were calculated for all three stains investigated (D2-40-tumor to D2-40-normal ratio, CD31-tumor to CD31-normal ratio, and CD34-tumor to CD34-normal ratio). Testing using two-tailed Spearman’s rho yielded no significant correlation between vascular density in surrounding tissue and tumor enhancement (It1 /It1-fat ratio) at the first measurement after contrast medium administration. For all immunohistochemical markers used, the Kruskal–Wallis test revealed no statistically significant differences in capillary density between breast lesions with delayed enhancement or a plateau and lesions showing washout (D2-40: p = 0.173; CD31: p = 0.647; CD34: p = 0.515). 4. Discussion Three immunohistological stains, CD34, CD31, and D2-40, were used to determine capillary density in breast lesions and surrounding tissue. The average numbers of vessels in tumors and surrounding tissue found in our study are in line with formerly published results obtained using the same immunohistochemical stains for detecting tumor neoangiogenesis [7,9,15,16]. Teifke at al. found the microvessel density (MVD) of malignant lesions (stained with CD34) to be between 19 and 50 microvessels/mm2 , which is in the range of the MVD in our study (32–37 microvessels/mm2 ). The rather lower MVD found in our study might be attributable to the relative high frequency of lobular cancers in our study that show variable degree of neoangiogenesis. Slight differences may be due to the use of different methods for determining vascular density. For instance, Buadu et al. [15] divided breast lesions into four groups based on vascular density. A similar approach was used by Su et al. [9], who distinguished three groups of tumors based on microvessel density (MVD) – tumors with high, intermediate, and low MVD. Müller-Schimpfle et al. [7] determined capillary density in the tumor center, in the tumor margin, and in the surrounding tissue; in contrast to our approach, they counted three high-power fields at 200× magnification (in our study: five high-power fields at 400× magnification). Because of the different methods used for quantifying neoangiogenesis, our results cannot easily be compared with the reported results. Correlation between MR-imaging and histopathology yielded a statistically significant association between vessel counts using
CD34 and tumor enhancement for all 66 breast lesions. Several investigators tried to identify possible effects of tumor capillary density on tumor enhancement on Gd-DTPA-enhanced MRM [6–9,16–20]. Teifke et al. [16] also observed a significant association between tumor capillary density determined with CD34 and early enhancement of the tumor center for both benign and malignant breast lesions. The correlation coefficient was higher for breast cancer compared with benign lesions. In our study, separate analysis of benign and malignant lesions failed to reveal a significant association between tumor enhancement and capillary density. Our results are in agreement with the study of Buadu et al. [15]. They also found the intensity of lesion enhancement one minute after contrast administration to correlate with the number of tumor vessels stained with CD34. In addition, they showed tumor capillary density as well as contrast enhancement to be heterogeneously distributed in tumorous lesions. Supporting the experience of many investigators [15,21–23] that the degree of contrast enhancement alone is insufficient to differentiate benign from malignant breast lesions, they stated that breast lesion characterization should also include the course of enhancement and the MR morphology of focal lesions. We found no statistically significant correlation between tumor capillary density after staining with CD31 and tumor enhancement. Knopp et al. and Müller-Schimpfle et al. [6,7] also used CD31 for immunohistological staining to investigate the effect of capillary density of breast tumors on enhancement characteristics in dynamic MRM. In contrast to our results, Knopp et al. observed a significantly faster signal increase in malignant breast lesions compared to benign lesions, but no significant association between capillary density and tumor signal intensity one minute after contrast medium administration. Müller-Schimpfle et al., on the other hand, found a slightly significant correlation between peripheral enhancement of breast lesions and capillary density in the lesion margin, while the association between enhancement of the tumor center and capillary density did not reach significance. Su et al. [9], using CD31 staining in a total of 71 breast cancers, observed no significant correlation in the tumor periphery or center, which is in agreement with our findings. It should be noted that a direct comparison of these results is not possible as different image postprocessing techniques were used by Su et al. and in the present study. Although no definite answer can be given, it may be assumed that some of the discrepancy between the study results is attributable to the use of different methods for determining vessel
A. Poellinger et al. / European Journal of Radiology 83 (2014) 2129–2136
2133
Fig. 1. Photomicrographs (original magnification, ×200) show invasive ductal carcinoma (left column) and surrounding tissue (right column) with different immunohistochemical markers: H&E staining showing atypical glandular tumor infiltrates with numerous mitotic figures and a lymphocytic inflammatory infiltrate (a) and connective tissue with low cellularity in the surrounding tissue (b). Immunohistochemical staining with D2-40 displaying some lymph vessels in the tumor (c) and the surrounding tissue (d). CD31 staining of the tumor with multiple microvessels (e) and only few vessels in the surrounding tissue (f). CD34 staining of the ductal carcinoma shows several microvessels (the dark blue structure in the middle of the image is an artifact) (g). The surrounding tissue also exhibits some microvessels (h).
density and MR signal intensities as well as to the use of different statistical tests. In our study, staining with the endothelial marker D2-40 for identifying and counting lymphatic vessels revealed no significant
correlation with MR contrast enhancement. To date, published studies investigating lymphatic capillary density in breast lesions have not correlated this parameter with MR contrast enhancement. Rather this parameter was used to investigate the possible
2134
A. Poellinger et al. / European Journal of Radiology 83 (2014) 2129–2136
Fig. 2. Photomicrographs (A) of lesion samples of the left breast of a 25-years old woman and dynamic MRI images (B). (A) Photomicrographs (original magnification, 200×) show mastitis (left column) and surrounding tissue (right column) with different immunohistochemical markers: (a) H&E staining exhibits a cell-rich inflammatory response. Microscopically the milk ducts can be hardly discerned due to massive inflammation and cell proliferation. (b) Connective tissue surrounding the lesion. (c) Immunohistochemical staining with D2-40 displays a lymph vessel in the inflamed area (upper right corner) and (d) a rather large lymph vessel the surrounding tissue. (e) CD31 staining of the mastitis area with multiple microvessels and (f) only few vessels in the surrounding tissue. (g) CD34 staining of the sample shows several microvessels. (h) The surrounding tissue also exhibits some microvessels. (B): Corresponding dynamic MRI: from top to bottom: non-contrast enhanced T1-weighted coronal scan followed by scans at one, three and five minutes after contrast administration. The bottom image displays subtraction between the images at 3 min and the non-contrast enhanced scan.
diagnostic relevance of lymphatic vessel density in breast tumors in terms of metastatic spread, recurrence-free survival, and overall survival [11,12,24]. El-Gohary et al. and Mohammed et al. [11,12] found a significant association between intratumoral lymphatic vessel density and the presence of distant metastasis, which in turn correlated with a shorter overall survival. In our study, we identified lymphatic vessels using D2-40 in 19 of 41 breast cancers (46%) and in 17 of 25 benign lesions (68%). As expected, lymphatic capillary density was much smaller than the number of intratumoral blood capillaries. This finding is in agreement with the results of El-Gohary and Mohammed. We identified on average less than one lymphatic vessel per tumor, for both benign and malignant tumors. In addition, we also investigated capillary density in fat tissue surrounding breast lesions (but clearly outside the lesion) and correlated this histopathological parameter with early contrast
enhancement of peritumoral fat. These investigations revealed no statistically significant correlations for any of the three stains used. Only one of the research groups investigating capillary density and contrast enhancement also investigated capillary density in the tissue surrounding breast lesions [7]. In contrast to our results, they found a statistically significant correlation between capillary density around breast lesions and MR signal intensity. Note that, when interpreting these results, our analysis only included specimens (n = 40) in which the surrounding tissue was exclusively composed of fat. This approach was necessary as the relative contrast enhancement was based on the surrounding fat. Therefore, 26 lesions with heterogeneous surrounding tissue could not be included in this analysis. A comparison between our results and those reported is difficult because of the use of different histopathological methods and the use of different formulas for calculating relative signal enhancement.
A. Poellinger et al. / European Journal of Radiology 83 (2014) 2129–2136
2135
4.2. Conclusion Of the immunohistochemical stains used in the present study, CD34 appears to be most suitable to identify a possible correlation between capillary density and MR contrast enhancement of breast lesions. To obtain more reliable results, further studies using a standardized methodology are necessary to clearly demonstrate an association between capillary density and contrast enhancement in breast tumors and surrounding tissue. Once such data are available, a strong correlation between capillary density and enhancement might be interpreted to confirm lesion sampling. Conversely, a weak correlation, along with discordant imaging and histopathology findings, might be one more criterion in support of repeating the biopsy. Nevertheless, an interdisciplinary tumor conference continues to be indispensible for a comprehensive evaluation and correlation of imaging findings and histological results and timely identification of false-negative findings. Funding Fig. 3. Statistical association between the CD34 vessel counts and tumor enhancement (t1/t0 ratio). For benign and malignant lesions taken together (n = 66), two-tailed Spearman’s rho showed a significant correlation between tumor enhancement (It1 /It0 ratio) at the time of the first postcontrast measurement and the tumor vessel count using CD34 (r = 0.259, p = 0.035).
4.1. Limitations of the study Our study has some methodological limitations: 1. The fact that only lesions suspected for malignancy were submitted to histopathological verification results in a selection bias. This bias mainly affects the signal intensity time course in benign breast lesions and explains the high proportion of intensity-time curves with washout or a plateau phase we observed in benign lesions. This bias has no direct effect on the correlation between capillary density and contrast enhancement in tumors and surrounding tissue. Hence, the effect is negligible for the purpose of the present study. 2. The maximum delay between the detection of a suspicious breast lesion by MRM and histological verification was 90 days. This is critical in view of the average growth rate of breast cancers of 0.6 mm per month, depending on the patient’s age and the initial tumor size [25]. However, the median delay between MRM and histological verification was 9 days. Only for 5 of the 66 lesions the delay was longer than 30 days. Therefore, the risk of a possible statistical distortion resulting from differences in tumor size between the time of mammography and the time of histology again appears to be small. 3. Another limitation of our study is that the signal intensities in tumors and surrounding tissue were only assessed visually in subjectively placed ROIs, which limits reproducibility. However, experienced researchers as Kuhl et al. [26] consider visual assessment of ROIs and signal intensity-time curves to be sufficiently reproducible. Mammograms and pathologic specimens were assessed by only one radiologist and one pathologist. However, both readers were blinded to the results of the other. 4. Our study is also limited by the fact that only 40 of the 66 cases were included in the analysis of capillary density and contrast enhancement of surrounding fatty tissue as the remaining 26 cases did not meet the criteria (lesion surrounded by fat). 5. Finally, the composition of the benign lesion group was very heterogeneous consisting e.g. of five cases of mastitis and two cases of mastopathy. On the other hand there was a high percentage of invasive lobular carcinomas in the malignant group that exhibited a lesser degree of MVD. This heterogeneity might have weakened the correlation between MVD and signal intensity.
No funding. References [1] Akita A, Tanimoto A, Jinno H, Kameyama K, Kuribayashi S. The clinical value of bilateral breast MR imaging: is it worth performing on patients showing suspicious microcalcifications on mammography? Eur Radiol 2009;19(9):2089–96. [2] Peters NH, Meeuwis C, Bakker CJ, et al. Feasibility of MRI-guided largecore-needle biopsy of suspicious breast lesions at 3 T. Eur Radiol 2009;19(7):1639–44. [3] Lee JM, Kaplan JB, Murray MP, et al. Imaging histologic discordance at MRI-guided 9-gauge vacuum-assisted breast biopsy. AJR Am J Roentgenol 2007;189(4):852–9. [4] Liberman L, Bracero N, Morris E, Thornton C, Dershaw DD. MRI-guided 9-gauge vacuum-assisted breast biopsy: initial clinical experience. AJR Am J Roentgenol 2005;185(1):183–93. [5] Ikeda O, Nishimura R, Miyayama H, et al. Evaluation of tumor angiogenesis using dynamic enhanced magnetic resonance imaging: comparison of plasma vascular endothelial growth factor, hemodynamic, and pharmacokinetic parameters. Acta Radiol 2004;45(4):446–52. [6] Knopp MV, Weiss E, Sinn HP, et al. Pathophysiologic basis of contrast enhancement in breast tumors. J Magn Reson Imaging 1999;10(3):260–6. [7] Muller-Schimpfle M, Noack F, Oettling G, et al. Influence of histopathological factors on dynamic MR mammography. Rofo 2000;172(11):894–900. [8] Stomper PC, Winston JS, Herman S, Klippenstein DL, Arredondo MA, Blumenson LE. Angiogenesis and dynamic MR imaging gadolinium enhancement of malignant and benign breast lesions. Breast Cancer Res Treat 1997;45(1):39–46. [9] Su MY, Cheung YC, Fruehauf JP, et al. Correlation of dynamic contrast enhancement MRI parameters with microvessel density and VEGF for assessment of angiogenesis in breast cancer. J Magn Reson Imaging 2003;18(4):467–77. [10] Marinho VF, Metze K, Sanches FS, Rocha GF, Gobbi H. Lymph vascular invasion in invasive mammary carcinomas identified by the endothelial lymphatic marker D2-40 is associated with other indicators of poor prognosis. BMC Cancer 2008;8:64. [11] Mohammed RA, Ellis IO, Elsheikh S, Paish EC, Martin SG. Lymphatic and angiogenic characteristics in breast cancer: morphometric analysis and prognostic implications. Breast Cancer Res Treat 2009;113(2):261–73. [12] El-Gohary YM, Metwally G, Saad RS, Robinson MJ, Mesko T, Poppiti RJ. Prognostic significance of intratumoral and peritumoral lymphatic density and blood vessel density in invasive breast carcinomas. Am J Clin Pathol 2008;129(4):578–86. [13] Pusztaszeri MP, Seelentag W, Bosman FT. Immunohistochemical expression of endothelial markers CD31, CD34, von Willebrand factor, and Fli-1 in normal human tissues. J Histochem Cytochem 2006;54(4):385–95. [14] Mohammed RA, Green A, El-Shikh S, Paish EC, Ellis IO, Martin SG. Prognostic significance of vascular endothelial cell growth factors -A, -C and -D in breast cancer and their relationship with angio- and lymphangiogenesis. Br J Cancer 2007;96(7):1092–100. [15] Buadu LD, Murakami J, Murayama S, et al. Breast lesions: correlation of contrast medium enhancement patterns on MR images with histopathologic findings and tumor angiogenesis. Radiology 1996;200(3):639–49. [16] Teifke A, Behr O, Schmidt M, et al. Dynamic MR imaging of breast lesions: correlation with microvessel distribution pattern and histologic characteristics of prognosis. Radiology 2006;239(2):351–60. [17] Buadu LD, Murakami J, Murayama S, et al. Patterns of peripheral enhancement in breast masses: correlation of findings on contrast medium enhanced MRI with histologic features and tumor angiogenesis. J Comput Assist Tomogr 1997;21(3):421–30.
2136
A. Poellinger et al. / European Journal of Radiology 83 (2014) 2129–2136
[18] Ikeda O, Nishimura R, Miyayama H, et al. Magnetic resonance evaluation of the presence of an extensive intraductal component in breast cancer. Acta Radiol 2004;45(7):721–5. [19] Oshida K, Nagashima T, Ueda T, et al. Pharmacokinetic analysis of ductal carcinoma in situ of the breast using dynamic MR mammography. Eur Radiol 2005;15(7):1353–60. [20] Van den Eynden GG, Van der Auwera I, Van Laere SJ, et al. Comparison of molecular determinants of angiogenesis and lymphangiogenesis in lymph node metastases and in primary tumours of patients with breast cancer. J Pathol 2007;213(1):56–64. [21] Bick U. Typical and unusual findings in MR mammography. Rofo 2000;172(5): 415–28. [22] Siegmann KC, Muller-Schimpfle M, Schick F, et al. MR imaging-detected breast lesions: histopathologic correlation of lesion characteristics and signal intensity data. AJR Am J Roentgenol 2002;178(6):1403–9.
[23] Wiener JI, Schilling KJ, Adami C, Obuchowski NA. Assessment of suspected breast cancer by MRI: a prospective clinical trial using a combined kinetic and morphologic analysis. AJR Am J Roentgenol 2005;184(3):878–86. [24] Van der Auwera I, Van Laere SJ, Van den Eynden GG, et al. Increased angiogenesis and lymphangiogenesis in inflammatory versus noninflammatory breast cancer by real-time reverse transcriptase-PCR gene expression quantification. Clin Cancer Res 2004;10(23):7965–71. [25] Weedon-Fekjaer H, Lindqvist BH, Vatten LJ, Aalen OO, Tretli S. Breast cancer tumor growth estimated through mammography screening data. Breast Cancer Res 2008;10(3):R41. [26] Kuhl CK, Mielcareck P, Klaschik S, et al. Dynamic breast MR imaging: are signal intensity time course data useful for differential diagnosis of enhancing lesions? Radiology 1999;211(1):101–10.
