Pathologic correlates of false positive breast magnetic resonance imaging findings: which lesions warrant biopsy?

The American Journal of Surgery 190 (2005) 633– 640

Presentation

Pathologic correlates of false positive breast magnetic resonance imaging findings: which lesions warrant biopsy? Samantha A. Langer, M.D.a, Kathleen C. Horst, M.D.b, Debra M. Ikeda, M.D.c, Bruce L. Daniel, M.D.c, Christina S. Kong, M.D.d, Frederick M. Dirbas, M.D.a,* b

a Department of Surgery, Stanford Cancer Center and Stanford University School of Medicine, Stanford, CA 94305, USA Department of Radiation Oncology, Stanford Cancer Center and Stanford University School of Medicine, Stanford, CA 94305, USA c Department of Radiology, Stanford Cancer Center and Stanford University School of Medicine, Stanford, CA 94305, USA d Department of Pathology, Stanford Cancer Center and Stanford University School of Medicine, Stanford, CA 94305, USA

Manuscript received June 5, 2005; revised manuscript June 10, 2005 Presented at the Sixth Annual Meeting of the American Society of Breast Surgeons, Los Angeles, California, March 16 –20, 2005

Abstract Background: Contrast-enhanced breast magnetic resonance imaging (MRI) is highly sensitive for breast cancer. However, adoption of breast MRI is hampered by frequent false positive (FP) findings. Though ultimately proven benign, these suspicious findings require biopsy due to abnormal morphology and/or kinetic enhancement curves that simulate malignancy on MRI. We hypothesized that analysis of a series of FP MRI findings could reveal a pattern of association between certain “suspicious” lesions and benign disease that might help avoid unnecessary biopsy of such lesions in the future. Methods: A retrospective chart review identified women undergoing breast MRI between June 1995 and March 2002 with FP findings identified by MRI alone. Lesions were retrospectively characterized according to an MRI Breast Imaging–Reporting and Data System lexicon and matched to pathology. Results: Twenty-two women were identified with 29 FP lesions. Morphology revealed 1 focus (3.5%), 5 masses less than 5 mm (17%), 11 masses greater than 5 mm (38%), 1 (3.5%) linear enhancement, and 11 (38%) non–mass-like enhancement. Kinetic curves were suspicious in 15 (52%). Histology demonstrated 20 (69%) variants of normal tissue and 9 (31%) benign masses. MRI lesions less than 5 mm (n ⫽ 6, 20.5%) were small, well-delineated nodules of benign breast tissue. Conclusion: Suspicious MRI lesions less than 5 mm often represent benign breast tissue and could potentially undergo surveillance instead of biopsy. © 2005 Excerpta Medica Inc. All rights reserved. Keywords: Breast magnetic resonance imaging; False positive breast MRI findings; MRI-guided biopsy

Contrast-enhanced breast magnetic resonance imaging (MRI) is a highly sensitive test for identifying breast carcinoma and, to a lesser extent, ductal carcinoma in situ (DCIS). Further, breast MRI is able to demonstrate malignancy invisible to clinical examination, mammography, and ultrasound. Therefore, MRI is now used in a variety of clinical settings to supplement, and in some instances to precede, more traditional methods of breast evaluation

* Corresponding author. Stanford Cancer Center, 875 Blake Wilbur Dr., CC2235, Stanford, CA 94305. Tel.: ⫹1-650-723-6779; fax: ⫹1-650724-4137. E-mail address: [email protected]

[1–3]. Principal criteria for identifying suspicious lesions with MRI are lesion morphology and enhancement kinetics. Breast MRI is emerging as the preferred imaging study for screening women at very high risk for developing breast cancer and for women with axillary metastases with unknown primary breast lesions [2]. Nevertheless, clinical benefit in other settings is more variable. Reasons commonly cited for hesitancy in administering the test are high cost, lack of diagnostic information beyond that obtainable with a clinical breast examination and conventional imaging, and frequent false positive (FP) findings. FP findings are particularly problematic and significantly offset potential benefits from breast MRI. FP findings prolong patient evaluation and increase pa-

0002-9610/05/$ – see front matter © 2005 Excerpta Medica Inc. All rights reserved. doi:10.1016/j.amjsurg.2005.06.030

634

S.A. Langer et al. / The American Journal of Surgery 190 (2005) 633– 640

tient anxiety. They also typically lead to additional imaging studies, including ultrasound and repeat MRI. Moreover, they may lead to unnecessary biopsies and, in some situations, may even lead to unnecessary mastectomies. If FP findings were reduced, MRI would be a more useful adjunctive diagnostic imaging tool. This study retrospectively evaluated a cohort of women who had abnormalities demonstrated solely on MRI that were subsequently found to represent benign processes by surgical biopsy. Our goal was to carefully correlate these false positive MRI findings with histology. The specific aim was to define a pattern of morphologic characteristics and kinetic enhancement curves that might more accurately have predicted the benign processes that were eventually found. The hypothesis was that by analyzing a series of FP MRI findings we would be able to reveal an association between certain “suspicious” lesions based on morphology and kinetic enhancement and benign breast disease. If this were demonstrated, lesions with such MRI characteristics could potentially be approached with close surveillance rather than biopsy in the future.

Methods Identification of patient cohort We performed an institutional review board–approved, retrospective, chart review within a single surgeon’s practice between June 1995 and March 2002 to identify women who underwent breast MRI. Once they were identified, their individual clinical breast examinations and matched imaging studies were carefully reviewed to determine the subset of women who had a breast lesion detected solely by MRI that was interpreted as suspicious but was later proven to represent benign histology by surgical biopsy. These defined our FP findings. Once this cohort was identified, histories were reviewed in detail for chief complaint, physical findings, results from any and all breast imaging studies, diagnostic procedures, surgical pathology, and clinical follow-up. The more detailed review allowed us to confirm that women in the study group had a likely benign ipsilateral clinical breast examination, standard 2-view screening or diagnostic mammography, and focused ultrasound (if performed) with MRI alone responsible for identifying a suspect lesion. Because our intent was to determine the full clinical impact of administering breast MRI, women were also included in the study even if the MRI suggested a benign finding and the patient requested a biopsy of this previously occult lesion. Biopsy results were carefully reviewed to confirm the absence of atypical hyperplasia, carcinoma-in-situ (CIS), or other suspicious/malignant findings to exclude women whose MRI-guided biopsy results might have altered management recommendations, such as initiation of tamoxifen.

MRI MRI was performed in a 1.5-T scanner (Echospeed; GE Medical Systems, Milwaukee, WI). All images were obtained in the prone position. Since the study spanned a 7-year period, there were slight alterations in equipment and technique. Examinations were performed with non-phased array surface coil (Medrad, Pittsburgh, PA), and with a phased array breast coil (MRI Devices, Waukesha, WI). Whole breast rapid dynamic MR images and high-spatialresolution fat-nulled MR images were contemporaneously acquired by using a combination of dynamic 3-dimensional (3D) spiral MRI to obtain initial kinetic enhancement curves during the wash-in phase of gadolinium (GD) contrast. Intravenous GD (Gadoteridol, Bracco Diagnostics, Princeton, NY or Magnevist, Berlex, Berlin, Germany) at a dose of 0.1 mmol/kg was infused as a rapid bolus at a rate of 2 to 3 mL/s using a power injector (Spectris, Medrad). Immediately following this injection, high-spatial-resolution transfer (3DSSMT) imaging was done to collect information regarding morphology followed by additional dynamic 3D spiral MRI to obtain delayed kinetic enhancement curves. All MRI scans in this report were performed prior to the development of a standardized reporting system. Accordingly, it was necessary to retrospectively review all MRI scans and classify lesions using a standardized lexicon. We used a modification of the recently developed, American College of Radiology (ACR) MRI Breast Imaging–Reporting and Data System (BI-RADS) lexicon [4]. MRI findings were retrospectively classified into 5 morphologic subtypes: foci (entities ⬍2 mm), masses less than 5 mm, masses ⱖ5 mm, linear enhancement, and non–mass-like enhancement. Kinetic enhancement curves were coded based on initial and delayed enhancement patterns. These curves were classified as highly suspicious (rapid initial uptake with washout or plateau), indeterminate suspicion (rapid initial uptake with sustained late phase), or low suspicion (slow uptake with a persistent late phase). At a minimum, each patient discussed the suspicious findings seen on breast MRI with their surgeon (F.M.D.). Other members of each patient’s health care team participated in this decision making process to a greater or lesser degree. In the early part of the study period, an MRI wirelocalized surgical biopsy was the initial management recommendation for lesions interpreted as indeterminate or suspicious by radiologists specializing in breast MRI. During the latter part of the study period, we recommended that suspicious MRI lesions be evaluated with focused ultrasound prior to MRI-guided wire-localized biopsy. Others have shown that focused ultrasound can often identify discrete MRI lesions [5]. These can then be managed more expeditiously with ultrasound-guided percutaneous biopsy or ultrasound-guided wire localization biopsy. Women in the current study had lesions that were not seen on ultrasound and, accordingly, were encouraged to pursue MRI-

S.A. Langer et al. / The American Journal of Surgery 190 (2005) 633– 640 Table 1 Demographics Total patients No. of patients with FP MRI No. of MRI guided biopsies Mean age, years (range) Family history of cancer Patients with personal history of cancer

635

less than 5 mm were subsequently re-reviewed by the study pathologist. 110 22 (25%) 29 49 (24–69) 3 (14%) 11 (52%)

guided wire-localized surgical biopsy to exclude malignancy. In all women, the workup based on clinical examination, imaging, and percutaneous biopsy was completed to the greatest extent possible before surgical intervention. MRI wire-localized biopsy technique and excision All patients were informed of risks related to MRIguided wire placement. Free-hand, sterile, wire localization was performed in an open MRI magnet, 0.5-T Signa-SP scanner (GE Medical, Milwaukee, WI) in the prone position. Preliminary T1-weighted fast spin-echo images were taken with a fiducial marker adherent to the breast. MRIcompatible needles were then placed adjacent to the area of concern. Enhancement in the localization was confirmed with the intravenous administration of GD contrast. Once confirmed, 0.2 mL of methylene blue dye (Mayne Pharmaceutical Inc., Paramus, NJ) was instilled, followed by the placement of an MRI-compatible hook wire (E-Z-EM Inc., Westbury, NY). Conventional 2-view mammograms were performed after hook-wire placement to assist with surgical excision. Most excisional biopsies were performed using local anesthesia with sedation. Specimen radiographs were taken at the time of operation to confirm the removal of the entire hook wire. Immediate postoperative MRI was not routinely obtained to confirm removal of the targeted lesion. Tissue analysis All biopsy specimens underwent gross and microscopic evaluation using standard tissue processing techniques. Specimens measuring up to 4.5 cm were submitted completely for histologic examination, while those measuring larger than 4.5 cm were sampled but not initially submitted in toto. Gross examination of the specimens was performed by trained pathology assistants or by surgical pathology residents. Tissue was fixed in formalin then serially sectioned at 3-mm intervals prior to obtaining specimen radiographs. These radiographs guided specimen sampling in cases where the material was not completely submitted for processing. Hematoxylin and eosin stained slides of 4-␮m thick sections were prepared from the paraffin-embedded tissue blocks. Cases were signed out by general surgical pathologists. Slides from MRI-detected lesions measuring

Results Between June 1995 and March 2002, 110 women in a single surgeon’s practice underwent 144 consecutive MRI scans. In this study, 33 (23%) MRI scans revealed true positive lesions (risk lesions, CIS, or invasive carcinoma); 76 (53%) MRI scans were truly negative based on histology or follow-up; 32 (22%) MRI studies demonstrated lesions that were ultimately found to be FP findings; 3 (2%) MRI scans missed confirmed DCIS or invasive carcinoma, detected by either clinical breast examination or conventional imaging, and therefore were considered to be false-negative studies. Of the 32 FP MRI studies, we identified 22 scans performed on 22 patients in whom the MRI was the only modality that identified a finding resulting in biopsy. This cohort of 22 women (20%) (mean age 49 years, range 24 – 69 years), the focus of this study, were noted to have 29 lesions (range 1–3 lesions per patient). These patients underwent MRI-directed hook-wire localized excisional biopsy that subsequently demonstrated benign breast tissue. In all cases, the breast MRI was the only modality that identified a suspicious lesion or a probably benign lesion for which the patient requested excision (Table 1). Various physicians from the patients’ health care team ordered MRI scans. Indications for obtaining breast MRI (Table 2) included patient request, high-risk patient screening, axillary metastasis with unknown primary, or the local staging of known breast malignancy. Biopsies were recommended for lesions seen on MRI predominantly because of suspicious or indeterminate kinetic curves in 23 (80%) lesions, suspicious morphology in the absence of abnormal kinetic enhancement curves in 2 (7%) lesions, probably benign lesions associated with suspicious MRI findings elsewhere in the same scan for 3 (10%) lesions, and probably benign findings in the setting of a strong family history of breast cancer in 1 (3%) lesion. Standard 2-view screening or diagnostic mammography was performed on 21 women (96%). One woman declined

Table 2 Indication for MRI Indication

No. (%)

Patient request Vague physical findings Vague imaging findings Personal history of cancer Family history of cancer BRCA gene mutation Unknown primary Nipple discharge Suspicious fine-needle aspiration

8 (36%) 4 (18%) 4 (18%) 0 1 (4.5%) 1 (4.5%) 1 (4.5%) 2 (9%) 1 (4.5%)

636

S.A. Langer et al. / The American Journal of Surgery 190 (2005) 633– 640 Table 5 False positive results

Table 3 Pre-MRI workup Type

No. (%)

Physical exam, mammogram, ultrasound Physical exam and mammogram Physical exam and ultrasound

14 (64) 7 (32) 1 (4)

because of concern over possible injury to her saline implants. Focused ultrasound was performed either before or after MRI in 15 women (68%) (Table 3). As noted earlier, ultrasound did not identify any lesions seen on MRI in our current cohort and was a criterion for entry in this study. A retrospective review and classification of the 29 lesions seen on MRI according to a modified ACR MRI BI-RADS lexicon revealed a spectrum of findings. Five morphologic subtypes were observed: a focus in 1 (3.5%), mass less than 5 mm in 5 (17%), mass ⱖ5 mm in 11 (38%), linear enhancement in 1 (3.5%), and non-mass enhancement in 11 (38%). Kinetic enhancement curves of the 29 lesions were coded as highly suspicious in 15 (52%) cases, of indeterminate suspicion in 8 (28%), and of low suspicion in 6 (21%) (Table 4). Histologic findings of the 29 lesions revealed variants of normal breast tissue in 20 (69%) (1 proliferative fibrocystic change [PFCC], 15 nonproliferative fibrocystic change [NPFCC], 1 lactational changes, 1 sclerosing adenosis, 1 benign duct with calcification, 1 dense fibro-connective tissue) and 9 (31%) specific benign masses (1 lymph node, 4 papillomas, 3 post-biopsy changes, 1 radial scar). MRI findings and tissue correlates are summarized in Table 5. Using a modification of the ACR MRI BI-RADS classification scheme, lesions were then divided into 4 groups: lesions measuring less than 5 mm (n ⫽ 6, 20.5%); lesions ⱖ5 mm (n ⫽ 11, 38%); linear/ductal enhancement (n ⫽ 1, 3.5%); and non-mass enhancement (n ⫽ 11, 38%). Biopsy of the 6 lesions (foci, n ⫽ 1; and mass ⬍5 mm, n ⫽ 5) associated with highly suspicious enhancement curves revealed simple variants of normal breast tissue. In contrast, 6 (55%) of the 11 lesions ⱖ5 mm associated with variable enhancement curves were discrete benign entities. Only a single finding fell in the linear enhancement group. Table 4 False positive MRI findings No. (%) Morphology Focus Mass ⬍5 mm Mass ⱖ5 mm Non–mass-like enhancement Linear enhancement Kinetic enhancement curves High suspicion Indeterminate Low suspicion

1 (3.5) 5 (17) 11 (38) 11 (38) 1 (3.5) 15 (52) 8 (28) 6 (21)

Morphologic type

No. (%)

Kinetic suspicion

Tissue type 1 nonproliferative fibrocystic change 3 nonproliferative fibrocystic change 1 proliferative fibrocystic change 1 benign duct calcification 2 nonproliferative fibrocystic change 3 nonproliferative fibrocystic change 1 benign lymph node 1 radial scar 2 papilloma 2 papilloma 1 sclerosing adenosis 2 nonproliferative fibrocystic change 3 nonproliferative fibrocystic change 1 nonproliferative fibrocystic change 3 post-biopsy changes 1 dense fibroconnective tissue 1 lactational change

Focus

1 (3.5)

High

Mass ⬍5 mm

5 (17)

High High High

Mass ⱖ5 mm

11 (38)

Low High

Linear Non–mass-like enhancement

1 (3.5) 11 (38)

Low High High Indeterminate High Low Indeterminate High Indeterminate High

Indeterminate

The 11 lesions with non-mass enhancement had heterogeneous kinetic enhancement curves, rather than a consistent pattern. Three (27%) of these lesions were associated with a discrete benign entities. There was no correlation between detection of these lesions and time points within the study. In contrast, of the 33 MRI scans performed on 24 patients that revealed true positive findings, 21 (88%) of 24 patients were noted to have physical examination findings suggestive of malignancy. Fourteen (67%) of these 21 patients were noted to have a palpable breast mass or abnormal skin changes. Seven of these 21 (33%) patients were noted to have a biopsy-confirmed, pathologic axillary lymph node in the absence of a palpable breast mass. Of the 7 patients with a pathologic lymph node and positive MRI finding, a primary breast lesion was also identified by either mammography or ultrasound in 4 (57%). In the remaining 3 patients, MRI alone visualized the occult primary lesion; these averaged 2.4 cm (range .8 to 4 cm). Three of the 24 patients with true positive MRI findings (13%) were noted to have normal physical examination findings. Two patients had an abnormal mammogram and ultrasound in addition to MRI. One patient, a known BRCA1 carrier, was found to have a .5 cm abnormality by MRI alone that was ultimately found associated with high-grade DCIS. Of these 24 patients with true positive MRI findings, a

S.A. Langer et al. / The American Journal of Surgery 190 (2005) 633– 640

single patient had an MRI finding less than 5 mm that was found to represent invasive breast cancer. This patient, who had metastatic cancer in an ipsilateral axillary lymph node, underwent MRI early in our experience and did not have a focused ultrasound examination following detection of the lesion with MRI. Accordingly, we do not know whether this lesion was apparent on MRI alone.

Comments The majority of women diagnosed with breast cancer initially present with an abnormal finding on either clinical breast examination or mammography. The ratio of malignancies detected by imaging alone compared with clinical breast examination has risen dramatically in the 4 decades since the development of mammography and the introduction of regular mammographic screening. Randomized trials have found that mammographic screening of women between 40 and 70 years of age reduces mortality by 25% to 63%. This has led the American Cancer Society to recommend annual screening of women over the age of 40 years [6,7]. Although mammography can successfully detect occult malignancy, it also produces many FP findings and can lead to benign biopsies. While results vary between institutions, approximately 25% to 29% of mammographically directed surgical biopsies are positive for cancer, while 71% to 75% show variants of benign tissue [8]. This leads to claims that the greatest risk from screening mammography is unnecessary biopsy [9]. Breast MRI was introduced as an adjunct to physical examination and mammographic screening approximately 15 years ago. It was assumed that this new imaging modality would not only improve mammographic sensitivity, but would also improve specificity. In fact, MRI identifies malignancy invisible to mammography in 16% to 29% of cases [10]. Currently, MRI is considered the best screening tool available for women under the age of 40 with a genetic predisposition to breast cancer [11]. MRI has also become the procedure of choice for women with axillary metastases of unknown primary [3]. In this setting, MRI can identify occult primary lesions and facilitate breast conservation while avoiding unnecessary mastectomy. While other routine indications are less clear, MRI has sometimes been used to identify vague but suspicious lesions encountered on physical examination or conventional imaging; to facilitate breast conservation efforts of early breast cancer prior to initial excision or re-excision; to monitor response to neoadjuvant chemotherapy; to guide surgical excision of locally advanced tumors; and to assess chest wall involvement in patients who have had recurrences after mastectomy. As with mammography, however, MRI is associated with many FP findings. Not only do FP findings extend the length of the workup, but they can lead to additional imaging studies and increased patient anxiety. More importantly,

637

FP MRI findings may lead to unnecessary biopsies and, on occasion, even to unnecessary mastectomy. The latter situation may be more likely with anxious patients or if interpreted by physicians less accustomed to incorporating breast MRI into their practice. While results vary between institutions and among patient populations, recent studies indicate that benign findings occur in approximately 62% of MRI-directed biopsy cases [12,13]. The combination of high cost paired with modest test specificity has been a major contributor to the slow adoption of MRI into clinical practice. In attempt to clarify features of benign and malignant lesions on MRI and thereby reduce FP findings, investigators have looked prospectively at both lesion architecture and kinetic enhancement curves. Architectural features were first used to predict histology [14,15]. Classifying regions of interest based on time-signal intensity curves further improved prediction of histology [16,17]. Combining these 2 descriptive methods, architecture and enhancement kinetics, with standardizing terminology through a new lexicon, has led to the development of an MRI BI-RADS classification scheme [4,15,18,19]. The goal of this new framework is to standardize the reporting system among institutions. To better understand the sensitivity and specificity achieved at our institution, we undertook a retrospective study of consecutive series of women undergoing breast MRI over a 7-year period. This cohort included some of the earliest efforts at utilizing breast MRI for both imaging and wire localization prior to surgical biopsy. In this series of 110 women we found FP findings in 20% with an overall sensitivity of 92% and specificity of 70% [20]. Recognizing the deleterious consequences of false positive studies in our sample, as well as that of others, we questioned what steps could be taken to minimize these FP findings. Accordingly, we analyzed the MRI findings of all women with negative results on open surgical biopsy. The aim was to identify consistent morphologic features or kinetic enhancement patterns among our known FP lesions in order to identify similar lesions in the future that could be observed rather than undergo biopsy. Findings from this exploratory study suggest that suspicious lesions measuring less than 5 mm identified only by breast MRI are often associated with benign variants of breast tissue, even if the MRI enhancement pattern is highly suspicious. While these lesions did not represent discrete histologic entities that matched the discrete MRI findings in this group, they were characteristically discrete nodules of fibrocystic change that were surrounded by fatty tissue. In contrast, 9 (39%) FP lesions measuring ⱖ5 mm in size correlated to a discrete benign entity, such as a radial scar, papilloma, or lymph node, or prior biopsy changes. Normally, it is not uncommon for these types of previously unrecognized, clinical findings to undergo biopsy when discovered within the breast by physical examination, mammography, or ultrasound. We identified only 1 lesion in the ductal/linear enhancement group and therefore no prediction can be made regarding management with observation versus biopsy for

638

S.A. Langer et al. / The American Journal of Surgery 190 (2005) 633– 640

this class. Among the group of lesions with non-mass enhancement, there was great heterogeneity among lesions, and for this group as well no definitive recommendations can be made regarding observation versus biopsy. When comparing the 22 (20%) patients with FP MRI findings to patients to the 24 (22%) women with true positive MRI findings we found appreciable differences in clinical examination findings as well as conventional imaging studies. In the majority of patients (96%) with true positive MRI findings, an abnormality was noted by physical examination, mammography, or ultrasound or by a combination of modalities prior to the MRI. Additionally, the average true positive MRI abnormality measured 3.1 cm (range .5–10 cm). In only 4 instances did the true positive MRI lesion measure 10 mm or less. In 1 case, the lesion measured 5 mm: this patient had a pathologic ipsilateral axillary node. This patient was not evaluated by ultrasound before or after MRI and therefore we can not be sure that the abnormal lesion was detected by MRI alone. Recognizing the dilemma in establishing cut-offs for observation versus biopsy, Liberman et al reviewed a series of 89 high-risk women who had lesions identified on breast MRI that were interpreted as probably benign and who initially underwent observation rather than biopsy [21]. This study demonstrated that 7% to 10% of women were subsequently found to have DCIS or invasive ductal carcinoma. However, if all women with benign MRI findings underwent biopsy, then the proportion of women requiring biopsy would have increased substantially from 16% to 40%. Additionally, this would have decreased the positive predictive value of breast MRI in this cohort from 24% to 16%. These authors echo our concern that establishing rigid management guidelines for or against routine biopsy of lower risk MRI lesions may lead to decreased sensitivity and ultimately lead to a delay in diagnosis of breast cancer. Other investigators have suggested that suspicious, small lesions seen on MRI may have a lower likelihood of representing a malignancy than previously thought. Siegmann et al demonstrated that the positive predictive value of MRI for lesions ⱕ1 cm was 27.6%, compared with a positive predictive value of 45.5% for lesions greater than 1 cm [12]. Gibbs et al reported that the specificity of MRI in 17 lesions less than 1 cm was only 41%. In contrast, the specificity of ultrasound for the 11 lesions in this group that were studied with ultrasound was 91% [22]. This latter observation supports our recommendation, as well as that of others, that suspicious discrete MRI findings be evaluated with subsequent focused ultrasound in order to help guide management. There are several potential explanations why the lesions we identified as being smaller than 5 mm were found to be associated with discrete nodules of benign breast tissue despite suspicious enhancement curves. It is possible that at the time of wire placement by the radiology team under MRI guidance, the lesion was not localized properly for surgical removal [23]. The potential for improper localization is increased with MRI-guided wire placement as com-

pared to stereotactic guided wire placement. Radiologists placing the MRI guidewire have to stop just short of placing the needle into the lesion to prevent wire induced MRI artifact from obscuring the targeted lesion on confirmatory MRI images. However, this possible source of error seems unlikely since MR images obtained immediately after the guidewire placement confirmed successful wire placement and localization. Following MRI-guided wire placement, all patients undergo 2-view mammograms to assist with surgical planning for excision. This extra step could potentially result in wire migration and ultimately interfere with accurate localization. Should the wire be pushed into or pulled out of the breast, slightly altering the relationship of the wire to the lesion, this would not be recognized by the radiology team since the mammogram cannot visualize a lesion detected by MRI only. Again, this seems unlikely as it is standard practice for the radiology team to obtain multiple confirmatory mammographic images following the more commonly performed stereotactic guided wire localizations. This type of wire migration is not typically encountered. Once the wire is placed, another possible source of error might occur in the operating room. It is possible that the surgical team did not successfully remove the wire with adequate surrounding breast tissue containing the less than 5 mm lesion. Since it is not possible to perform a MRI specimen image (there is no way to perfuse tissue with GD ex vivo), we cannot confirm that the lesion visualized by MRI has been removed. However, it seems unlikely that all of these lesions could have been left in the breast following successful wire localization. Although we did not perform immediate postoperative confirmatory MRI scans, none of the patients in this series developed breast cancer or detectable lesions at the original biopsy site with a mean follow-up of 55 months (range 39 to 90 months). Furthermore, it is possible that MRI lesions measuring less than 5 mm were in fact removed but could not be successfully identified by pathologists at the time of either gross or histologic examination due to their small size. In addition, the specimen radiographs that are obtained at the time of gross examination to aid in specimen sampling may not show MRI-detected abnormalities. The localization wire or blue dye instilled at the site of concern by the radiologists during wire localization is helpful in directing tissue sampling by the surgeon. However, the guidewires are sometimes displaced before the specimen is received in pathology; if not displaced, the guidewire is still typically removed before serial sectioning. Additionally, the blue dye diffuses over a wide area during formalin fixation, leading to nonspecific staining. At our institution, specimens up to 4.5 cm in size are submitted completely for histologic examination, while those larger than 4.5 cm are sampled based on gross and radiographic examination of the specimen serially sectioned at 3-mm intervals. Even when the specimens are optimally processed, very small lesions measuring less

S.A. Langer et al. / The American Journal of Surgery 190 (2005) 633– 640

than 2 mm can be missed since the material submitted for paraffin embedding is generally 3 mm in thickness and typically only 1 slide is prepared from each block. In this scenario, results from our and other institutions may be incorrectly reporting low specificity rates since some of these small lesions may represent discrete clinical entities or very early malignancies. Methods to permanently localize small MRI lesions and ensure proper identification of the region of interest, such as placing a clip adjacent to the area of concern, may be considered. While these concerns deserve consideration, only 1 of the 6 lesions in our group under 5 mm was less than 2 mm in size, and it would seem unusual that all lesions smaller than 5 mm as seen on MRI failed to show a relationship with any discrete histologic entity. Providing that MRI lesions were properly localized preoperatively, removed at the time of surgery and properly sampled and reviewed by pathologists, it is possible that the lesions seen on MRI were merely artifacts of imaging and did not reflect a distinct entity. One explanation for this artifact is the known phenomenon that normal breast tissue enhancement varies at different times of the menstrual cycle. Regional or discrete enhancement may be a function of asymmetric, endogenous or exogenous, hormonal influences [24]. Ideally, all MRI studies would be performed at the time of least hormonal stimulation during days 6 through 16 of the menstrual cycle [25]. However, the routine synchronization of MRI studies with the menstrual period is not universally performed for logistical reasons or biologic uncertainty. In our cohort, 18 of our patients (82%) were premenopausal at the time of their MR imaging. None of the postmenopausal women in this study were using hormone replacement therapy. Blocking hormonal influence to minimize FP MRI findings has been described. Heinig et al demonstrated a significant improvement in nonspecific enhancement MRI patterns in a small group of women treated for 8 weeks with tamoxifen [26]. None of the patients in this study were treated in this manner.

Conclusion At present, there are no other reports specifically investigating a series of FP MRI studies and comparing MRI morphology and kinetic enhancement curves to histopathology. We investigated a consecutive series of women with FP lesions identified on MRI and found that lesions identified solely on breast MRI, less than 5 mm in size, despite suspicious enhancement curves, represented discrete nodules of benign breast tissue. While one could consider managing patients in this group with clinical observation and serial MRI scans instead of performing surgical biopsy, the results from this exploratory study requires validation in a larger data set. If validation studies find a significant frequency of malignancy in this cohort of lesions smaller than 5 mm, then a delay in

639

biopsy would lead to a delay in diagnosis of breast cancer and should be avoided. On the other hand, if studies of a larger cohort of women undergoing MRI-directed breast biopsy confirm that these lesions infrequently represent risk lesions, precursor lesions, or frank malignancy, with particular attention to women with known axillary metastases, then it may be possible to cautiously observe women with these lesions, incorporating an open discussion with the patient instead of performing an immediate tissue biopsy. This might safely improve breast MRI specificity, enhancing the overall utility of this powerful breast imaging method.

References [1] Brekelmans CT, Seynaeve C, Bartels CC, et al. Effectiveness of breast cancer surveillance in BRCA1/2 gene mutation carriers and women with high familial risk. J Clin Oncol 2001;19:924 –30. [2] Orel SG, Weinstein SP, Schnall MD, et al. Breast MR imaging in patients with axillary node metastases and unknown primary malignancy. Radiology 1999;212:543–9. [3] Olson JA Jr, Morris EA, Van Zee KJ, et al. Magnetic resonance imaging facilitates breast conservation for occult breast cancer. Ann Surg Oncol 2000;7:411–5. [4] American College of Radiology (ACR) BI-RADS Magnetic Resonance Imaging. ACR Breast Imaging Reporting and Data Systems, Breast Imaging Atlas. 4th ed. Reston, VA: ACR; 2003. [5] LaTrenta LR, Menell JH, Morris EA, et al. Breast lesions detected with MR imaging: utility and histopathologic importance of identification with US. Radiology 2003;227:856 – 61. [6] Tabar L, Vitak B, Chen HH, et al. Beyond randomized controlled trials: organized mammographic screening substantially reduces breast carcinoma mortality. Cancer 2001;91:1724 –31. [7] Nystrom L, Andersson I, Bjurstam N, et al. Long-term effects of mammography screening: updated overview of the Swedish randomised trials. Lancet 2002;359:909 –19. [8] Goscin CP, Berman CG, Clark RA. Magnetic resonance imaging of the breast. Cancer Control 2001;8:399 – 406. [9] Elmore JG, Barton MB, Moceri VM, et al. Ten-year risk of false positive screening mammograms and clinical breast examinations. N Engl J Med 1998;338:1089 –96. [10] Brown J, Smith RC, Lee CH. Incidental enhancing lesions found on MR imaging of the breast. AJR 2001;176:1249 –54. [11] Kriege M, Brekelmans CT, Boetes C, et al. Efficacy of MRI and mammography for breast-cancer screening in women with a familial or genetic predisposition. N Engl J Med 2004;351:427–37. [12] Siegmann KC, Muller-Schimpfle M, Schick F, et al. MR imagingdetected breast lesions: histopathologic correlation of lesion characteristics and signal intensity data. AJR 2002;178:1403–9. [13] Cheung HS, Tse GM, Lai SY, et al. Relationship between lesion size and signal enhancement on subtraction fat-suppressed MR imaging of the breast. Magn Reson Imaging 2004;22:1259 – 64. [14] Nunes LW, Schnall MD, Orel SG, et al. Breast MR imaging: interpretation model. Radiology 1997;202:833– 41. [15] Nunes LW, Schnall MD, Orel SG. Update of breast MR imaging architectural interpretation model. Radiology 2001;219:484 –94. [16] Kuhl CK, Mielcareck P, Klaschik S, et al. Dynamic breast MR imaging: are signal intensity time course data useful for differential diagnosis of enhancing lesions? Radiology 1999;211:101–10. [17] Leong CS, Daniel BL, Herfkens RJ, et al. Characterization of breast lesion morphology with delayed 3DSSMT: an adjunct to dynamic breast MRI. J Magn Reson Imaging 2000;11:87–96.

640

S.A. Langer et al. / The American Journal of Surgery 190 (2005) 633– 640

[18] Schnall MD, Rosten S, Englander S, et al. A combined architectural and kinetic interpretation model for breast MR images. Acad Radiol 2001;8:591–7. [19] Ikeda DM, Hylton NM, Kinkel K, et al. Development, standardization, and testing of a lexicon for reporting contrast-enhanced breast magnetic resonance imaging studies. J Magn Reson Imaging 2001;13:889 –95. [20] Dirbas F, Millet-Serrano A, Horst K, et al. MRI in the surgical management of women with benign and malignant breast disease. Presented at: the Pacific Coast Surgical Association Scientific Program; February 17, 2004; Wailea, HI. Abstract 3J, page 109. [21] Liberman L, Morris EA, Benton CL, et al. Probably benign lesions at breast magnetic resonance imaging: preliminary experience in highrisk women. Cancer 2003;98:377– 88. [22] Gibbs P, Liney GP, Lowry M, et al. Differentiation of benign and

[23]

[24]

[25] [26]

malignant sub-1 cm breast lesions using dynamic contrast enhanced MRI. Breast 2004;13:115–21. Daniel BL, Birdwell RL, Ikeda DM, et al. Breast lesion localization: a freehand, interactive MR imaging-guided technique. Radiology 1998;207:455– 63. Kuhl CK, Bieling HB, Gieseke J, et al. Healthy premenopausal breast parenchyma in dynamic contrast-enhanced MR imaging of the breast: normal contrast medium enhancement and cyclical-phase dependency. Radiology 1997;203:137– 44. Heywang-Kobrunner S, Beck R. Contrast-Enhanced MRI of the Breast. New York, NY: Springer; 1996. Heinig A, Lampe D, Kolbl H, et al. Suppression of unspecific enhancement on breast magnetic resonance imaging (MRI) by antiestrogen medication. Tumori 2002;88:215–23.