| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Institute of Diagnostic and Interventional Radiology, University Hospital
Jena, Jena 07747, Germany.
2 Present address: Institut für Diagnostische, Interventionelle und
Pädiatrische Radiologie, Inselspital, CH-3010 Bern, Switzerland.
3 Present address: Institute of Radiology, Suedharz-Hospital, Nordhausen,
Germany.
Received March 3, 2005;
accepted after revision August 10, 2005.
Address correspondence to D. R. Fischer
(dorothee.fischer{at}insel.ch).
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. The study included 132 histologically verified lesions (71 malignant lesions, 10 pure carcinoma in situ, and 51 benign lesions) enhancing on dynamic breast MRI before biopsy. The lesions were evaluated by three radiologists in a double-blinded manner. The presence of an adjacent vessel was supposed if at least two observers voted positively.
RESULTS. Sixty-one (85.9%) of 71 malignant lesions, six (60%) of 10 carcinomas in situ, and 10 (19.6%) of 51 benign lesions were associated with an adjacent vessel, which differed significantly (p < 0.001) between benign and malignant lesions (the latter with and without including pure carcinoma in situ), leading to a positive predictive value of 85.9% (87% including pure carcinomas in situ), a negative predictive value of 80.4% (74.5% including pure carcinomas in situ), an accuracy of 83.2% (81.6% including pure carcinomas in situ), a sensitivity of 85.9% (82.7% including carcinomas in situ), and a specificity of 80.4% for this sign concerning malignancy.
CONCLUSION. The presence of an adjacent vessel seen on subtraction images promises to be a good marker for malignancy and can therefore help increase the specificity of breast MRI.
Keywords: angiogenesis • breast • breast cancer • breast MRI • MRI • oncology
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Combining these ideas, both to increase the specificity of breast MRI and supposing an increased vascularity in a breast having malignancy, this study aimed to find out whether a prominent enlarged adjacent vessel leading to an enhancing lesion and seen on subtraction images occurs significantly more often in malignancies, so that an additional diagnostic sign could be found for the routine diagnosis of breast MR images (Figs. 1, 2, 3, 4, and 5).
|
|
|
|
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Patients included in this study either were referred to the gynecologic department because of suspicious findings at palpation, sonography, or mammography, or presented themselves for a follow-up examination after breast cancer therapy. The patients' ages ranged from 16 to 80 years (mean age, 57 years).
Chemotherapy had been given to 13 patients before examination. Of these 13 patients, six had received chemotherapy from 1993 to 2002, for a prior diagnosis of breast cancer and one for a non-Hodgkin's lymphoma. Neoadjuvant chemotherapy was performed before the MRI examination in seven patients. A history of a core biopsy was given in 20 patients. After examination, all lesions were histologically verified by lumpectomy, mastectomy, or core biopsy at the Institute of Pathology at the University Hospital Jena.
MRI
A predefined examination protocol was used. All MR images were obtained on
a 1.5-T imager (Symphony, Siemens Medical Solutions) using a double breast
coil with the patient lying in the prone position. Examinations followed the
same imaging protocol. According to our quality criteria, routine examinations
are performed in patients who have not received hormone replacement therapy in
the past 4-6 weeks and who are not in days 4-20 of the menstrual cycle
[8-10].
For the dynamic study, multislice 2D FLASH (fast low-angle shot) T1-weighted sequence images were obtained with the following parameters: TR/TE, 113/4.6; flip angle, 80°; field of view, 350 x 350 mm; matrix, 384 x 384; slice thickness, 3.0 or 4.0 mm; gap, 0.4 mm; axial orientation; 33 slices covering both breasts. After acquisition of a native scan, 0.1 mmol/kg of body weight of gadopentetate dimeglumine (Magnevist, Schering) was administered IV as a rapid bolus within 10 seconds followed by a 20-mL saline flush. Thirty-five seconds after the bolus injection and saline administration, dynamic scanning was continued with the same sequence parameters and under identical tuning conditions at 1-minute intervals for a total of 8 minutes. After the dynamic scanning, axial T2-weighted turbo spin-echo images (8,900/207; field of view, 350 x 350; matrix, 512 x 512) were obtained in identical slice positions in the axial plane. Unenhanced images of the dynamic study were subtracted from the contrast-enhanced dynamic images.
Image Analysis
Whether an adjacent vessel was present in 132 histologically verified
lesions was evaluated on breast MRI subtraction images by three radiologists
who were blinded to the histopathologic outcome, blinded to each other's
results, and interpreting images under the same conditions. If at least two
observers judged the lesion to be associated with an adjacent vessel, an
adjacent vessel was supposed to exist. The possibility of a significantly
higher occurrence of an adjacent vessel in malignant lesions was evaluated
using Pearson's chi-square and Fisher's exact tests.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Malignant lesions measured 5-97 mm (mean, 24 mm), benign lesions measured 3-100 mm (mean, 13 mm), and pure carcinoma in situ lesions measured 6-44 mm (mean, 17 mm); sizes were measured on subtraction images. Lesions showed either a focal, clumped, masslike, or segmental enhancement.
On the basis of the mean score of the three observers, in 61 (85.9%) of 71 malignancies and in six (60%) of 10 pure carcinomas in situ, an adjacent vessel leading to the enhancing lesion was found. In contrast, in only 10 (19.6%) of 51 benign lesions could an adjacent vessel be confirmed.
In the 61 malignancies with an adjacent vessel, 32 lesions were judged by all three observers to show an adjacent vessel, and 29 lesions were judged to be associated with an adjacent vessel by two observers. In three of 10 benign lesions, two of the three observers saw an adjacent vessel; and in seven of 10 benign lesions, the three observers stated an adjacent vessel was present. Of the six pure carcinomas in situ, three were judged to be associated with an adjacent vessel by two observers, and three lesions were so judged by all three observers independently.
The presence of an adjacent vessel was significantly different (p < 0.001) in benign and malignant lesions (calculation both including and not including pure carcinomas in situ with the malignant lesions, using Pearson's chi-square test). This finding has a positive predictive value of 85.9% (87% when including pure carcinomas in situ), a negative predictive value of 80.4% (74.5% when including pure carcinomas in situ), an accuracy of 83.2% (81.6% when including pure carcinomas in situ), a sensitivity of 85.9% (82.7% when including pure carcinomas in situ), and a specificity of 80.4%.
The sizes of malignant and benign lesions with and without an adjacent vessel are shown in Table 1.
|
Of 14 malignant lesions 
1 cm, nine were associated with an adjacent
vessel, three of 31 benign lesions 1 cm were associated with an adjacent
vessel, and one of three pure carcinoma in situ lesions 10 mm was
associated with an adjacent vessel. This finding differed significantly
between small malignant and benign lesions (p < 0.001), both when
including and when not including carcinomas in situ, using Fisher's exact
test.
Of 16 lesions in 13 patients with a history of chemotherapy (including a remote history of chemotherapy for a prior diagnosis of breast cancer), 11 lesions in nine patients revealed invasive cancer, all showing an adjacent vessel; three lesions in three patients revealed DCIS, two of which did not show an adjacent vessel; and two benign lesions in two patients did not show an adjacent vessel. Lesions of patients with a remote history of chemotherapy (1993-2002) showed an adjacent vessel in five of seven cases, whereas seven of nine lesions in patients after neoadjuvant chemotherapy were associated with an adjacent vessel.
In 24 lesions that only one observer judged positive for an adjacent vessel, 16 were benign, seven were malignant, and one was DCIS with all 24 lesions ranging in size from 3-24 mm (mean, 11.4 mm).
The 10 false-positive lesions were 7-100 mm (mean, 25.2 mm) and had a histology of fibrosis, atypical ductal hyperplasia, fibroadenoma, fibrosis and adenosis, and juvenile fibroadenoma. The 10 false-negative lesions measured 5-24 mm (mean, 12.7 mm) and included a lymphoma and a mucinous carcinoma.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Because an adjacent vessel was also observed in a minority of benign lesions in our study, we assume that, because of the increasing metabolic demand of growing benign lesions, those lesions can also increase their blood supply.
Possibly an adjacent vessel more often leads to a lesion (benign and malignant) from the medial side because the breast is being supplied by the internal rather than by the lateral thoracic artery [7]. The question from which direction the vessel nourishes the tumor has not been answered in our study because such a finding would be of no further diagnostic value. Also, the fact that the density of vessels in breast cancer depends on age [15] could be an explanation of the higher occurrence of adjacent vessels in young women with breast cancer, a possibility that was not considered in this study.
Chemotherapy either in the past or before an examination seems not to have an influence on the occurrence of an adjacent vessel in invasive malignancies: Of 16 lesions in 13 patients with a history of chemotherapy (seven of them having had neoadjuvant chemotherapy before MRI), 11 lesions in nine patients revealed invasive cancer and all showed an adjacent vessel, whereas three lesions in three patients revealed DCIS, of which two did not show an adjacent vessel.
On the other hand, the real effect of chemotherapy on the visualization of vessels can only be judged by scanning the patient before the initiation of chemotherapy, choosing a vessel at that time, and then following the visibility of this vessel throughout the course of chemotherapy. In such a case, problems would occur because of changes of the breast position.
In our study, the mean size of lesions associated with an adjacent vessel was 25.4 mm, but it was also observed that an adjacent vessel in lesions smaller than or equal to 10 mm occurs significantly more often in malignant than in benign lesions (p < 0.001 both including and not including carcinomas in situ).
On the other hand, 10 false-negative lesions (Fig. 4) (mean size, 12.7 mm) and 10 false-positive lesions (Fig. 5) (mean size, 25.2 mm, which is more than twice the size of the false-negative lesions) support the hypothesis that a significant correlation exists between breast tumor size and the number of vessels detected on contrast-enhanced breast MRI [16], possibly in both benign and malignant histology.
This study was purely subjective. No vessel size or thickness was defined as a threshold parameter for an adjacent vessel, so that sometimes the decision as to the presence of an adjacent vessel was difficult; in 24 lesions, 16 of which were benign at histology, one of the three observers believed an adjacent vessel was present.
Although these values are subjective, both positive (85.9%) and negative (80.4%) predictive values for the presence of an adjacent vessel were promising. This observation of a prominent vessel more likely accompanying a malignant than a benign lesion might also be true in mammography, which would be an additional diagnostic tool for that technique as well. This hypothesis should be verified in a further study.
With regard to breast MRI, a computer-assisted analysis could of course improve and simplify analysis. Maximum intensity projection might better visualize adjacent vessels [17], especially when the images are rotated. In addition, motion artifacts can hamper the interpretation of subtraction images and therefore make impossible a clear statement regarding the existence of an adjacent vessel.
In this study, 2D FLASH T1-weighted images were obtained. In contrast to high-resolution 3D data sets in cases of especially thin adjacent vessels, 2D FLASH T1-weighted images might sometimes lead to false-negative results, so possibly our data cannot be reproduced in a different imaging protocol; this possibility should be evaluated [13].
Furthermore, interobserver reliability in this study was not considered, because only mean values given by the three examiners were analyzed (at least two thirds of the positive diagnoses resulted in the assumption of an adjacent vessel being present). Because of the study design, we could not be expected to classify malignant lesions (including nonenhancing types of carcinoma in situ), which are occult on MR images because they do not enhance.
In conclusion, the sign of an adjacent vessel reveals a positive predictive value of 85.9% (87% when pure carcinomas in situ are included), a negative predictive value of 80.4% (74.5% when pure carcinomas in situ are included), an accuracy of 83.2% (81.6% when pure carcinomas in situ are included), a sensitivity of 85.9% (82.7% when carcinomas in situ are included), and a specificity of 80.4% for indicating malignancy. An adjacent vessel occurs significantly more often (p < 0.001) in malignant lesions (including small lesions; p < 0.001). Therefore, an adjacent vessel is a promising sign in addition to other well-known signs [1, 2] for routine breast MRI diagnosis.
The presence of an enlarged adjacent vessel seen in subtraction images seems to be a good marker for malignancy and can therefore help increase the specificity of dynamic breast MRI because its presence significantly differs in benign and malignant lesions (p < 0.001). Its occurrence can be observed significantly more often in malignant lesions than in benign lesions less than or equal to 1 cm (p < 0.001).
|
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 |
