Magnetic resonance-guided, vacuum-assisted breast biopsy

Results from a European multicenter study of 538 lesions


  • Study coordinator: Sylvia H. Heywang-Kobrunner.



The objective of this study was to determine the accuracy, reproducibility, and clinical value of magnetic resonance (MR)-guided, vacuum-assisted breast biopsy (MR-VAB) in a prospective, multicenter study.


In 5 European centers, MR-VAB was performed or attempted on 538 suspicious lesions that were visible or could targeted only by MR imaging (MRI). Verification of malignant or borderline lesions included reexcision of the biopsy cavity. Benign biopsy results were verified by retrospective correlation of histology with preinterventional and postinterventional MRI studies. Follow-up of 24–48 months (median, 32 months) was available for 491 of 538 patients.


MR-VAB was unsuccessful or was not completed in 21 of 538 patients, for which an immediate repeat biopsy was recommended. Five hundred seventeen of 538 performed VAB procedures (96%) were successful. Histology yielded 138 (27%) malignancies, 17 (3%) atypical ductal hyperplasias, and 362 (70%) benign entities. No false-negative diagnoses occurred among the 517 successful MR-VAB procedures. The positive predictive value of VAB depended on patient preselection, which differed according to the indication for the initial MRI study.


The results of this study indicated that MR-VAB offers excellent accuracy. Small lesion size did not prove to be a limitation. Cancer 2006. © 2006 American Cancer Society.

Contrast-enhanced magnetic resonance imaging (CE-MRI) currently is the most sensitive additional imaging method for the detection of invasive breast carcinoma. Important additional information may be gained when MRI studies are performed for appropriate indications.1–11

Because of the limited specificity of CE-MRI, minimal invasive breast biopsy is especially important to avoid unnecessary surgical diagnostic biopsy. MR-guided percutaneous biopsy has become possible because of the development of special, so-called breast biopsy coils.12–15 However, MR-guided core-needle biopsy (CNB) has not proven to be sufficiently accurate and, thus, has not been recommended generally for the work-up of MR-detected lesions.16, 17 Compared with MR-guided CNB, MR-guided, vacuum-assisted biopsy (MR-VAB), which was introduced in 1997,18 offers several important advantages. Its main advantage concerns the acquisition of a larger tissue volume. This allows for a reduction in sampling error, which is important for the histopathologic diagnosis of small in situ malignancies or borderline lesions. Tissue shift by bleeding is avoided with an MR-VAB, because blood continuously is suctioned. Tissue shift during needle insertion also may be compensated, because a large area of tissue (1.5–2.0 cm in greatest dimension) is removed by VAB. Finally, correct biopsy results can be verified by direct visualization of lesion removal on the postinterventional images.

Until now, there have been no data on the reproducibility of MR-guided breast interventions in a multicenter setting from a large series of patients. In October 1998, a large multicenter study of MR-guided breast biopsy was started. The study was supported by the European Commission (Biomed-2 Project BMH 4-CT 98-3741). Interim results from that study were published in 2002.19 The current report summarizes the final results, including follow-up data on 538 MR-VAB interventions. The objectives of this multicenter study were to optimize the procedure20 and to assess the accuracy, reliability, and reproducibility of the method and its clinical value.



The selection criterion for the MR-guided biopsy was the presence of ≥ 1 suspicious lesions either that were visible by CE-MRI alone (Groups 2–4) or that could be localized in 3-dimensions by MRI only (Group 1). The patients were assigned to one of the following groups, according to the indication for vacuum biopsy and the indication of the initial MRI that showed the MR-detected lesion(s).

Group 1 included 6 patients who had large breast tumors after they received neoadjuvant chemotherapy. The tumor residuals were visible mammographically and by ultrasound. However, mammography and ultrasound cannot reliably distinguish between granulation tissue or scarring within these tumors and viable cells that persist. In theses patients, MR-VAB was required to target the contrast-enhancing areas, which presumably represented areas of viable tumor tissue. The remaining 113 lesions were suspected on one mammographic view only but could not be localized on further mammographic views or on repeat ultrasound studies. Thus, those 113 lesions could not be targeted 3-dimensionally by conventional imaging.

Group 2 included patients with lesions (usually additional indeterminate or suspicious foci) detected by MRI alone on an MR study that was obtained preoperatively for staging before planned breast-conserving therapy to exclude multicentricity in patients who had one mammographically or sonographically suspicious lesion. Only those additional foci were worked up, as that histology was important for treatment planning. For example, if multicentricity already was proven on one additional focus, then further foci were not biopsied, because the patient then was a candidate for mastectomy. In addition, foci close to the primary lesion usually were excised surgically together with the primary lesion; thus, those patients did not undergo MR-VAB or MR-guided wire localization.

Group 3 included patients with MR-detected lesions who underwent the initial MRI because of scarring. These included four subgroups, as follows: Group 3a, status postsurgery for breast carcinoma; Group 3b, diagnostic problems because of significant scarring after surgery for a benign lesion; Group 3c, diagnostic problems not associated with the scar; and Group 3d, patients with axillary metastases of suspected breast origin in whom mammography and ultrasound studies were negative.

Group 4 included patients with MR-detected lesions who underwent the initial MRI because of an increased risk based on family history (the lifetime risk for these patients was > 15–30% according to the model of Claus et al.21), or a history of histologically proven atypical ductal hyperplasia (ADH) or lobular carcinoma in situ (LCIS), or a history of contralateral breast carcinoma.

Recruited Patients and Reasons for Not Undergoing VAB

Initially, 578 patients with 649 lesions had been referred for MR-VAB. In 80 patients, the lesion could not be reproduced on the preinterventional planning MRI. Thus, those 80 patients did not undergo a biopsy. In most of these patients, the lesion disappeared because of hormone changes (changes in menstrual cycle; n = 43 patients) or withdrawal of hormone-replacement therapy for at least 4 weeks (n = 28 patients); in other patients, the lesion proved to be an artifact and, thus, did not require biopsy (n = 9 patients). In 10 of the remaining 569 indicated MR-VAB procedures, MR-VAB was not performed for reasons that were not associated to the biopsy coil or the procedure itself. These included breakdown of MR unit (n = 2 patients; this breakdowns concerned technical problems of the MR units that were not associated with the biopsy procedure), withdrawal of patient consent (n = 3 patients), severe artifacts because of patient motion (n = 3 patients) on the initial planning MRI sequences, and operator failure (n = 2 patients) concerning incorrect transfer of the indicated lesion coordinates to the aiming device.

In another 21 of 569 patients (3.6%), the procedure was not performed because of limitations associated with the procedure. Four patients were too large and did not fit into the magnet. In 11 patients, the lesion could not be approached by the biopsy needle because of problems of access. Access is limited in an area of 0–2 cm from the chest wall far medially, as reported previously.20 In addition, large breasts with lesions close to the nipple should not be positioned too deeply within the coil. The breast was too small to perform VAB in two patients. For VAB, a thickness of the compressed breast of at least 3 cm is necessary. For four lesions, compression that was too strong was considered responsible for the absence of enhancement, because those lesions were visible again on a follow-up MRI without compression. The 21 patients described above were rescheduled for MR-VAB using different patient positioning, less compression, and MR-guided wire localization or computed tomography (CT)-guided wire localization followed by surgery. The finding that compression that is too strong may suppress enhancement has been described previously.14 In the 21 patients who could not undergo the initial VAB, eventually, 8 carcinomas and 13 benign lesions were confirmed. Finally, 538 lesions were eligible for vacuum biopsy.

Biopsy Procedure

In all patients, VAB was performed under the condition that the coagulation parameters were normal. Patients provided informed consent before they underwent the procedure. The VAB was performed considering the optimum phase in the menstrual cycle (Days 7–17) or after hormone-replacement therapy was stopped for 4–6 weeks whenever this was possible and justified.

All interventions were performed either on an Impact Expert (1.0 T; Siemens, Erlangen, Germany) or a Vision (1.5 T; Siemens) MR scanner using the 11-G Mammotome® (Biopsys, Irvine, CA) VAB needle. Using the special biopsy device (Siemens and Epoxonic, Munich, Germany) the patient lies prone on top of the patient table, which includes cushions and the biopsy device. The breast is fixed in the mediolateral direction by two compression plates, which consist of flexible bars. These can be spread apart to allow optimum access of the biopsy probe from medially or from laterally at variable angles. The technical details and descriptions of the procedure of MR-VAB have been published previously.17–20 In this position, the patient's breast is scanned before and after intravenous injection of 0.15 mmol gadolinium-diethylene-triamine pentaacetic acid/kg (fast low-angle shot, 3-dimensional technique; TR, 14 msec; TE, 7 msec; flip angle, 25; slice thickness, 2 mm). Based on these preinterventional planning MRIs, the lesion that is to be biopsied is identified, and its coordinates are calculated. The calculated coordinates of the lesion are transmitted to the aiming device of the biopsy unit. After the patient receives skin anesthesia, the needle path is cut by a precutter according to these coordinates. The precutter is a very sharp knife that imitates a needle with a thickness of 4 mm. Then, a thin substitute needle is placed into this path. The substitute needle, which also is made out of titanium, can be imaged without artifacts. After the correct position of the substitute needle is verified, it is replaced by the biopsy needle; and, after the patient receives additional deep local anesthesia, VAB is performed outside the magnet, and at least 20 specimens are acquired. After the VAB is finished, the breast is reimaged before and after another contrast injection to check whether the intervention was successful.

Verification of Biopsy Results

Our standard procedure is to correlate the histologic results retrospectively with CE-MR imaging before and after MR-VAB. VAB was considered “successful” if the lesion was removed completely (Fig. 1) or partially (reduction ≥ 50%) according to the postinterventional MRIs and if the diagnosis of preinterventional and postinterventional imaging agreed with the histopathologic result of the VAB. If the lesion was visible with little or no change, then VAB was considered “unsuccessful,” and rebiopsy was recommended. In all patients in whom the assessment was considered impaired and histology was not obviously compatible with imaging, early follow-up (after 2–4 days), using the breast biopsy coil for imaging was recommended. This examination could be converted immediately into a repeat MR-VAB if the lesion proved to be persistent. If a clear assessment of complete or partial removal of the lesion was not possible (for example, in procedures in which a relevant amount of blood was seen in or around the cavity), but histology was compatible with the imaging results (e.g., imaging and histology showed a focal lesion of similar shape and size), then the result of VAB was considered “successful, but assessment impaired,” and follow-up after 6 months was recommended.

Figure 1.

These mammographic images were obtained from a patient 1 year after she received breast-conserving therapy for invasive ductal carcinoma. (A,B) Mammography shows a density (arrow) in the outer quadrant of the right breast that is visible on the craniocaudal mammogram (A) with a correlate in the mediolateral-oblique view (B). The preinterventional magnetic resonance image (MRI) reveals an enhancing lesion (arrow) within the scar that could not be differentiated from scar tissue on mammography. (C) A precontrast image with transverse orientation (acquired with a fast low-angle shot, 3-dimensional technique is shown; image parameters: TR, 14 msec; TE, 7 msec; flip angle, 25; slice thickness, 2 mm). (D) A postcontrast image that was obtained with the same acquisition technique by scanning after intravenous injection of 0.15 mmol gadmium-diethylenediamine tetraacetic acid/kg. (E) A subtraction image between the precontrast and postcontrast images. (F) The substitute needle has been advanced 2 cm beyond the lesion (arrow). This is necessary, because the substitute needle is as long as the biopsy needle, but the biopsy chamber is located 2 cm before the tip of the biopsy needle. With this technique, the lesion is centered at the acquisition chamber of the biopsy needle. On the precontrast scan that was obtained after the biopsy, the biopsy cavity with some bleeding is visible. The hypointense area within the biopsy cavity represents residual local anesthesia. (G) On the postcontrast image (H) and on the subtraction slice (I) that was obtained after the biopsy, no residual enhancement is seen, indicating complete macroscopic removal of the lesion. Histology revealed a nonspecific intramammary lymph node.

All patients who had histologic results of malignant tumor or ADH underwent surgical reexcision after sonographically, mammographically, or MR-guided wire localization of the cavity.19 All patients who had benign diagnoses by VAB were given a recommendation to undergo follow-up CE-MRI. In patients who had lesions that appeared completely removed or that corresponded to a specific histology (e.g., fibroadenoma or benign lymph node), the recommended follow-up was 12 months. In all other patients, follow-up was recommended at 6 months and 12 months. Some patients attended follow-up after longer intervals than recommended. Follow-up data were available for 491 of 538 biopsies (91%).


Technical Success

At the end of Year 1, several optimizations concerning technical improvements had been introduced (e.g. implementation of a medial access, verification of the distance between the bars of the compression paddle). These procedural optimizations were made together with the technology partner by using the input of all investigators and have been published previously in detail.20 Because of a problem of personnel, one institution where breast MRI was used in a very limited number of patients could not continue to recruit and, thus, stopped participating in the study.

According to the postinterventional scan and correlation of imaging and histopathology, 517 of 538 procedures (96%) were considered successful. In 21 of 538 procedures, MR-VAB was considered unsuccessful (n = 13 procedures) or uncertain (e.g. assessment impaired and histology not clearly compatible with imaging; n = 8 procedures). In these patients, immediate rebiopsy was recommended. The reasons for unsuccessful or uncertain biopsies included termination without a representative biopsy because of breakdown of the MR unit during the procedure (n = 1 procedure; defect of the positioning table), because of a technical problem of the biopsy probe (n = 1 procedure; defect of the vacuum suction), because of operator error (n = 3 procedures; wrong transfer of coordinates), and because of strong patient motion (n = 2 procedures). Bleeding during the intervention caused early termination in 2 procedures (without acquisition of sufficient tissue) or led to strongly impaired or impossible assessment of representative biopsy (n = 4 procedures). In two procedures, strong postinterventional enhancement impaired assessment, and histology was not clearly compatible. In three procedures, the lesion was displaced by bleeding and, thus, was not sampled. In another three procedures, the lesions were not sampled for unknown reasons. In the latter six patients, the postinterventional MRIs clearly showed that the enhancing lesion persisted beside the cavity. In all 21 procedures described above, rebiopsy (by MR-VAB, MR-guided or CT-guided needle localization, and surgery) was recommended. Unfortunately, one of the patients who received an explicit recommendation for further work-up was lost to follow-up (change of patient's address).

Assessment was considered impaired in another 12 successful MR-VAB procedures. These patients were followed, because histology was compatible with the preinterventional imaging. The assessment was impaired by bleeding (n = 10 patients) or by patient movement (n = 2 patients). To date, follow-up, which was available from 10 of 12 procedures, has shown that the lesions were sampled successfully. Two patients were lost to follow-up.

Overall, 52% of the lesions measured < 1 cm. The success rate (96% for lesions < 1 cm, 97% for lesions > 1 cm) did not depend on the lesion size. The results did not vary significantly between the participating institutions.

Clinical Results

The final histologies obtained in 517 successful MR-VAB procedures are listed in Table 1. All malignancies and borderline lesions were referred for surgical (re)excision. The diagnoses achieved by VAB were verified, yielding the exact same histopathologic result in 147 of 155 patients. Five of 17 patients with ADH were upgraded to ductal carcinoma in situ (DCIS), and 3 of 64 patients with DCIS (5%) were upgraded to invasive carcinoma. In 21 patients who underwent VAB procedures that revealed invasive carcinoma (n = 6 patients), DCIS (n = 13 patients), or ADH (n = 2 patients), no remaining malignancy or ADH was found within the reexcised tissue, suggesting that complete removal with VAB was achieved. All of these lesions measured < 10 mm.

Table 1. Final Histology Results from 517 Magnetic Resonance-Guided Vacuum-Assisted Breast Biopsy Procedures
Lesion/biopsy typeNo. of lesionsMean ± SD lesion size (mm)aNo. of lesions measuring ≤ 1 cma
  • SD: standard deviation; NA: not applicable.

  • a

    Lesion size was measured by magnetic resonance imaging; diffusely enhancing tissue (n = 42) was excluded from the calculation of size.

  • b

    The size of normal breast tissue was not applicable, because it usually is not indicated in the pathology report and was not correlated routinely in all institutions.

Invasive carcinoma7414 ± 731
Ductal carcinoma in situ6410 ± 436
Atypical ductal hyperplasia1710 ± 511
Proliferative fibrocystic changes15510 ± 689
Adenosis5911 ± 631
Fibroadenoma5710 ± 429
Papilloma1512 ± 77
Inflammation2410 ± 69
Lymph nodes89 ± 46
Normal breast tissueb25NANA
Other1911 ± 813

In the 362 lesions in which successful MR-VAB yielded benign results, no further histopathologic work-up was considered necessary. To date, the patients with 316 of those 362 benign lesions have been followed for 24–48 months. Forty-six lesions were lost to follow-up. To our knowledge, rebiopsies during follow-up only occurred in 3 followed patients because of some clinical, mammographic finding in the same quadrant (n = 1 patient) or because of a persistent part of the enhancing lesion that seemed to increase. Rebiopsy eventually confirmed benign, nontumerous changes, as verified by a previous MR-VAB procedure. To date, follow-up and (if performed) histology of rebiopsy confirmed the benign results of MR-VAB in these lesions.

In one patient, who had results that we did not count as false negative, rebiopsy of an enhancing residual benign lesion (a focal area of adenosis) was performed 1 year after MR-VAB and confirmed the above-mentioned diagnosis of MR-VAB. However, histology in this patient also revealed an incidental invasive carcinoma that measured 2–3 mm, which (in the described location) was occult to imaging. It was located at the margin of a large surgical specimen, remote from the site of VAB and remote from the benign enhancing residual, which was the reason for the rebiopsy.

Value of MRI and MR-VAB for Various Indication Groups

The results from MR-VAB were analyzed according to the various indication groups and are shown in Table 2. Indication Group 1 included 6 breast carcinomas after neoadjuvant chemotherapy as well as 113 lesions that had been visible on 1 mammographic view only with no clear correlate on ultrasound or on further mammographic views. Eighteen invasive carcinomas, 5 DCIS (20% malignancy), and 7 atypias (6%) were found among these 113 lesions. In the remaining 83 lesions (74%), benign changes were found. In Group 2, which included patients for preoperative staging, 36 additional contrast-enhancing lesions were found by MRI only. Invasive carcinoma was verified in 10 of those patients, DCIS was confirmed in 3 patients (36% malignancy), and atypia was confirmed in another 3 patients (8%). In the remaining 20 of 36 patients (56%), the additional lesion proved to be benign. In Group 3, a very high percentage of malignancy (42%; 32 of 77 lesions) was observed among patients after they received breast-conserving therapy. The same was true for lesions that were detected in patients who initially had been referred for exploration of an unknown primary tumor. In this indication group, 5 of 10 lesions (50%) proved malignant, yielding a high positive predictive value (PPV) of 50%. Among patients who underwent an MRI study to assess diagnostically disturbing scarring after benign surgery or surgeries and among patients who underwent the initial MRI for other diagnostic problems without a known increased risk of malignancy, the yield of malignancies (+ atypias) was somewhat lower: 8 of 53 malignancies (+ 0 atypias) = 15% or 16 of 80 malignancies (+ 1 atypia) = 20 (+ 1%), respectively. Indication Group 4 included lesions in high-risk patients (n = 59 patients), e.g. high family risk of breast carcinoma or proven BRCA1 or BRCA2 mutation, lesions in patients with atypias confirmed by previous surgery (n = 10 patients), or lesions in patients who had a history of contralateral breast carcinoma (n = 73 patients). Sixteen of 59 lesions (27%) in patients who had a high family risk proved malignant, and another 2 lesions were borderline (ADH; 3%). The remaining 41 lesions (70%) were proven benign. In 3 of 10 patients (30%) who underwent MRI because a previous biopsy had yielded a diagnosis of ADH or LCIS, malignancy was proven by MR-VAB; in 1 other patient (10%), ADH was verified. In 17 of 73 patients (23%) who underwent breast MRI after contralateral breast carcinoma, malignancy was confirmed by MR-VAB; in another 3 lesions (4%), atypia was detected, whereas 53 lesions (73%) proved to be benign.

Table 2. Percent of Lesions with Malignancy and Atypical Ductal Hyperplasia among Various Indication Groups (n = 517 lesions)
Indication groupNo. of lesionsResults (%)
MalignantMalignant and ADHBenign
  1. ADH: atypical ductal hyperplasia.

Group 1    
 After neoadjuvant chemotherapy6100
 Mammographically detected113202674
Group 2: Multicentricity (preoperative patients)36364456
Group 3: Diagnostic problem    
 After breast-conserving surgery77424258
 After benign surgery53151585
 Search for primary tumor10505050
Group 4: High genetic risk    
 Family risk59273070
 After ADH10304060
 After contralateral carcinoma73232773

Patient Tolerance and Side Effects

VAB revealed important side effects in only a few patients. A hematoma measuring > 3 cm occurred in 19 patients. Six patients had complications from bleeding, which could be stopped by a compression bandage in three patients. In one patient, a surgical suture was necessary; in two other patients, the hematoma had to be removed by surgical intervention; and, in two patients, bleeding led to an interruption of the biopsy. In 1 patient, infection developed 6 weeks after biopsy, and the patient had to be treated by antibiotics. In six patients, a vasovagal reaction was induced. One other patient was hospitalized because of hyperventilation after the examination. The majority of patients suffered more from positioning than from the biopsy itself, because they judged it uncomfortable to remain still in the prone position during the whole biopsy procedure, which, including preinterventional and postinterventional imaging, lasted for approximately 70 minutes for 1 lesion and 90 minutes for 2 lesions.


Because of the limited specificity of CE-MRI, the rate of malignancy varies between 15% and 50%. Among MR-detected lesions, percutaneous biopsy may help to avoid unnecessary surgery for benign MR-detected lesions.1

Since so-called “MR breast biopsy coils” have become available, both MR-guided wire localization and MR-guided CNB are feasible. MR-guided preoperative wire localization is available in many breast centers where CE-MRI is performed.

Although MR-guided CNB it is available, it has not become a routine procedure. The main problem with MR-guided CNB concerns the fact that the procedure cannot be monitored by MRI; because, in closed magnets, the lesion cannot be approached while the patient is in the bore, and even MR-compatible probes may cause significant artifacts that obscure the lesion. Thus, errors in targeting or lesion shifts during needle insertion may go undetected because of local anesthetics or bleeding. Furthermore, because of the small size of the tissue cores, in most patients, the biopsy site cannot be identified, even in retrospect. Because lesions that measure > 1 cm often can be detected by ultrasound in retrospect and then can be biopsied under ultrasound guidance, MR-guided percutaneous interventions would be highly desirable to assess lesions that measure < 1 cm. However, based on the above-mentioned problems, MR-guided CNB has not been recommended for the work-up of small lesions.16, 17

Based on its different principle, MR-VAB is suited better for the work-up of MR-detected lesions than MR-guided CNB. This is because of the acquisition of a larger tissue volume that promises to reduce sampling error, a feature that is particularly important in the diagnosis of early-stage breast carcinoma and its precursors. The possibility of acquiring a larger volume and the use of continuous suction may compensate for some inaccuracies inherent in MR targeting, which cannot be monitored directly in closed MR units. Finally, direct visualization of complete or partial removal of the enhancing lesion should allow an accurate assessment of the precision of the biopsy. To date, no large studies and no multicenter studies have been performed to test the reproducibility and the accuracy of MR-guided, minimally invasive interventions.

In the current study, despite an existing indication, MR-VAB was not performed in 10 patients for reasons that were not associated to the procedure itself. In another 21 patients, however, MR-VAB was indicated but could not be performed because of limitations associated with the procedure. Reasons for early termination included limitations that became obvious during or directly after positioning: The patient needs to fit into the magnet on top of the breast biopsy coil. Furthermore, there is a dead space of approximately 2 cm medially, close to the chest wall, that often may not be accessed medially, because it usually cannot be positioned deep enough in the breast coil to be reached by the biopsy needle. Finally, a minimum breast thickness of 3 cm is required to perform VAB (this also is a limitation of stereotactic VAB). In those patients for whom MR-VAB is not possible, MR-guided wire localization can be performed. The proportion of procedures that could not be performed because of limitations inherent in the procedure amounted to 3%.

Five hundred thirty-eight procedures could be performed, including 517 procedures (96%) that were successful. All unsuccessful or uncertain procedures could be identified according to the postinterventional imaging compared with histopathologic results. In these patients, rebiopsy was recommended. Using this careful approach, an excellent accuracy could be achieved in all patients who had MR-VAB procedures that were considered successful.

Benign diagnoses after a successful MR-VAB procedure proved very reliable (negative predictive value, > 99%) and correctly allowed the avoidance of surgery in the vast majority of patients who had benign lesions. Whenever malignancy was proven, optimized planning of the oncosurgical approach became possible. For example, among patients who had proven malignant lesions in different quadrants, mastectomy could be planned. If a second focus proved benign, then the patient underwent excision that was limited to the main focus.

Compared with open surgical biopsy, which has to be combined with MR-guided needle localization in MR-detected lesions, MR-VAB is less time consuming than the combined procedure (MR-guided wire localization and open surgery). However, whenever a single malignant or borderline lesion was confirmed by MR-VAB, excision of the cavity after VAB remained necessary; because, even if complete removal of an enhancing lesion was documented, residual microscopic or nonenhancing malignancy could not be excluded. The results of this study confirmed that only a few of the lesions were completely removed histologically. Therefore, surgical excision of all malignancies needs to be recommended.

An upgrade was noted in 5 of 17 patients who had a diagnosis of ADH; and, in 3 of 64 patients who had a diagnosis of DCIS (5%), an invasive carcinoma was found at reexcision. Upgrades after percutaneous biopsies are common because of sampling error and occur far more frequently with CNB than with VAB, which is also true in stereotactic percutaneous biopsies. The rate of upgrades of ADH to DCIS or invasive breast carcinoma in the current study with MR-VAB was similar to the rate reported for stereotactic VAB, which was approximately 24%.22, 23 The rate of upgrades from DCIS to invasive breast carcinoma in the current study of MR-VAB appeared to be lower than the rate reported for stereotactically guided VAB, which is approximately 12%.23

MR-VAB was tolerated fairly well. The number of significant side effects was low (bleeding, n = 8 patients; hematoma, n = 19 patients; infection, n = 1 patient; vasovagal reactions, n = 6 patients; hyperventilation episode, n = 1 patient). The most widely reported and sometimes severe side effect was neck pain from lying prone. Pain medication and sedation should be considered in patients who report problems lying prone or who have known problems of their cervical spine. No significant scarring (visible clinically or on mammography) was noted among the patients in this study.

When the use of MR-VAB was examined for the different indication groups, MR-VAB was helpful in solving the problem of an enhancing lesion in all indication groups. If a benign lesion was confirmed by a successful MR-VAB, then surgical biopsy no longer was considered necessary. Whenever a malignant or borderline lesion was documented, surgical excision remained necessary; however, in numerous lesions, the definitive surgical approach concerning the extent of surgery could be optimized by proving or excluding multicentricity. When comparing the different prevalence of malignancy among the enhancing lesions in the different indication groups, the usefulness of the initial MRI appeared to vary with the indication. MRI may be quite useful for detecting a small unsuspected malignancy, an unsuspected recurrence, a second focus, or an unexpectedly large extent of malignancy. Furthermore, MRI may be helpful in confirming the absence of a malignancy in mammographically and sonographically difficult tissue. Conversely, the detection of large numbers of enhancing benign lesions (false-positive results) may pose a significant problem. When comparing the indications for the initial MRI, the greatest proportion of malignancy among enhancing lesions occurred in patients who underwent MRI to search for a primary tumor (50%), followed by the indications status post breast-conserving therapy (42%), preoperative staging (36%), a history of histologically proven ADH (30%), family risk (27%), history of contralateral breast carcinoma (23%), mammographically detected architectural distortion (20%), and impaired assessment after benign surgery (15%). Considering the decreasing prevalence of malignancy among patients with those indications, it appears to be sensible predominantly to use CE-MRI in those patients who have diagnostic problems associated with a significant risk of malignancy. If an enhancing lesion is detected by CE-MRI, then MR-VAB appears to be useful for most of the indications described above, because follow-up may be associated with the risk of missing an early malignancy. Knowledge of the varying prevalence of malignancy in the individual patient who has an indeterminate lesion may support a decision for an intervention (in patients with high-risk indications), or it may support the decision for follow-up (in patients with low-risk indications).

Overall, MR-VAB appears to be a valuable new alternative for the work-up of MR-detected lesions. With MR-VAB, surgical biopsy may be avoided by the majority of patients who have benign lesions, and definitive therapy may be optimized. The current results show that MR-VAB is a reliable method, although it may not be feasible in some patients (who cannot lie still, etc.) or in certain locations (breast thickness too small, one area medially close to the chest wall). When it is performed diligently with a critical review of preinterventional and postinterventional images, including correlation with histology, the accuracy of MR-VAB is at least comparable to the accuracy of MR-guided localization followed by surgery.16, 17, 24 The accuracy of MR-VAB is not limited by lesion size. Even very small lesions can be diagnosed accurately. The method proved to be reproducible at institutions that had sufficient training and experience. The fact that MR-VAB can be concentrated in highly specialized centers where indeterminate lesions can be worked up may be an advantage. After histologic diagnosis of the MR-detected lesion is available, the patients either may go back their referring physician or can be treated definitely in a small or midsized breast center. Although MR-VAB offers a new alternative for the work-up of MR-detected lesions, a sensible selection of the indications for which MRI may be useful remains important to limit the number of lesions that require further work-up and, thus, lead to additional costs and unnecessary patient anxiety.