| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Diagnostic Radiology, Graduate School of Medicine, Tohoku
University, Sendai, Miyagi, Japan.
2 Present address: Section of Breast Imaging, Department of Radiology, 5841 S.
Maryland Ave., MC2026, Chicago, IL 60637.
3 Department of Surgical Oncology, Graduate School of Medicine, Tohoku
University, Sendai, Miyagi, Japan.
4 Department of Pathology, Graduate School of Medicine, Tohoku University,
Sendai, Miyagi, Japan.
Received May 23, 2005;
accepted after revision August 26, 2005.
Address correspondence to A. Shimauchi.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. Sixty-nine patients with histologically proven invasive carcinoma underwent MDCT and MRI. Retrospectively, two radiologists performed a blinded review of the MDCT and MRI. Cases with intraductal component enhancement were classified into two morphologic types: ductal extension and segmental distribution. The lengths of the main tumor and of the intraductal component were measured in cases with ductal extension. For cases with segmental distribution, only the maximum length of the tumor was measured.
RESULTS. The sensitivity, specificity, and accuracy in detecting the intraductal component were 61%, 88%, and 71%, respectively, using MDCT; and 75%, 88%, and 80%, respectively, using MRI. MRI revealed the presence of the intraductal component with significantly higher sensitivity than did MDCT (p = 0.031). In the analysis of the length of the intraductal component in cases with ductal extension, which had relatively small intraductal components, underestimation by 15 mm or more was significantly less frequent with MRI than with MDCT (p = 0.008). There was no significant difference between MDCT and MRI in the evaluation of the maximum length of tumors in cases with segmental distribution, which had relatively large intraductal components.
CONCLUSION. Compared with MDCT, MRI revealed the presence of the intraductal component with higher sensitivity and equivalent specificity. In cases with ductal extension, MRI is more precise than MDCT for determination of the margin for surgical removal, with less underestimation of the extent of the intraductal component.
Keywords: breast cancer • MDCT • MRI • radiologic-pathologic correlation
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
For years the extent of breast cancers has been studied using MRI [4-8]. Some authors have shown that MRI can detect the intraductal component more accurately than can conventional techniques [9-11]. However, several reports have shown that CT can also be used to accurately assess the extent of breast cancer [12-14]. Two studies that compared the use of single-detector helical CT and MRI concluded that these methods were equally effective for assessing the extent of breast cancer [15, 16]. In the former of these two studies, a 0.5-T imaging unit was used whereas in the latter study two different populations were used for each of the two methods. In recent years MDCT has become commercially available. The greater spatial and temporal resolution of MDCT images compared with those obtained using single-detector helical CT suggests that the extent of breast cancer might be evaluated more accurately with MDCT. Although a recent study reported that MDCT could be used to distinguish benign lesions from carcinomas [17], the extent of the involvement of intraductal components was not evaluated in that study. Therefore, the accuracy of using MDCT to examine intraductal components of breast cancer remains uncertain.
In this study, we compared the accuracy of MDCT and MRI for evaluating intraductal components of breast cancer in the same population.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Among the 69 patients, 48 (70%) complained of the presence of a self-palpable mass, one (1%) complained of nipple discharge, 18 (26%) lesions were detected during screening (17 palpable and one nonpalpable), one lesion (1%) was detected on CT incidentally, and one lesion (1%) was detected during follow-up for contralateral breast cancer. Fine-needle aspiration or core needle biopsy of suspicious lesions was performed for all patients. Thereafter, the patients underwent MRI and MDCT to evaluate the local extent of the breast cancer. The median interval between MDCT and MRI was 11 days (range, 0-46 days). After the preoperative examinations, breast-conserving surgery was performed on all patients. Informed consent for the MDCT and MRI examinations was obtained from all patients. The median interval between MDCT and MRI examinations and surgery was 3 days (range, 2-16 days).
MDCT and MRI Protocol
MDCT examinations were performed with a LightSpeed Qx/i (GE Healthcare).
Contrast-enhanced scans were obtained from the level of the axilla to the
lower edge of the breast with a collimation of 1.25 mm and a pitch of 6.
Nonionic contrast material (100 mL of Omnipaque [iohexol] 300; Daiichi
Pharmaceutical) was injected IV at a rate of 4.0 mL/s. All patients underwent
two phases of scanning that started 40 and 65 seconds after the injection of
the contrast material.
Other scan parameters were as follows: acquisition time, 0.8 s/rotation; image matrix, 512 x 512; field of view, 32-42 cm; tube voltage, 120 kV; and tube current, 250 mA. Data were reconstructed at 0.6-mm increments. The initial 25 patients underwent MDCT in the prone position, whereas the remaining 44 patients were in the supine position to simulate surgical positioning.
MRI examinations were performed using a 1.5-T imaging unit (Magnetom Vision; Siemens Medical Solutions). All patients underwent MRI in the prone position using a dedicated breast coil. After obtaining bilateral fat-saturated T1-weighted images (TR/TE, 700/12) and T2-weighted images (TR/TE, 8,000/60) of the breasts, a T1-weighted 3D fast low-angle shot (FLASH) gradient-echo sequence (TR/TE, 8.1/4.0; flip angle, 25°; matrix, 256 x 256; section thickness, 2 mm without intersection gap; acquisition time, 87 seconds) was performed before and after the injection of the contrast material. A dynamic study in the sagittal plane was performed three times at the initiation of the IV injection of 0.1 mmol/kg gadopentetate dimeglumine (Magnevist, Schering) or 0.1 mmol/kg gadodiamide (Omniscan; Daiichi Pharmaceutical) at a rate of 2 mL/s. After the examination, unenhanced images were subtracted from the corresponding contrast-enhanced images on a pixel-by-pixel basis. The images were transferred to a workstation (Advantage Windows, software version 4.0; GE Healthcare) for analysis.
Histopathologic Investigation of Tumors
Surgical specimens were fixed in a 10% formaldehyde solution and cut into
serial 5-mm-thick slices. Slices for the whole of each specimen (9-72 slices;
median, 31) were embedded individually in paraffin wax to investigate the
microscopic features. From every paraffin wax-embedded slice, one or several
5-µm-thick sections, were cut and stained with H and E. For each patient,
the histopathologic extent of the carcinoma in all slices was determined by
pathologists who specialize in pathology of the breast. The histopathologic
extent of the carcinoma was recorded as a topographical map.
Using only histopathologic information, the microscopic features of the cancer, including the presence or absence of an intraductal component, the length of the main invasive tumor, and the length of the intraductal component were recorded. In this study, the presence of even a small intraductal component was considered positive. The intraductal component within the main mass was not included in measurements of the intraductal component. The areas of the intraductal component and of the entire lesion were measured on the topographical map, and the ratio between the two parameters was calculated.
In this study, invasive tumors with an intraductal component were classified into two morphologic types. Tumors that comprised a main tumor and an outward intraductal extension were classified as the ductal extension type (Fig. 1A); and in the absence of a dominant mass, tumors with an intraductal component that were poorly defined were classified as the segmental type (Fig. 1B).
|
|
Image Analysis
Two radiologists with expertise in breast imaging retrospectively reviewed
the MDCT and MRI images by consensus. The radiologists viewed the MDCT and MRI
images at different times with an interval of at least 1 month. They
interpreted the images with the benefit of a brief clinical history and
knowledge of the histopathologic findings but without knowledge of the
mammography and sonography results. In addition to source images,
maximum-intensity-projection and multiplanar reconstruction images were
reviewed.
First, the presence of an intraductal component of the invasive cancer was determined. Because intraductal components tend to extend along the duct lobular system, the pattern of distribution of lesion enhancement was taken into consideration. Non-masslike enhancement in clumped or ductal form that extended from the main tumor (ductal extension type) and enhancement in a segmental form (segmental type) were regarded as the intraductal component of the tumor according to BI-RADS [18] and Liberman et al. [19]. For ductal extension-type tumors, the intraductal component was regarded as positive irrespective of the length of the enhancement. Multiple round or oval masses that were distributed diffusely were regarded as benign [20]. The aforementioned method of morphologic typing of invasive tumors was performed for both the MDCT and MRI images (Figs. 1A and 1B).
|
|
|
Correlation Between Radiologic and Histopathologic Findings
To evaluate the sensitivity, specificity, and accuracy of MDCT and MRI for
detecting intraductal components of invasive breast carcinomas, we correlated
the images with the topographical maps. In cases with an intraductal
component, we then examined the concordance of the radiologic and
histopathologic findings with regard to the morphologic typing of tumors
(ductal extension or segmental).
When the radiologic and histopathologic findings were in agreement, the measured lengths of the main tumor and the intraductal component using each method were compared (Figs. 2A, 2B, and 2C). The intraductal component within the main mass was not included in the measurements. In cases with segmental distribution, the maximum length of the entire lesion measured using each method was compared.
Statistical Analysis
The sensitivity, specificity, and accuracy of the determination of the
presence of intraductal components and the accuracy of the measured lengths of
intraductal components and the maximum length of the tumor were analyzed using
the McNemar test. The ratio of the area of the intraductal component to that
of the entire lesion was compared between ductal extension- and segmental-type
tumors using a Mann-Whitney test. The statistical analysis was performed using
SPSS version 11.0 (Statistical Package for the Social Sciences) for Windows
(Microsoft). Statistical significance was inferred for p values less
than 0.05.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Detection of the Intraductal Component Using MDCT and MRI
All 69 cases of invasive carcinoma were depicted by both MDCT and MRI
(Figs. 3A,
3B,
3C,
3D,
4A,
4B, and
4C). The sensitivity,
specificity, and accuracy in detecting an intraductal component with MDCT were
61%, 88%, 71%, respectively; the values of the same parameters for MRI were
75%, 88%, and 80%, respectively (Table
2). In six patients, intraductal components were detected on MRI
but not on MDCT, and a significant difference occurred between the sensitivity
of MRI and that of MDCT (p = 0.031). Among 44 cases with a
histopathologically proven intraductal component, 27 were detected by both
MDCT and MRI. Eleven cases with intraductal components that were not detected
by either MDCT or MRI and the six cases with intraductal components that were
detected only by MRI were revealed histopathologically to be of the ductal
extension type.
|
|
|
|
|
|
|
|
Comparison of Tumor and Intraductal Component Using MDCT and MRI
The morphologic typing of tumors using MDCT and MRI agreed in each of 44
cases and corresponded to the histopathologic findings in all but two cases.
In the 42 cases for which the radiology- and histopathology-based findings
were in agreement, the lengths of the main tumor and the intraductal component
in cases with ductal extension and the maximum length of the tumor in cases
with segmental distribution were measured using MDCT and MRI
(Table 1). The correlation
coefficient between the length of the main tumor measured by MDCT and that
determined histopathologically was 0.72, whereas that measured by MRI was
0.70; both MDCT and MRI generally underestimated the length of the intraductal
component. In cases with segmental distribution, the correlation coefficient
between the maximum length of the tumor measured by MDCT and that determined
by histopathology was 0.59, and the correlation coefficient between the
lengths determined by MRI and by histopathology was 0.57.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
In this study, there were six cases in which MRI revealed an intraductal component but MDCT did not; there were no cases in which MDCT revealed an intraductal component that was undetectable with MRI. Differences in the sizes of the intraductal components, which depended on the morphologic type of the tumor, might have resulted in the discrepancy between MRI and MDCT. Indeed, all the lesions detected by MRI but not by MDCT were classified as ductal extension type, and the area of the intraductal component in this type of tumor was significantly smaller than that in tumors with a segmental distribution. Furthermore, the greater contrast obtained with MRI by using the subtraction technique probably contributed to the higher sensitivity of this method in detecting small intraductal components (subtraction was not performed for the MDCT images). Moreover, there are reports that ductal carcinoma in situ is enhanced at a slower rate than invasive carcinoma [22-24]. Therefore, in some cases the intraductal component might have become enhanced more slowly than the main tumor. This would explain why three-phase dynamic MRI revealed the presence of small intraductal enhancement with greater sensitivity than did two-phase MDCT.
Two other factors might have influenced the diagnostic accuracy of MRI and MDCT in this study. First, the use of different histopathologic criteria for the intraductal component might have contributed to the relatively lower sensitivity of MDCT in our study as compared with the results of previous studies. In most previous studies [9, 12-14], the intraductal component was defined as positive if it was longer than 1 cm. By contrast, we judged intraductal components as positive even if we could observe only a very small intraductal lesion in the histologic sections.
Second, our interpretation of the radiologic images might have affected our quantification of the specificity. Nakahara et al. [25] and Tamaki et al. [16] reported that the specificity of MRI was lower than that of CT for detecting intraductal components. The authors of the former study speculated that the detection of subtle enhancements of benign fibrocystic changes by MRI might lead to overestimation. The equivalent specificities of MRI and MDCT that we observed might be attributable to our interpretation of the images: We excluded diffuse smooth, round, or oval masses as benign [18] and took into account that intraductal components tend to extend from the primary invasive carcinoma along the mammary ductal tree [26].
Length of the Intraductal Component Measured Using MDCT and MRI
The preoperative radiologic evaluation of the length of intraductal
components is generally used to determine the margin for surgical removal. In
the present study, radiology-based measurements of intraductal components
underestimated the extent of the lesions in most cases. Because each specimen
was cut into 5-mm-thick slices, the lengths determined by histopathology using
the topographical maps might have been overestimated. This might have been the
cause of the discrepancies between the lengths determined using radiologic
images and those determined from the topographical maps. A 5-mm-thick slice is
used to evaluate intraductal components pathologically in our institute
because the detection rate of intraductal components was comparable to thinner
sections in the former study
[27]. It should be thinner to
correlate the radiologic and pathologic measurement more exactly; however, it
is technically difficult and impractical to cut into uniform thickness less
than 5 mm to make a topographical map.
By contrast, the radiologically determined lengths of the main tumors corresponded well with the lengths obtained histopathologically in the present study. Therefore, the point at which the intraductal component originated from the main tumor appears to correspond in the radiology- and histopathology-based measurements. Given this correspondence, we can regard the radiologically and histopathologically determined lengths of the intraductal component as the distances of the radiologic and histopathologic margins, respectively (the peripheral end points of the intraductal component), from the common reference point—that is, from the point of origin of the intraductal component at the main tumor (Figs. 2A, 2B, and 2C).
In our institution, surgeons err on the side of caution by using a surgical margin that is 20 mm outside the radiologically determined tumor margin. If the difference between radiologically and pathologically determined tumor lengths or the radiologic underestimation of this parameter is less than 15 mm (that is, the gap between the pathologically determined and surgical margins is greater than 5 mm; see Fig. 2A), the surgical margin is regarded as negative. When the underestimation is in the range of 15 to 20 mm (that is, the gap between the pathologically determined and surgical margins is less than 5 mm; see Fig. 2B), the surgical margin is negative but close to the tumor mass. Even with a safety zone of 20 mm, the surgical margin is positive if the magnitude of underestimation is greater (see Fig. 2C). In the present study, underestimation of the length of intraductal components by 15 mm or more was significantly less frequent with MRI than with MDCT (Table 3). Therefore, MRI was more accurate than MDCT for evaluating the length of intraductal components. We propose that MRI is better than MDCT for evaluating tumors with ductal extension, especially those with relatively small intraductal components. This does not apply to segmental-type tumors with relatively large intraductal components because MDCT and MRI were equally accurate in measuring the length of this type of tumor.
|
Advantages and Disadvantages of MDCT
The most undesirable feature of MDCT is exposure to radiation. The dose of
radiation at the skin surface for the acquisition of an MDCT image during a
single breath-hold was 
26 mGy
[17], which is almost
equivalent to the dose reported for conventional CT. However, the safety of
the exposure of the contralateral breast to radiation during CT has not been
verified. This could limit the repeat scanning of the breast that would be
possible with MRI, which would be preferable to improve the depiction of the
intraductal component.
However, the advantages of breast MDCT over MRI include a faster scanning speed, the availability of the supine position (which allows surgical simulation), and simultaneous scanning of both breasts (which allows the affected breast to be compared with the contralateral breast) [14, 17, 28].
Limitations of This Study
Our study included primarily palpable, symptomatic breast cancers rather
than screen-detected cases. This is probably because our population does not
include noninvasive cancer, but it might have affected the proportion of cases
with an intraductal component.
According to previous studies [12, 14, 28], a suitable injection rate for contrast material is 2-3 mL/min, whereas the optimal scan timing is every 40 or 60 seconds during the early phase and every 120-180 seconds during the late phase after injection. Our protocol (4 mL/min and scans at 40 and 65 seconds) might not have been optimal for the detection of the intraductal component.
Another potential problem is our assumption that the origin of the intraductal component was the same in the radiologic and histopathologic examinations. Although we cannot prove that this was the case, this assumption is reasonable considering the close correspondence between the radiologically and pathologically determined lengths of the main tumor.
Conclusion
In summary, compared with MDCT, MRI revealed the presence of the
intraductal component of invasive tumors with higher sensitivity and with
equivalent specificity. In cases with ductal extension, MRI is more precise
than MDCT for the preoperative determination of the margin for surgical
removal, because MRI is less likely to underestimate the extent of the
intraductal component.
Acknowledgments
We gratefully acknowledge Noriaki Ohuchi and Hiroyuki Abe for their
assistance in manuscript preparation.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  E. J. Stern Seek and You Shall Find Am. J. Roentgenol., August 1, 2006; 187(2): 265 - 265. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
20 mm, the surgical margin is
positive.
