1. The Midwest Surgical Association
Predicting cancer on excision of atypical ductal hyperplasia
Erin Doren, B.S.a
, Melissa Hulvat, M.D.a
, Jonathan Norton, Ph.D.b
, Prabha Rajan, M.D.c
,
Sharfi Sarker, M.D.a
, Gerard Aranha, M.D.a
, Katharine Yao, M.D.a,
*
a
Department of Surgery, Loyola University Medical Center, 2160 S. First Ave., Maywood, IL, 60153, USA
b
Cardinal Bernardin Cancer Center, Loyola University Medical Center, 2160 S. First Ave., Maywood, IL, USA
c
Department of Pathology, Loyola University Medical Center, 2160 S. First Ave., Maywood, IL, USA
Manuscript received November 26, 2007; revised manuscript November 27, 2007
Abstract
Background: There are no specific histopathologic factors that allow identification of patients with
atypical ductal hyperplasia (ADH) who will have cancer on final excision.
Methods: This was a retrospective study of all patients who had ADH on biopsy followed by excision
from 1999 to 2006.
Results: Fifty-one patients were found to have ADH on core biopsy. Eight (15.7%) patients had invasive
carcinoma on surgical excision, 9 (17.5%) had ductal carcinoma-in-situ (DCIS), 21 (41.5%) had ADH,
4 (8%) patients had atypical lobular hyperplasia, and 9 (17.5%) had benign tumors. The grade of atypia on
the core biopsy was mild in 13 (25%) patients, moderate in 22 (43%), and marked in 16 (32%). On
multivariate analysis of histopathologic factors, the grade of atypia was the only significant variable that
predicted a diagnosis of cancer on final surgical excision (P ϭ .001).
Conclusions: The grade of atypia correlated with the presence of cancer on surgical excision. © 2008
Elsevier Inc. All rights reserved.
Keywords: Breast cancer; Atypical ductal hyperplasia
The standard of care for a diagnosis of atypical ductal
hyperplasia (ADH) on core biopsy is surgical excision to
exclude cancer. Numerous studies have demonstrated that
cancer can be found in 11% to 47% of cases of ADH [1–14].
Variability in cancer rates depends on size of the core
biopsy needle, histologic criteria for ADH versus ductal
carcinoma-in-situ (DCIS), and threshold for excision.
Several studies have examined certain clinicopathologic
factors that predict cancer in subsequent excised speci-
mens [7,10,15] but none are used routinely in practice.
Identifying pertinent clinical and pathologic factors that are
associated with the presence of either invasive or non-
invasive cancer will help surgeons counsel patients on the
need for a surgical procedure when the diagnosis of ADH
has been made. The aim of the current study was to identify
any clinicopathologic factors that would predict the pres-
ence of DCIS or invasive cancer.
Methods
A search of the pathology database at Loyola University
Medical Center for cases of ADH from between 1999 and
2006 revealed 90 cases. The study included only patients
who were diagnosed with ADH on core biopsy and then
underwent a subsequent surgical excision at Loyola Univer-
sity Medical Center; outside core biopsy or outside exci-
sions were excluded. After excluding patients with previous
breast cancers, papillary lesions, and atypical lobular hyper-
plasia, the final study number was 51 cases. All core biopsies
were performed using an 11-gauge needle with stereotactic
guidance and a vacuum-assisted device (Mammotome; John-
son and Johnson, New Jersey). The number of core biopsies
taken was not available but is estimated to be between 8 and
12 per case. Clinical data were obtained from the medical
record and included patient’s age, menopausal status, Gail
model risk, and family history of breast cancer. Mammo-
grams from all patients were reviewed for size of calcifica-
tions, density of the breast, and the Breast Imaging and
Reporting Data System (BIRADS) classification of the
mammographic findings. Density of the breast was classi-
fied into 4 categories (fatty, scattered densities, heteroge-
nous, and extremely dense) according to each individual
* Corresponding author. Tel.: ϩ1-708-327-3433; fax: ϩ1-708-327-3565.
E-mail address: kyao@lumc.edu
The American Journal of Surgery 195 (2008) 358–362
0002-9610/08/$ – see front matter © 2008 Elsevier Inc. All rights reserved.
doi:10.1016/j.amjsurg.2007.11.008

2. radiologist’s reading. All histological slides from the orig-
inal core biopsy and subsequent surgical excision were
reviewed in a blinded fashion by a single pathologist for the
following histopathologic variables: number of ADH foci,
number of ducts involved with ADH, size of ADH, grade of
atypia and architectural pattern (micropapillary vs non-mi-
cropapillary). The distinction between ADH and low-grade
DCIS of the breast was based on both quantitative and
qualitative criteria, the former referring to the extent of
proliferative abnormality (Ͼ2 mm) and the latter referring
to microscopic architectural and cytologic details. ADH was
classified into 3 different grades as mild, moderate, and
marked (Fig. 1). This classification is based on escalating
scale of quantitative and qualitative changes in cyto-archi-
tectural features beginning with the most mild type and it is
similar to methods previously published by Black and Cha-
bon [16] and Wellings et al [17] for proliferative breast
lesions.
Multivariate logistic regression analysis was used to deter-
mine which clinical, radiologic and pathologic factors corre-
lated with a diagnosis of noninvasive or invasive cancer on
final surgical excision. This study was approved by the Insti-
tutional Review Board at Loyola University Medical Center.
Results
A total of 51 patients were found to have ADH on
stereotactic core biopsy and then underwent surgical exci-
sion. Demographics of these patients are listed in Table 1.
All but 4 of the patients had calcifications noted in the core
specimens. Seventeen (33%) of the patients had either DCIS
or invasive cancer on final excision and 25 (49%) had some
form of atypia (Fig. 2). When clinical, radiologic, and
pathologic variables were examined on multivariate analy-
sis, only the grade of atypia was found to significantly
predict a diagnosis of either non-invasive or invasive cancer
on final excision (Table 2). The grade of atypia on the core
biopsy was mild in 13 (25%) patients, moderate in 22
(43%), and marked in 16 (32%). The correlation between
grade of atypia and cancer is broken down by the specific
grade, with marked atypia demonstrating the highest cancer
rate (75%) (Fig. 3).
Of the 9 DCIS cases found on final excision, 4 were
classified as grade I DCIS and 5 were classified as grade II
DCIS; one DCIS had necrosis. Of the 8 invasive cancers
found on final excision, 5 were ductal carcinomas, 1 was
lobular carcinoma, 1 was tubular carcinoma, and 1 was
papillary cancer. All but 1 case were estrogen receptor–
positive and 1 case had tumor-positive nodes.
Comments
This study demonstrated that of all the clinical, radio-
logic, and pathologic factors we examined the only signif-
icant factor that predicted DCIS or invasive cancer on final
excision was the grade of atypia. Of the 51 patients studied,
16 (32%) had marked atypia on core biopsy, of which 75%
had cancer on final excision. Twenty-two patients had mod-
erate atypia, of which 18% had cancer. Specific criteria to
distinguish the level of atypia depend on quantitative and
qualitative criteria focusing on the extent of proliferation,
microscopic architectural and cytologic features, and the
size of the ADH component. Nonetheless, there can still be
considerable variability in diagnosing a lesion as ADH
versus DCIS [18] or “mild” ADH versus hyperplasia. In-
deed, in this study only 1 pathologist examined the slides so
we do not know if these results are truly reproducible. In
many cases of marked ADH, the cells possess all the char-
acteristics for DCIS, but unless there is greater than 2 mm
involvement, it is called ADH.
Two previous studies have examined the severity of
atypia and its relation to findings on final excision. O’Hea et
Fig. 1. Pathologic criteria for mild, moderate, and marked ADH. ADH ϭ
atypical ductal hyperplasia.
Table 1
Demographics of the patients with ADH on core biopsy
Characteristic Description
Age (y) Mean 59 (range 41–91)
Menopausal status
Premenopausal 10 (20%)
Postmenopausal 41 (80%)
Family history (first-degree relatives) 5 (10%)
Density on MGM
Fattry 21 (41%)
Scattered densities 18 (35%)
Heterogenous 6 (12%)
Extremely dense 6 (12%)
Gail risk 1.67 (range 0.6–7.1)
BIRADS classification
BIRADS 3 1 (2%)
BIRADS 4 47 (92%)
BIRADS 5 3 (6%)
ADH ϭ atypical ductal hyperplasia; MGM ϭ mammogram; BIRADS ϭ
Breast Imaging and Reporting Data System.
Fig. 2. Pathologic results in 51 patients excised for ADH. ADH ϭ atypical
ductal hyperplasia; ALH ϭ atypical lobular hyperplasia; DCIS ϭ ductal
carcinoma-in-situ.
359E. Doren et al. / The American Journal of Surgery 195 (2008) 358–362

3. al [7] examined 27 patients with atypia found on core
biopsy and found that 8 patients initially labeled as “mild
atypia” were not actually ADH cases at all according to
Page and Rogers [15] criteria. None of these patients had
cancer on subsequent excision. Adrales et al [19] found that
marked atypia along with family history and residual calci-
fications after Mammotome biopsy predicted benign versus
malignant histology. These studies did not specify exact
criteria for “mild” versus “marked” as we have in this study.
In addition to the grade of atypia, we examined the number
of foci, the number of ducts, size of ADH, and architectural
pattern. Only the grade of atypia predicted cancer on final
excision in our study but others have shown that the number
of ducts or number of foci of ADH did predict cancer on
final excision [19]. Perhaps with larger numbers we would
see these associations.
One issue is on what exactly the biopsy is being per-
formed in the first place. Were the target calcifications
adequately sampled? Were calcifications the actual target
lesion or was a density or mass seen? Four of our patients
did not have calcifications noted in the specimen cores and
1 of these patients did have cancer on final excision. Fur-
thermore, 5 patients had an initial biopsy of a mass or
“density” that was not the typical mammographic presenta-
tion of ADH. This may explain why we had such a high
cancer rate on final excision. At the beginning of the study
we searched the pathology database for ADH seen on core
biopsy, not for mammographic findings. Perhaps the biopsy
was performed on the edge of a cancer and that was why
ADH was only seen. It is imperative that the surgeon verify
concordancy of the biopsy result with mammographic find-
ings to insure that the target lesion was adequately sampled.
In addition, an adequate number of core specimens should
be taken—between 8 and 12, which was the practice at our
institution. Lastly, a proper biopsy device should be used,
ie, a 9- to 11-gauge needle with vacuum assistance [9–14].
A significant strength of our study was that all biopsies were
performed with an 11-gauge vacuum-assisted Mammotome
needle.
ADH is considered a high risk lesion and falls between
benign hyperplasia and ductal carcinoma in situ when ex-
amined on breast cancer models of tumor progression. Al-
though some studies have suggested that ADH is a direct
precursor to cancer [20], it is generally believed that ADH
is a marker for increased risk of breast cancer, usually
reported at 4 to 5 times that of the general public and in
those with family history, 11 times that of the general public
[21], although some studies dispute whether family history
influences the risk of cancer [22]. In the classic study by
Page et al [23], 12% of 150 women with ADH developed
invasive cancer at an average period of 16 years, and 44%
Table 2
Multivariate analysis of clinical and pathologic variables examined and
significance in predicting cancer on final excision
Variable Variable description P value
Age (y) Mean 59 (range 41–91) NS
Size of calcifications (cm) 1.4 (range .4–3.6) NS
No. of foci of ADH Mean 1.14 (range 1–2) NS
Size of ADH (mm) Mean 0.67 (range 0–3) NS
No. of ducts Mean 2.98 (range 1–10) NS
Grade of atypia .001
Mild 13 (25%)
Moderate 22 (43%)
Marked 16 (32%)
Architectural pattern NS
Micropapillary 17 (33%)
Non-micorpapillary 34 (67%)
ADH ϭ atypical ductal hyperplasia; NS ϭ not statistically significant.
Fig. 3. Grade of atypia and DCIS/carcinoma on final excision. DCIS ϭ ductal carcinoma in situ.
360 E. Doren et al. / The American Journal of Surgery 195 (2008) 358–362

4. were in the contralateral breast, suggesting that ADH is not
a direct precursor. However, the study only looked at inva-
sive cancer occurrence and not DCIS. In a recent study from
the Mayo Clinic, approximately 20% of patients with ADH
developed cancer over a follow-up period of 13.7years [22].
The investigators did not find an association with family
history. Nonetheless, just as there are different levels of
severity of invasive or non-invasive cancer, there exists a
way to “grade” ADH into mild, moderate, and marked
categories. This classification scheme does not imply that
ADH is a direct precursor of cancer, but it may enable
clinicians to stratify patients into different “levels of risk,”
which can be useful when discussing a diagnosis of ADH
with patients and explaining the need for further surgical
excision. Indeed, Degnim et al [22] showed that another
pathologic factor of ADH, the number of foci, conferred
a much higher risk of developing cancer over long-term
follow-up.
Subsequent excision of ADH found on a core biopsy
reveals either invasive or non-invasive cancer in 11% to
47% of cases [1–14]. In our study we found cancer in 17
(33%) patients, 8 with invasive cancer and 9 with DCIS.
Our numbers appear on the high end of the spectrum, which
calls into question whether some ADH lesions, especially
the “marked” ADH lesions, would have been classified as
low-grade DCIS at another institution or by another pathol-
ogist. As mentioned previously, distinguishing ADH from
DCIS can be difficult and disagreement among pathologists
does exist [18]. An experienced breast pathologist re-re-
viewed all of the slides in this study and used strict criteria
(see Methods) to determine the overall diagnosis of ADH
versus cancer. However, to validate this grading system will
require further evaluation by other pathologists and institu-
tions to show reproducibility of our system and interob-
server concordancy. At this point we do not recommend
using these criteria exclusively to make decisions about
further surgical excision for a core biopsy diagnosis of
ADH. Indeed, in our study 1 patient with mild ADH had
invasive cancer on further excision. Based on these re-
sults, we still advise patients with all grades of ADH on
core biopsy to undergo surgical excision. With further
studies, perhaps we will be able to use our grading
system in the future to exclude some ADH patients from
surgical excision.
References
[1] Moore MM, Hargett CW III, Hanks JB, et al. Association of breast
cancer with the finding of atypical ductal hyperplasia at core breast
biopsy. Ann Surg 1997;225:726–33.
[2] Liberman L, Smolkin JH, Dershaw DD, et al. Calcification retrieval
at stereotactic, 11 gauge directional, vacuum-assisted breast biopsy.
Radiology 1998;208:251–60.
[3] Gadzala DE, Caderbom Bolton JS, et al. Appropriate management of
atypical ductal hyperplasia diagnosed by stereotactic core needle
biopsy. Ann Surg Oncol 1997;4:283–6.
[4] Jackman RJ, Nowels KW, Shepard MJ, et al. Stereotaxic large core
needle biopsy of 450 nonpalpable breast lesions with surgical corre-
lation in lesions with cancer of atypical hyperplasia. Radiology 1994;
193:91–5.
[5] Brown TA, Wall JW, Christensen EP, et al. Atypical hyperplasia in
the era of stereotactic core needle biopsy. J Surg Oncol 1998;67:
168–73.
[6] Lin PH, Clyde JC, Bates DM, et al. Accuracy of stereotactic core
needle biopsy in atpyical ductal hyperplasia. Am J Surg 1998;175:
380–2.
[7] O’Hea BJ, Tornos C. Mild ductal atypia after large core needle biopsy
of the breast: is surgical excision always necessary? Surgery 2000;
128:738–43.
[8] Sneige N, Lim SC, Whitman GJ, et al. Atypical ductal hyperplasia
diagnosis by directional vacuum assisted stereotactic biopsy of breast
microcalcifications. Am J Clin Pathol 2003;119:248–53.
[9] Liberman L, Hann LE, Dershaw DD, et al. Mammographic findings after
stereotactic 14-gauge vacuum biopsy. Radiology 1997;203:343–7.
[10] Burbank F. Mammographic findings after 14-gauge automated needle
and 14-gauge directional, vacuum assisted stereotactic breast biop-
sies. Radiology 1997;204:153–6.
[11] Jackman RJ, Burbank F, Parker SH, et al. Accuracy of sampling
ductal carcinoma in situ by three stereotactic breast biopsy methods.
Radiology 1998;209:197–8.
[12] Jackman RJ, Burbank F, Parker SH, et al. Atypical ductal hyperplasia
diagnosed by 11-gauge, directional, vacuum assisted breast biopsy:
how often is cancer found at surgery? Radiology 1997;205:325.
[13] Winchester DJ, Bernstein JR, Jeske JM, et al. Upstaging of ADH after
vacuum-assisted 11 gauge stereotactic core needle biopsy. Arch Surg
2003;138:619–22.
[14] Sohn V, Arthurs Z, Herbert G, et al. Atypical ductal hyperplasia:
improved accuracy with the 11 gauge vacuum assisted versus the
14-gauge core biopsy needle. Ann Surg Oncol 2007;14:2497–501.
[15] Page DL, Rogers LW. Combined histologic and cytologic criteria for
the diagnosis of mammary atypical ductal hyperplasia. Hum Pathol
1992;23:1095–7.
[16] Black EM, Chabon AB. In-situ carcinoma of the breast. Pathol Annu
1969;4:185–210.
[17] Wellings SR, Jensen HM, Marcum RG. An atlas of subgross pathol-
ogy of human breast with special reference to possible precancerous
lesions. J Natl Cancer Inst 1975;55:231–73.
[18] Rosai J. Borderline epithelial lesions of the breast. Am J Surg Pathol
1991;15:209–21.
[19] Adrales G, Turk P, Wallace T, et al. Is surgical excision necessary for
atypical ductal hyperplasia of the breast diagnosed by mammotome?
Am J Surg 2000;180:313–5.
[20] Larson PS, de las Morenas A, Cerda SR, et al. Quantiative analysis of
allele imbalance supports atypical ductal hyperplasia lesions as direct
breast cancer precursors. J Pathol 2006;209:307–16.
[21] Dupont WD, Page DL. Risk factors for breast cancer in women with
proliferative breast disease. N Engl J Med 1985;312:146–51.
[22] Degnim AC, Visscher DW, Berman HK, et al. Stratification of breast
cancer risk in women with atpyia: a Mayo cohort study. J Clin Oncol
2007;25:2671–7.
[23] Page DL, Dupont WD, Rogers LW, et al. Atypical hyperplastic
lesions of the female breast. Cancer 1985;55:2698–2708.
Discussion
Dr. Elizabeth A. Mittendorf (Houston, TX): The au-
thors have attempted to employ a more rigorous objective
and standardized pathologic assessment of core biopsy sam-
ples that are initially identified as ADH in order to identify
any potential clinical pathologic features that could predict
finding carcinoma on subsequent excisional biopies. In your
study population, it does seem to be effective in better
stratifying patients with ADH with respect to the likelihood
of identifying DCIS or invasive cancer. In order for this to
be broadly applicable and have clinical utility, your findings
would need to be vertified. So my question to you is whether
or not the additional pathologists, either at your institution
or an outside institution have had the opportunity to attempt
to employ this pathologic standard? In addition, have you
given any consideration to validating your scoring system
by obtaining slides from another outside institution? The
Gail model risk in your patient population was 1.67, so let’s
361E. Doren et al. / The American Journal of Surgery 195 (2008) 358–362

5. call it 1.7, which, by the NSABP P1 trial, indicates a group
of patients who are at increased risk for breast cancer. I
think you’d be hard-pressed not to ever excise 1.7 Gail risk
model patient with an ADH diagnosis, so if you can just
comment on that.
One final question, could identify for us some potential
future directions with respect to molecular profiling for some
of these atypical lesions, as opposed to invasive cancers.
Dr. Yao: When we did this study, we had one breast-
trained pathologist who was working with us on this. But
since the inception of this study, there has been another
breast-trained pathologist who has come to our institution,
so our plan is to have this other pathologist get trained up in
this method and look at these cases, as well, so we can show
that this grading system is reproducible. And I think it is a
great idea to branch it out to other institutions.
With respect to the Gail model, most of our patients were
postmenopausal, and the Gail model reflects that. I think it
is definitely something you need to take into consideration
when you are looking at these high-risk patients and the
need for further surgery. Molecular profiling of ADH le-
sions would be great and wonderful, but it’s very difficult to
do. We have been studying receptors in the lab this on
atypical ductal hyperplasia lesions, and it’s very difficult to
get the tissue. Fresh tissue is a totally separate issue, but just
looking at paraffin-imbedded tissue, once you take your
slices from the block just to make your diagnosis, you’ve
lost your ADH. So we’ve had a lot of difficulties in obtain-
ing specimens, but I agree it would be a great project.
Dr. Dennis E. Johnson (York, PA): Your rate of cancer
was 33% in these biopsies. Our institution has looked at our
rate of cancer in ADH, and it’s much less than that, prob-
ably more like 10% to 15%, which leads me to the question,
which is, how extensive are you sampling these abnormal
areas? How many cores are being taken? Is MRI being
employed to evaluate any of these patients prior to trials?
Dr. Yao: At our institution, anywhere from 8 to 12 cores
are taken, and, unfortunately, our radiologists do not specify
the exact number that they obtain. So all I can say is, we
know it’s anywhere from 8 to 12. I think the other question
is, what exactly are they biopsying and are they actually
biopsying what they were supposed to be biopsying, and
most of our patients had calcifications. But some patients
did have some density, so we have to ask ourselves, maybe
that was a cancerous lesion to begin with and not just the
typical presentation of atypical ductal hyperplasia that is
calcifications.
We have not used MRI to evaluate these patients, but I’m
hoping that maybe in the future we might start a study
looking at that.
Dr. Donn M. Schroder (Grosse Pointe, MI): At what
point will you make the recommendation with ADH not to
do an excisional breast biopsy based on a core? Certainly
you can have a core needle biopsy that has only one of the
cores that shows mild ADH or you could have eight of your
cores all show mild ADH. They are both termed mild ADH.
At what point do you say, “we’re satisfied with this in our
core, we don’t need to proceed to an excisional breast
biopsy”?
Dr. Yao: This is the whole reason why I did this study,
because I was getting a lot of reports coming back as mild
or focal atypia and questioning the need for further excision,
but you can see in our study, one of the patients who had
mild atypia ended up having a cancer, and it was an invasive
cancer. Now that patient, on retrospect, was one of the
patients that actually had a density on mammograms, so we
have to question whether there was really cancer there, to
begin with. I think you have to look at that and look at your
concordency rate, but I think we need to show that this
system is reproducible and that multiple pathologists can
grade these lesions accurately, and it would be great to get
another institution’s input, as well.
362 E. Doren et al. / The American Journal of Surgery 195 (2008) 358–362