European Radiology

, Volume 16, Issue 8, pp 1811–1817

MRI-guided needle localization of suspicious breast lesions: results of a freehand technique

Authors

  • M. A. A. J. van den Bosch
    • Department of RadiologyStanford University Medical Center
    • Department of RadiologyUniversity Medical Center Utrecht
  • B. L. Daniel
    • Department of RadiologyStanford University Medical Center
  • S. Pal
    • Department of RadiologyStanford University Medical Center
  • K. W. Nowels
    • Department of PathologyStanford University Medical Center
  • R. L. Birdwell
    • Department of RadiologyStanford University Medical Center
  • S. S. Jeffrey
    • Department of SurgeryStanford University Medical Center
    • Department of RadiologyStanford University Medical Center
    • Department of RadiologyStanford Center for Advanced Medicine
Breast

DOI: 10.1007/s00330-006-0214-5

Cite this article as:
van den Bosch, M.A.A.J., Daniel, B.L., Pal, S. et al. Eur Radiol (2006) 16: 1811. doi:10.1007/s00330-006-0214-5
  • 98 Views

Abstract

Magnetic resonance imaging (MRI) can detect clinically and mammographically occult breast lesions. In this study we report the results of MRI-guided needle localization of suspicious breast lesions by using a freehand technique. Preoperative MRI-guided single-needle localization was performed in 220 patients with 304 MRI-only breast lesions at our hospital between January 1997 and July 2004. Procedures were performed in an open 0.5-T Signa-SP imager allowing real-time monitoring, with patient in prone position, by using a dedicated breast coil. MRI-compatible hookwires were placed in a noncompressed breast by using a freehand technique. MRI findings were correlated with pathology and follow-up. MRI-guided needle localization was performed for a single lesion in 150 patients, for two lesions in 56 patients, and for three lesions in 14 patients. Histopathologic analysis of these 304 lesions showed 104 (34%) malignant lesions, 51 (17%) high-risk lesions, and 149 (49%) benign lesions. The overall lesion size ranged from 2.0–65.0 mm (mean 11.2 mm). No direct complications occurred. Follow-up MRI in 54 patients showed that two (3.7%) lesions were missed by surgical biopsy. MRI-guided freehand needle localization is accurate and allows localization of lesions anterior in the breast, the axillary region, and near the chest wall.

Keywords

Magnetic resonance imaging (MRI)BreastNeedle localizationFreehand technique

Introduction

Magnetic resonance imaging (MRI) of the breast has become a well-established method for detection of invasive breast carcinoma [13]. It is a more sensitive method than conventional mammography or ultrasound (US) for detection of invasive tumors, with a sensitivity approaching 100% [4, 5]. However, the specificity of MRI of the breast is rather variable, ranging from 40% to 95% depending on patient selection and imaging technique [6]. Today, MRI of the breast is mainly used as a valuable adjunct to conventional imaging modalities for analysis of questionable breast lesions, for follow-up after breast-conserving therapy, to identify breast cancer in high-risk patients, and for preoperative assessment of patients with known breast cancer to rule out multicentric or contralateral disease [3, 710].

Because of its high sensitivity, MRI of the breast may reveal suspicious breast lesions that are clinically and mammographically occult [11, 12]. To allow histopathologic analysis of these lesions, MRI-guided tissue sampling is required. The gold standard for sampling MRI-detected lesions is preoperative MRI-guided needle localization and hookwire placement followed by surgical biopsy [13, 14]. Two different techniques for MRI-guided needle localization have been reported: stereotaxic or freehand [1114, 15]. Most previous reported studies focused on the first method whereas the freehand technique has only been described in a small study population [15]. The freehand technique has several potential advantages. It is a simple and versatile technique that is similar to freehand lesion localization with US or computed tomography (CT). Furthermore, because it can be used without specialized grid devices, it allows localization of lesions throughout the breast, including those near the chest wall, in the axillary region, and anterior in the breast.

The purpose of this study was to report our experience with MRI-guided needle localization for surgical biopsy of suspicious breast lesions using the freehand technique in a 0.5-T open magnet.

Materials and methods

Patients

We retrospectively reviewed all preoperative MRI-guided needle localizations performed in 220 patients with 304 breast lesions at our hospital between January 1997 and July 2004 in women aged 29–71 years (mean age, 49 years). Localization was performed for a single lesion in 150 patients, for two lesions in 56 patients, and for three lesions in 14 patients. In all cases the lesions were localized using a single wire. Patients who underwent bracket localization (n=26) were excluded and are reported elsewhere. Also, the first 17 patients who underwent needle localization in our clinic in 1997 were excluded because their results previously have been reported. The study was performed in accordance with a protocol approved by our institutional panel on human subjects.

MRI of the breast: indications and technique

Of the 220 patients, 48 (22%) underwent MRI because of increased breast cancer risk, including patients with a positive family history, biopsy-proven diagnosis of atypical ductal hyperplasia, lobular carcinoma in situ, ductal carcinoma in situ (DCIS), or with prior invasive breast cancer. MRI was performed in 115 (52%) patients for problem solving, including questionable findings at physical examination or on mammography or sonography or occult carcinoma consistent with breast primary carcinoma in an axillary lymph node. The remaining 57 (26%) underwent MRI for staging of known breast cancer.

All patients underwent unilateral diagnostic breast MRI on a separate day prior to biopsy (1.5T Echospeed Scanner, GE Medical Systems, Milwaukee, WI, USA; phased-array breast coil, MRI devices, Waukesha, WI, USA). The protocol included axial unenhanced T1-weighted spin-echo imaging and rapid dynamic imaging [three-dimensional (3D) spiral MRI] after intravenous contrast injection, followed by high spatial resolution 3D gradient echo imaging, water-selective spectral-spatial excitation (3D-SSMT) and an additional 4 min of dynamic washout data (3D spiral MRI identical to the initial dynamic scan), a technique that has been described elsewhere in detail (Fig. 1) [7, 8].
https://static-content.springer.com/image/art%3A10.1007%2Fs00330-006-0214-5/MediaObjects/330_2006_214_Fig1_HTML.jpg
Fig. 1

Sagittal contrast-enhanced high-resolution 3DSSMT image showing a suspicious enhancing mass in the lower outer quadrant of the left breast of a 53-year-old female patient

Before the day of the needle localization procedure, diagnostic MR images were reformatted in the axial plane to facilitate selection of the skin entry site and optimal needle trajectory. Needle trajectory was based on the desired surgical approach and the location of any structures to be avoided, such as implants. The trajectory was usually planned so that the needle remained within the axial plane although a full range of angled paths within that plane were available. This was done to ensure full visualization of the needle on axial-guidance images. In general, a lateral approach was planned, but a medial or circumareolar approach was also possible. In two cases a trajectory that was tangential to the surface of an implant was planned.

MRI-guided needle localization was performed using a 0.5 T vertically open, horizontal field scanner (Signa-SP, General Electric Medical Systems) with the patient in the prone position by using a phased-array breast coil (Fig. 2a,b). All needle localizations were performed by using a freehand technique. In this method all steps up to and including initial needle placement were performed prior to contrast administration because the preferential enhancement of suspicious breast lesions over background breast tissue is brief. In detail, the technique is started by fixing a lifesaver-shaped aqueous fiducial marker (Radionics, Z Medical Inc., Baltimore, MD, USA) to the skin of the breast overlying the approximate position of the lesion. The position of the marker relative to the target was determined by comparing unenhanced axial T1 fast spin echo (FSE) images (TR/TE/ETL=300/13/4, 24 s per stack of three 5-mm-thick slices) with axial T1 spin echo and axially reformatted contrast-enhanced 3D-SSMT images from the preliminary diagnostic MRI [16, 17]. The target lesion was identified on the nonenhanced T1 FSE images by using breast architecture as a map. To determine optimal needle entry site on the skin, the marker was repositioned and reimaged until both the center of the fiducial marker and the center of the suspected target were visualized in the same axial image. The skin entry site was marked with a pen, sterilized with Betadine, draped, infiltrated with 1% lidocaine, and nicked with an MRI-compatible disposable scalpel (Microsurgical Techniques, Fort Collins, CO, USA). An MRI-compatible needle [20-gauge MR imaging histology needle (E-Z-Em, Westbury, NY, USA) or 21-gauge Kopans MREYE needle (Cook, Bloomington, IN, USA)] was inserted into the marked skin entry site and directed along the planned needle approach. The needle was inserted, imaged, and interactively repositioned until T1 FSE images revealed the needle tip at the edge of the target (Fig. 3a–d). The 0.5 T open imager allowed the radiologist continuous access without moving the patient. Enhancement of the target was then confirmed using rapid, water-specific imaging by three-point Dixon gradient echo imaging (150/12.8, 19.8, 26.8; 90° flip angle; 5-mm section thickness; 256×128 matrix) after intravenous infusion of 0.1 mmol/kg contrast material (Gd-DTPA, Magnevist, Berlex, Wayne, NJ, USA). Three-point Dixon imaging was used because conventional fat-saturation techniques are unsatisfactory due to the eccentric location of the breast in the magnet bore, the inability to achieve robust shim in the open MR scanner, and the relatively poor separation between the spectral peaks of fat and water at 0.5 T (only 72 Hz) [16]. Once it was confirmed that the needle had reached the target lesion, 0.2 ml of methylene blue dye was injected according to the customary protocol for all preoperative hookwire localizations at our institution, and an MR-compatible hookwire was deployed. Diagnostic whole-breast MRI was then repeated at the conclusion of the procedure to document the final trajectory of the hookwire with respect to other breast structures. Immediately after the MRI, all patients underwent additional imaging with conventional mammography to independently document the position of the hookwire before surgery.
https://static-content.springer.com/image/art%3A10.1007%2Fs00330-006-0214-5/MediaObjects/330_2006_214_Fig2_HTML.jpg
Fig. 2

The open 0.5-T MR imager (Signa-SP; General Electric Medical Systems, Milwaukee, Wisconsin). b: In the open magnet, the patient is placed prone on a dedicated phased-array breast coil, and the radiologist has direct access to the breast for needle placement

https://static-content.springer.com/image/art%3A10.1007%2Fs00330-006-0214-5/MediaObjects/330_2006_214_Fig3_HTML.jpg
Fig. 3

MRI-guided needle localization of the same patient with the suspicious enhancing lesion in the lower outer quadrant of the left breast by using a 0.5-T open MR imager. a) Axial T1-weighted fast spin echo image from the initial set of guidance images showed the fuducial marker and identification of the lesion’s top as a hypointense area (arrow). b) Axial T1-weighted fast spin echo image obtained during initial needle placement shows the tip of the needle posterior of the lesion. c) Axial T1-weighted fast spin echo image showing the needle direction after repositioning, the needle tip is advanced towards the mass. d) Gadolinium-enhanced axial three-point Dixon gradient-echo image shows the enhancing mass just approximate to the needle tip, before the wire is deployed

Analysis

All preprocedural diagnostic MR images were interpreted by experienced breast radiologists. According to the MRI Breast Imaging Reporting and Data System (BI-RADS) lexicon, lesions were scored on morphologic and kinetic characteristics and classified on a scale of 0–5 [18]. Numeric categories were the following: 0, needs additional imaging evaluation; 1, normal; 2, benign; 3, probably benign, recommend 6-month follow-up MRI; 4, suspicious; 5, highly suggestive of malignancy; 6, proven malignancy. Subsequently, maximum lesion size and location of the lesion in the breast was assessed. Because postprocedural MRI was not routinely performed during the study period but was ordered at the discretion of the attending surgeon, only postprocedural MRI examinations that were available were reviewed to assess lesion retrieval. Finally, medical records were reviewed to determine histologic findings and complications. The positive predictive value of MRI-guided needle localization was defined as the number of cancers found at MRI-guided needle localization divided by the total number of lesions that underwent MRI-guided needle localization. Imaging and histologic findings were considered concordant if the histologic findings provided a sufficient explanation for the imaging features. Statistical analyses were performed by using statistics software packet SPSS version 10.0.

Results

Between 1 January 1997 and 31 July 2004, a total of 304 lesions were localized in 220 patients. All patients had a diagnostic MRI performed prior to the date of biopsy at our clinic (Table 1). MRI-guided needle localization was performed for a single lesion in 150 patients, for two lesions in 56 patients, and for three lesions in 14 patients. The incidence of MRI-guided localizations performed at our clinic increased substantially each year (Table 2).
Table 1

Indication for magnetic resonance imaging (MRI) examination in 220 patients

 

Number

Increased breast cancer risk

48 (22%)

Positive family history

34

History of high-risk lesion or cancer

14

Problem solving

115 (52%)

Palpable abnormality

36

Questionable lesion mammogram/ultrasonography

70

Positive axillary node, unknown primary

9

Staging of known breast cancer

57 (26%)

Table 2

Number of magnetic resonance imaging (MRI)-guided needle localizations per year

Year

Number of needle localizations

1997

6

1998

10

1999

21

2000

29

2001

45

2002

63

2003

68

2004a

62

Total

304

aIn this year only the needle localizations performed between 1 January and 31 July (7 months) were included

In all patients the hookwire was placed through the breast lesion, with the tip of the wire within 1 cm of the edge of the lesion. Procedure time, from selection of skin entry site to final deployment of the wire, averaged 20 min. Histopathologic analysis showed 104 (34%) malignant lesions, 51 (17%) high-risk lesions, and 149 (49%) benign lesions. Of the 104 malignant lesions, 38 (37%) were invasive carcinoma (22 invasive ductal carcinoma, eight infiltrating lobular carcinoma, five tubular carcinoma, and three B-cell lymphoma), 36 (35%) were pure DCIS, and 30 (28%) were a combination of in situ and infiltrating carcinoma (Table 3). Of the 51 high-risk lesions, 22 (43%) were intraductal papilloma, 14 (28%) were atypical ductal hyperplasia, 11 (21%) were radial scar, and four (8%) were lobular carcinoma in situ (LCIS). Of the 149 benign lesions, 120 (81%) were focal fibrocystic change, 14 (9%) were fibroadenoma, ten (7%) were sclerosing adenosis, and five (3%) were lymph nodes (Table 3). Overall lesion size ranged from 2.0 to 65.0 mm (mean 11.2 mm).
Table 3

Histopathologic findings of the 304 breast lesions

 

Number

Malignant lesions

104 (34%)

Invasive ductal carcinoma

22

Infiltrating lobular carcinoma

8

Tubular carcinoma

5

Lymphoma

3

Pure ductal carcinoma in situ (DCIS)

36

Combination of DCIS and invasive carcinoma

30

High-risk lesions

51 (17%)

Intraductal papilloma

22

Atypical ductal hyperplasia

14

Radial scar

11

Lobular carcinoma in situ

4

Benign lesions

149 (49%)

Fibrocystic change

120

Fibroadenoma

14

Sclerosing adenosis

10

Lymph node

5

Total

304 (100%)

Overall positive predictive value of MRI-guided freehand needle localization technique was 34% (104/304). When patients were stratified according to their MRI examination indication, the highest proportion of malignant lesions was found in patients that underwent MRI because of staging for known breast cancer (n=57); 32/75 (43%) lesions proved to be malignant (Table 4). The highest proportion of benign lesions was found in high-risk patients (n=48) who were screened by MRI; in this group a total of 37/68 (55%) lesions proved to be benign (Table 4).
Table 4

Histopathologic findings stratified according to indication for magnetic resonance imaging (MRI) examination

 

High risk screening: 48 patients

Problem solving: 115 patients

Staging of known cancer: 57 patients

Total

Malignant lesions

22 (32%)

50 (31%)

32 (43%)

104 (34%)

High-risk lesions

9 (13%)

31 (19%)

11 (14%)

51 (17%)

Benign lesions

37 (55%)

80 (50%)

32 (43%)

149 (49%)

Total lesions

68 (100%)

161 (100%)

75 (100%)

304 (100%)

Follow-up of 82 patients with 104 malignant lesions showed that 77 patients (94%) were primarily treated by lumpectomy, and five (6%) underwent modified radical mastectomy. The lumpectomy specimens of 77 patients showed tumor-free margins on pathology in 33 (43%) and at least one margin with tumor invasion in 44 (57%) cases. Of the 33 patients with negative margins, nine (27%) underwent follow-up MRI, confirming lesion removal and no tumor recurrence in all cases over a period ranging from 8 to 61 months (mean MRI follow-up time: 23 months). Of the 44 patients with positive resection margins, 17 (39%) underwent additional reexcision with tumor-free margins on pathology, 17 (39%) underwent modified radical mastectomy, and ten (23%) were lost to follow-up. Of the 40 patients with 51 high-risk lesions on pathology, eight (20%) underwent follow-up MRI, confirming lesion removal in all cases. No other suspicious lesions were diagnosed within the follow-up period ranging from 6 to 70 months (mean 24 months). Of the 98 patients with 149 benign lesions on pathology, 37 (38%) underwent follow-up MRI imaging. Follow-up time ranged from 2 to 63 months (mean 19 months). MRI findings confirmed complete lesion removal in 35/37 (95%) of the patients but also showed that the lesion was missed in 2/37 (5%) patients. One of these patients underwent additional needle localization and surgical reexcision. Follow-up MRI in this case confirmed lesion retrieval, and pathology showed fibrocystic change. In the other case a conservative strategy was chosen, and follow-up MRI showed unchanged lesion size and shape for more than 12 months.

Discussion

MRI of the breast detects suspicious lesions that are clinically and mammographically occult [12, 13]. The capability to localize and biopsy these lesions is a crucial part of a breast MRI program [1922]. In this study we presented the results of MRI-guided needle localization of suspicious MRI-only lesions by using a freehand technique. Patients were placed prone on a platform-type breast coil in an open 0.5 T magnet. With the freehand technique the needle is incrementally advanced toward the target lesion; the number of manipulations required is dependent on the depth of the lesion in the breast and the density of the breast parenchyma.

Feedback of needle position relative to target lesion is acquired by sets of five contiguous rapid axial MR images obtained each time the needle is repositioned. With this method needles were successfully positioned in 304 breast lesions. Accuracy was confirmed by contrast-enhanced MRI, showing contact between the hookwire and the enhancing lesion in all cases. In total, 104 (34%) lesions proved to be malignant. This is in concordance with the results of previous mammographically, US-, or MRI-guided needle localization studies reporting breast cancers to be present in 12–55% of cases [1214, 23]. Of the 104 malignant lesions, 38 (37%) were invasive carcinoma, 36 (35%) were pure DCIS, and 30 (28%) were a combination of DCIS and invasive carcinoma. Compared with other investigations we had a relatively high proportion of DCIS lesions in our study [1114, 19, 24]. The most probable explanation for this discrepancy is the fact that most lesions in our study were MRI-detected-only lesions. The size of lesions localized in our study ranged from 2 mm to 65 mm (mean 11 mm), indicating that the spatial resolution was sufficient to allow successful preoperative wire localizations, even from small lesions with diameters of only 2 mm. In these cases the size of the MRI-compatible needle was as large as the lesion itself. It is therefore important during initial needle placement to stop just short of placing the needle in the lesion, to allow visualization of lesion enhancement after intravenous Gadolinium injection.

Since 1992 several MRI-guided needle localization techniques have been reported to allow histopathologic workup of suspicious MRI-detected breast lesions [1114, 19, 2426]. In general, the systems are based on the same principle: the breast is compressed in the sagittal plane by perforated compression plates (grid) that allow horizontal needle insertion. These grid techniques proved to have several limitations. The studies of Heywang-Kobrunner et al. and Orel et al. showed that by using a grid, several parts of the breast were difficult to access, especially posterior localized lesions near the chest wall and the axillary tail [11, 14]. Also, lesions localized in the retroareolar region were difficult to target because the anterior part of the breast could not be adequately stabilized. Another problem addressed by the same studies is the fact that any arrangement of holes within the grid limits access to areas located between the holes, which may be of special importance when localizing small lesions. Another shortcoming more recently stressed by a study of Morris et al. is that both breasts can, with most systems, only be approached from the lateral position, causing longer tissue penetration and suboptimal localization of lesions located medial in the breast [12]. Furthermore, Kuhl et al. reported that breast compression by the stereotactic system might interfere with contrast material enhancement of breast cancers. They described two breast cancer cases were enhancement was reduced to such a degree that cancer was indistinguishable from the remainder of the breast parenchyma whereas the normal contrast material enhancement of the tissue adjacent to the tumor was preserved [13].

In contrast to techniques that use a grid for needle localization, the freehand method allowed needle placement with a freely chosen insertion angle, enabling localization of lesions throughout the breast from near the chest wall to those superficially located near the nipple, as well as localization of lesions in patients with silicon implants [15]. Furthermore, because of the open magnet, needle localization is allowed without the need to move the patient in and out of the magnet. Direct monitoring of the procedure is possible, which increases accuracy because lesion shift caused by needle insertion, application of local anesthetic, or bleeding can be detected, and subsequent needle-tract adjustment can be made. Another benefit of the freehand technique with the patient in prone position is that the configuration of the breast during the procedure is comparable with the configuration of the breast during the preprocedural diagnostic MRI. This allows proper identification of the target lesion in the beginning of the needle localization on the nonenhanced T1-weighted images by using the breast architecture as map. The importance of the possibility of proper target-lesion identification is stressed by the results of previous studies showing that variation of lesion appearance after contrast enhancement can occur in patients between preprocedural diagnostic MRI and the images obtained during localization because of compression by the grid or varying hormonal and inflammatory changes [12, 27]. As a consequence, it is estimated that, in general, the localization procedure has to be canceled in 5% of patients due to changing enhancement of the lesion or surrounding breast tissue. Unfortunately, because of our retrospective study design and the lack of registration of unperformed procedures, we were unable to assess how many procedures were canceled in our patients during the time frame of our study.

To best utilize the time window between enhancement of the breast lesion compared with the surrounding parenchyma, all steps preceding initial needle placement must be performed prior to intravenous contrast injection. In two patients with benign findings on pathology, a follow-up MRI showed that the lesion localized with a wire using MRI guidance was not removed after surgical excision. It has been reported that guide wires may migrate before biopsy, and it is possible that the failure was due to guide wire migration.

Unfortunately, lack of Gadolinium-enhanced-specimen MRI techniques makes it difficult to confirm lesion retrieval after MRI-guided needle localization and surgical excision biopsy. Placement of an MRI-compatible and mammographically visible clip that can be left behind after needle localization have been suggested so specimen radiographs could document retrieval of the clip. Postoperative follow-up MRI can also be used to document successful lesions excision. In our study, follow-up MRI in 54 patients showed complete lesion removal in 52 patients whereas in two (3.7%), the lesion was missed. This is in accordance with failure rates for stereotactic, mammographically guided needle localization, ranging from 0–18% [23]. In a previous study on MRI-guided needle localization of 101 lesions by using a grid, Morris et al. reported a failure rate of 3% [12]. The reason why the lesion localized with a wire in these two cases was initially not removed after surgery remains unanswered. It has been reported that guide wires may migrate before excisional biopsy [13, 23]. We therefore postulated that the failure was due to guide-wire migration. Therefore, short term follow-up MRI 6 months or earlier after surgery is recommended to ensure lesion retrieval if MRI-guided localization yields benign findings or high-risk lesion [13, 28].

A possible disadvantage of the freehand technique is that only an open MRI system allows real-time monitoring of the procedure and appropriate insertion and adjustment of the guiding needle without the need to move the patient in and out of the magnet. As a consequence, an open MRI system has to be available in the clinic were the freehand technique is used for lesion localization. Today, open MRI systems become increasingly commercially available but are still associated with high costs, and lower field strengths resulting in less optimal image quality [29]. It is possible to perform MRI-guided needle localizations of suspicious breast lesions with the freehand technique in a closed-bore magnet, but then the patient has to be moved in and out of the scanner for needle insertion and position control, which results in increased procedure time. Another possible limitation of the freehand technique is possible motion of the target lesion in the noncompressed breast. However, in our study, substantial breast motion did not occur during the procedure because of the weight of the patient’s prone body fixes the chest wall to the coil platform. When less-critical breast motion did occur, the rapid interactive imaging approach allowed redirecting of the needle toward the lesion. We experienced that less-critical tissue motion typically occurred in women with dense fibroglandular breast tissue, and it could be hypothesized that these patients should preferably be localized with a grid system.

MRI-guided freehand needle localization is an easy to perform and accurate technique for localization of suspicious MRI-detected breast lesions. In our study, 34% of the lesions proved to be malignant, which is in concordance with the results of previously reported techniques using conventional, i.e., mammography or US guidance, or MRI-guided needle localization techniques that used a grid. A main advantage of the freehand technique in an open MRI system over methods that use a grid is that localization of lesions throughout the breast is allowed, including those located posterior near the chest wall or anterior in the retroareolar region.

Copyright information

© Springer-Verlag 2006