| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Department of Diagnostic Radiology, University Hospital Tübingen,
Hoppe-Seyler-Str. 3, 72076 Tübingen, Germany.
2 Department of Obstetrics and Gynecology, University Hospital Tübingen,
Schleichstr. 4, 72076 Tübingen, Germany.
3 Institute of Pathology, University Hospital Tübingen, Liebermeisterstr.
8, 72076 Tübingen, Germany.
Received July 30, 2001;
accepted after revision December 27, 2001.
Address correspondence to K. C. Siegmann.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. We performed 51 MR imaging—guided breast interventions (41 preoperative lesion localizations and 10 large-core needle biopsies) in 45 patients with exclusively MR imaging—detected lesions. All patients had previously undergone diagnostic dynamic contrast-enhanced MR imaging of the breast with a double breast coil at 1.0 T (n = 36) or 1.5 T (n = 15). The diagnostic MR images were evaluated on a workstation. Lesion morphology (size, shape, margin type, enhancement pattern), signal intensity parameters (time to peak enhancement, maximum slope of enhancement curve, washout, relative water content), and scores analogous to the Breast Imaging Reporting and Data System (BI-RADS) categories were correlated with histology.
RESULTS. Histology revealed malignancy in 37.3% (19/51) of the lesions. The positive predictive value for malignancy of exclusively MR imaging—detectable lesions increased as the analogous BI-RADS category increased. Late inhomogeneous contrast enhancement was the only morphologic criterion that was statistically significantly correlated with malignancy. Malignant and benign lesions did not differ significantly in any of the quantitatively evaluated signal intensity parameters. Carcinomas showed a tendency toward faster and stronger enhancement and stronger washout.
CONCLUSION. The classification of exclusively MR imaging—detectable breast lesions according to a combination of morphologic and perfusion parameters including the late enhancement pattern helps identify the lesions for which interventional MR imaging is required. Quantitative signal intensity data alone do not suffice.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
If a suspicious lesion is detectable exclusively by MR imaging and is not visible on mammography or sonography—even retrospectively, then the most exact way to obtain a histologic diagnosis is through an MR imaging—guided intervention, which can be performed using different techniques [16,17,18,19,20,21,22,23]. Because of its limited specificity for detection of malignant breast lesions, MR imaging reveals a considerable number of enhancing lesions that are benign. These MR findings result in a number of unnecessary biopsies. To decrease the number of unnecessary biopsies, we attempted to find a more precise way to differentiate between false-positive enhancing lesions and true-positive malignancies. Therefore, the purpose of this study was to correlate qualitative and quantitative MR imaging characteristics of exclusively MR imaging—detectable lesions with histology.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
In all patients, high-quality two-view mammography and 7.5-MHz sonography of both breasts were performed before MR imaging. Diagnostic MR imaging of the breast was performed 2-77 days (mean, 19.8 days) before the intervention. Lesions were imaged during either the second or the third week of the patient's menstrual cycle in those who were premenopausal. Indications for the diagnostic MR imaging were equivocal findings on mammography and sonography—that is, cancer could not be ruled out because of dense parenchyma and heterogeneous echogenicity with dorsal shadowing in patients either with clinical symptoms (skin retraction or palpable lump) or with a family history of breast cancer (20/51 [39.2%]); preoperative exclusion of multifocal disease or contralateral breast cancer in patients with known lesions (14/51 [27.5%]); follow-up after breast-conserving therapy (12/51 [23.5%]); and follow-up of indeterminate lesions detected by previous MR imaging (5/51 [9.8%]) [26].
Imaging
Diagnostic MR imaging was performed at 1.0 T (n = 36) or 1.5 T
(n = 15) (Magnetom Expert or Vision; Siemens Medical Systems,
Erlangen, Germany) using a double breast coil with the patient in a prone
position. Breast motion was prevented by cushioning material. The protocol at
1.0 T included a T1-weighted dynamic three-dimensional fast low-angle shot
sequence (3D FLASH) in the coronal plane with seven dynamic studies
(Table 2). At 1.5 T the 3D
FLASH sequence in the coronal plane was repeated eight times
(Table 2). The temporal
resolution was 60 and 85 sec per dynamic study at 1.5 and 1.0 T, respectively.
In all patients, a bolus of contrast medium (gadopentetate dimeglumine) was IV
administered at a dose of 0.16 mmol/kg body weight, followed by 20 mL saline
solution (0.9%). Additionally, a fat-suppressed inversion recovery sequence in
the axial plane (parameters at 1.0 T: TR/TE, 6200/60; inversion time, 150
msec; flip angle, 120°; parameters at 1.5 T: 5600/60; inversion time, 150
msec; flip angle, 180°) was acquired in all patients.
|
Interventional MR imaging (Magnetom Open; Siemens Medical Systems) of one breast was performed with a surface coil and the patient in a semiprone position at 1.0 T (n = 45) or 0.2 T (n = 6). For data acquisition at 1.0 T, a dynamic contrast-enhanced (gadopentetate dimeglumine; dose, 0.16 mmol/kg body weight) T1-weighted 3D FLASH sequence in the sagittal plane (13/6; flip angle, 50°) was performed. To localize lesions, we used additional gradient-echo sequences in the axial or coronal plane. At 0.2 T, a dynamic contrast-enhanced T1-weighted three-dimensional fast imaging with steady-state precession sequence (30/10.4; flip angle, 40°) in the coronal orientation and additional sequences in the sagittal orientation were obtained.
Interventional Procedure
We performed 51 breast interventions: 41 were preoperative localizations
and 10 were large-core needle biopsies. Lesion localization was performed
using an MR imaging—compatible metal coil (Cook, Moenchengladbach,
Germany) with (n = 16) or without (n = 4) the injection of a
charcoal—gadopentetate dimeglumine suspension; a 20-gauge hookwire
(Somatex, Berlin, Germany) was used in 20 patients. In one patient, a mark on
the skin was made with a pen because of the superficial location of the
lesion. A self-designed, perforated plate with integrated markers
(Fig. 1) ensured sufficient
needle guidance and breast fixation
[16]. Core biopsies
(n = 10) were performed outside the magnet after exact placement of
an MR imaging—compatible 13-gauge coaxial needle (Somatex) using an MR
imaging—incompatible 14-gauge high-speed core biopsy needle (Bard,
Karlsruhe, Germany). Breast fixation and coaxial needle guidance were achieved
using a commercially available perforated plate that can be adjusted in all
directions.
|
Qualitative Lesion Characteristics
All lesions were prospectively analyzed regarding their qualitative
characteristics. Lesions were divided into two groups: small (
10 mm) and
large (>10 mm). Additionally, the following lesion characteristics were
investigated separately on the first and last contrast-enhanced series: lesion
shape (characterized as regular [oval, round, or polygonal] or as irregular
[linear, branching, or stellate]), margin type (ill-defined or well-defined),
and homogeneity of contrast medium enhancement (homogeneous or heterogeneous).
Finally, the positive predictive values for malignancy with regard to lesion
size, lesion shape, margin type, and homogeneity of contrast medium
enhancement were determined.
Quantitative Lesion Characteristics
The diagnostic dynamic contrast-enhanced breast MR images were evaluated
prospectively on a workstation using a software application (MT-DYNA; MeVis
Technology, Bremen, Germany). A region of interest focusing on the area of
strongest early contrast enhancement within the lesion was chosen to measure
the signal intensity values in T1-weighted unenhanced (SI1) and
contrast-enhanced (SI2-SI7 or
SI2-SI8) series. With this information, we analyzed the
following MR imaging parameters. First, we recorded the time in minutes from
the administration of the contrast medium to the maximum signal intensity,
which we refer to as the time to peak enhancement (Tp). Second, we
determined the maximum slope of the enhancement curve (Smax). This
value was calculated as the relative lesion enhancement (related to maximum
lesion enhancement) per minute. In all cases, the maximum slope was reached
either between the unenhanced and first contrast-enhanced series or between
the first and second contrast-enhanced series. The highest value (either
SI2-SI1 or SI3-SI2) was chosen for
the calculation of the maximum slope using the following equation:
 |
Third, we analyzed the percentage of washout—the change of the
relative lesion enhancement (related to the maximum lesion enhancement) from
the maximum initial enhancement to the last contrast-enhanced
series—using the following equation:
 |
Within the first set of brackets, the lesion enhancement of the last study (SIlast - SI1) is calculated as a percentage of the maximum enhancement (SImax - SI1). The second set of brackets shows the initial enhancement, defined as the maximum enhancement within the first 3 min after contrast medium injection, calculated as the percentage of the maximum lesion enhancement. Subtracting the initial enhancement from the enhancement of the last study yields the percentage of washout. If the enhancement decreases during the dynamic study, then the resulting value is negative.
The enhancement parameters have been calculated as relative lesion enhancement related to the maximum lesion enhancement. This method is the most exact way of comparing lesion data acquired at different field strengths (1.0 and 1.5 T). According to the acquisition time, the time resolution was restricted to 60 and 85 sec at 1.5 and 1.0 T, respectively. The measured signal intensity data were assigned to the center of each dynamic series because these Fourier data include the most important information for the signal intensity of the data set.
In addition, a T2-weighted inversion recovery sequence was used to measure the relative water content of each lesion. Values were calculated by dividing the lesion signal intensity by the signal intensity of the major pectoral muscle. If the lesion had the same signal intensity as the surrounding tissue, then it could not be defined within the parenchyma of the breast and, therefore, a reference measurement was performed at the supposed localization.
Lesion Categorization
In accordance with the BI-RADS system
[24], which was developed by
the American College of Radiology to categorize mammographically detected
findings, we divided all MR imaging—detected lesions into different
groups prospectively. Depending on various parameters (shape, margin type,
enhancement pattern, and kinetics), each lesion was assigned a score that is
analogous to one of the following BI-RADS categories: probably benign finding
(category 3), suspicious abnormality (category 4), and highly suggestive of
malignancy (category 5). To establish an objective and reproducible scoring
system, we modified the system described by Fischer et al.
[10,
25] as shown in
Table 1. This modified scale
can be used to categorize three morphologic and two enhancement dynamics; the
total score can range from 0 to 8 points. Lesion scores of 3, 4, and 5-8
points are analogous to BI-RADS category 3, 4, and 5, respectively
[26]. The positive predictive
value for malignancy depending on the analogous BI-RADS category was
evaluated.
Histopathology
All specimens were analyzed using a thin (5 µm) slice and H and E
staining by an experienced breast pathologist who was unaware of the MR
imaging findings. The lesions were classified according to the classifications
published by Tavassoli [27].
Malignancies were classified following the TNM classification system
established by the Union Internationale Contre le Cancer
[28].
If the presence of invasiveness was questionable, an immunohistochemical analysis with antibodies against actin was performed. Invasive carcinoma was diagnosed if actin-positive myoepithelial cells were missing. In cases of intraductal epithelial proliferations, the presence of cytokeratin 5/6 expression was used to exclude atypical ductal hyperplasia and ductal carcinoma in situ. The tissue was further tested for smooth muscle actin by antibody tests.
Statistics
MR imaging signal intensity data (time to peak enhancement, maximum slope
of enhancement curve, washout, relative water content) were tested for a
standard distribution by the Kolmogorov-Smirnov test. After validating the
gaussian distribution, we evaluated the benign and malignant lesions for
significant differences in enhancement kinetics and relative water content
using the Student's t test for independent samples.
The chi-square test for independent variables was used to assess the lesion qualities (size, shape, margin type, homogeneity of contrast enhancement) and the analogous BI-RADS category to determine whether any of the lesion qualities showed a significant correlation with the histologic result.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
In cases of large-core needle biopsies, a precise sampling was made possible by imaging control of the coaxial needle placement. An example of a lesion before and after MR imaging—guided biopsy is shown in Figure 2A,2B,2C,2D. In the majority of the lesion localizations (27/41 [65.9%]), exact positioning—placement of the marking material immediately adjacent to the lesion—could be achieved (Fig. 3A,3B,3C,3D). In 31.7% (13/41) of the cases, the coil or wire tip was within 10 mm from the lesion. In one intervention (2.4%), the marking material was placed more than 10 mm from the target.
|
|
|
|
|
|
|
|
Histology
Histology revealed malignancy in 35.3% (18/51) of the cases. The most
frequent malignant findings were invasive ductal carcinoma (9/18), followed by
invasive lobular carcinoma (6/18), ductal carcinoma in situ (2/18), and mixed
lobular and ductal invasive carcinoma (1/18). In most of the cases, benign
findings were proliferative changes (75.8% [25/33]) (i.e., lobular carcinoma
in situ, atypical or non-atypical ductal or lobular hyperplasia, fibroadenoma,
papilloma, and sclerosing adenosis) or inflammatory changes (12.1%
[4/33]).
Qualitative Lesion Characteristics
As shown in Table 3, the
overall positive predictive value of exclusively MR imaging—detectable
lesions was 35.3%. The positive predictive values of small (10 mm) and of
large (>10 mm) lesions were 27.6% and 45.5%, respectively. Lesions with
well-defined margins had a higher positive predictive value for malignancy
(44%) than those with ill-defined margins (26.9%). Round, oval, or polygonal
lesions were malignant in 31.3% of the cases, and irregularly shaped lesions
were malignant in 37.1%. Lesions that showed homogeneous late enhancement were
malignant in 20% of the cases, whereas half of the lesions (50%) with
heterogeneous late enhancement were malignant. As opposed to these findings,
early homogeneous lesion enhancement was more frequently associated with
malignancy (42.9%) than heterogeneous early enhancement (32.4%).
|
Late heterogeneous contrast enhancement was the only parameter that was significantly (p < 0.05) associated with malignant histology (Table 3). Large lesions and lesions with well-defined margins tended to be associated with malignancy, although the p values for these criteria were not statistically significant. Other grouping variables such as homogeneity of early enhancement and lesion shape were not related to the histologic diagnosis.
Signal Intensity Parameters
All the quantitatively evaluated signal intensity parameters (time to peak
enhancement, maximum slope of the enhancement curve, washout, relative water
content) showed gaussian distribution by means of the Kolmogorov-Smirnov test.
As illustrated in Table 4, malignant and benign lesions did not differ significantly in any of the
enhancement parameters and the relative water content. However, malignant
lesions showed a higher maximum slope of the enhancement curve
(Fig. 4), reached the mean
signal intensity peak earlier (Fig.
5), and had a stronger loss of enhancement (washout) from the
initial signal intensity peak to the last contrast-enhanced measurement
(Fig. 6). Nevertheless, there
is a considerable overlap of both parameters.
|
|
|
|
Lesion Categorization
The positive predictive value for malignancy of exclusively MR
imaging—detectable lesions increases as the BI-RADS category increases:
from 0% (0/4) for category 3 lesions and 29.3% (12/41) for category 4 lesions
to 100% (6/6) for category 5 findings. Although the p value is less
than 0.05 (p = 0.001), the chi-square test is not valid because more
than 20% of the squares showed expected values of less than 5.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
In our study, late heterogeneous lesion enhancement (>6 min after gadopentetate dimeglumine injection) correlated significantly with malignancy. This correlation can be explained by the washout phenomenon of malignant lesions, which shows an irregular enhancement pattern within the lesion that subsequently becomes heterogeneous. Other single lesion characteristics are not as helpful in distinguishing malignant from benign disease in exclusively MR imaging—detectable lesions. For a better understanding of the high positive predictive value of well-defined margins, one has to consider MR image resolution. Stomper et al. [29] stated that the analysis of margins of focal enhancing areas is of less value than analysis of margins in mammograms because MR images do not have as high a resolution as film-screen mammograms. The shape of small lesions is difficult to judge for the same reason. Most enhancing foci in our study measured less than 10 mm.
Thorough analysis of lesion signal intensity data in correlation with histology seemed promising for the differentiation of benign from malignant breast lesions. However, in our study no quantitatively evaluated signal intensity parameter differed significantly between malignant and benign lesions. Similar results have been reported by Stomper et al. [29] and Orel et al. [30] who evaluated MR imaging of the breast in patients with known palpable or mammographically detected lesions. Although enhancement and washout tended to be more rapid for carcinomas, these groups of researchers found a considerable overlap in signal intensity and enhancement characteristics of malignant and benign lesions.
These findings do not coincide with the results of two previous studies. Kaiser and Zeitler [31] and Gribbestad et al. [32] reported that all carcinomas could be differentiated from benign lesions by early signal enhancement in a series of 25 and 18 dynamic contrast-enhanced breast MR examinations, respectively. The difference between our results and their figures may be explained by the frequent occurrence of proliferative changes among the benign samples of our series. An increase in signal intensity correlates with vascularization, which again reflects proliferative activity and does not necessarily imply malignancy. Orel et al. [30] also had several "young" fibroadenomas in their series that showed marked and rapid enhancement. On the other hand, only eight lesions in the series investigated by Kaiser and Zeitler [31] were fibroadenomas or proliferative dysplasias, and only one fibroadenoma of the juvenile type was among the benign findings in the patient population of Gribbestad et al. [32]. In a larger study, Kaiser and Mittelmeier [33] reported on a series of 226 dynamic contrast-enhanced MR examinations in which histologically proven fibrocystic changes (proliferative and nonproliferative) showed a significantly lower increase in signal intensity than carcinomas. Nevertheless, these researchers also reported that lesion enhancement increased with increasing proliferative activity in benign lesions. In some cases of proliferative changes, the signal pattern was even suspicious for malignancy.
Moreover, nearly half of the invasive cancers in our series were carcinomas of the lobular type. This type of cancer in particular can present with atypical or even no enhancement [34]. The higher incidence of lobular cancer in exclusively MR imaging—detectable lesions is compatible with difficulty in detecting these lesions on mammography and sonography.
As shown by the analysis of T2-weighted signal intensity data, the water content of malignancies did not differ significantly from that of benign lesions. Therefore, at least in exclusively MR imaging—detected lesions, this parameter does not improve the distinction between malignant and benign lesions; these findings have also been reported by Orel et al. [30]. However, Kuhl et al. [35] could distinguish between fibroadenomas and breast cancers on the basis of the water content of the lesions using a T2-weighted turbo spin-echo sequence with fat suppression.
Although the amplitude of signal intensity does not lead to more precise identification of breast carcinomas, Kuhl et al. [36] stated that the shape of the time—signal intensity curve seems to be important in differentiating benign from malignant lesions. These authors distinguished three different curve shapes: continuous enhancement, plateau, and washout. Fifty-seven percent of all carcinomas in their study showed a washout curve. This finding correlates to our study in which 66.7% of all carcinomas showed washout. Nevertheless, 45.5% of all the benign lesions in our study also had a loss of enhancement that was greater than 10% from the initial signal intensity increase to the last measurement.
Assigning a score to lesions as a synopsis of all lesion features and using the score to determine the analogous BI-RADS category seem to be helpful in assessing the likelihood of malignancy in exclusively MR imaging—detectable lesions. This system could ease the decision about the further diagnostic or interventional course. The score applied in this study was described by Fischer et al. [10, 25] and helps to achieve a more objective categorization of MR imaging—detected lesions.
In conclusion, our results emphasize that MR imaging of the breast can reveal lesions that are occult at mammography and sonography. Approximately one third of these lesions detected exclusively on MR imaging are malignant. In these cases, signal intensity data alone proved to be not helpful in defining malignant and benign abnormalities. However, classifying lesions into BI-RADS categories according to a published score system is helpful in identifying the lesions for which interventional MR imaging is required.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  E. E. Deurloo, S. H. Muller, J. L. Peterse, A. P. E. Besnard, and K. G. A. Gilhuijs Clinically and Mammographically Occult Breast Lesions on MR Images: Potential Effect of Computerized Assessment on Clinical Reading Radiology, March 1, 2005; 234(3): 693 - 701. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. L. Pereira, J. Fritz, C. W. Koenig, F. Maurer, P. Boehm, A. Badke, M. Mueller-Schimpfle, M. Bitzer, and C. D. Claussen Preoperative Marking of Musculoskeletal Tumors Guided by Magnetic Resonance Imaging J. Bone Joint Surg. Am., August 1, 2004; 86(8): 1761 - 1767. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
