Flat Epithelial Atypia on Core Needle Biopsy

Which is the Right Management?

Abstract

The clinical significance and management (surgical excision vs. follow-up) of the patients with the diagnosis of flat epithelial atypia (FEA) on core needle biopsy (CNB) are actually under discussion. Using standardized criteria and precise terminology, we analyzed retrospectively our CNB diagnosis of FEA, dividing patients with pure FEA as the most advanced pathologic lesion from patients with FEA associated to atypical ductal hyperplasia (FEA+ADH). Both the categories were correlated with radiologic data and findings on subsequent surgery. We evaluated 875 core needle biopsies (11-gauge stereotactic vacuum-assisted procedure), performed over a 5-year period. A CNB diagnosis of pure FEA was made in 33/875 (3.7%) cases; in other 11 (1.2%) cases we observed the coexistence of FEA and ADH. Subsequent surgical excisions were available in 20/33 pure FEA and in 10/11 FEA+ADH: of the 20 patients with pure FEA on CNB, none had either ductal carcinoma in situ or invasive carcinoma in their excisional biopsy, whereas 3/10 (30%) FEA+ADH on CNB showed, at subsequent surgery, more advanced lesions (2 ductal carcinoma in situ, 1 invasive carcinoma). Our results suggest that patients with an 11-gauge vacuum-assisted CNB diagnosis of pure FEA (especially if related to a small radiologic target, completely or almost completely removed by the needle biopsy procedure) could be spared surgical excision and managed with close radiologic follow-up.

As defined by World Health Organization Working Group on Pathology and Genetics of Tumors of the Breast,³² flat epithelial atypia (FEA) is an “intraductal alteration characterized by replacement of the native epithelial cells by a single or 3 to 5 layers of mildly atypical cells.” The same entity is reported in the literature with many different terms including clinging carcinoma-monomorphous type,³ columnar alteration with prominent apical snouts and secretion,¹² atypical cystic lobules,^17,26 columnar cell change with atypia, and columnar cell hyperplasia with atypia.³⁰ Owing to low-grade cytologic atypia and absence of architectural abnormality, for many years FEA remained unrecognized or misinterpreted as a part of fibrocystic disease, blunt duct adenosis or even as normal breast tissue. Recently, there has been a growing interest in this lesion with the increasing frequency of its detection in breast biopsy performed for mammographically detected microcalcifications.

Although the biologic and clinical significance of FEA is still far to be completely elucidated, at present we do know that flat atypia frequently coexists with several types of low-grade carcinoma [micropapillary and cribriform ductal carcinoma in situ (DCIS), invasive well-differentiated/tubular carcinoma, lobular neoplasia (LN)]^{1,12,13,26,28} and shows the same genetic alterations of eventually associated ductal carcinomas—both invasive and in situ^23,31: these findings lead to conclude, as stated by Schnitt,²⁹ that “at least some flat atypias may well represent either a precursor of or the earliest morphologic manifestation of low-grade DCIS, and being a precursor to invasive carcinoma, particularly tubular carcinoma.”

From a prognostic point of view, the few available studies^4,11,21 concerning the natural history of FEA suggest a very low risk of neoplastic progression for this lesion: therefore, flat atypia can be considered a marker of slightly increased risk for subsequent development of invasive breast carcinoma²¹ or, in other terms, an indolent, nonobligate precursor of ductal carcinoma.¹⁹

A more immediate clinical dilemma concerns the management of the patients with diagnosis of FEA on core needle biopsy (CNB): to excise or not to excise? The current dominant recommendation for surgery after a needle biopsy diagnosis of FEA^16,25,29 is based on limited data^5,7,14,15,24 coming from small studies, almost all published in abstract form, in which subsequent excision showed a more advanced lesion, either DCIS or invasive carcinoma, in approximately one-third of cases. These studies, however, lack a standardized terminology, including in their series cases with pure FEA and cases with FEA plus more advanced lesions with architectural atypia in form of arcades, tufting or cribriform arrangement, as we expect in atypical ductal hyperplasia (ADH): so we do not know how many of the reported cases are pure FEA and how many include ADH too. At our knowledge only few studies concerning the significance of flat atypia on CNB have used a precise and nonambiguous terminology, distinguishing pure FEA from FEA associated to ADH (FEA+ADH)¹⁸ or accepting in their analysis only cases of pure, well-defined, FEA.²¹

The aim of this study was to evaluate the frequency, the association with ADH and the significance of FEA in our mammotome CNB by correlations with radiologic data and subsequent surgical excisions.

MATERIALS AND METHODS

At our hospital, the Ospedale Civile Maggiore (Verona, Italy), 875 CNBs were performed from May 2003 to October 2008, using an 11-gauge stereotactic vacuum-assisted instrument (Mammotome). The files of the Pathology Department were searched for diagnosis of FEA and all slides were reviewed by authors (Q.P., A.P.). As a part of our routine protocol, we cut 3 hematoxylin and eosin slides at different levels per block. In presence of FEA we performed 3 more levels and some unstained sections were saved for potential immunohistochemistry.

According to established criteria,^21,30,32 we included in our analysis cases of pure FEA as the most advanced lesion (variably distended acini lined by 1 to several layers of monotonous, mildly atypical, cuboidal to columnar cells growing in an exclusively, real “flat” pattern, with complete absence of intraluminal proliferation with architectural atypia) and cases in which FEA was associated to concomitant ADH (in a background of flat atypia even the presence of a single atypical intraluminal structure as arcade, bar, roman bridge, tuft, or cribriform-micropapillary formation was considered enough for a diagnosis of concomitant ADH). Cases of FEA associated to DCIS, invasive carcinoma or LN (atypical lobular hyperplasia and lobular carcinoma in situ) were excluded from the study.

Moreover, the clinical and radiologic records of the cases included were reviewed to establish: (1) mammographic findings (calcifications vs. mass), (2) breast imaging reporting and data system classification of the lesion,² (3) lesion size on mammography, and (4) entity of lesion removed on mammograms performed after needle biopsy: lesion entirely or almost entirely removed (≥90%) versus lesion only partially removed (<90%).

Finally, the pathology files were searched for subsequent surgical procedures: for all cases available, slides were reviewed and final diagnosis on excision biopsy was recorded and compared with the needle biopsy findings.

RESULTS

A CNB diagnosis of pure FEA (Fig. 1) was made in 33/875 cases (3.7%), whereas in other 11 cases (1.2%) we observed the coexistence of FEA and ADH (Fig. 2). All the patients were women, ranging from 38 to 83 years old. For pure FEA the mammographic alterations leading to CNB were calcifications in 30/33 cases (90.9%) and mass in the remaining 3 cases (9.1%); all cases of FEA+ADH underwent CNB for calcifications.

FIGURE 1.: Pure flat atypia. A, Variably distended acini with epithelial cells in flat growth pattern, intraluminal calcifications and secretion (low power). B and C, Distended acini lined by 1 to 3 layers of monotonous cells, with complete absence of architectural atypia (no arcades, bridges, micropapillae, or cribriform structures) (medium power). D, Acinar unit lined by 3 to 5 layers of monotonous, mildly atypical, cuboidal to columnar cells, with relatively round (rather than elongated) nuclei and loss of polarization (high power).

FIGURE 2.: Coexistence of flat atypia and atypical ductal hyperplasia. A and B, In a background of flat atypia, presence of initial intraluminal atypical proliferation in the form of 1 to 2 rigid arcades or micropapillary-cribriform structures (high/A and medium/B power). C and D, More advanced intraluminal proliferation with architectural abnormalities in the form of bridges, tufts, and micropapillary protrusions, again in the setting of flat atypia (high/C and medium/D power).

Subsequent surgical excisions were available in 20/33 (60.6%) pure FEA and in 10/11 (90.9%) FEA+ADH. Table 1 summarizes the results: none of the 20 patients with pure FEA on CNB had either DCIS or invasive carcinoma in their excisional biopsy (4 cases were totally negative, 10 had residual pure FEA, 5 had FEA and LN, 1 had FEA and ADH). Three of 10 (30%) FEA+ADH on CNB showed, at subsequent surgery, more advanced lesions (2 DCIS, 1 invasive carcinoma).

TABLE 1: Findings on Surgical Excision After Core Needle Biopsy Diagnosis of Pure FEA or FEA+ADH

Tables 2 and 3 describe in detail the clinicoradiologic features respectively of patients with pure FEA and of patients with FEA+ADH, who underwent surgical excision. The median age of patients was 47.5 years in pure FEA and 50 years in FEA+ADH. In both categories most of the cases presented lesions with a low level of radiologic suspicion (breast imaging reporting and data system 3 in 18/20 cases of pure FEA and in 7/10 cases of FEA+ADH), entirely or almost entirely removed (14/19 cases of pure FEA and 7/10 cases of FEA+ADH) by mammotome. Noteworthy, in the 3 cases (no. 2, 3, and 6) of FEA+ADH, subsequently upgraded to DCIS or invasive carcinoma on excisional biopsy, the radiologic target was entirely or almost entirely removed.

TABLE 2: Summary of Clinicoradiologic Data of Patients Who Underwent Subsequent Surgery After Initial Core Needle Biopsy of Pure Flat Epithelial Atypia

TABLE 3: Summary of Clinicoradiologic Data of Patients Who Underwent Subsequent Surgery After Initial Core Needle Biopsy of Flat Epithelial Atypia Associated to Atypical Ductal Hyperplasia

The radiologic size range was 2 to 25 mm (median: 5 mm) for patients with pure FEA and 5 to 15 mm (median 10 mm) for patients with FEA+ADH.

DISCUSSION

Although limited by its retrospective nature and small number of cases, the current work may offer some contributions to the assessment of the clinical and biologic significance of FEA and to the management of the patients with CNB diagnosis of FEA.

Pure FEA was recognized in 3.7% (33/875) of our CNB: a similar frequency (3.6%) was reported by other authors.²¹ Clinically, we confirm that almost all (90.9%) patients with pure FEA underwent needle core biopsy for calcifications, as previously noted by Kunju and Kleer¹⁸, Martel et al²¹: in these studies CNB was performed because of mammographically detected calcifications in 86% and 73% of their pure FEAs, respectively. Interestingly, in our series, patients with pure FEA, compared with those with FEA+ADH, were slightly younger (median age: 47.5 vs. 50 y) and had smaller radiologic lesions (median size: 5 vs. 10 mm). A similar (51 vs. 53 y) difference in median age between patients with pure FEA and patients with FEA+ADH was already observed.¹⁸ These data (younger age and smaller lesions), together with those coming from observational studies,^{1,12,13,26,28} support the idea that FEA represents one of the earliest recognizable intraepithelial neoplastic alteration of mammary tissue and its correct inclusion as DIN 1A lesion into the DIN system of the 2003 World Health Organization classification.³²

Actually, the management (follow-up vs. surgical excision) of patients with a CNB diagnosis of FEA is under discussion. Most studies about flat atypia detected on CNB^5,7,14,15,24 used a nonuniform, variable terminology, including in their series not only pure FEA, but also more advanced proliferative lesions characterized by intraluminal protrusions in form of bridges, arcades, or micropapillae: these studies included ADH among flat lesions, making impossible to establish how many of the reported cases were truly pure FEA and how many were indeed ADH. The current dominant recommendation for surgery after a needle biopsy diagnosis of FEA^16,25,29 is based on such confusing studies. Our work may contribute to clarify this topic, analyzing separately patients with pure FEA and those with FEA+ADH.

Three of 10 (30%) FEA associated to ADH on our CNB showed, at subsequent surgery, more advanced lesions (2 DCIS, 1 invasive carcinoma). Not surprisingly, this value is very close to the range (10% to 27%) of underestimation rate reported in the literature^{6,10,20,22,27} for CNB diagnosis of ADH with 11-gauge vacuum-assisted procedure. As known, using automated biopsy gun methods and smaller (14-gauge) needles, the underdiagnosis rates of DCIS or invasive carcinoma, after a CNB diagnosis of ADH, are substantially higher.¹⁰ Interestingly, in all our 3 cases upgraded, the related mammographic abnormalities (calcifications) were entirely or almost entirely removed by the core needle procedure. Thus, in our experience, a CNB diagnosis of FEA associated to ADH (even if carried out with complete or almost complete removal of the radiologic target) shows a risk of upgrading to carcinomatous lesions, in subsequent surgery, similar to that reported in the literature for ADH. So, as it is suggested for the latter,¹⁶ it is prudent to recommend surgical excision for patients with CNB diagnosis of FEA associated to ADH.

Data concerning our CNB diagnosis of pure FEA seem to be more interesting: as a matter of fact, none of the 20 patients with pure FEA on CNB had either DCIS or invasive carcinoma in their excisional biopsy. The recent study by Kunju and Kleer¹⁸ (which used the same diagnostic criteria and terminology to distinguish pure FEA from FEA+ADH) achieved different results: 3 of their 12 pure FEA on CNB, were upgraded to DCIS (1 case) or invasive carcinoma (2 cases) at subsequent surgery. On the basis of that finding, the authors suggest follow-up excision in all the patients with a CNB diagnosis of pure FEA. However, there are some relevant differences between the study by Kunju and Kleer¹⁸ and the current one, which at least in part may explain the different results. First, we used an 11-gauge needle with vacuum-assisted procedure, whereas the study by Kunju and Kleer was performed using a 14-gauge needle (with a biopsy technique not clearly explained). As known, the 11-gauge vacuum-assisted needle method provides larger tissue specimens, with much higher average specimen weights (96 mg vs. 40 mg for 14-gauge vacuum-assisted vs. 17 mg for 14-gauge automated gun).^8,9 Second, we know that, in most (14/19, 73.6%) of our CNB diagnosis of pure FEA, the related radiologic target was entirely or almost entirely removed by mammotome: the work by Kunju and Kleer lacks radiologic correlations and it does not provide informations about the amount of target removed. So, the discrepancies between these 2 studies may be due in large part to the more extensive tissue sampling and to the greater likelihood of complete removal of the radiologic target allowed by 11-gauge vacuum-assisted needle procedure.

The study by Martel et al²¹ used the same terminology for flat atypia (pure flat pattern of growth, uncontaminated by more complex and atypical intraluminal proliferations), but it is conceptually different from the current one. It is basically a follow-up study based on a retrospective assessment and identification of FEA as the most advanced pathologic alteration in core biopsies performed before the widespread recognition of this lesion. Owing to the absence of the diagnostic entity of FEA at the time of initial diagnosis, no immediate (within 3 mo) surgery was performed in the large majority (58/63) of their cases. They found an infiltrating carcinoma in 9/63 patients (7 ipsilateral—absolute risk 11.1%, 2 controlateral—absolute risk 3.2%) with a mean interval between the initial CNB and subsequent surgery of 3.7 years for ipsilateral tumors and of 7 years for controlateral tumors. Only 5 of their patients underwent surgery within less than 3 months of the initial CNB and, as occurred in our study, none had either DCIS or invasive carcinoma in the excisional biopsies.

The necessity of performing surgical excision after a CNB diagnosis of pure FEA is questioned by our findings. The management decision in a given patient with a CNB diagnosis of pure FEA need to be based on many factors: pathologic (standardized criteria and common terminology for the diagnosis), clinicoradiologic (size and characteristics of the lesion, entity of target removal, presence or absence of other lesions, concordance or nonconcordance between histologic findings and radiologic data) and technical (caliber of the needles and method—vacuum-assisted vs. automated gun—of the needle biopsy adopted). Thus, a rational decision about the management of the single patients should be taken as a multidisciplinary task. It seems from our experience that women with a CNB diagnosis of pure FEA (obtained using an 11-gauge vacuum-assisted instrument, often with radiologic target of small size, entirely or almost entirely removed by the core needle procedure) may be spared surgical excision. Larger studies using the same diagnostic criteria and terminology are needed to confirm our results and to fully address this issue. In the meantime, we agree with the methodologic approach proposed by Martel et al²¹: when pure FEA is encountered on CNB, 3 additional levels of section are needed to exclude the presence of a more advanced lesion. If additional levels fail to show more advanced lesions and there are not radiologic features of concern, an excisional biopsy is not mandatory: the patient could be managed with close radiologic follow-up (mammography every 6 mo for 2 to 3 y) for early detection of any clinically occult carcinoma potentially missed nearby the pure FEA.

In summary, the use of standardized criteria and precise terminology allows to separate on CNB patients with pure FEA from patients presenting FEA+ADH. The subsequent management, taking always into account the radiologic characters of the lesion and the technical features of the needle biopsy procedure, could be different: for the first we suggest close follow-up, for the second surgical excision.

ACKNOWLEDGMENT

The authors thank Dr Stefano Gobbo for imaging assistance.