Wo m e n ’s I m a g i n g • O r i g i n a l R e s e a r c h

Chikarmane et al.
BI-RADS Analysis of Nonmass Enhancement on MRI

Women’s Imaging
Original Research

Revisiting Nonmass Enhancement

Downloaded from www.ajronline.org by 2.177.167.142 on 10/31/23 from IP address 2.177.167.142. Copyright ARRS. For personal use only; all rights reserved

in Breast MRI: Analysis of

Outcomes and Follow-Up Using the
Updated BI-RADS Atlas
Sona A. Chikarmane1,2 OBJECTIVE. The purpose of this study is to assess the frequency of reclassification of
Aya Y. Michaels1 nonmass enhancement (NME) as background parenchymal enhancement (BPE) and to deter-
Catherine S. Giess1,2 mine positive predictive values (PPVs) of NME descriptors using the revised BI-RADS atlas.
MATERIALS AND METHODS. A retrospective review of our institution’s MRI data-
Chikarmane SA, Michaels AY, Giess CS base from January 1, 2009, through March 30, 2012, identified 6220 contrast-enhanced breast
MRI examinations. All findings prospectively assessed as NME and rated as BI-RADS cate-
gories 3, 4, and 5 (n = 386) were rereviewed in consensus by two radiologists who were blind-
ed to pathologic findings with the fifth edition of the BI-RADS lexicon. Findings considered
as postsurgical, associated with known cancers, NME given a BI-RADS category 3 assess-
ment before 2009, previously biopsied, and those reclassified as BPE, focus, or mass were ex-
cluded (n = 181). Medical records were reviewed for demographics and outcomes.
RESULTS. Two hundred five women were included (mean age, 48.8 years; range, 21–84
years). Seventy-seven of 386 findings (20.0%) were reclassified as BPE, and patients with BPE
were younger than those with NME (mean age, 43.9 years; range, 31–62 years) (p = 0.003).
Pathology results for 144 of 205 (70.2%) patients included 52 malignant, 11 high-risk, and 81
benign lesions. The highest PPVs for distribution patterns were 34.5% (10/29) for segmental
and 100.0% (3/3) for diffuse distribution. The highest PPVs for internal enhancement patterns
were 36.7% (11/30) for clustered ring enhancement and 27.5% (11/40) for clumped enhance-
ment. No difference for NME malignancy rate was noted according to BPE (10/52 [19.2%]
marked or moderate; 42/153 [27.5%] mild or minimal, p = 0.24). Thirty-two percent (17/52) of
malignant NMEs had high T2 signal.
CONCLUSION. Careful assessment of findings as BPE versus NME can improve PPVs,
particularly in younger women. Although clustered ring enhancement had one of the study’s
highest PPVs, this number falls below previously published rates. Reliance on T2 signal as a
Keywords: breast, clustered ring enhancement, MRI, benign feature may be misleading, because one-third of malignancies had T2 signal.
nonmass enhancement
reast MRI is the most sensitive of NME include distribution and internal en-

DOI:10.2214/AJR.17.18086
Received February 13, 2017; accepted after revision
April 26, 2017.
Based on a presentation at the Radiological Society of
North America 2016 annual meeting, Chicago, IL.
1
Department of Radiology, Brigham and Women’s
Hospital, Harvard Medical School, 75 Francis St, Boston,
MA 02115. Address correspondence to S. A. Chikarmane
(schikarmane@partners.org).
2
Department of Imaging, Dana-Farber Cancer Institute,
Harvard Medical School, Boston, MA.
AJR 2017; 209:1178–1184
0361–803X/17/2095–1178
© American Roentgen Ray Society
B
modality for detecting breast
cancers [1–4]. To standardize the
terminology and reporting used
when describing findings detected on breast
imaging, the American College of Radiology
(ACR) developed the BI-RADS lexicon ini-
tially for mammography and ultrasound and
subsequently added breast MRI [5]. Per the
ACR BI-RADS atlas for breast MRI, mor-
phologic categories for findings detected at
breast MRI include mass, nonmass enhance-
ment (NME), and focus. NME is defined as
an enhancing area that is not a mass and can
be distinguished from background parenchy-
mal enhancement (BPE). NME may contain
interspersed fat and extend over a small or
large region in the breast [6]. Characteristics
hancement patterns, internal T2 signal on
unenhanced imaging, and kinetic curves [5].
The 5th edition of the ACR BI-RADS lex-
icon was published in 2013 [5] with specific
changes made to the distribution and inter-
nal enhancement patterns of NME [6]. With
regard to distribution, the word “ductal” was
replaced with “linear” to describe enhance-
ment in a line, with or without branching.
The remaining distribution descriptors were
not changed. Clustered ring enhancement
(CRE) was added as one of the four inter-
nal enhancement patterns. Stippled, reticular,
and dendritic patterns were removed from
the lexicon.
CRE is defined as thin rings of enhance-
ment clustered around ducts. The pathophysi-

1178 AJR:209, November 2017

 BI-RADS Analysis of Nonmass Enhancement on MRI

TABLE 1: Review of Literature on Clustered Ring Enhancement (CRE)

Total No. of No. of CREs/Total No. of Malignant No. of Malignant CREs/No. PPV2 of CRE,
Study Biopsied NMEs No. of NMEs (%) NMEs/Total (%) of Malignant NMEs (%) No./Total (%)
Tozaki et al. [7] 61 23/61 (38) 35/61 (57) 22/35 (63) 22/23 (96)
Tozaki and Fukuda [8] 30 8/30 (27) 18/30 (60) 8/18 (44) 8/8 (100)
Downloaded from www.ajronline.org by 2.177.167.142 on 10/31/23 from IP address 2.177.167.142. Copyright ARRS. For personal use only; all rights reserved

Sakamoto et al. [9] 102 39/102 (38) 10/102 (10) 2/10 (20) 2/39 (5)
Yuen et al. [11] 70a 42/70 (60) 57/70 (81) 42/57 (74) 42/42 (100)
Uematsu and Kasami [10] 124 66/124 (53) 85/124 (69) 51/85 (60) 51/66 (77)
Chikarmane et al. (presented at RSNA 2016) 144 30/144 (21) 50/144 (35) 11/50 (22) 11/30 (37)
Total 531 208/531 (39) 255/531 (48) 136/255 (53) 136/208 (65)
Note—NME = nonmass enhancement, PPV2 = positive predictive value 2 (percentage of cancers in cases referred for biopsy), RSNA 2016 = Radiological Society of North
America 2016 annual meeting.
aSegmental only.

ologic process of CRE is thought to be sec-
ondary to the washout phenomena of an
intraductal carcinoma with an abundant vas-
cular supply, in contrast to the gradual en-
hancement of the adjacent periductal stroma
[7, 8]. Early studies published on CRE had ex-
tremely high positive predictive values (PPVs;
96–100%) (Table 1); however, these were per-
formed before the BI-RADS atlas was updat-
ed and had small numbers of subjects [7–11].
In addition, these studies assessed CRE as
either a subset of heterogeneous internal en-
hancement or as categorically present versus
not present, and not as its own separate en-
hancement pattern, as described in the updat-
ed lexicon [7–12]. To date and to our knowl-
edge, no study has assessed the PPV of CRE
as a separate enhancement pattern.
Another change to the BI-RADS lexicon for
breast MRI included the introduction of BPE
[5]. Although commonly used in clinical prac-
tice, BPE was recently formalized in the lex-
icon and is defined as the degree of normal
parenchymal tissue enhancement on the first
contrast-enhanced MRI sequence. BPE is re-
ported as minimal, mild, moderate, or marked
[5]. BPE differs from the amount of fibroglan-
dular tissue, which can be assessed as almost
entirely fat, scattered fibroglandular tissue, het-
erogeneous fibroglandular tissue, and extreme
fibroglandular tissue. The amount of BPE does
not directly correlate with the amount of fibro-
glandular tissue; for example, a patient with ex-
tremely dense fibroglandular tissue may or may
not have moderate or marked BPE [5].
Per the lexicon, BPE may be symmetric
versus asymmetric. Symmetric enhancement
is seen as mirror-image enhancement of both
breasts and is suggestive of a benign pro-
cess. Asymmetric enhancement, on the oth-
er hand, is enhancement that is more promi-
nent on one side or area when compared with
the other breast. Diagnostic dilemmas may 6220 consecutive contrast-enhanced examinations
occur when BPE presents as asymmetric en- in 3802 women. All breast MRI examinations were
hancement, multiple enhancing foci, or larg- initially interpreted at the time of clinical care by
er focal areas, because these may be inter- one of 14 dedicated breast fellowship–trained radi-
preted as NME and therefore managed as a ologists with 5–25 years of experience. Initial image
focal lesion [13]. interpretation was performed according to the first
Our study objectives were to assess the edition of the BI-RADS MRI lexicon. Therefore, le-
frequency of reclassification of NME as BPE sions that had been interpreted as “nonmasslike en-
and to determine the PPVs of NME descrip- hancement” (which corresponds to NME in the 2nd
tors using the revised BI-RADS atlas [5]. edition of the BI-RADS MRI lexicon) at the time of
initial interpretation and were assessed as BI-RADS
Materials and Methods categories 3, 4, and 5 were retrospectively identi-
Study Population fied through a search of our breast MRI database.
This study was approved by an institutional re- This review yielded 386 examinations with a find-
view board (Dana-Farber Cancer Institute protocol ing of NME prospectively assessed as BI-RADS
number 07-234) and is HIPAA compliant. A retro- categories 3, 4, and 5. All imaging of the 386 ex-
spective review of our breast MRI database from aminations was then rereviewed by two breast fel-
January 1, 2009, through March 30, 2012, found lowship–trained radiologists (with 2 and 20 years of
6220 Contrast-enhanced
breast MRI examinations
in 3802 women
386 Examinations in
386 women prospectively
assessed as NME and BI-RADS
categories 3, 4, and 5
104 Patients for
- Postsurgical
- Associated with known
Excluded cancers
- Probably benign NME with
initial assessment before
77 Reclassified as study period
Excluded
background parenchymal - Previously biopsied
enhancement - Reclassified as focus or mass
Final study group:
205 findings of NME in
205 women
Fig. 1—Flowchart for study population (205 findings of nonmass enhancement [NME] in 205 women).

AJR:209, November 2017 1179

 Chikarmane et al.

(GE Healthcare), HDX 3-T (GE Healthcare), or

Trio 3-T (Siemens Healthcare) unit and a dedi-
cated breast surface coil (16-channel, eight-chan-
nel, or seven-channel). Sequences include a three-
plane localizing sequence, axial fat-suppressed
T2-weighted fast spin-echo or T2-weighted STIR
sequence, and axial gradient-echo T1-weighted
Downloaded from www.ajronline.org by 2.177.167.142 on 10/31/23 from IP address 2.177.167.142. Copyright ARRS. For personal use only; all rights reserved

non–fat-suppressed sequence before contrast ma-

terial administration. Dynamic T1-weighted fat-
suppressed 3D gradient-echo sequences are then
performed before and four times after (at approxi-
mately 90-second intervals) IV administration of
0.1 mmol/kg contrast material (Magnevist, Bay-
er) in the axial plane. A T1-weighted 3D gradient-
echo delayed contrast-enhanced sequence is per-
formed in the sagittal plane at approximately 5
minutes after initial contrast agent injection. Be-
fore 2010, dynamic images were acquired in either
the sagittal or axial plane, with delayed imaging
performed in the orthogonal plane. Postprocess-
ing, including subtraction axial images, maximum
A B intensity projections, and computer-aided diag-
Fig. 2—38-year-old woman with history of radiation therapy for mediastinal non-Hodgkin lymphoma 20 years prior. nosis (CADstream, Sectra; or VersaVue, iCAD),
A and B, Axial (A) and sagittal (B) screening breast MR images show focal homogeneous nonmass is routinely used. The same imaging protocol is
enhancement (arrows) in 12 o’clock axis posteriorly. MRI-guided core needle biopsy (not shown) revealed
used for both screening and diagnostic indica-
densely fibrous breast tissue with apocrine cysts and pseudoangiomatous stromal hyperplasia, considered
concordant. Patient’s condition has been stable on imaging to date. tions. Breast MRI is performed with five magnets
at our institution (two 3-T units and three 1.5-T
experience) who were blinded to radiology reports, (n = 12), and findings associated with known cancers units). Patients are assigned magnets on the basis
patients’ clinical information (age and risk factors), (n = 9) were excluded. On consensus review by the of availability of magnet time.
and pathologic findings for the retrospective review same two breast fellowship–trained radiologists, 77
and were informed only of the prospectively affected of the 386 cases (20.0%) were reclassified as BPE, MR Image Interpretation
breast (left or right). The entire breast MRI study was using the definitions set forth by the ACR BI-RADS For the 205 examinations included in the study
reviewed, including both breasts and all available se- 5th edition [5]. This left a final study population of population, the following information was collected
quences. Cases of NME (n = 104) assessed as proba- 205 findings of NME in 205 women (Fig. 1). during image review: BI-RADS NME distribution
bly benign (BI-RADS category 3) on an index exam- (focal, linear, segmental, regional, multiple region-
ination performed before the study period (n = 33), MRI Protocol al, and diffuse), BI-RADS internal enhancement
postsurgical changes (n = 26), findings reclassified as The current breast MRI protocol at our insti- pattern (homogeneous, heterogeneous, clumped, or
a focus or mass (n = 24), previously biopsied findings tution includes prone imaging with a Signa 1.5-T CRE) assessed on both axial and sagittal imaging,
and associated T2 signal characteristics of the le-
sion (with T2 signal defined as that of water signal).
Kinetic information was obtained via visual assess-
ment of the dynamic series.
After imaging review, the electronic medical
record and pathology reports from all cases in-
cluded in the final study population were reviewed
by at least one breast fellowship–trained radiolo-
gist. Data collected for the study included patient
age, family history (first-degree relative with a his-
tory of breast cancer or first-degree relative with a
history of germline mutation), and personal his-
tory of germline mutation or breast cancer as indi-
cated in the history, at subsequent follow-up imag-
ing (records available to December 31, 2016), and
at histologic examination of biopsy specimens.
A B
Fig. 3—57-year-old woman (from outside hospital) who presented with decreasing size of left breast. Biopsy and Pathologic Assessments
A and B, Diagnostic problem-solving contrast-enhanced axial (A) and sagittal (B) breast MR images show
diffuse heterogeneous nonmass enhancement (arrows) in left breast. Subsequent ultrasound-guided biopsy For patients undergoing biopsy, concordance
revealed node-negative grade 1 out of 3 invasive lobular carcinoma. with imaging was determined. Potentially discor-

1180 AJR:209, November 2017

 BI-RADS Analysis of Nonmass Enhancement on MRI
TABLE 2: Imaging Characteristics of 144 Pathology-Proven Cases of Nonmass Enhancement and Their Positive
­Predictive Values (PPVs)
Total
Descriptor (n = 205)
Distribution
Segmental 29
Downloaded from www.ajronline.org by 2.177.167.142 on 10/31/23 from IP address 2.177.167.142. Copyright ARRS. For personal use only; all rights reserved
Linear 53
Focal 102
Regional 18
Diffuse 3 3
Internal enhancement pattern
Homogeneous 66
Heterogeneous 69
Clumped 40
Clustered ring 30
Note—Except where noted otherwise, data are number of lesions. PPV2 = percentage of cancers in cases referred for biopsy, PPV3 = percentage of cancers in cases
that actually undergo biopsy, NA = not applicable.
aThese lesions were high-risk abnormalities without upgrades as confirmed on final surgical pathologic analysis.
dant results were discussed at a routine biweekly
radiology-pathology conference. The standard prac-
tice at our institution was to recommend excision
for any histopathologic result believed to be discor-
dant with the level of suspicion on the basis of ini-
tial imaging. At our institution, surgical excision is
recommended for atypical ductal hyperplasia, radial
scar, complex sclerosing lesions, any papilloma with
atypia, lobular carcinoma in situ with pleomorphic
or variant features, and fibroepithelial lesions.
Statistical Analysis
Statistical analysis, including the determination
of malignancy rates, significant values (p < 0.05),
and 95% CIs, was performed with software (SAS
version 9.04, SAS Institute; and Prism 7, Graph-
Pad). PPVs were calculated as recommended by
the 5th edition of the ACR BI-RADS lexicon for
breast MRI and included PPV1 (percentage of can-
cers in recalled cases), PPV2 (percentage of can-
cers in cases referred for biopsy), and PPV3 (per-
centage of cancers in cases that actually undergo
biopsy) [5]. As per the BI-RADS lexicon, the PPV1
and PPV2 for breast MRI are essentially the same.
Categoric variables were compared with use of the
chi-square test. Age was compared between study
patients and those excluded for BPE by means of
one-way ANOVA. A p < 0.05 was considered indic-
ative of a statistically significant difference.
Results
Two hundred five women with 205 find-
ings of NME were included in the study;
their mean age was 48.8 years (range, 21–84
years). The mean age of women with NME
was significantly older than those who were
AJR:209, November 2017
Total Biopsied Benign High Risk Malignant
(n = 144) (n = 81) (n = 11)a (n = 52)
25 15 0 10
38 22 4 12
66 37 5 24
12 7 2 3 48 (20–95)
0 0 3 NA
37 21 3 13
49 27 5 17
33 19 3 11
25 14 0 11
excluded as having BPE (43.9 years; range,
31–62 years; p = 0.003). Thirty-seven percent
(76/205) of patients had a personal history of
breast cancer, and 22.9% (47/205) had a fami-
ly history of breast cancer. Twenty-eight of 205
(13.6%) were BRCA mutation carriers, and two
of 205 (0.01%) were Li-Fraumeni mutation car-
riers. Of the 205 examinations, 98 (47.8%) were
performed for screening indications and 107
(52.2%) were performed for diagnostic indica-
tions. Diagnostic indications (n = 107) includ-
ed problem solving for an equivocal mammo-
graphic or ultrasound finding (n = 29), problem
solving for clinical symptoms (n = 29), follow-
up for a probably benign or BI-RADS category
3 separate finding in the same breast (n = 23),
workup to evaluate the extent of disease for re-
cently diagnosed breast cancer (n = 18), evalu-
ation of chemotherapy response (n = 3), postbi-
opsy follow-up for another finding (n = 3), and
MRI for unknown primary cancer (n = 2). Of
the 205 cases, the amount of fibroglandular tis-
sue was as follows: 14 fatty, 63 scattered, 98
heterogeneous, and 30 dense. BPE was as fol-
lows: 57 minimal, 96 mild, 38 moderate, and 14
marked background.
Pathology results were available for 144 of
205 (70.2%) cases, of which 52 (36.1%) were
malignant, 11 (7.6%) were high risk (confirmed
on surgical excision, without upgrade), and the
remaining 81 (56.3%) were benign. Of the 61
remaining cases not sampled, there has been no
evidence of disease to date (April 1, 2017) on
follow-up imaging in all patients. The malig-
nant cases (n = 52) were invasive ductal carci-
noma (n = 12; 23.0%), ductal carcinoma in situ
Size (mm),
Mean (Range) PPV2 (%) PPV3 (%)
36 (10–80) 34.5 40.0
14 (6–27) 22.6 31.5
15 (6–34) 23.5 36.4
16.7 25.0
100.0 100.0
15 (4–71) 19.7 32.4
22 (6–77) 24.6 34.7
30 (11–70) 27.5 33.3
33 9–80 36.7 44.0
(n = 28; 53.8%), invasive lobular carcinoma (n =
8; 15.4%), invasive carcinoma with mixed lobu-
lar and ductal features (n = 3; 5.8%), and inva-
sive tubular carcinoma (n = 1; 2.0%).
Table 2 shows the pathologic profiles of
the NME distributions and internal enhance-
ment patterns. The most common distribu-
tion pattern was focal (49.7%; 102/205) (Fig.
2), followed by linear (25.6%; 53/205); the
most common internal enhancement pat-
tern was heterogeneous (Fig. 3) (33.7%;
69/205), followed closely by homogeneous
(32.2%; 66/205). Of the distributions, the
highest PPVs (PPV2 and PPV3) were found
in diffuse distributions (n = 3; 100% PPV2
and 100% PPV3); however, the number of
cases was small. All three cases of diffuse
NME were in patients presenting with clini-
cal findings, including erythema and nipple
inversion (n = 1) or a shrinking breast (n =
2), who underwent diagnostic breast MRI
for problem-solving clinical examination.
Ultrasound-guided biopsy revealed invasive
lobular carcinoma (Fig. 3) (n = 2, both with
shrinking breast) and invasive ductal carci-
noma (n = 1, erythema, nipple inversion).
The next highest PPV for distribution af-
ter diffuse was that of segmental distribu-
tion, which was 34.5% for PPV2 and 40.0%
for PPV3. The internal enhancement pattern
with the highest PPV2 and PPV3 was CRE,
with PPV2 of 36.7% and PPV3 of 44.0%.
This was followed by clumped enhancement,
with PPV2 of 27.5% and PPV3 of 33.3%.
The PPV2 for the detection of malignancy
was 32.9% (31/94) for a curve showing a fast
1181
 Chikarmane et al.
Downloaded from www.ajronline.org by 2.177.167.142 on 10/31/23 from IP address 2.177.167.142. Copyright ARRS. For personal use only; all rights reserved

A B C
Fig. 4—61-year-old woman with recently diagnosed left breast cancer who presented for evaluation of extent of disease.
A and B, Sagittal (A) and axial (B) contrast-enhanced breast MR images show segmental clustered ring enhancement (arrows) in right breast.
C, MRI-guided biopsy shows high-grade ductal carcinoma in situ (arrow).

initial increase, 19.2% (15/78) for a curve show- was the lowest of all internal enhancement pat- and invasive carcinoma with mixed features
ing a medium initial increase, and 18.1% (6/33) terns. Of the 30 findings described as CRE, (n = 1). Of the remaining five cases of CRE that
for a curve showing a slow initial increase. 25 were biopsied, yielding 11 malignant (Fig. did not undergo biopsy, all have been negative
The PPV2 for the detection of malignancy was 4) and 14 benign results (Fig. 5). The benign on imaging to date (April 1, 2017).
24.0% (40/166) for a curve showing a persistent results (n = 14) included duct ectasia and peri- Seventy-seven of 386 (20.0%) cases were ex-
delayed phase, 28.5% (6/21) for a curve show- ductal fibrosis (n = 1), silicone granulomas (n = cluded as BPE on rereview (Fig. 6). Of these
ing a plateau delayed phase, and 33.3% (6/18) 1), sclerosing adenosis (n = 4), granulomatous 77 cases, one (1.3%) was positive for malig-
for a curve showing a washout delayed phase. mastitis (n = 1), multiple intraductal papillomas nancy on review of pathologic examination.
The frequency of CRE was 14.6% (30/205) (n = 2), apocrine cysts (n = 1), epithelial hyper- The malignancy was in a 65-year-old woman
(mean size, 33 mm; range, 9–80 mm), which plasia (n = 2), pseudoangiomatous stromal hy- with BRCA1 mutation and a personal history of
perplasia (n = 1), and normal breast tissue (n = ovarian cancer who had been placed in short-
1). Malignant abnormalities included invasive interval follow-up for linear NME in the left
ductal carcinoma (n = 5), ductal carcinoma in upper outer breast. At 12-month follow-up, the
situ (n = 3), invasive lobular carcinoma (n = 2), finding was thought to be more conspicuous
and MRI-guided core needle biopsy revealed
solid-type high-nuclear-grade ductal carcino-
ma in situ; the mastectomy specimen showed
no invasive component, and the patient was
node negative. If these 77 cases of BPE were
included in the analysis, the overall malignancy
rate of NME would have been 18.0% (51/282)
versus 24.4% (50/205) (p = 0.09). The malig-
nancies (not including cases reclassified as
BPE) stratified by degree of BPE were 19.2%
(10/52) marked or moderate and 26.1% (40/113)
mild or minimal, with no significant difference
between malignancy rates according to degree
of BPE (p = 0.35).
T2-hyperintense signal was found in 78
A B (38%) of the 205 cases of NME; of these 78
Fig. 5—61-year-old woman with benign clustered ring enhancement (CRE) in left breast. Diagnostic MRI was cases, 56 were biopsied, 17 (30.3%) were
performed after discordant ultrasound-guided biopsy (not shown) result for palpable finding in left breast. found to be malignant, and 39 (69.7%) were
A, Axial contrast-enhanced image of left breast shows focal CRE in subareolar 12 o’clock region (arrow).
B, On delayed sagittal image, CRE (arrow) appears more apparent. MRI-guided wire localization showed duct found to be benign. Of the 17 malignant cases,
ectasia with periductal fibrosis and chronic inflammation. more than half were ductal carcinoma in situ

1182 AJR:209, November 2017

 BI-RADS Analysis of Nonmass Enhancement on MRI
Downloaded from www.ajronline.org by 2.177.167.142 on 10/31/23 from IP address 2.177.167.142. Copyright ARRS. For personal use only; all rights reserved
Fig. 6—36-year-old woman with no symptoms but with history of Hodgkin lymphoma and mediastinal mantle
radiation. Screening examination revealed nonmass enhancement (NME) reclassified as benign parenchymal
enhancement (BPE).
A and B, Axial (A) and sagittal (B) MR images show NME (arrows) in right upper outer quadrant. NME was
described as focal and homogeneous and was reclassified as BPE on review. MRI-guided breast biopsy (not
shown) revealed benign breast tissue.
(53%; 9/17) (Fig. 7) and the remaining were
invasive. There was no significant difference
in T2 signal of malignancies by distribution
or internal enhancement patterns (p > 0.05).
Discussion
In 2013, the ACR BI-RADS lexicon was
updated to include CRE as one of four inter-
nal enhancement patterns, defined as minute
rings of enhancement clustered around ducts
[5]. The lexicon states that because of bet-
ter spatial resolution of modern scanners, we
are now better able to appreciate this pattern.
All currently published articles evaluating
the PPVs of malignancy for CRE were per-
formed before the publication of the updat-
ed lexicon (Table 1). The BI-RADS lexicon
explicitly states that more research needs to
be performed on the PPV of CRE. Therefore,
this study of 205 findings of NME in 205
women provides important data on the util-
ity of the updated BI-RADS lexicon termi-
nology for NME, particularly that of CRE.
Early studies in 2006 by Tozaki et al. [7]
and Tozaki and Fukuda [8] found extremely
high PPVs for CRE (96–100%); however, those
studies had small numbers of both total biop-
sied NMEs (Table 1) and CREs. The high PPV
of CRE was redemonstrated with higher to-
tal numbers by Yuen et al. [11] in 2008 (PPV,
100%); however, all 70 cases analyzed in that
study were of segmental distribution and there-
fore showed inherent bias toward malignan-
cy. Our results reflect the results of the highest
number of total biopsied NMEs (n = 144) of all
studies published to date (Table 1).
More important, however, the previously
published studies analyzed CRE as either a
subset of heterogeneous enhancement or cat-
egorically as present or not present in NME;
therefore, the findings did not analyze CRE
AJR:209, November 2017
A
as a separate internal enhancement pattern, as
it is described in the updated BI-RADS atlas
[5]. Uematsu and Kasami [10] in 2012 report-
ed CRE as being present in 66 of 124 (53%)
total biopsied NMEs; similar numbers have
been reported by Tozaki et al. [7] and Saka-
moto et al. [9] (38% each), who also described
CRE as categorically present or not present in
their biopsied NMEs. Our study of 144 biop-
sied NMEs, of which 20% were CRE, is the
first study to report CRE as a separate inter-
nal enhancement pattern. Our lower frequen-
cy of CRE when compared with other studies
(20% vs 27–60%; Table 1) likely reflects our
analysis of CRE as a separate pattern.
The discrepancy of our lower frequency of
CRE compared with prior studies that ana-
lyzed CRE as present or not present raises the
question of whether CRE should be assessed
differently than as suggested by the BI-RADS
lexicon [5]. Anecdotally, CRE can be difficult
to differentiate from other internal enhance-
ment patterns and may be present with either
clumped or heterogeneous NME. In 2006, To-
zaki et al. [7] first described two types of CRE:
one that showed clumped enhancement with
enhancing masses or foci constituting a ring-
like enhancement pattern and heterogeneous
enhancement inside of which minute ring en-
hancement is clustered. More recently, Machi-
da et al. [12] reported hypointense areas and
clustered rings as major structural elements
of heterogeneous enhancement and evaluated
them as one collected internal enhancement
pattern of heterogeneous structure. Given these
prior analyses, CRE may be better assessed as
categorically present or not present, to reflect
clinical practice, and possibly be listed as an as-
sociated feature in the BI-RADS atlas.
BPE, when presenting as large focal areas,
may be interpreted as NME and sometimes
B
is then managed as a focal lesion [13]. In our
study, 20% (77/386) of cases prospectively as-
sessed as NME were reclassified as BPE on im-
aging review, with one malignancy (1/77; 1.3%)
classified as ductal carcinoma in situ with-
out invasion. The mean age of these excluded
women was significantly lower than the mean
age of those whose findings were interpreted as
true NME (p < 0.01). If these 77 cases of BPE
were included in the analysis, the overall ma-
lignancy rate of NME would have been lower
at 18.0% (included) compared with 24.4% (ex-
cluded). The improvement in malignancy rate
suggests that these cases can be safely reas-
sessed as BPE, while maintaining cancer de-
tection and reducing false-positives. There was
no significant difference between malignancy
rates by degree of BPE (p = 0.24), suggesting
that when true NME was identified, the degree
of BPE did not influence cancer detection.
Literature on breast MRI states that breast
carcinomas have short T2 relaxation times and
therefore appear hypointense on T2-weighted
imaging [14]; however, many of these stud-
ies reflected mass enhancement and not NME
[15]. Our study shows a 32% malignancy rate
for T2-hyperintense NME, which corrobo-
rates prior studies by Uematsu and Kasami
[10] and Baltzer et al. [16]. The latter article
reported that 9% (11/124) of NME cases had
associated T2 signal, of which 100% were ma-
lignant [10]. Associated T2 signal likely repre-
sents associated edema, necrosis, or lymphatic
involvement in NME and therefore should not
be assumed to be a benign finding because the
associated signal intensity may be an impor-
tant indicator of malignancy [17].
Our study has limitations, including that
it is a retrospective study from a single in-
1183

Downloaded from www.ajronline.org by 2.177.167.142 on 10/31/23 from IP address 2.177.167.142. Copyright ARRS. For personal use only; all rights reserved
A
Fig. 7—46-year-old woman positive for BRCA1
mutation with history of right breast cancer in 2004
who presented for screening breast MRI.
A and B, Contrast-enhanced axial (A) and sagittal
(B) MR images show linear clumped nonmass
enhancement (arrows) in left breast.
C, T2-weighted sagittal MR image shows associated
high T2 signal (arrow). MRI-guided biopsy (not
shown) revealed high-grade ductal carcinoma in situ.
stitution. Images were prospectively read by
multiple breast imagers, with variations in
completeness of lexicon characterization and
classification. Only examinations prospec-
tively assessed as BI-RADS categories 3, 4,
and 5 were rereviewed. Kinetic assessment
was done visually and not via computer-aid-
ed detection. Finally, in a tertiary care cen-
ter, optimal follow-up is challenging because
some patients continue care at local facilities
and we are not linked to a tumor registry.
In conclusion, although CRE had the high-
est PPV for malignancy in our study, the over-
all PPVs for this enhancement pattern vary in
the literature. Further research is needed to as-
sess whether CRE should be evaluated as a sep-
arate internal enhancement pattern versus be-
ing evaluated as categorically present or not
present, to reflect true clinical practice. Care-
ful assessment of BPE versus NME can im-
prove PPVs, particularly in younger women.
Finally, reliance on T2 signal as a benign fea-
ture of NME may be misleading because ap-
proximately one-third of our malignant NMEs
had T2 signal.
References
1. Kuhl CK, Schrading S, Leutner CC, et al. Mam-
mography, breast ultrasound, and magnetic reso-
nance imaging for surveillance of women at high
familial risk for breast cancer. J Clin Oncol 2005;
23:8469–8476
1184
Chikarmane et al.
B C
2. Kuhl CK, Schmutzler RK, Leutner CC, et al. Breast
MR imaging screening in 192 women proved or
suspected to be carriers of a breast cancer suscepti-
bility gene: preliminary results. R­ adiology 2000;
215:267–279
3. Lehman CD. Role of MRI in screening women at
high risk for breast cancer. J Magn Reson Imaging
2006; 24:964–970
4. Kriege M, Brekelmans CTM, Boetes C, et al. Ef-
ficacy of MRI and mammography for breast-can-
cer screening in women with a familial or genetic
predisposition. N Engl J Med 2004; 351:427–437
5. D’Orsi CJ, Sickles EA, Mendelson EB, et al. ACR
BI-RADS Atlas, Breast Imaging Reporting and
Data System. Reston, VA: American College of
Radiology, 2013
6. Edwards SD, Lipson JA, Ikeda DM, Lee JM. Up-
dates and revisions to the BI-RADS magnetic
resonance imaging lexicon. Magn Reson Imaging
Clin N Am 2013; 21:483–493
7. Tozaki M, Igarashi T, Fukuda K. Breast MRI using
the VIBE sequence: clustered ring enhancement in
the differential diagnosis of lesions showing non-
masslike enhancement. AJR 2006; 187:313–321
8. Tozaki M, Fukuda K. High-spatial-resolution
MRI of non-masslike breast lesions: interpreta-
tion model based on BI-RADS MRI descriptors.
AJR 2006; 187:330–337
9. Sakamoto N, Tozaki M, Higa K, et al. Categoriza-
tion of non-mass-like breast lesions detected by
MRI. Breast Cancer 2008; 15:241–246
10. Uematsu T, Kasami M. High-spatial-resolution 3-T
breast MRI of nonmasslike enhancement lesions:
an analysis of their features as significant predictors
of malignancy. AJR 2012; 198:1223–1230
11. Yuen S, Uematsu T, Masako K, Uchida Y,
Nishimura T. Segmental enhancement on breast
MR images: differential diagnosis and diagnostic
strategy. Eur Radiol 2008; 18:2067–2075
12. Machida Y, Shimauchi A, Tozaki M, Kuroki Y,
Yoshida T, Fukuma E. Descriptors of malignant
non-mass enhancement of breast MRI. Acad
­Radiol 2016; 23:687–695
13. Giess CS, Yeh ED, Raza S, Birdwell RL. Back-
ground parenchymal enhancement at breast MR
imaging: normal patterns, diagnostic challenges,
and potential for false-positive and false-negative
interpretation. RadioGraphics 2014; 34:234–247
14. Westra C, Dialani V, Mehta TS, Eisenberg RL.
Using T2-weighted sequences to more accurately
characterize breast masses seen on MRI. AJR
2014; 202:[web]W183–W190
15. Yuen S, Uematsu T, Kasami M, et al. Breast carci-
nomas with strong high-signal intensity on
T2-weighted MR images: pathological character-
istics and differential diagnosis. J Magn Reson
Imaging 2007; 25:502–510
16. Baltzer PAT, Dietzel M, Kaiser WA. Nonmass le-
sions in magnetic resonance imaging of the breast.
J Comput Assist Tomogr 2011; 35:361–366
17. Uematsu T. Focal breast edema associated with ma-
lignancy on T2-weighted images of breast MRI:
peritumoral edema, prepectoral edema, and subcu-
taneous edema. Breast Cancer 2015; 22:66–70
AJR:209, November 2017