Background Parenchymal Enhancement on Breast MRI: Impact on Diagnostic Performance

Transcription

1 Women s Imaging Original Research DeMartini et al. Parenchymal Enhancement on Breast MRI Women s Imaging Original Research Wendy B. DeMartini 1,2 Franklin Liu 3 Sue Peacock 1,2 Peter R. Eby 4 Robert L. Gutierrez 1,2 Constance D. Lehman 1,2 DeMartini WB, Liu F, Peacock S, Eby PR, Gutierrez RL, Lehman CD Keywords: breast, MRI, parenchymal enhancement DOI: /AJR Received December 7, 2010; accepted after revision August 30, Presented at RSNA 2008, Chicago, IL. 1 Department of Radiology, Seattle Cancer Care Alliance, 825 Eastlake Ave East, Seattle, WA Address correspondence to W. B. DeMartini (wdemarti@uw.edu). 2 Department of Radiology, University of Washington Medical Center, Seattle, WA. 3 Southdale Medical Center, Edina, MN. 4 Department of Radiology, Virginia Mason Medical Center, Seattle, WA. WEB This is a Web exclusive article. AJR 2012; 198:W373 W X/12/1984 W373 American Roentgen Ray Society Background Parenchymal Enhancement on Breast MRI: Impact on Diagnostic Performance OBJECTIVE. Breast density is documented to reduce sensitivity and specificity of mammography. However, little is known regarding the effect of normal background parenchymal enhancement on accuracy of breast MRI. The purpose of this study was to evaluate the effect of background parenchymal enhancement on MRI diagnostic performance. MATERIALS AND METHODS. A review of our established MRI data identified all women undergoing breast MRI from March 1, 2006, through June 30, Prospectively reported background parenchymal enhancement categories of minimal, mild, moderate, or marked (anticipated MRI lexicon definitions) and assessments were extracted from the database for each patient. Outcomes were determined by pathologic analysis, imaging, and linkage with the regional tumor registry with a minimum of 24 months of follow-up. Patients were dichotomized into categories of minimal or mild versus moderate or marked background parenchymal enhancement. Associations with patient age, abnormal interpretation rate, positive biopsy rate, cancer yield, sensitivity, and specificity were compared using chi-square and z score tests. RESULTS. The study cohort included 736 women. Moderate or marked background parenchymal enhancement was significantly more frequent among patients younger than 50 years compared with those 50 years old and older (39.7% vs 18.9%; p < ). Moderate or marked background parenchymal enhancement was also associated with a higher abnormal interpretation rate compared with minimal or mild background parenchymal enhancement (30.5% vs 23.3%; p = 0.046). Positive biopsy rate, cancer yield, sensitivity, and specificity were not significantly different according to background parenchymal enhancement category. CONCLUSION. Increased background parenchymal enhancement on breast MRI is associated with younger patient age and higher abnormal interpretation rate. However, it is not related to significant differences in positive biopsy rate, cancer yield, sensitivity, or specificity of MRI. M RI is a very valuable tool for the detection and characterization of breast carcinoma. The identification of breast cancer using MRI is largely dependent on the enhancement of malignant lesions after IV administration of contrast agent, resulting in visible areas with both morphologic and kinetic features. However, enhancement on MRI is not limited to malignancies or discrete benign lesions of the breast. It has been recognized that normal breast parenchyma enhances after IV administration of contrast agent, a phenomenon now termed background parenchymal enhancement. Such physiologic background parenchymal enhancement has been described in premenopausal women, in whom it can fluctuate with the menstrual cycle [1, 2], and has been found to increase in postmenopausal women undergoing hormone replacement therapy [3, 4]. Some researchers have hypothesized that background parenchymal enhancement may decrease the sensitivity of breast MRI by obscuring enhancing malignancies or may diminish the specificity of this tool by causing enhancement patterns that overlap with the appearance of cancers. Recognizing background parenchymal enhancement and distinguishing it from suspicious findings are potentially important in reducing false-positive assessments of breast MRI examinations while maintaining high sensitivity. As such, the American College of Radiology (ACR) MRI Committee plans to incorporate the assessment of background parenchymal enhancement into the next version of the published lexicon (Morris E, written communication, 2010). To date, studies of background parenchymal enhancement have largely been limited to describing its occurrence and imaging AJR:198, April 2012 W373

2 DeMartini et al. patterns. There have been few investigations of the impact on MRI diagnostic performance for this potentially important imaging entity. Thus, the purpose of this study was to determine whether background parenchymal enhancement affects key measures of breast MRI performance. In particular, we sought to evaluate whether the extent of background parenchymal enhancement, when characterized using anticipated categories, effects breast MRI diagnostic outcomes. Materials and Methods Study Population This study was approved by our institutional review board and was HIPAA compliant. A review of our prospectively populated MRI data was performed to identify all women 18 years old or older undergoing breast MRI examinations at our institution between March 1, 2006, and June 30, None of the patients in this group were excluded from the study. For women with multiple examinations in the study interval, only the most recent MRI examination was included in the analysis set. Thus, each patient contributed a single MRI examination to the study. For the study cohort, breast MRI examination and patient outcomes data were obtained from the Consortium Oncology Data Integration project (IR no. 5586, protocol 1833E). Consortium Oncology Data Integration is a solid tumor clinical research database developed and maintained by the Fred Hutchinson Cancer Research Center in collaboration with the University of Washington. Data in Consortium Oncology Data Integration have been obtained in accordance with applicable human subjects laws and regulations, including any requiring informed consent. There are many sources of data for Consortium Oncology Data Integration, including our prospectively recorded breast MRI data forms, our institutional pathology database, and the Cancer Surveillance System regional tumor registry. The Cancer Surveillance System is a part of the National Cancer Institute s Surveillance, Epidemiology, and End Results program and collects population-based data on the incidence, treatment, A C and follow-up on all newly diagnosed cancers (except nonmelanoma skin cancers) occurring in residents of the Puget Sound region. This registry has achieved case ascertainment of 95% or higher completeness according to the North American Association of Central Cancer Registries [5]. Breast MRI Interpretations and Patient Outcomes At the time of clinical interpretation, all breast MRI examinations were assessed by one of four fellowship-trained breast imagers who had from 1 to 10 years of breast MRI experience. For each MRI examination, background parenchymal enhancement was prospectively assigned one of four categories in accordance with the anticipated BI- RADS MRI lexicon classification system: minimal ( 25% enhancement of glandular tissue) (Fig. 1A), mild (25 50% enhancement of glandular tissue) (Fig. 1B), moderate (50 75% enhancement of glandular tissue) (Fig. 1C), or marked (> 75% enhancement of glandular tissue) (Fig. 1D). A combination of the unenhanced, initial contrast-enhanced, Fig. 1 Examples of different degrees of background parenchymal enhancement in four patients. A, 63-year-old woman with family history of breast cancer. Contrast-enhanced axial bilateral breast maximum-intensity-projection (MIP) MRI performed as high-risk screening shows minimal background parenchymal enhancement. B, 42-year-old woman with history of multiple biopsies. Contrast-enhanced axial bilateral breast MIP MRI performed as high-risk screening shows atypical ductal hyperplasia and mild background parenchymal enhancement. C, 27-year-old woman with strong family history of breast and ovarian cancer. Contrast-enhanced axial bilateral breast MIP MRI performed for high-risk screening shows moderate background parenchymal enhancement. D, 26-year-old woman with history of chest radiation for cancer. Contrast-enhanced axial bilateral breast MIP MRI performed for high-risk screening shows marked background parenchymal enhancement. B D W374 AJR:198, April 2012

3 Parenchymal Enhancement on Breast MRI Fig year-old woman with strong family history of breast cancer. Image is example of false-positive finding with moderate background parenchymal enhancement. Bilateral maximum-intensity-projection breast MRI shows enhancing 5-mm oval mass (arrow) in left breast at 9 o clock position. Pathology results indicated benign fibroadenoma. Fig year-old woman with newly diagnosed invasive ductal carcinoma (dashed arrow) in right breast. Image is example of true-positive and false-positive findings with marked background parenchymal enhancement. Bilateral breast maximum-intensityprojection MRI shows lobular masses in left breast at 11 o clock (solid arrow) and 5 o clock (circle) positions. Biopsy of left 11 o clock position mass yielded intraductal carcinoma, whereas biopsy of left 5 o clock position mass showed fibroadenoma. A Fig year-old woman with family history of ovarian cancer and recent biopsy revealing lobular carcinoma in situ, atypical ductal hyperplasia, and radial scar in left breast. Image is example of false-positive finding with marked background parenchymal enhancement. Bilateral breast maximum-intensity-projection MRI shows 9-mm lobular mass (arrow) in right breast at 12 o clock position. Pathology results indicated fibroadenoma. Note two oval masses (circle) in right posterolateral breast and axilla, which were shown on ultrasound to be lymph nodes. Fig year-old woman referred from outside institution. A, Example of false-negative finding with mild background parenchymal enhancement. Bilateral breast maximum-intensity-projection MRI shows known newly diagnosed right breast cancer (arrow) at 1 o clock position. Left breast was determined to be negative and assessed as category 1. B and C, Two days later, mammogram was performed, for which patient was overdue. Mammograms in craniocaudal (B) and mediolateral oblique (C) spot views revealed spiculated mass in right breast (data not shown), corresponding to known biopsy-proven cancer, and grouped amorphous calcifications (oval) in left breast at 1 o clock position. These calcifications were biopsied, and pathologic results indicated ductal carcinoma in situ. B C AJR:198, April 2012 W375

4 DeMartini et al. subtraction, and maximum-intensity-projection images was typically used to determine background parenchymal enhancement. MRI lesion features, assessments, and recommendations in accordance with the current MRI lexicon [6] were also prospectively recorded at the time of clinical interpretation, and all MRI variables were subsequently entered into our Consortium Oncology Data Integration database. These imaging variables, as well as patient age and clinical indication for undergoing breast MRI, were extracted from the electronic database for this study. Benign versus malignant outcomes were determined by biopsy, imaging surveillance, and data from Consortium Oncology Data Integration with a minimum of 24 months of follow-up. Final histopathologic outcomes for high-risk lesions at core needle biopsy (i.e., atypical ductal or lobular hyperplasia, lobular carcinoma in situ, radial scar, or phyllodes tumor) were based on subsequent benign (no upgrade to malignancy) or malignant (upgrade to malignancy) surgical biopsy results. Any invasive carcinoma or ductal carcinoma in situ histology was categorized as malignant. Follow-up or pathologic findings were confirmed in all study patients using clinical and tumor registry databases. C A Statistical Analyses All statistical outcome definitions were derived from the fourth edition of the ACR Atlas Follow-Up and Outcome Monitoring section [6]. Abnormal interpretation rate was defined as the percentage of all study examinations receiving initial category 0, 3, 4, or 5 assessments. categories 0 and 3 were included as abnormal interpretations because, from the index examination, each resulted in a recommendation for additional immediate ( category 0) or short-term follow-up ( category 3) imaging. This reflects the fact that breast MRI serves as an amalgam of study types, acting as both a screening and a diagnostic study. A negative (nonactionable) MRI examination was defined as one with a final category 1, 2, 3, or 6 assessment. A positive (actionable) MRI examination was defined Fig year-old asymptomatic woman with history of prior right breast cancer status post lumpectomy and radiation therapy. A, Example of false-negative finding in patient with marked background parenchymal enhancement. Maximum-intensity-projection (MIP) image is shown. Bilateral breast MRI was performed for screening. No suspicious enhancing abnormalities were detected. Right breast was given category 2 assessment for benign changes of prior lumpectomy, and left breast was given category 1 (negative). B, Eight months after screening MRI, scheduled mammogram (craniocaudal spot magnification view shown) revealed fine linear and pleomorphic calcifications (oval) in right breast. Biopsy of these calcifications indicated invasive ductal carcinoma. C and D, Follow-up breast MRI was performed 7 days after mammogram to evaluate for extent of disease (approximately 8 months after original screening MRI). MIP (C) and axial (D) subtraction images show significant change compared with screening MRI obtained previously, with new heterogeneous nonmasslike enhancement involving majority of right breast, as above, corresponding to biopsy-proven malignancy. B D W376 AJR:198, April 2012

5 Parenchymal Enhancement on Breast MRI as one with a final category 4 or 5 assessment. True-positive results were defined as positive MRI examinations with tissue diagnoses of malignancy during the follow-up interval. False-positive results were defined as positive MRI examinations with no tissue diagnoses of malignancy during the follow-up interval (Figs. 2 4). True-negative results were defined as negative MRI examinations with no tissue diagnoses of malignancy during the follow-up interval. False-negative results were defined as negative MRI examinations with tissue diagnoses of malignancy during the follow-up interval (Figs. 5 and 6). Positive biopsy rate was defined as the percentage of biopsies performed that resulted in a tissue diagnosis of cancer. Cancer yield was defined as the percentage of biopsy-proven cancers resulting from a positive MRI examination. Because each patient contributed one examination to the study, results at patient and examination levels are equivalent. Our study was not an evaluation of the diagnostic performance of breast MRI for detecting known biopsy-proven category 6 lesions. We assessed the diagnostic performance of this tool for identifying and excluding malignancy not known or suspected before breast MRI. An examination was assigned the highest-order actionable category according to a hierarchy (i.e., category 5 > 4 > 0 > 3 > 6 > 2 > 1). Thus, an examination with category 6 as the highest-order assessment had no suspicious finding in addition to the known malignancy. Category 1 or 2 n = 278 Category 1 or 2 n = 279 n = 0 Category 3 n = 20 Category 3 n = 23 n = 0 Minimal or Mild Parenchymal Enhancement n = 536 Category 0 n = 12 Patients were dichotomized into categories of minimal or mild versus moderate or marked background parenchymal enhancement for data analyses. For patients in the minimal or mild versus moderate or marked groups, associations with age younger than 50 years versus 50 years old or older were determined and compared using chi-square analyses. Menopausal status was not available for the study cohort at the time of our analyses. Thus, age 50 years old or older served as a surrogate for postmenopausal status. Abnormal interpretation rate, positive predictive value, and cancer yield were calculated and compared across groups using a chi-square test. Sensitivity and specificity were determined and compared across groups using the z score statistic. MRI Technique With the exception of MRI examinations for newly diagnosed breast cancers, we attempt to perform breast MRI in premenopausal women on days 7 14 of their menstrual cycle, although we do not currently have data regarding the frequency with which this timing is achieved. All breast MRI examinations, including those for implant evaluation, were performed with contrast agent at our institution. MRI examinations were performed with the patient prone in a 1.5-T commercially available system (LX 1.5T scanner, GE Healthcare) using a dedicated breast coil (GE HD 8-channel Breast Array coil, GE Healthcare). During the study period, two scanning protocols were used as clinical practice and technology evolved. All protocols Category 4 or 5 n = 93 Category 4 or 5 n = 101 n = 44 Total No. of Women in Study n = 736 Initial Assessment Category 6 n = 133 Category 6 n = 133 n = 2 Final Assessment Outcome Category 1 or 2 n = 94 Category 1 or 2 n = 94 n = 1 Category 3 n = 6 Category 3 n = 10 n = 1 were consistent with the guidelines established by the International Breast MRI Consortium and by the ACR Imaging Network MRI trials. Imaging sequences included unenhanced and at least two contrast-enhanced T1-weighted fat-suppressed 3D fast spoiled gradient recalled series. From March 2006 through June 2006, scans were performed in the axial plane with the following parameters: TR/TE, 6.2/3; flip angle, 10 ; FOV, cm; slice thickness, 2.2 mm; and matrix size From July 2006, scans were performed in the axial plane with the following parameters: TR/TE, 5.5/2.7; flip angle, 10 ; FOV, cm; slice thickness, 1.6 mm; and matrix size, For all protocols, initial contrast-enhanced acquisitions were centered at 1.5 minutes after contrast agent administration. Delayed acquisitions were centered between 4.5 and 7.5 minutes after contrast agent administration, depending on the protocol. Gadodiamide (Omniscan, Mallinckrodt) contrast agent was used (0.1 mmol/ kg dose power-injected at 2 ml/s) followed by a 10- ml saline flush. All MRI examinations were processed using a commercially available computer-aided evaluation system (CADstream, Confirma) and were reviewed on high-resolution PACS monitors. Results A total of 736 women 18 years old or older (mean age, 52 years; range, years) underwent breast MRI from March 1, 2006, to June 30, 2007, and they comprise the study Moderate or Marked Parenchymal Enhancement n = 200 Category 0 n = 12 Category 4 or 5 n = 43 Category 4 or 5 n = 51 n = 20 Category 6 n = 45 Category 6 n = 45 n = 0 Fig. 7 Flowchart of study patients by background parenchymal enhancement category with initial and final assessments and malignant outcomes, according to Standards of Reporting of Diagnostic Accuracy recommendations. In group with minimal or mild background parenchymal enhancement, initial assessment was category 0 in 12 patients. After additional imaging, final assessment was category 2 in one patient, category 3 in three patients, and category 4 in eight patients. In group with moderate or marked background parenchymal enhancement, initial assessment was category 0 in 12 patients. After additional imaging, final assessment was category 3 in four patients and category 4 in eight patients. Two malignancies occurring in patients with category 6 assessments were not biopsy-proven category 6 lesions known before breast MRI but were separate additional malignancies. AJR:198, April 2012 W377

6 DeMartini et al. population. None of the 736 patients was excluded. Clinical indications for breast MRI were high-risk screening (368/736 [50.0%]), evaluation of the extent of disease in patients with a recent diagnosis of breast cancer (279/736 [37.9%]), problem solving (32/736 [4.3%]), short-term follow-up (26/736 [3.5%]), evaluation of response to neoadjuvant treatment (23/736 [3.2%]), evaluation of silicone breast implants (5/736 [0.7%]) and other (3/736 [0.4%]). Figure 7 is a flowchart of the study patients categorized into minimal or mild versus moderate or marked enhancement groups, with initial and final assessments and malignant outcomes, according to the Standards of Reporting of Diagnostic Accuracy recommendations. Outcomes were malignant for 68 (9%) and benign for 668 (91%) of the 736 patients. The 68 malignancies were 27 invasive ductal carcinomas, six invasive lobular carcinomas, three cases of mixed invasive ductal carcinoma and invasive lobular carcinoma, 29 cases of ductal carcinoma in situ, and three miscellaneous other breast cancers. Four of the malignancies were false-negative MRI cases and are described later in this article. The overall frequencies of background parenchymal enhancement categories were 64.3% for minimal (n = 473), 8.6% for mild (n = 63), 20.0% for moderate (n = 147), and 7.2% for marked (n = 53) enhancement. The extent of background parenchymal enhancement varied significantly by age (Table 1). Moderate or marked background parenchymal enhancement was significantly more frequent among women younger than 50 years (116/292 [39.7%]) than among women 50 years old or older (84/444 [18.9%]) (p < ). The abnormal interpretation rate for women in the moderate or marked background parenchymal enhancement category (30.5% [61/200]) was significantly higher than that for women in the minimal or mild category (23.3% [125/536]) (Table 2; p = 0.046). Positive biopsy rates were not affected by background parenchymal enhancement, with a proportion of cancer at biopsy of 44.4% (20/45) in the moderate or marked background parenchymal enhancement category compared with 47.3% (44/93) in the minimal or mild background parenchymal enhancement category (p = 0.75). There was also no significant difference in cancer yield according to background parenchymal enhancement, with a 10.0% (20/200) yield among women in the moderate or marked category TABLE 1: Background Parenchymal Enhancement and Association With Age Patient Age (y) TABLE 2: Breast MRI Performance Measures as a Function of Background Parenchymal Enhancement Performance Measure compared with 8.2% (44/536) in the minimal or mild group (p = 0.44). Sensitivity was not significantly different across background parenchymal enhancement categories, with results of 90.9% (20/22) in the moderate or marked group and 95.7% (44/46) in the minimal or mild background parenchymal enhancement group (p = 0.82). Four breast cancers occurred during the 2-year follow-up after negative breast MRI examinations and were false-negative cases. In three cases (two in the minimal or mild enhancement group and one in the moderate or marked enhancement group), biopsy of suspicious calcifications seen at mammography revealed ductal carcinoma in situ in two patients (Fig. 5) and invasive ductal carcinoma in one (Fig. 6) patient. In the last case (moderate or marked enhancement group), skin punch biopsy based on physical examination findings revealed Paget disease of the nipple. Specificity was slightly lower in women with moderate or marked background parenchymal enhancement (82.6% [147/178]) than in women with minimal or mild background parenchymal enhancement (88.4% [433/490]), but the difference was not statistically significant (p = 0.07). Background Parenchymal Enhancement Minimal or Mild Background Parenchymal Enhancement Minimal or Mild Moderate or Marked Moderate or Marked < (176/292) 39.7 (116/292) (360/444) 18.9 (84/444) Note Data are % (no. of patients/total). p < (chi-square test) for all comparisons. Abnormal interpretation rate 23.3 (125/536) 30.5 (61/200) a Positive biopsy rate 47.3 (44/93) 44.4 (20/45) 0.75 a Cancer yield 8.2 (44/536) 10.0 (20/200) 0.44 a Sensitivity 95.7 (44/46) 90.9 (20/22) 0.82 b Specificity 88.4 (433/490) 82.6 (147/178) 0.07 b Note Data are % (no. of patients/total). a Chi-square test. b z score. Discussion Akin to mammographic breast density, background parenchymal enhancement on MRI is a tissue characteristic that varies widely among patients. Importantly, although fibroglandular tissue volume on breast MRI has been linked to mammographic density and is not dependent on contrast agent administration [7], background parenchymal enhancement is an imaging characteristic of normal breast parenchyma defined by the amount of fibroglandular breast tissue that enhances. Normal background parenchymal enhancement has been described to have the appearance of diffuse or nodular enhancement patterns, which can vary depending on the phase of menstrual cycle [1], has been shown to be increased in women undergoing hormone replacement therapy [3, 4], and also has been found to be increased in the lactating breast [8]. The amount of background parenchymal enhancement that occurs is thought to be associated with endogenous hormone levels. The underlying tissue cause of background parenchymal enhancement has yet to be fully characterized but may be related to the increased vascular permeability associated with estrogen and the elevated metabolic activity associated with progesterone [9]. Multiple studies have revealed the negative effect of increased mammographic breast density on the sensitivity and specificity of breast cancer detection [10 19]. Thus, although breast density and background parenchymal enhancement are not clearly linked, many have wondered whether the tissue characteristic of background parenchymal enhancement might have a similar important effect on the diagnostic accuracy of breast p W378 AJR:198, April 2012

7 Parenchymal Enhancement on Breast MRI MRI. In particular, it has been hypothesized that background parenchymal enhancement could cause false-negatives by obscuring malignancies, or could result in false-positive results by mimicking the appearance of breast cancer. As such, recognizing background parenchymal enhancement and categorizing it as minimal, mild, moderate, or marked in extent is anticipated to be a component of the updated breast MRI lexicon. However, there are currently sparse data regarding background parenchymal enhancement and its impact on breast MRI performance. Two studies have very recently investigated the impact of background parenchymal enhancement on diagnostic performance [20, 21]. Hambly et al. [20] evaluated the effect of background parenchymal enhancement on short-term follow-up and cancer detection rates in 250 baseline high-risk screening MRI examinations. Uematsu and colleagues [21] assessed the impact of background parenchymal enhancement on accuracy for determining extent of malignancy in 146 MRI examinations performed for newly diagnosed breast cancer. These important preliminary studies are augmented by our investigation. Our study of 736 MRI examinations is the largest to date, to our knowledge, and includes a spectrum of indications for breast MRI such that our results should be more generalizable to broad clinical practice. Furthermore, we have performed linkage to a regional tumor registry to accurately determine true- and falsepositive and -negative results with a minimum of 24 months of follow-up. We found that background parenchymal enhancement was influenced by age, with women younger than 50 years having more extensive background parenchymal enhancement compared with older women. A similar inverse correlation of age with breast density has been well established in mammography [22]. Although we were not able to evaluate the relationship between background parenchymal enhancement and mammographic breast density in our study, it is important to note that background parenchymal enhancement and amount of fibroglandular tissue are not necessarily associated. Indeed, two recent studies by Ko et al. [19] and Cubuk et al. [23] specifically found no correlation between background parenchymal enhancement and mammographic density. We found that increased background parenchymal enhancement was associated with a significantly higher abnormal interpretation rate for MRI, with women in the moderate or marked category more likely to be given initial assessments leading to additional imaging or biopsy ( category 0, 3, 4, or 5). This is in keeping with the results of Hambly et al. [20] regarding the significantly greater frequency of abnormal interpretations in their study, measured in particular by category 3 assessments, in women with greater than minimal background parenchymal enhancement. In our study, although a higher number of women in the moderate or marked background parenchymal enhancement group were initially assessed as having abnormal findings, our measures of final outcome of positive biopsy rate, cancer yield, sensitivity, and specificity were not significantly different between the background parenchymal enhancement groups. This is also similar to the results of Hambly and colleagues [20], who found no difference in the positive biopsy rate or cancer yield based on background parenchymal enhancement in their study. Their study included only patients who underwent biopsy, and sensitivity and specificity were not assessed. Our study has limitations. It is a single-institution investigation in the academic setting, with fellowship-trained breast imagers experienced in breast MRI interpretation. Our results may not be generalizable to less-established practices. In addition, there may be a selection bias in including only the most recent MRI examinations in the study interval for patients with more than one MRI examination. Background parenchymal enhancement may change over time, such as by decreasing in women with breast cancer treated with neoadjuvant or adjuvant systemic therapy. Furthermore, we were not able to evaluate the relationships between background parenchymal enhancement and other variables such as mammographic breast density, menopausal status, or timing relative to the menstrual cycle in premenopausal women. Other than MRI performed for newly diagnosed breast cancers, we attempt to perform MRI examinations in premenopausal women on days 7 14 of their menstrual cycle, but we do not currently have data regarding the frequency with which this timing is achieved. These are important topics that warrant further investigation. In summary, our study shows that moderate or marked background parenchymal enhancement resulted in more assessments, leading to additional imaging or biopsy, but that these initial findings did not impact final diagnostic performance outcomes when compared with women with minimal or mild background parenchymal enhancement. These results are encouraging and suggest that breast MRI maintains its utility as a diagnostic tool even in the presence of more extensive background parenchymal enhancement. In studies to date, background parenchymal enhancement does not appear to have the impact on MRI diagnostic accuracy that breast density has on mammographic performance. Further studies of this key breast MRI tissue characteristic across institutions and clinical indications are warranted. References 1. 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 cyclicalphase dependency. Radiology 1997; 203: Müller-Schimpfle M, Ohmenhauser K, Stoll P, Dietz K, Claussen CD. Menstrual cycle and age: influence on parenchymal contrast medium enhancement in MR imaging of the breast. Radiology 1997; 203: Reichenbach JR, Przetak C, Klinger G, Kaiser WA. Assessment of breast tissue changes on hormonal replacement therapy using MRI: a pilot study. J Comput Assist Tomogr 1999; 23: Delille JP, Slanetz PJ, Yeh ED, Kopans DB, Halpern EF, Garrido L. Hormone replacement therapy in postmenopausal women: breast tissue perfusion determined with MR imaging initial observations. Radiology 2005; 235: North American Association of Central Cancer Registries. North American Association of Central Cancer Registries Website. Accessed August 22, D Orsi CJ, Mendelson EB, Ikeda DM, et al. Breast Imaging Reporting and Data System: ACR breast imaging atlas. Reston, VA: American College of Radiology, Wei J, Chan HP, Helvie MA, et al. Correlation between mammographic density and volumetric fibroglandular tissue estimated on breast MR images. Med Phys 2004; 31: Talele AC, Slanetz PJ, Edmister WB, Yeh ED, Kopans DB. The lactating breast: MRI findings and literature review. Breast J 2003; 9: Söderqvist G, Isaksson E, von Schoultz B, Carlstrom K, Tani E, Skoog L. Proliferation of breast epithelial cells in healthy women during the menstrual cycle. Am J Obstet Gynecol 1997; 176: Ciatto S, Visioli C, Paci E, Zappa M. Breast density as a determinant of interval cancer at mammographic screening. Br J Cancer 2004; 90: Moskowitz M, Gartside P, McLaughlin C. Mammographic patterns as markers for high-risk benign breast disease and incident cancers. Radiology 1980; 134: Threatt B, Norbeck JM, Ullman NS, Kummer R, AJR:198, April 2012 W379

8 DeMartini et al. Roselle P. Association between mammographic study in Nijmegen, The Netherlands. J Epidemiol 20. Hambly NM, Liberman L, Dershaw DD, Brennan parenchymal pattern classification and incidence Community Health 1998; 52: S, Morris EA. Background parenchymal enhance- of breast cancer. Cancer 1980; 45: Kolb TM, Lichy J, Newhouse JH. Comparison of ment on baseline screening breast MRI: impact 13. Boyd NF, Guo H, Martin LJ, et al. Mammograph- the performance of screening mammography, phys- on biopsy rate and short-interval follow-up. AJR ic density and the risk and detection of breast can- ical examination, and breast US and evaluation of 2011; 196: cer. N Engl J Med 2007; 356: factors that influence them: an analysis of 27, Uematsu T, Kasami M, Watanabe J. Does the de- 14. Whitehead J, Carlile T, Kopecky KJ, et al. Wolfe mammographic parenchymal patterns: a study of the masking hypothesis of Egan and Mosteller. Cancer 1985; 56: Bird RE, Wallace TW, Yankaskas BC. Analysis of cancers missed at screening mammography. Radiology 1992; 184: van Gils CH, Otten JD, Verbeek AL, Hendriks JH, Holland R. Effect of mammographic breast density on breast cancer screening performance: a patient evaluations. Radiology 2002; 225: Mandelson MT, Oestreicher N, Porter PL, et al. Breast density as a predictor of mammographic detection: comparison of interval- and screen-detected cancers. J Natl Cancer Inst 2000; 92: Ko ES, Lee BH, Choi HY, Kim RB, Noh WC. Background enhancement in breast MR: correlation with breast density in mammography and background echotexture in ultrasound. Eur J Radiol 2011; 80: gree of background enhancement in breast MRI affect the detection and staging of breast cancer? Eur Radiol 2011; 21: Kopans D. Breast imaging, 3rd ed. Philadelphia, PA: Lippincott Williams and Wilkins, Cubuk R, Tasali N, Narin B, Keskiner F, Celik L, Guney S. Correlation between breast density in mammography and background enhancement in MR mammography. Radiol Med (Torino) 2010; 115: W380 AJR:198, April 2012