This document summarizes a study that evaluated imaging and histopathologic differences between benign and malignant breast lesions initially assessed as BI-RADS category 3 (probably benign) but later upgraded during imaging surveillance. Of the 1,017 lesions initially assessed as BI-RADS 3, 60 (5.9%) were upgraded after a mean follow-up of 54.1 months. Fifteen of the 60 upgraded lesions (25%) proved to be malignant. There was no significant difference in malignancy rates between different types of upgraded lesions (masses, asymmetries, calcifications). While calcifications made up most upgraded lesions, most proved benign, suggesting decreased surveillance of calcifications may be appropriate.

1. ORIGINAL ARTICLE
Imaging and Histopathologic Features of BI-RADS 3
Lesions Upgraded during Imaging Surveillance
Aya Michaels, MD,* Chris SungWon Chung, MD, PhD,*,†
Robyn L. Birdwell,
MD, FACR,* Elisabeth P. Frost, MD,* and Catherine S. Giess, MD*
*Department of Radiology Brigham and Women’s Hospital, Harvard Medical School, Boston,
Massachusetts; †
Mid-Atlantic Permanente Medical Group, Kaiser Permanente Capitol Hill Medical
Center, Washington, DC
n Abstract: To evaluate imaging and histopathologic differences between screen-detected benign and malignant upgraded
lesions initially assessed as BI-RADS 3 at diagnostic evaluation. An IRB approved retrospective review of the mammography
data base from January 1, 2004 to December 31, 2008 identified 1,188 (1.07%) of 110,776 screening examinations assessed
as BI-RADS 3 following diagnostic evaluation at our academic center (staffed by breast specialists) or our outpatient center
(staffed by general radiologists), 1,017 with at least 24 months follow-up or biopsy. Sixty (5.9%) BI-RADS 3 lesions were
upgraded to BI-RADS 4 or 5 during imaging surveillance (study population). Prospective reports, patient demographics, and
clinical outcomes were abstracted from the longitudinal medical record. Mean patient age was 54.1 years (range 35–85).
Lesions consisted of 7 masses, 12 focal asymmetries and 41 calcifications. Fifteen (25%) of 60 lesions upgraded from initial
BI-RADS 3 assessment were malignant (1.47% of total; 15/1,017 BI-RADS 3 studies). Malignancy rates by upgraded lesion
type showed no significant difference: Thirty-three (73.3%) of 45 benign upgraded lesions were calcifications compared to
8 (53.3%) of 15 malignant upgraded lesions (p = 0.202). Twelve (26.7%) of 45 benign upgraded lesions were masses or focal
asymmetries, compared to 7 (46.7%) of 15 upgraded malignant lesions (p = 0.202). Six (85.7%) of 7 malignant upgraded
masses/focal asymmetries had no US correlate at initial BI-RADS 3 assessment compared to 7 (58.3%) of 12 benign upgraded
masses/focal asymmetries (p = 0.33). Breast-imaging specialists interpreted 21 studies, 3 (14.3%) malignant; general radiolo-
gists interpreted 39 studies, 12 (30.8%) malignant (p = 0.218). There was no significant difference in malignancy rate among
different types of upgraded mammographic lesions, nor depending on subspecialty interpretation versus nonsubspecialist
interpretation. Although calcifications made up a majority of upgraded lesions, most were benign, suggesting that decreased
surveillance of calcifications may be appropriate. n
Key Words: BI-RADS 3, probably benign, upgraded lesions
The “probably benign” assignment was introduced
to mammography in the early 1990s and became
widely accepted following several large prospective
clinical studies that demonstrated its reduction in false-
positive biopsies, while maintaining a high detection
rate of early cancer (1–4). Probably benign lesions are
placed in category 3 according to the ACR BI-RADS
lexicon and should carry a less than 2% likelihood of
malignancy (5,6). These lesions are closely monitored
so that if later proven to be malignant, they should be
detected at an early stage (1–7). Prior to the “probably
benign” classification, additional surgical consultations
and biopsies for benign disease represented the major
costs of screening (8). Thus, proper implementation of
BI-RADS 3 assessment can increase the cost-effective-
ness and efficacy of screening mammography.
According to large prospective trials, mammo-
graphic findings suitable for short-term follow-up
include the following: (i) Noncalcified solid masses
with a round, oval, or gently lobular contour and
margins that are predominantly circumscribed. (ii)
Clustered tiny calcifications that are round or oval in
morphology. (iii) Focal asymmetric densities without
associated calcifications, architectural distortion, or
suspicious ultrasound findings (1–4). By definition,
lesions showing imaging change at diagnostic assess-
ment are unstable, and thus, traditionally should be
excluded from a probably benign categorization (1–3).
In actual practice, however, many lesions not fulfilling
Address correspondence and reprint requests to: Aya Michaels, MD,
Department of Radiology Brigham and Women’s Hospital, Harvard Medical
School, 75 Francis St. Boston, MA 02115, USA, or e-mail: aymichaels@
partners.org
DOI: 10.1111/tbj.12677
© 2016 Wiley Periodicals, Inc., 1075-122X/16
The Breast Journal, Volume 23 Number 1, 2017 10–16

2. the above criteria still are categorized as “probably
benign.” Several papers have evaluated the appropri-
ate use of BI-RADS 3 and have found that most
lesions that were upgraded showed signs of interval
change at time of original probably benign assessment
and/or did not adhere to the morphologic criteria out-
lined above (9,10).
Some potential explanations for the disparity
between strict adherence to established criteria and
actual clinical practice include variations in individual
radiology practice and experience versus interpretive
error. For instance, the newest edition of the BI-RADS
atlas suggests that a “probably benign” interpretation
for other types of lesions may be appropriate “. . ..if the
radiologist has personal experience to justify a watch-
ful-waiting approach, preferably involving observation
of a sufficient number of cases of an additional mam-
mographic finding to suggest a likelihood of malignancy
within the defined (≤2%) probably benign range” (6).
Such lesions may have demonstrated change at the time
of initial BI-RADS 3 assessment and include vascular
calcifications, calcifications suggestive of early, evolving
fat necrosis in a patient postbiopsy, or findings thought
to be more visible due to technical reasons (11).
Appropriate use of BI-RADS 3 has thus generated
much debate. A review of types of lesions placed into
the BI-RADS 3 category, as well as the frequency of,
and histopathologic findings for, upgraded lesions can
potentially improve practice performance. As part of a
radiology quality assurance initiative, we performed a
retrospective evaluation of mammographic lesions ini-
tially assessed as BI-RADS 3, but upgraded to BI-
RADS 4 or 5 during imaging surveillance, in order to
determine if there are imaging or histopathologic dif-
ferences in mammographic appearance of benign ver-
sus malignant lesions.
MATERIALS AND METHODS
An IRB approved, retrospective review of our mam-
mography data base from January 1, 2004 to Decem-
ber 12, 2008, identified 1,188 (1.07%) of 110,776
screening examinations that received BI-RADS 3 assess-
ment following diagnostic evaluation: 2.28% (952/
41,743) at our ambulatory center and 0.34% (236/
69,033) at our main academic medical center. There
were 1,017 cases with at least 24 months follow-up or
biopsy. Among these, 1,017 cases were 60 lesions
(5.9%) that were upgraded to BI-RADS 4 or 5 during
imaging surveillance and subsequently underwent
breast biopsy, representing the study population. Mean
patient age was 54.1 years (range 35–85 years).
Mammograms were performed with digital tech-
nique on Senographe DS and Senographe Essential
equipment (GE Healthcare, Little Chalfont, Buckin-
ghamshire, UK). During this study period, mammo-
grams at the teaching hospital were interpreted by
dedicated breast-imaging specialists, and mammograms
at the ambulatory center were interpreted by non-
breast-imaging specialists. In both clinical settings,
ultrasound examinations were performed by the radiol-
ogist with a handheld high-resolution 5–12 MHz linear
transducer (iu22 machine; Philips, Bothell, WA).
In our practice, we did not place any lesions
directly from screening examination into the BI-RADS
3 category. Each case underwent diagnostic examina-
tion. Imaging criteria used for BI-RADS 3 assignment
were according to established criteria (6). Appropriate
BI-RADS 3 classifications included, (i) Noncalcified
circumscribed masses with round, oval or gently lobu-
lated shape; (ii) Clustered calcifications that were
round or oval on magnification images; (iii) Focal
asymmetries or one-view asymmetries; (iv) Coarse cal-
cifications suggestive of dystrophic calcifications; (v)
Miscellaneous findings such as suspected fat necrosis
or suspected biopsy change.
The longitudinal medical record was retrospectively
reviewed to determine patient age, lesion type, the
breast subspecialty status of the prospective interpret-
ing radiologist, and biopsy and clinical outcomes. Sta-
tistical analysis was performed using the Fisher’s exact
test with two-tailed p-value (www.graphpad.com).
RESULTS
Of all 1,017 lesions assessed as BI-RADS 3, 542
(53.3%) were calcifications, 269 (26.5%) were focal
asymmetries, 190 (18.7%) were masses, and 16
(1.6%) were miscellaneous findings. The 60 upgraded
lesions (study population) consisted of seven (11.7%)
masses, 12 (20.0%) focal asymmetries (FAD), and 41
(68.3%) calcifications.
Fifteen (25%) of 60 lesions upgraded from initial
BI-RADS 3 assessment were malignant (1.47% of
total; 15/1,017 BI-RADS 3 examinations). Of these
15 malignancies, 8 (53.3%) were calcifications, 4
(26.7%) were focal asymmetries, and 3 (20.0%) were
masses (Fig. 1). Final pathology showed that 4
(26.7%) of 15 lesions were ductal carcinoma in situ,
10 (66.7%) were stage 1 invasive cancers (8 N0 and 2
BI-RADS 3 Lesions Upgraded during Surveillance • 11

3. NX), and 1 (6.7%) case was a stage 2 invasive cancer
(N1a). Imaging and pathologic features are detailed in
Table 1. The remaining 45 (75%) of 60 upgraded
lesions were benign: 33 (73.3%) calcifications, 4
(8.9%) masses, and 8 (17.8%) focal asymmetries.
Benign histologic and imaging findings are detailed in
Table 2. The number of months to upgrade is also
outlined in these tables.
There was no significant difference in malignancy
rate according to type of upgraded lesion. Thirty-three
(73.3%) of 45 benign upgraded lesions were calcifica-
tions (Fig. 2), compared to 8 (53.3%) of 15 malignant
upgraded lesions (p = 0.202). 12 (26.7%) of 45
benign upgraded lesions were masses or focal asym-
metries, compared to 7 (46.7%) of 15 upgraded
malignant lesions (p = 0.202). All 12 benign focal
asymmetries and masses underwent US, 5 (41.7%)
with findings. All 7 of the malignant focal
asymmetries and masses underwent diagnostic US, 6
(85.7%) with negative findings (Fig. 3). There was no
significant difference in the likelihood of an US corre-
late in benign compared to malignant masses or focal
asymmetries (p = 0.33). The overall percentage of
malignant lesions among BI-RADS 3 lesions did not
vary significantly by lesion type: 8 (1.5%) of 542 cal-
cifications, 4 (1.5%) of 269 focal asymmetries, and 3
(1.6%) of 190 masses (p = 1.0).
Breast-imaging specialists interpreted 21 (35%) of
60 upgraded lesions, with 3 (14.3%) malignancies,
compared to general radiologists who interpreted 39
(65%) of 60 upgraded lesions with 12 (30.8%) malig-
nancies (p = 0.218). Both specialists and generalists
followed more calcifications as compared with focal
asymmetries/masses. Fifteen of 21 (71.4%) lesions fol-
lowed by specialists were calcifications (14 benign, 1
malignant). Twenty-six of 39 (66.7%) lesions fol-
lowed by generalists were calcifications (19 benign, 7
malignant). Six masses/focal asymmetries were fol-
lowed by breast specialists (4 benign, 2 malignant),
compared to 13 masses/focal asymmetries followed by
generalists (8 benign, 5 malignant). The mean interval
time between BI-RADS 3 diagnosis and subsequent
biopsy recommendation was 12 months (range 6–
24 months). For benign upgrades, the mean interval
time was 12.2 months, and for malignant upgrades,
11.2 months. Five (33.3%) of 15 malignancies were
upgraded and subsequently biopsied at the 6-month
exam; an additional 6 (40.0%) were diagnosed at
12 months.
Figure 1. Percentage of malignant calcifications, masses, and
focal asymmetric densities.
Table 1. Imaging and Histopathologic Features of Malignant BI-RADS 3 Upgrades
Lesion type Pathologic size Prospective description US findings Histopathologic findings
Time to diagnosis
(months) Stage
Mass 0.6 cm Circumscribed Negative IDC, well differentiated 18 1A
Mass 1.1 cm Negative IDC, poorly differentiated 6 1A
Mass 1.2 cm Fat containing, oval IDC, moderately differentiated 6 1A
Mass 1.4 cm Oval, “probable fat necrosis” IDC/ILC, moderately differentiated 12 1A
Focal asymmetry 0.2 cm Negative IDC, moderately differentiated 12 1A
Focal asymmetry 1.4 cm Negative IDC, moderately differentiated 24 2A
Focal asymmetry Negative DCIS, low grade 12 0
Calcifications Scattered, some appear vascular DCIS, high grade 6 0
Calcifications 0.6 cm Punctate IDC, moderately differentiated 12 1A
Calcifications Microinvasion Coarse DCIS, high grade, microinvasion 6 1A
Calcifications Punctate DCIS, high grade 6 0
Calcifications 0.9 cm Punctate IDC, moderately differentiated 12 1A
Calcifications 1.0 cm Grouped, probably vascular IDC, moderately differentiated 24 1A
Calcifications 1.0 cm Scattered coarse, round, punctate IDC, poorly differentiated 12 1A
Calcifications Scattered, heterogeneous,
punctate, round
DCIS, high grade 24 0
Size given only for invasive malignancy.
IDC, invasive ductal cancer; DCIS, ductal carcinoma in situ; ILC, invasive lobular carcinoma.
12 • michaels et al.

4. DISCUSSION
Although the percentage of lesions given a BI-
RADS 3 assessment varies by practice (12–15), it
should be small, given widespread mammographic
screening and the prevalence of prior studies. Our
percentage of probably benign lesions within a general
screening population (1.07%) was somewhat lower
than the previously reported range of 2–11% (1–
4,9,12,13). This may be because our institution is a
large breast-imaging center with a minority of our
Table 2. Imaging and Histologic Features of Benign BI-RADS 3 Upgrades
Lesion type Prospective description US finding Time to biopsy Histologic findings
Mass Round, oval Negative 12 FA
Mass Round/oval, obscured “Possible lymph
node versus cyst”
6 Cyst
Mass Round/oval with
obscured margins.
Few cysts 9 Sclerosing adenosis, epithelial hyperplasia
Mass Round/oval, circumscribed Negative 12 Interlobular fibrosis, LCIS, PASH
Focal Asymmetry Negative 6 Radial sclerosing lesion, epithelial
hyperplasia, adenosis with CCC
Focal asymmetry Negative 24 FA
Focal asymmetry Negative 24 Benign breast tissue
Focal asymmetry Negative 24 FA
Focal asymmetry Negative 6 FA, apocrine cysts
Focal asymmetry Cyst with internal echoes 6 Benign cyst
Focal asymmetry Complicated cyst
with thick wall
6 Benign cyst
Focal asymmetry Complicated cyst 12 Apocrine cysts
Calcification Punctate, round 18 Benign (per outside hospital report)
Calcification Punctate, round.
Diffuse/scattered distribution
6 Benign breast epithelium
Calcification Punctate, round 6 FA
Calcification Coarse, clustered/grouped 6 FA
Calcification “Faint,” clustered/grouped 6 Apocrine cysts
Calcification Amorphous/indistinct, grouped 12 Normal breast tissue
Calcification Punctate, round, grouped 12 ADH
Calcification Punctate, round, grouped 6 CCC
Calcification Punctate, round, grouped 22 Apocrine cysts
Calcification Coarse, grouped 9 Sclerosing papilloma
Calcification Punctate, round 15 Benign breast tissue, CCC
Calcification Amorphous, grouped 6 Papilloma, cysts
Calcification Amorphous, grouped 6 Cysts, CCC, Epithelial hyperplasia,
sclerosing adenosis, PASH
Calcification Coarse 6 Fibrous breast tissue with stromal Ca++
Calcification Amorphous, grouped 6 Apocrine cysts with calcium oxalate
Calcification “probable MOC,” grouped 12 Apocrine cysts, CCC, papillary hyperplasia
Calcification Punctate, round, grouped 12 ALH
Calcification Coarse, grouped 6 FA
Calcification Coarse 6 ADH
Calcification Punctate, round, grouped 24 Columnar altered hyperplastic duct
Calcification Punctate, round. “Probably
vascular”
6 Adenosis with mucin
Calcification “Likely MOC” 18 Benign breast tissue
Calcification Coarse, heterogeneous, grouped 6 Benign breast tissue, cysts
Calcification Punctate, round. Likely MOC 18 Benign breast tissue
Calcification Likely MOC 24 Interlobular fibrosis, fibromatoid
change, apocrine metaplasia,
sclerosing adenosis, cysts
Calcification Punctate, round 6 FA with stromal Ca++
Calcification Amorphous, coarse and punctate 24 ADH
Calcification Amorphous, indistinct, grouped 6 Benign epithelium, focal epithelial hyperplasia
Calcification Amorphous, indistinct 12 ADH
Calcification Punctate, round. In multiple groups 6 ALH, CCC, cysts, sclerosing adenosis
Calcification Punctate, round, grouped 6 Adenosis with secretory changes
and psammoma bodies
Calcification Punctate, round 18 Apocrine cysts, adenosis, FA
Calcification Punctate, round, grouped 18 Normal breast tissue, CCC
FA, fibroadenoma; CCC, columnar cell changes; MOC, milk of calcium; ADH, atypical ductal hyperplasia; PASH, pseudoangiomatous stromal hyperplasia; ALH, atypical lobar hyper-
plasia; LCIS, lobar carcinoma in situ.
BI-RADS 3 Lesions Upgraded during Surveillance • 13

5. patients presenting as a baseline or from other imag-
ing centers. Most patients who present for screening
had multiple prior studies available in our system for
comparison. Thus, low BI-RADS 3 usage was not
unexpected.
Compliance with follow-up mammographic imaging
is an important consideration when assigning BI-RADS
3, although this is not the primary focus of this manu-
script. The issue of compliance in the BI-RADS 3 cohort
is discussed further in another paper from our institu-
tion (11). In summary, compliance rates for follow-up
in patients assigned to BI-RADS 3 decreased over time.
At 6, 12, 18, and 24 months, compliance rates were
83.3%, 75.9%, 54.8%, and 53.9%, respectively.
In addition, although BI-RADS 3 lesions overall
should have a less than 2% chance of malignancy (1–
4,6), there are no practice benchmarks for the percent-
age of BI-RADS 3 cases upgraded during surveillance
and the subsequent malignancy rate. Review of a
practice’s upgraded BI-RADS 3 lesions may yield valu-
able lessons on practice performance, and can be a valu-
able quality assurance initiative. In our study interval,
1.07% of screen-detected cases were assigned BI-RADS
3 after diagnostic evaluation. The total percentage of BI-
RADS 3 cases that proved malignant was 1.47%, nearly
three times higher compared to the National Cancer
Institute’s 0.43% cancer detection rate of all screening
exams from the same 5-year time period (16). After the
assignment of a BI-RADS 3 category, 5.9% of BI-RADS
3 lesions were upgraded to B4 or B5 during surveillance.
This is within others’ reported ranges of 0.9–7.9% (1–
4). Our malignancy rate for upgraded lesions was 25%,
compared to published malignancy rates in BI-RADS 3
upgrades ranging from 10% to 56% (1–4).
Overall, there was no significant difference in the
malignancy rate according to lesion type placed into
the BI-RADS 3 category. However, masses and focal
asymmetries may be considered potentially more signifi-
cant than calcifications, since most often malignant focal
asymmetries and masses represent invasive malignancies
rather than in situ malignancies. Of interest, many
benign and malignant focal asymmetries and masses ini-
tially assessed as probably benign had no US correlate,
and the lack of an ultrasound correlate had no predictive
value for malignancy. It is important to note that even if
the ultrasound examination is normal, the image modal-
ity depicting the most suspicious-looking feature should
be used to make the final assessment (13,17).
In this study, calcifications comprised more than
half of all BI-RADS 3 lesions, and also the majority of
(a) (b)
Figure 2. A 85-year-old female with calcified fibroadenoma. (a) “Punctate and round calcifications” under magnification view given a BI-
RADS 3 assessment. (b) At 6 months, heterogeneous calcifications had increased and lesion was upgraded to BI-RADS 4.
Figure 3. A 68-year-old female with intraductal carcinoma. Patient
presented with solitary oval mass on screening mammogram.
Ultrasound exam was negative. BI-RADS 3 was assigned. At 6-
month follow-up exam, ultrasound exam was positive (not shown).
Mass was upgraded to BI-RADS 4.
14 • michaels et al.

6. upgraded lesions. However, most were benign on
biopsy. This finding is in agreement with previous
studies (1,2,4). In Sickles’ study (1), 38.8% of BI-
RADS 3 lesions were calcifications, but only one
(0.1%) proved to be malignant. In Vizcaino’s study
(4), 59% of BI-RADS 3 lesions were calcifications,
and only one was malignant. Similarly, in the series of
Helvie et al. (2), 52% of BI-RADS 3 lesions were cal-
cifications, and none proved malignant. A large study
conducted by Varas et al. (7), reported different find-
ings; in that study, calcifications made up a smaller
percentage of BI-RADS 3 lesions (19%) but had a
greater likelihood of malignancy (3.8%) compared to
masses (1.4%) or focal asymmetries (1.9%). These
differing findings presumably represent variable prac-
tice settings.
There has been extensive debate recently surround-
ing the possibility of mammographic over-diagnosis of
breast cancer, particularly ductal carcinoma in situ,
which is predominantly a mammographically detected
lesion (18–20). Some authors have attempted to deter-
mine predictors for high-grade in situ disease or con-
current invasive disease in malignant calcifications
based on various imaging parameters, with variable
results (21–23). The calcifications in these studies were
of suspicious morphology. Reassessing grouped, punc-
tate or round calcifications as benign rather than
probably benign may be an area for improving BI-
RADS 3 utilization with minimal loss of cancer detec-
tion, because multiple studies, including ours, have
shown a very low malignancy rate for these types of
lesions. In the opinion of one author (24), “. . .clus-
tered calcifications must be categorized when found.
Either they appear benign and the patient can continue
routine annual screening, or they appear indeterminate
or malignant, in which case intervention is indicated.”
While this opinion may represent one extreme, perhaps
such lesions can be followed by annual screening mam-
mograms as a better understanding of the natural his-
tory of these lesions evolves.
In a recent opinion piece (25) on improving screen-
ing mammography performance in the USA, Hall sug-
gested ceasing recall and biopsy of low suspicion,
nonsegmental, amorphous calcifications, arguing that
when not benign, such calcifications usually represent
atypia or low-grade malignant disease, usually change
slowly (thereby detectable on subsequent screenings),
and do not usually dedifferentiate into higher grade
lesions. An analysis of calcifications placed into fol-
low-up but with benign outcomes could yield valuable
information useful for decreasing unnecessary surveil-
lance of calcifications, and may improve practice
specificity. This is an area of future study for our
practice.
In our study, breast-imaging specialists had lower
malignancy upgrade rates than nonbreast-imaging spe-
cialists, although the difference was not statistically
significant. Other studies (20,21) similarly have
demonstrated more accurate performance by breast-
imaging specialists in screening and diagnostic mam-
mography compared to nonbreast-imaging specialists.
The volume of cases interpreted, interpretation of
both screening and diagnostic mammography, and
review of case outcomes are all important factors in
improving performance accuracy for radiologists,
whether they are breast specialists or have a more
general radiology practice.
Our study had some limitations. Because our study
was retrospective, the rationale for BI-RADS 3 assess-
ment was not always apparent from the prospective
report. Differences in interpretation among radiolo-
gists can yield differences in BI-RADS assessment and
management. However, this study was not done to
evaluate observer variability in BI-RADS 3 assessment,
but to determine frequency and outcomes of upgrade,
and evaluate any imaging differences among upgraded
lesions. Because upgraded lesions should comprise a
small minority of lesions assessed as BI-RADS 3, our
series is relatively small. The small sample size in our
series limits statistical analysis of the malignancy rate
among different types of mammographic lesions.
In conclusion, practice benchmarks for BI-RADS 3
upgrade rates and malignancy rates in upgraded
lesions are not well established, and our audit data
adds to the existing literature on these characteristics
of BI-RADS 3 mammography practice. More data
from other radiology groups would be helpful to
determine an optimal target range for these practice
measures. Traditionally, interval change precluded
lesions from being categorized as “probably benign,”
and imaging change is an important but nonspecific
feature of malignancy. Yet, change or increased con-
spicuity may sometimes be balanced against morphol-
ogy and clinical experience in lesion assessment and
management. Evaluation of a practice’s upgraded BI-
RADS 3 lesions, both benign and malignant, can
improve practice performance. Decreasing surveillance
of probably benign calcifications could decrease the
number of benign cases placed into short-term imag-
ing surveillance and decrease benign biopsies.
BI-RADS 3 Lesions Upgraded during Surveillance • 15

7. DISCLOSURE
No disclosures from any of the authors.
REFERENCES
1. Sickles EA. Periodic mammographic follow-up of probably
benign lesions: results in 3,184 consecutive cases. Radiology
1991;179:463–8.
2. Helvie MA, Pennes DR, Rebner M, Adler DD. Mammo-
graphic follow-up of low-suspicion lesions: compliance rate and
diagnostic yield. Radiology 1991;178:155–8.
3. Varas X, Leborgne F, Leborgne JH. Nonpalpable, probably
benign lesions: role of follow-up mammography. Radiology
1992;184:409–14.
4. Vizcaıno I, Gadea L, Andreo L, et al. Short-term follow-up
results in 795 nonpalpable probably benign lesions detected at
screening mammography. Radiology 2001;219:475–83.
5. Leung J, Sickles EA. The probably benign assessment. Radiol
Clin North Am 2007;45:773–89.
6. D’Orsi CJ, Sickles EA, Mendelson EB, et al. ACR BI-RADSâ
Atlas, Breast Imaging Reporting and Data System. Reston, VA:
American College of Radiology, 2013.
7. Varas X, Leborgne JH, Leborgne F, Mezzera J, Jaumandreu
S, Leborgne F. Revisiting the follow-up for BI-RADS category 3
lesions. AJR 2002;179:691–5.
8. Cyrlak D. Induced costs of low-cost screening mammography.
Radiology 1988;168:661–3.
9. Rosen EL, Baker JA, Soo MS. Malignant lesions initially sub-
jected to short-term mammographic follow-up. Radiology
2002;223:221–8.
10. Lehman CD, Rutter CM, Eby PR, White E, Buist DS, Taplin
SH. Lesion and patient characteristics associated with malignancy
after a probably benign finding on community practice mammogra-
phy. Am J Roentgenol 2008;190:511–5.
11. Chung CS, Giess CS, Gombos EC, et al. Patient compliance
and diagnostic yield of 18-month unilateral follow-up in surveillance
of probably benign mammographic lesions. Am J Roentgenol
2014;202:922–7.
12. Baum JK, Hanna LG, Acharyya S, et al. Use of BI-RADS 3-
probably benign category in the American College of Radiology
Imaging Network Digital Mammographic Imaging Screening Trial.
Radiology 2011;260:61–7.
13. Yameen S, Romano PS, Pettinger M, et al. Frequency and
predictive value of a mammographic recommendation for short-
interval follow-up. J Natl Cancer Inst 2003;95:429–36.
14. Kerlikowske K, Smith-Bindman R, Abraham LA, et al.
Breast cancer yield for screening mammographic examinations with
recommendation for short-interval follow-up. Radiology
2005;234:684–92.
15. Monticciolo DL, Caplan LS. The American College of Radi-
ology’s BI-RADS 3 classification in a nationwide screening program:
current assessment and comparison with earlier use. Breast J
2004;10:106–10.
16. NCI-funded Breast Cancer Surveillance Consortium
(HHSN261201100031C). Breast Cancer Surveillance Consortium.
Available at: http://breastscreening.cancer.gov/statistics/benchmarks/
screening/2009/table 4.html (accessed 10/11/15).
17. Boyer B, Canale S, Arfi-Rouche J, Monzani Q, Khaled W,
Balleyguier C. Variability and errors when applying the BIRADS
mammography classification. Eur J Radiol 2013;82:388–97.
18. Coldman A, Phillips N. Incidence of breast cancer and esti-
mates of overdiagnosis after the initiation of a population-based
mammography screening program. CMAJ 2013;185:492–8.
19. Bleyer A, Welch G. Effect of three decades of screening
mammography on breast-cancer incidence. NEJM 2012;367:1998–
2012.
20. Kopans DB, Webb ML, Cady B. The 20-year effort to reduce
access to mammography screening. Cancer 2014;120:2792–9.
21. Stomper PC, Geradts J, Edge SB, Levine EG. Mammo-
graphic predictors of the presence and size of invasive carcinomas
associated with malignant microcalcification lesions without a mass.
AJR 2003;181:1679–84.
22. Hofvind S, Iversen BF, Eriksen L, et al. Mammographic
morphology and distribution of calcifications in ductal carcinoma
in situ diagnosed in organized screening. Acta Rad 2011;52:481–7.
23. Slanetz PJ, Giardino AA, Oyama T, et al. Mammographic
appearance of ductal carcinoma in situ does not reliably predict his-
tologic subtype. Breast 2001;7:417–21.
24. Rubin E. Six-month follow-up: an alternate view. Radiology
1999;213:15–8.
25. Hall FM. Screening mammography guidelines: an alternative
proactive approach. Radiology 2014;273:646–51.
16 • michaels et al.