Clinicopathologic Implications of “Flat Epithelial Atypia” in Core Needle Biopsy Specimens of the Breast

z Image 2 z Spectrum of atypical ductal hyperplasia seen in our core needle biopsy cases. A , B , C , and D , Atypical ductal hyperplasia characterized by complex architectural atypia (formation of rigid bars, bridges, and micropapillary projections) and cytologic atypia ( A , H&E, ×40; B

Flat epithelial atypia (FEA) is an emerging entity of uncertain clinical significance, and outcome data are sparse. The aim of this study was to evaluate the clinicopathologic significance of this entity for proper management. All core needle biopsy (CNB) specimens diagnosed as atypical ductal hyperplasia (ADH) from January 2006 to April 2008 were retrieved. H&E-stained slides of 5 levels on each case were reviewed. The differences in upstaging in subsequent excisions in the FEA and ADH group (31/189 [16.4%]) vs the pure FEA group (5/35 [14%]) and pure FEA (5/35 [14%]) vs pure ADH (5/45 [11%]) were not statistically significant. We observed that FEA evolved into ADH at the same site at an average of 3 to 4 levels. Our study concludes that there is an association of FEA with ADH on multiple levels of CNB specimens, and follow-up surgical excision findings for FEA are clinically significant.
Flat epithelial atypia (FEA) is a newly emerging entity of uncertain clinical significance, known in the past to pathologists as columnar cell alterations with apical snouts and secretions with atypia; atypical cystic lobules; ductal intraepithelial neoplasia, flat type; and clinging carcinoma. 1 Columnar cell alterations are classified into 2 categories: columnar cell change (CCC) and columnar cell hyperplasia. The vast majority of these lesions are considered benign, and the clinical significance of these entities is under debate. Some of them manifest with cytologic or architectural atypia or both. When architectural atypia is associated, a careful search for atypical ductal hyperplasia (ADH) or for low-grade ductal carcinoma in situ (DCIS) needs to be performed.
FEA is characterized by the dilated ducts lined by 1 or 2 to 3 layers of atypical cuboidal or columnar cells. The cells retain the apical snouts, and the acini contain flocculent secretions and may be associated with calcifications. 2-4 FEA differs from CCC by the presence of cytologic atypia and from ADH by the absence of complex architectural atypia. The clinical significance of FEA is beginning to emerge in the literature.
Retrospective studies have shown a high association of these lesions with putative precursor lesions such as lobular neoplasia, low-grade DCIS, and tubular carcinoma, which includes the described Rosen triad. [5][6][7][8][9] Based on the results, it was suggested that FEA potentially has a "precursor" role in the development of low-grade DCIS. The role of FEA as a risk lesion needs to be understood, and the question of whether women with FEA should have a follow-up surgical excision remains unanswered.
The outcome data on the incidence of FEA found in core needle biopsy (CNB) specimens are limited, and the reported frequency of upstaging of pure FEA to cancer in the follow-up resection specimens in a single study was shown to be 21%. 10 A high association of FEA is also shown with ADH, suggesting the association of these lesions in the low-grade DCIS pathway. 11 As the biologic mechanism of FEA as a presumed precursor is emerging, establishing the significance of this entity in CNB specimens is critically important. The objectives of this study were to determine the association, if any, of FEA with ADH on multiple levels of core breast biopsy specimens and determine the clinical significance of FEA with regard to follow-up surgical breast excisions to determine the proper management for these lesions.

Materials and Methods
Institutional review board approval was obtained for our study. All CNB specimens diagnosed as ADH (included pure FEA, FEA and ADH, or pure ADH) at our institution for a 2-year period (January 2006-April 2008) were retrieved from our pathology files. H&E-stained slides of 5 levels on each case were reviewed, as per our departmental protocol for core biopsy specimens. 12 The diagnostic criteria for FEA used in this study were those adopted by the World Health Organization Working Group on the Pathology and Genetics of Tumours of the Breast. 4 FEA is defined as "acini lined by monomorphic cells that exhibit low-grade cytologic atypia, with apical snouts and with or without calcifications." [2][3][4] The cells are evenly spaced and single layered with enlarged round or ovoid nuclei with inconspicuous nucleoli and abundant eosinophilic cytoplasm; the cells sometimes are stratified (more than 2 cell layers) 3 with loss of polarity but lack complex architectural patterns such as rigid bars, bridges, and micropapillary projections.
CCC, on the other hand, is composed of uniform, ovoid to elongated nuclei oriented perpendicular to the basement membrane; nucleoli are absent or inconspicuous. FEA is distinguished from CCC by the presence of round, ovoid nuclei with or without prominent nucleoli zImage 1z.
ADH was defined in our study as lesions manifesting with architectural and cytologic atypia that fall short of the diagnosis of DCIS. This definition encompasses lesions that do not entirely involve a single duct. The architectural changes are represented as focal establishment of roman arches, cribriform pattern, or micropapillary projections zImage 2z. FEA is distinguished from ADH by the absence of the aforementioned complex architectural features. Although FEA shows cells that resemble those seen in low-grade DCIS, it differs from it by the absence of the aforementioned combined complex architectural and cytologic features that involve the duct entirely.
Radiology records were reviewed to ascertain the reason for biopsy and the method of biopsy. At our institution, magnetic resonance imaging-guided biopsies are performed with 9-gauge, vacuum-assisted needles with a 12-mm-long trough; ultrasound-guided and stereotactic-guided biopsies are routinely performed with 14-gauge, non-vacuum-assisted needles and sometimes with 14-and 12-gauge vacuumassisted needles. Each CNB specimen was evaluated for the presence of microcalcifications and the relationships between FEA and ADH. Because 5 levels were examined for each case, we evaluated and tabulated each level for the presence of FEA and ADH with reference to their respective positions in the tissue levels. As part of our routine histology protocol, we cut 10 blanks-2 before level 1, 2 after level 1, 2 after level 2, 2 after level 3, and 2 after level 4.
The follow-up resection reports were reviewed for presence of a "significant lesion," which, in our study, included DCIS, invasive carcinoma (IC), or both. ADH (which was pure FEA or ADH and FEA) and the presence of other highrisk lesions were also noted. High-risk lesions were defined in our study as nonmalignant lesions that predispose patients to an increased risk of developing breast cancer.

Statistical Analysis
The χ 2 test was used to compare 2 proportions and generate the P values.

Results
The total number of CNBs performed from 2006 to 2008 was 8,054. More than 99% (8,051/8,054) were stereotacticguided and fewer than 1% (3/8,054) were ultrasound-or magnetic resonance imaging-guided biopsies. The incidence of ADH (included pure FEA, FEA and ADH, or pure ADH) was 3.7% (301/8,054) and 99% of these cases were biopsied for abnormal calcifications. The mean age of patients in the ADH group was 54 years (range, 29-83 years). The CNB cases associated with DCIS or IC were excluded from the study. For purposes of analysis, we divided the ADH cases into 3 groups: 1, ADH and FEA; 2, pure FEA; and 3, pure ADH.
Microcalcifications were seen in association with ADH in all 301 cases (100.0%). Besides ADH, microcalcifications were also seen in association with 4 cases of intraductal papilloma (1 case in Group 1 and 3 cases in Group 3) and 1 case of LN in Group 2.
In 17.3% of the cases in group 1 (37/214), we observed that FEA evolved into ADH at the same site at an average of 3 to 4 levels of 5 levels obtained as part of our routine protocol zFigure 1z and zImage 4z.
All surgical excisions for a diagnosis of ADH at our institution are entirely submitted for microscopic examination.
In group 3, follow-up was available for 94% of cases (45/48). In the follow-up resections, DCIS/IC was found in 11% (5/45), ADH (included pure FEA, FEA and ADH, or pure ADH) in 53% (24/45), and LN in 4% (2/45); the lesions were benign in 31% of cases (14/45) zTable 2z.The breakdown of DCIS and IC are shown in detail in Table 2. The most frequently encountered type of DCIS in all groups had low-grade morphologic features. Only 1 case of high-grade DCIS was encountered (in group 1).

Statistical Analysis
The difference in upstaging in group 2 was compared with those in group 1 and group 3. The differences in upstag-

Discussion
FEA is a constituent of the spectrum of columnar cell lesions (CCLs) that encompasses columnar cell change and columnar cell hyperplasia with atypia but is distinguished from ADH and DCIS by the absence of architectural atypia. 3 As the biologic and molecular silhouette of FEA is unraveling, with the rising association of FEA and low-grade neoplasia, 6,11 it is imperative to be familiar with this lesion in breast specimens.
Our study is one of the largest to highlight the significance of FEA in CNB specimens and also is one of the foremost to evaluate the significance of FEA with reference to the importance of examination of the appropriate number of tissue levels. Our data demonstrate that FEA evolves into ADH at the same site in 3 to 4 tissue levels in a significant number of cases (17.3%), indicative of the interrelationship of ADH and FEA, suggesting that FEA may be the surface depiction of an associated underlying significant architectural lesion. This study, along with the accruing literature on FEA, indicates that the finding of FEA in a CNB breast specimen has the same clinical significance as the finding of ADH in a CNB specimen. It is critically important to be able to properly recognize FEA in CNB levels so that patients will obtain proper surgical follow-up and management. Additional levels of CNB specimens that display FEA commonly demonstrate the criteria for ADH. Our data also show that when comparing the diagnosis of FEA with that of ADH in CNB specimens, there is no significant difference in the upstaging to carcinoma on the follow-up surgical resections. Recent molecular studies have shown 2 distinct pathways for the development of breast carcinoma, low-grade and highgrade. 11 Low-grade breast carcinomas strongly express estrogen receptor and have a good prognosis; included in this group are tubular, tubulo-lobular, cribriform, and lobular carcinoma subtypes. CCLs have been shown to be potential precursors in the pathway of development of low-grade carcinoma and are often seen in association with these low-grade malignant entities, in particular, FEA. Furthermore, the similarities that exist in FEA and low-grade DCIS at molecular and genetic levels suggest further that FEA is a nonobligate precursor for the development of low-grade breast carcinomas. A retrospective review of low-grade DCIS showed a significant association with the presence of FEA, particularly with the cribriform and micropapillary subtypes. 13 Low-grade DCIS associated with ADH showed a higher association of FEA, further confirming the association with the low-grade pathway. [13][14][15] In recent literature, LN is integrated as an important component of the family of low-grade neoplasms owing to its high frequency of association with CCLs; in particular, the association of FEA is reported to be up to 26%. 3,5,14 LN is also a known risk factor for breast carcinoma and may behave as a premalignant process requiring excisional biopsy. In our study, we observed the most common high-risk lesion encountered in all 3 groups included LN, indicating the close relation of LN and FEA.
The biologic significance of finding FEA in CNB specimens is meager, and the current management remains a dilemma. The clinical concern of whether patients with FEA found by CNB need to have follow-up excision, analogous to the management of patients with ADH diagnosed by CNB, remains to be determined. Data from a few small studies, some reported only in abstract form, indicate that follow-up excisional biopsy of FEA will reveal associated DCIS or IC in up to 30% of cases. 10 This study, although one of the largest, is still limited by the number of pure FEA  cases that have follow-up surgical excision. The data suggest that patients with a CNB diagnosis of FEA will have a 14% incidence of upstaging to DCIS or IC on resection, not statistically different from the incidence of upstaging with ADH. This mathematical observation reinforces the morphologic kinship between FEA and ADH seen on multiple tissue levels. In practice, if there is diagnostic uncertainty about the diagnosis for a CNB specimen, it would be prudent to examine further tissue sections or seek consultation if FEA is in the differential diagnosis.