Ductal carcinoma in situ of the breast with apocrine cytology: Definition of a borderline category

Ductal Carcinoma In Situ of the Breast With Apocrine Cytology: Definition of a Borderline Category FRANCES P. O’MALLEY, MB.* DAVID L. PAGE, MD, E. H. NELSON, MD, AND WILLIAM D. DUPONT, PHD to have malignant

potential from those that are truly benign. For the most part, the histologic patterns described by apocrine cells are those commonly seen in papillary apocrine change. However, when extensive (involving at least several lobular units), these apocrine proliferations often demonstrate some of the features of ductal carcinoma in situ (DCIS), ie, distension as well as deformity and cancerization of lobular units4 The most relevant feature is that they often lack the histologic pattern criteria of DCIS.5 Our approach accepts the fact that the pattern criteria present in most types of DCIS are absent in these apocrine lesions. Therefore, other criteria, specifically cytologic features and anatomic extent (ie, size), are highlighted in this study.

Papillary patterns of apoeriue cells are frequent in benign breast biopsies and pose little diagnosticdifficulty when the classic apocrine cytologyis present. Dilemmas in diagnosisdo occur, however,when the usual papillarypatterns of hyperplasticapocrine epithelium are absent in a hyperplasticlesion, particularlyif it is extensive,presents more complex patterns,and/or lacks usual nuclearfeatures.We classified lesions with nonclassicapocrine features into diagnosticcategories to separate those with malignantpotential from those that are trulybenign. We selected 54 such cases sent in consultationover a 2 year period (1989 to 1991). We applied two sets of criteria:nuclear patterns (usual, borderline, and as seen in ductal carcinomain situ) and extent of lesion (~4 mm, 4 to 8 mm, and >8 mm). The cases were categorizedinto diagnosticgroups defined by these criteriaof extent of lesion and cytologicfeatures.We believe that applicationof these criteriarecognixingborderlinefeatures to unusual,apocrinelesions fosters reproducibilityin diagnosis and thus predictabilityof clinicaloutcome. We found only 10 cases to fit into borderline categoriesafter use of the two criteria.The clinicalutilityof thissuggested diagnosticstratificationwill require follow-up studies for linkage of criteriato clinicaloutcomes. HUM PATHOL 25~164-168. Copyright0 1994 by W.B. SaundersCompany

In the human female breast changes regarded as hyperplastic or neoplastic with features of “apocrine” cytology are frequent.’ We regard apocrine features as a combination of cytoplasmic and nuclear characteristic~.*.~These features include a large cytoplasmic compartment that is richly eosinophilic and has fine granularity as well as a nucleus that is large, moderately vesicular, and usually has a prominent red nucleolus when stained with hematoxylin-eosin. These features are remarkably uniform throughout most foci, although an occasional nucleus may be enlarged. These classic features, however, are frequently absent when unusual histologic patterns are found. Dilemmas in diagnosis are particularly challenging when the usual papillary patterns of hyperplastic apocrine epithelium in a hyperplastic lesion are absent. In this study we have attempted to classify these more unusual and difficult apocrine lesions into useful diagnostic categories to separate those that might prove From the Departments of Pathology and Preventive Medicine, Vanderbilt University, Nashville, TN. Accepted for publication Sep tember 7, 1993. Present address: Victoria Hospital, London, Ontario, Canada. Supported in part by National Cancer Institutes grants no ROICA 50468 and 40517. Address correspondence and reprint requests to David L. Page, MD, Department of Pathology, C-3321 Medical Center North, Vanderbilt University, Nashville, TN 37232. Copyright 0 1994 by W.B. Saunders Company 00468177/94/2502-0008$5.00/0

MATERIALS

AND METHODS

Cases from our consultation files coded as DCIS with ape crine cytology, atypical ductal hyperplasia (ADH) with apocrine features, or apocrine lesions with unusual features were reviewed. These cases were selected from approximately 2,500 biopsy specimens submitted for consultation between 1989 and 1991. On histologic review the characteristics noted singly or in combination were (1) cellular populations with apocrine features, but lacking the classic vesicular and uniform nuclei of apocrine change; (2) involvement of lobular units with distortion and distension of spaces; (3) extent of lesion beyond a ductal-lobular unit; and (4) associated lesions, namely, complex sclerosing lesion, papilloma, sclerosing adenosis, ADH, noncomedo DCIS, and invasive mammary carcinoma. The ages of the patients also were noted. After reviewing all the cases coded as DCIS with apocrine cytology, ADH with apocrine cytology, ADH with apocrine features, and apocrine lesions with unusual cytology, we selected 54 cases suitable for inclusion in the study. These cases were categorized into diagnostic groups defined by size and nuclear features as illustrated in Figs 1 through 6 using the nuclear criteria outlined in Table 1. We used two criteria (nuclear features [Table l] and size) to define intersecting categories produced by defining features of each (Fig 7). As there were three groups of cytologic features and three groups of size, the resulting intersecting table has nine potential categories. This is analogous to the usual 2 X 2 table in which comparing two categories of two measures each produces a final table with four cells. The extent or size criteria were chosen recognizing that expanded, single lobular units may approach 4 mm and that most uniform, presumably neoplastic, cell populations over 8 mm in extent should be regarded as at least premalignant. In addition, these lesions of greater extent usually were associated with irregular distension and distortion of the involved base-

164

BORDERLINE

APOCRINE

LESIONS

OF THE BREAST

(O’Malley et al)

ment membrane-bound spaces, and therefore demonstrated growth features that characterize most lesions diagnosed as DCIS. The nine potential cells in our table have been resolved into seven final categories for the following reasons: we recognized that apocrine adenosis and papillary apocrine change with nuclei of the usual apocrine type could extend up to 8 mm in size. We only identified one case with these nuclear features that measured more than 8 mm (ie, “rare” category). Second, lesions with the typical nuclear features (comedo or non-comedo) of DCIS and measuring greater than 4 mm were regarded as DCIS, and a separate category for lesions between 4 to 8 mm in size was not thought necessary. Lesions smaller than this (<4 mm) having nuclei as in DCIS were separated as those most likely to represent limited disease (Fig 7). Resolving these categories along the size axis left three possible borderlines. We have only recognized those lesions greater than 4 mm with borderline nuclear features (see Table 1) as borderline DCIS lesions, ie, lesions 4 to 8 mm in size with a majority of the nuclei showing borderline nuclear features and’ lesions greater than 8 mm with a minority of the nuclei showing borderline features. Lesions greater than 8 mm but with a majority of borderline nuclei were classified as apocrine DCIS (Fig 7). We restricted the potential number of categories somewhat by recognizing the intermediate category of size only in the central cell with regard to nuclear features. Thus, the smaller lesions with benign. usual pattern apocrine nuclei formed a large group and were thought to extend to 8 mm in size, whereas nuclei with the features usually seen with DCIS were recognized as such when only 4 mm in size. This means that the upper left area of Fig 7 is inclusively benign with both benign features consistently present and the lower right is quite inclusive and easily diagnostic as malignant. However, the lower left category of normal nuclei and very large lesions and the upper right category of very small lesions with highly atypical nuclei are either rare and difficult to recognize as completely benign or malignant. The intent of this exercise is to define precise limits, which will allow reproducibility in allocating lesions to the various groups.

limited disease, were separated from the other apocrine DCIS categories. These three cases demonstrated patterns approximating those of cribriform DCIS and thus did not pose a diagnostic problem. Ten of the study cases fell into the two borderline apocrine DCIS categories (Fig 7), which were defined as follows: (I) lesions having a majority of cells with borderline nuclei (Table 1) but only measuring 4 to 8 mm in maximum dimension (seven cases), and (2) those that were much larger (>8 mm) but contained a majority of cells with classic apocrine nuclei admixed with nuclei with borderline cytologic features (type a borderline nuclei; Table 1) (three cases). Eight small lesions (<4 mm) with less than 50% borderline nuclei (type a or b) were separated from the borderline apocrine DCIS category on the basis of extent of lesion (apocrine adenosis or papillary apocrine change with atypia; Fig 7). There was only one case in this series that was extensive (>S mm) and showed apocrine change of the classic type throughout the lesion.. This solitary, remarkable case is descriptively designated only as “rare.” Benign epithelial changes that were commonly associated with both the borderline apocrine DCIS categories and the apocrine DCIS lesions were radial scars/complex sclerosing lesions, sclerosed papillomas, and ductal adenomas.” The cases of tumoral or aggregate adenosis in this series were a.ssociated with the apocrine DCIS or borderline apocrine DCIS lesions. Atypical ductal hyperplasia was present in two cases of apocrine DCIS and in one case of borderline apocrine DCIS. Small foci (<4 mm) of invasive mammary carcinoma of no special type were noted in one apocrine DCIS case and in one case of borderline apocrine DCIS.

RESULTS

Lesions of the breast with well-developed apocrine cytology are frequent and usually easily identifiable as benign or malignant using either pattern or cytologic features. Azzopardi reported the importance of cytology in diagnosing DCIS lesions, but discussed primarily pattern criteria.’ Using accepted pattern and cytologic criteria5 we found it difficult to assign reproducible diagnoses to a small group of proliferative apocrine lesions. The main difficulties with these lesions are the lack of pattern criteria associated with the diagnosis of DCIS and the presence of nuclear features that often are similar to or only slightly different from the usual, benign apocrine cells characteristically seen lining cystic spaces. During a review of these difficult apocrine lesions we attempted to classify these lesions in such a way as to allow the establishment of reproducible diagnostic categories. Our study identified a borderline group of cases (Fig 7). These borderline categories were defined in terms of extent of lesion and nuclear features. Lesions that were 4 to 8 mm in maximum dimension and had a majority of cells with nuclear characteristics in-

DISCUSSION

Fifty-four difficult apocrine lesions were selected from our consultation series of approximately 2,500 cases. Ages ranged from 29 to 91 years (mean age, 58.1 years). Using the nuclear criteria as outlined in Table 1 as well as the extent criteria, the cases were classified into seven main categories. Thirt.y-two (59%) of the cases were classified as apocrine DCIS and measured greater than 4 mm in maximum dimension, with the majority measuring at least 8 mm. Architectural distortion was identified in all these cases of apocrine DCIS. Cancerization of lobules was a prominent feature and expansion of lobular units was seen often. Intraluminal necrosis was identified in two of the six extensive DCIS lesions. The nuclear characteristics varied. The majority of nuclei had irregular nuclear membranes and coarse chromatin patterns similar to the nuclear features associated with usual DCIS. Three cases had usual DCIS nuclei (Table 1) but measured 2 to 4 mm, and as these most likely represent

165

166

BORDERLINE

Usual

~

or PAC

8mm

LESIONS

OF THE BREAST

TABLE 1.

Nuclei

Size

4mm

APOCRINE

~ Rare

Borderline

As in DCIS

AA or PAC with atypia (~50% borderline) nuclei: type a or b)

Limited APO DCIS

Borderline APO DCIS (~50% borderline nuclei: type a)

Nuclear Features of Apocrine Lesions

NuclearType Usual

Borderline APO DCIS (>50% borderline nuclei: type a or b)

(O’Malley et al)

apocrine

Borderline Type a

Type b APO DCIS Non-comedo DCIS

Features Regular, round, vesicular nuclei; single large nucleolus usually present Markedly enlarged nuclei in at least 25% of the cell population in a background of usual apocrine cells Majority of cells showing nuclei 2 to 3 times the size of the usual apocrine nuclei with slightly irregular nuclear membranes; 2 to 3 small nucleoli in some cells Irregular nuclear membrane, coarse chromatin pattern; multiple nucleoli

< cases of DCIS with apocrine cytology as those lesions that were characterized by narrow arcades of cells with cytoplasmic apocrine features, but with hyperchromatic nuclei rather than the more vesicular, regular, round nuclei of usual apocrine cells. We certainly agree with this definition, and such lesions are included in the DCIS categories. A clinicopathologic study of apocrine mammary carcinoma ’ included 55 cases of “apocrine DCIS”; however, the diagnostic criteria used to make this diagnosis were not discussed. We recognize that apocrine change in an otherwise diagnostic DCIS lesion does not have prognostic implications.” The utilization of size as a criterion for differentiating atypical nonlobular hyperplastic lesions from in situ carcinomas was pioneered by Tavassoli and Norris, who considered lesions less than 2 mm as ADH.” We believe that extent or size of these apocrine lesions may be important with respect to prognosis or likelihood of progression. For this reason we have separated three cases of apocrine DCIS measuring less than 4 mm from the 32 cases that were generally much more extensive. These small, limited cases of apocrine DCIS as well as the borderline apocrine DCIS lesions measuring 4 to 8 mm are most likely easily amenable to cure by local excision.” Our belief in the malignant potential of these borderline apocrine lesions rests on their similarity to indisputable DCIS lesions. There have been no epidemi-

FIQURE 7. Categorization of difficult apocrine lesions into diagnostic groups based on nuclear characteristics (along the horizontal axis) and extent of lesion (along the vertical axis). Cases of apocrine adenosis (AA) or papillary apocrine change (PAC) without atypia, although common, were excluded from this study, but would reside in the broadened category at the upper left. Only one case demonstrated usual apocrine nuclei but measured greater than 8 mm. As such a lesion is so rare, no descriptive term was given to this category. Note that the larger lesions with atypical nuclei are inclusively recognized as DCIS at the lower right (see text).

termediate between classic apocrine cells and nuclei usually seen in DCIS fell into one of these borderline categories. The second borderline group included lesions that were more than 8 mm in extent but that had a minority of cells with borderline nuclear features as described in Table 1. Thirty-two (59%) of the cases included in this study were classified as apocrine DCIS. The commonly utilized pattern and cytologic criteria used to make a diagnosis of DCIS could not be relied on for these ape crine lesions.5 Instead, we found the criteria of extent (>8 mm), distension and distortion of involved units (lobular, ductal, adenotic) , and intermediate nuclear characteristics as defined in Table 1 to be effective in accomplishing a stratification. Few pathologists have addressed the difficulty of assigning the diagnosis of DCIS to complex apocrine lesions. McDivitt et al* classified

FIGURE 1. Photomicrograph of typical apocrine nuclei as seen in usual apocrine change or metaplasia. original magnification x 1,250.)

(Hematoxylin-eosin

stain;

FIQURE 2. Nuclei typical of the borderline apocrine DCIS category, with slightly irregular nuclear borders and multiple small nucleoli. The granular nature of the cytoplasm typical of apocrine change is quite subtle. (Hematoxylin-eosin stain,. original magnification x 1,250.) FIGURE 3. Photomicrograph illustrating nuclei characteristic of DCIS (of comedo type) with marked nuclear enlargement, irregular nuclear margins, and a coarse chromatin pattern. Note the granular cytoplasm. (Hematoxylin-eosin stain: original magnification x 1,250.) FIGURE 4. Photomicrograph of cancerization of a lobule by low-grade apocrine DCIS in a single space (center left). This space is probably an intralobular terminal duct, (Hematoxylin-eosin stain; original magnification x60.) FIGURE 5. (a) Expansion of a lobule by borderline apocrine nuclei (type a, see Table 1). some of which are markedly enlarged. Note the residual and smaller, normal epithelial cells with darker cytoplasm (arrow). (Hematoxylin-eosin stain; original magnification x250.) (b) Borderline apocrine nuclei (type b, Table 1) with slight variation in nuclear and nuceolar size. (Hematoxylin-eosin stain; original magnification x250.) FIGURE 6. Limited apocrine DCIS. The maximum extent of the lesion is 3 mm. This pattern approximates that of cribriform DCIS centrally and thus does not pose a diagnostic problem, Note the uninvolved spaces at the left and right of the lesion (arrows). (Hematoxylin-eosin stain; original magnification x25.)

167

HUMAN

PATHOLOGY

Volume 25, No. 2 (February

ologic studies to date evaluating these lesions. One of our cases of apocrine DCIS and one case of borderline apocrine DCIS had adjacent small (2 to 3 mm) foci of invasive mammary carcinoma. Rare cases of invasive breast cancer associated with a full range of well-developed apocrine features in an apparent sequence from cyst lining, through in situ proliferations, through invasive carcinoma have been reported,‘“,‘” as well as series of invasive carcinomas with apocrine features.‘“.‘“.‘” However, there is no proven link of subsequent carcinoma development when the putative precursors are present alone. Wellings and Alpers” did not find a positive association between apocrine change and carcinoma, while Haagensen l8 and Naldoni et al’” maintain that there may be such an association. In this study our approach has been to develop criteria that reliably identify apocrine lesions on the borderline between those definitely benign and those indisputably malignant by currently available criteria. We believe this fosters diagnostic consistancy for these difficult lesions by recognizing that overlapping features obliterate a clear-cut margin between two diagnostic categories of “benign” and “malignant.” In the borderline category a combination of criteria encompassing cytology, pattern, and extent is most useful. Their utility rests in fostering replicability of diagnostic assignment and, hopefully, predictability of clinical outcome. This approach to categorization of these difficult apocrine lesions is at present untested. The clinical utility of this suggested approach needs to be tested by longterm follow-up studies.

REFERENCES 1. Bussolati G, Cattani MG, Gugliotta P, et al: Morphologic and functional aspects of apocrine metaplasia in dysplastic and neoplastic breast tissue. Ann N Y Acad Sci 464:262-274, 1986 2. Page DL, Vander Zwaag R, Rogers LW, et al: Relation between component parts of fibrocystic disease complex and breast cancer. J Nat1 Cancer lnst 61:1055-1063, 1978

168

1994)

3. Page DL, Anderson TJ, Rogers LW: Cysts and apocrine change, in Page DL, Anderson TJ (eds): Diagnostic Histopathology of the Breast. Edinburgh, UK, Churchill Livingstone. 1988. pp 43. 50 4. Fechner RE: Ductal carcinoma involving the lobule of the breast. Cancer 28:274-281, 1971 5. Page Dl., Anderson TJ, Rogers LW: Carcinoma in situ (CIS), in Page DL. Anderson TJ (eds): Diagnostic Histopathology of the Breast. Edinburgh, UK. Churchill Livingstone, 1988, pp 157192 6. Azzopardi JC. Salm R: Ductal adenoma of the breast: A lesion which can mimic carcinoma. J Pathol 144:11-23. 1984 7. Azzopardi J: Nomenclature of the microanatomy of the breast: Parts affected in different diseases: Normal structure and involution, in Problems in Breast Pathology. London, UK, Saunders. 1979, pp 1 l16 8. McDivitt RW, Stewart FW, BergJW: Tumors of the Breast. Atlas of Tumor Pathology. Second Series. Fascicle 2. Washington, DC, Armed Forces Institute of Pathology, 1968 9. Abati AD, Kimmel M, Rosen PP: Apocrine mammary carcinoma: A clinicopathologic study of 72 cases. Am J Clin Path01 94:371377, 1990 10. Mazoujian G: Apocrine carcinoma of the breast. Am J Clin Path01 94:485486. 1990 11. Tavassoli FA, Norris HJ: A comparison of the results of longterm follow-up for atypical intraductal hyperplasia and intraductal hyperplasia of the breast. Cancer 65:518-529, 1990 12. Lagios MD: Duct carcinoma in situ: Pathology and treatment. Surg Clin North Am 70:853-871. 1990 13. T&h J, Szimel I. Svastics E, et al: Significance of apocrine metaplasia in mammary carcinogenesis: A preliminary morphological and immunohistochemical study. Ann N Y Acad Sci 586:238251. 1990 14. Haagensen CD: Diseases of the Breast (ed 2): Philadelphia, PA, Saunders, 1971 15. d’Amore ESG, Terrier-Lacombe MJ, Travagli JP, et al: lnvasive apocrine carcinoma of the breast: A long-term follow-up study of 34 cases. Breast Cancer Res Treat 12:3744, 1988 16. Bundred NJ, Walker RA, Everington D, et al: Is apocrine differentiation in breast carcinoma of prognostic significance? Br J Cancer 62:11.3-l 17, 1990 17. Wellings SR, Alpers CE: Apocrine cystic metaplasia: Subgross pathology and prevalence in cancer-associated versus random autopsy breasts. HUM PATHOL 18:381, 1987 18. Haagensen DE, Jr: Is cystic disease related to breast cancer? Am J Surg Path01 15:687-694, 1991 19. Naldoni C, Costantini M, Dogliotti 1~. et al: Association of cyst type with risk factors for breast cancer and relapse rate in women with gross cystic disease of the breast. Cancer Res 52:17911795,1992