Virchows Arch (2003) 443:609–617 DOI 10.1007/s00428-003-0888-x

ORIGINAL ARTICLE

G. MacGrogan · F. A. Tavassoli

Central atypical papillomas of the breast: a clinicopathological study of 119 cases Received: 5 September 2002 / Accepted: 6 July 2003 / Published online: 13 September 2003
Springer-Verlag 2003

Abstract The clinicopathological features of central intraductal papillomas of the breast presenting with florid usual ductal hyperplasia or atypical ductal hyperplasia (ADH) were analyzed in a retrospective series of 119 patients, whose lesions were sent to the Armed Forces Institute of Pathology from 1976 to 1990. After histological review considering predefined morphological and quantitative criteria, the 119 central papillomas were classified into 22 papillomas with florid usual ductal hyperplasia (18%), 40 papillomas with focal atypia (34%), 24 atypical papillomas (20%) and 33 carcinomas arising in a papilloma (28%). After a median period of follow-up of 110 months, 16 recurrences (5 papillomas, 2 carcinomas arising in a papilloma, 4 ductal carcinomas in situ, 5 invasive carcinomas) occurred. No statistically significant difference was observed in relation to recurrence for the various categories of papillomas. The presence of epithelial hyperplasia, ADH or lobular neoplasia in the surrounding breast as well as infarction of the papilloma were significant predictive factors of recurrence (P=0.02 and P=0.005, respectively, log-rank test). The main reason for the observed low rate of significant recurrences in this series was that epithelial atypia (whether comprising 20% or 60% of the papillary lesion) was, in most of the cases, localized in a confined lesion that was completely excised. G. MacGrogan ()) Department of Pathology, Institut Bergoni, 229 Cours de l’Argonne, 33076 Bordeaux cedex, France e-mail: [email protected] Tel.: +33-55-6330436 Fax: +33-55-6330438 F. A. Tavassoli Department of Breast and Gynecology Pathologic, Armed Forces Institute of Pathology, Washington, DC, USA Present address: Yale University School of Medicine, Department of Pathology, New Haven, CT, USA

Keywords Breast papilloma · Epithelial atypia · Hyperplasia · Ductal carcinoma in situ · Invasive carcinoma

Introduction The discovery of atypical epithelial hyperplasia or carcinoma within an intraductal papilloma remains a matter of concern for the pathologist. For atypical intraductal hyperplasia (ADH) or ductal carcinoma in situ (DCIS) found within the breast proper, the relative and/or absolute risk for subsequent development of invasive carcinoma has been determined in several studies [7, 15]. The risk represented by the occurrence of such abnormalities within a confined lesion, such as a fibroadenoma or an intraductal papilloma, is still not fully established. Papillomas have been separated into two risk categories for the development of carcinoma: the central/ solitary type, for which the risk is low, and the peripheral/ multiple type, which is more frequently associated with DCIS or invasive carcinoma [2, 4, 5, 6]. In some cases, the distinction between solitary and multiple lesions is clear cut. In others, however, the distinction between the two types can reasonably be made only by threedimensional studies [5] and, in certain cases, any separation appears arbitrary on a routine breast biopsy. Over the past 20 years, intraductal papillomas have been described by morphological and immunohistochemical criteria acknowledging a proliferation of luminal epithelial and subjacent myoepithelial cells overlying fibrovascular stalks within the lumen of the duct system. These criteria have helped distinguish benign intraductal papilloma from those papillary carcinomas characterized by a complete absence of myoepithelial cells [1, 9]. There are, however, papillary carcinomas that show focal retention of the myoepithelial cell layer [9]. The epithelial component of an otherwise benign papilloma can be subject to a spectrum of morphological changes ranging from metaplasia to hyperplasia, ADH and in situ carcinoma. Given the absence of any definitions or guidelines

610

in the literature concerning the amount of ADH within a papilloma that would qualify it as a papillary carcinoma, one of the authors of the present study had proposed assessment of the quantity of ADH within these papillary lesions and their separation under different designations [14]. By using the terms “atypical papilloma” and “carcinoma arising in a papilloma,” this approach avoided the designation of papillary carcinoma for these lesions. Prior to this proposal, many pathologists diagnosed larger papillomas with such atypical foci as papillary carcinoma, and many women with such lesions had been managed by mastectomies. Using modifications of this quantitative approach, two separate studies have looked at the precancerous potential of proliferative atypia in papillary lesions [8, 10]. The presence of ADH within a papilloma and/or in the surrounding breast was associated with an increased risk of recurrence [10] or invasive carcinoma [8]. We undertook the present study to fully describe the morphological changes occurring within the epithelial component of central papillomas and to search for factors of prognostic significance concerning recurrence and progression.

Materials and methods Patient selection Retrieved from the files of the Armed Forces Institute of Pathology were 259 cases, coded as: solitary papilloma with focal atypia, atypical papilloma and carcinoma arising in a papilloma. These represented all the cases diagnosed as such over a 15-year period, from 1 January 1976 to 31 December 1990, and for which followup data was available. Patients’ charts were reviewed for clinical information and gross description of their lesions. After the preliminary review of charts and histological analysis, 140 cases were excluded . The main causes of exclusion from the study were mastectomy and death of patient within 6 months of surgery. A mean number of 8.1 (1 to 135) hematoxylin-eosin (H&E)-stained sections were available per case. After histological analysis, additional cases showing the following features were discarded: sclerosing papilloma as defined by an adenomatous nodule made of small tubules confined to a duct lumen; peripheral papilloma consisting of a papillary lesion originating in a lobular unit or papillomatosis as defined by multiple small papillary lesions originating in different lobular units; nodular sclerosing adenosis as defined by a large conglomerate of small parallel tubules involving one or several lobular units with small secondary papillary structures; extensive myoepithelial hyperplasia, qualifying the lesion as adenomyoepithelioma; intraductal or infiltrating carcinoma in the surrounding breast parenchyma. Histological definitions A papilloma was defined as central or solitary if it was localized in a ductal space with no connection to the terminal lobular unit and complex if three or more duct transections involved by such a papillary proliferation were visible on a tissue section. The epithelial atypia observed in the central papillomas of our series was comparable to the atypia observed in low grade DCIS; we generally do not encounter high-grade DCIS patterns or cytology in this setting. Because there are no standardized criteria for classifying solitary papillomas presenting with epithelial atypia and due to their highly polymorphous nature, we arbitrarily defined four categories at the beginning of our study: papillomas with focal

atypia, atypical papillomas, carcinoma arising in papilloma, papilloma with diffuse and extensive usual ductal hyperplasia. Classification into the first three categories was based on the presence of atypical cytological and architectural features and the extent of these features in the papilloma. The atypical cytological and architectural features consisted of: – An epithelial proliferation of ductal cell type, similar to atypical intraductal hyperplasia as defined by Tavassoli and Norris [15], e.g., an epithelial proliferation showing cytological atypia characterized by monotony and uniformity with either solid, cribriform (Fig. 1), micropapillary (Fig. 2) or stratified columnar cell (Fig. 3) architectural patterns. – An epithelial proliferation of apocrine cell type, in patterns similar to what has been described by Tavassoli and Norris [16] as atypical apocrine metaplasia, e.g., apocrine cells with a threefold variation in nuclear size and atypical apocrine hyperplasia, e.g., an apocrine cell population organized in a solid or cribriform pattern (Fig. 4d). – An epithelial proliferation with cytological features intermediate between ductal cells and apocrine cells, coined apocrine-like cells, lacking one or more of the classic apocrine features [13], often presenting with an amphophilic granular cytoplasm and round nuclei and arranged in cribriform (Fig. 5), solid (Fig. 4b, c), micropapillary, stratified spindle cell patterns. The central papilloma was designated papilloma with focal atypia if less than 10% of its entire surface in any given slide was involved by an atypical cell population as described in the preceding paragraph. If at least 10% but less than one-third of the papilloma’s entire surface was involved by the same atypical population it was named atypical papilloma (Fig. 4). The term carcinoma arising in a papilloma was used if one of the following characteristics was present: – If one-third up to 90% of its cross-section was involved by an atypical cell population as described for atypical papillomas. – If tumoral necrosis, characterized by curdled eosinophilic material mixed with nuclear debris, was present in the atypical cell population, regardless of the quantity of the atypical cells relative to the surface of the papilloma involved. With the exception of the immediately adjacent duct-extensions, abutting the wall of the papilloma, the surrounding breast parenchyma had to be devoid of intraductal carcinoma to qualify the lesion as carcinoma arising in a papilloma. Papilloma with florid “sheet-like” hyperplasia Even though this type of papilloma is not involved by atypia, we felt that it should be clearly defined and put in the differential diagnosis of the preceding categories because of its relative frequency and worrisome appearance to the inexperienced eye. At low power, this lesion had the appearance of a solid sheet of epithelial cells almost obscuring the lumen of a large duct, but at higher magnification, delicate fibrovascular stalks with overlying myoepithelial cells were apparent in the lesion. In the solid areas, the epithelial cells were arranged in streaming or whirling patterns; they were ovoid or spindle in shape with inconspicuous cytoplasmic margins, frequently overlapping bland nuclei (Fig. 6). Slit-like clear spaces punctuated the epithelial proliferation. Additional features evaluated on H&E sections included: microcalcifications in the papilloma; a lymphoid infiltrate with germinal centers, if present in the vicinity of the papilloma; usual ductal hyperplasia, ADH or lobular neoplasia (LN) in the breast parenchyma surrounding the papilloma. In 83 cases, paraffin-embedded tissue blocks were available. Immunostaining with smooth muscle actin antibody (SMA; Sigma Immunochemicals, St. Louis, Mo., 1/8000 po) was performed on all. Positivity was interpreted in reference to an adjacent control section in which normal rabbit serum was used as the secondary

611 antibody, and positive controls were also examined for each case. Each case was reviewed on a double-headed microscope by both authors and a final diagnosis was made after reaching consensus. Statistical analysis Clinical follow-up was obtained for the 119 patients from the referring pathologist, the surgeon or the patient. The relationship between morphological features and clinical outcome was determined in a univariate analysis by the log-rank test using the Kaplan Meier method (BMDP software, program 1L). Local recurrences (in the same breast) were the events taken into account for the statistical analysis. These included: DCIS, carcinoma arising in a papilloma and invasive carcinoma. Subsequent ipsilateral or contralateral papillomas (five papillomas) as well as one contralateral carcinoma were not considered as significant events for the present statistical analysis.

Results There were 119 women included in the study. The mean and median age was 56 years (19–83 years). For 77 patients, data on breast cancer history was available: a history of breast carcinoma involving first-degree relatives was present for 8 patients (10%). The right breast was preferentially involved and affected in 63 cases (57%), compared with 47 cases (43%) in the left breast. In a majority of cases, patients presented with a palpable mass [71 patients (60%)]. Other presenting symptoms included: – – – –

Abnormal mammogram, 16 patients (13%) Nipple discharge, 6 patients (5%) Nipple discharge and lump, 12 patients (10%) Nipple discharge and abnormal mammogram, 2 patients

The size of the central papillomas ranged from 0.3 cm to 7 cm; mean size:1.5 cm, median size: 1.2 cm. After histological review, the 119 central papillomas comprised: 40 papillomas with focal atypia (34%), 24 atypical papillomas (20%), 33 carcinomas arising in a papilloma (28%), 22 papillomas with florid hyperplasia (18%). The central papillomas were classified as localized in 81 cases (72%) and as complex in 32 cases (28%). Microcalcifications in the papillomas and lymphoid follicles surrounding the papillomas were present in 33 (28%) and 9 (8%) cases, respectively. Adjacent ducts and lobular units in the surrounding breast parenchyma were involved by usual ductal hyperplasia in 37 cases (31%), ADH in 19 cases (16%) and LN in 1 case (1%). The

Fig. 1 Focus of atypical ductal-type cells with monotonous small non-overlapping nuclei forming rosette-like structures within a papilloma. Bar 100 mm Fig. 2 Focus of atypical ductal-type cells forming micropapillary and cribriform structures within a papilloma. Space bar 100 mm Fig. 3 Focus of atypical ductal-type cells overlying papillary stalks arranged in a stratified columnar cell pattern. Space bar 100 mm

612

613

tively present in 24 and 12 papillomas. SMA immunostaining was positive in all cases analyzed and highlighted the presence of myoepithelial cells forming a continuous or discontinuous layer on the surface of the papillary stalks. Follow-up data

Fig. 5 Area of atypical apocrine-like cells with round monotonous non-overlapping nuclei organized around cribriform spaces containing eosinophilic secretions within a papilloma. Space bar 100 mm

adjacent ducts and lobules were devoid of any proliferative activity in 56 cases (47%). The most common architectural patterns of atypical epithelial proliferation within the papillomas were: a cribriform pattern in 67 cases, a solid pattern in 52 cases, a stratified columnar cell pattern in 15 cases and an adenomatous pattern in 12 cases. Other patterns less frequently encountered were of the micropapillary type and a pagetoid diffusion of the atypical cells underneath the epithelium of adjacent papillary fronds. Apocrine cells and apocrine-like cells were the most frequently proliferating cell type encountered in the atypical areas (40 and 47 cases, respectively), followed by, native ductal-type cells (37 cases). Other cell types less frequently encountered were clear cells (8 cases), cells identical to those encountered in lobular neoplasia (1 case) and hypersecretory cells (1 case). Each of these cell types could either form the entire atypical population, or be associated with one another. Squamous changes were observed in 2 cases, in combination with the more usual atypical patterns. Foci of tumoral necrosis or infarction (Fig. 7) were respec-

Fig. 4 Atypical central papilloma. a An atypical epithelial proliferation involving a portion of a central papilloma. b Same case, the atypical proliferation consists of apocrine-like cells arranged in solid sheets. c Same case, the apocrine-like cells show large eosinophilic cytoplasms, distinct cellular membranes and nonoverlapping slightly irregular nuclei. d Same case, apocrine cells with a threefold variation in nuclear size organized in cribriform structures intermixed with apocrine-like cells. e Same case, non atypical epthelial and myoepithelial cells overlying papillary stalks in another area of the papilloma. Space bars 100 mm

After a median period of 110 months 16 recurrences (five papillomas, two carcinomas arising in a papilloma, four DCIS, five invasive carcinomas) occurred. Nine recurrences were localized in the same area of the breast as the original papilloma, four recurrences in another area of the same breast and three recurrences in the other breast. At last follow-up, 103 patients were alive with no evidence of disease (87%), one patient died of ipsilateral invasive ductal carcinoma (1%), eight patients died of an unrelated cause (6%) and one patient died of an unknown cause (1%). Table 1 shows the characteristics of the central papillomas that recurred, the type of recurrence with the interval of time separating the original papilloma from the recurrence. Table 2 shows the different categories of central papillomas with follow-up data and their corresponding recurrences, considering only recurrences with carcinomatous features. When comparing the different categories: papilloma with florid hyperplasia, papilloma with focal atypia, atypical papilloma and carcinoma arising in a papilloma, there was no statistically different trend observed with relation to recurrence (log-rank test). Other characteristics of these central papillomas were also studied for their pertinence in relation to recurrences: complexity of the papilloma, presence of lymphoid follicles in the adjacent breast parenchyma, presence of microcalcifications in the papilloma, presence of a cribriform or a stratified columnar cell pattern, presence of apocrine, apocrine-like or ductal-type cells and presence of foci of tumoral necrosis. No significant statistical trend was observed for any of these features. On the contrary, the presence of ordinary hyperplasia and/or ADH or LN in the breast parenchyma surrounding the papilloma was significantly associated with recurrences in the same breast (P=0.02, log-rank test) (Table 3). Interestingly, the presence of infarction in the original papilloma was also associated with a significant increase in recurrences (P=0.005, log-rank test) (Table 4). The presence of infarction in the original papilloma was more frequently observed in older patients (P=0.06, Wilcoxon test) and in larger papillomas (P=0.09, Wilcoxon test). When considering only invasive carcinomas as significant recurrences, five invasive carcinomas occurred in the series with follow-up (4%); two in the same area as the original papilloma, two in another area of the same breast and one in the contralateral breast. Again, there was no statistically significant trend observed when comparing the various categories of papillomas with relation to recurrence, nor the various morphological features ob-

614

Fig. 6 a, b Central papilloma with florid usual ductal hyperplasia. The epithelial cells with overlapping coffee bean-shaped nuclei are arranged in swirling patterns. Space bar 100 mm

Fig. 7 Central papilloma with ischemic necrosis. Space bar 100 mm

served within the papillomas and in the surrounding breast parenchyma.

Discussion The purpose of our study was to look for morphological features in papillomas that could serve as prognostic

factors. The population studied retrospectively consisted of a series of papillomas sent to the Armed Forces Institute of Pathology because they presented with morphological features causing concern to the consulting pathologist. Among these papillomas, we distinguished four different categories. Our study did not show any statistically significant difference in terms of recurrence between these various categories, maybe because the number of recurrences was too small. Atypia occurring in papillomas appears to be a rare event, although no precise figure has been reported. Assembling a large group of such lesions is a difficult task, and, to our knowledge, this is the first large series of papillomas with atypia reported. The number of significant events that occurred after excluding patients who were treated by mastectomy, or patients with a personal history of breast cancer, was small: 11 events considering all lesions with carcinomatous features (both in situ and invasive) (8.4% ipsilateral, 0.8% contralateral) and 5 events considering only invasive carcinomas (3.3% ipsilateral, 0.8% contralateral). These rates are higher than those reported by Rosen in his review of the literature on 612 non-selected papillomas treated by excision only [11]. He found that recurrences, in the form of carcinoma, in situ or invasive, were reported in 23 of 612 cases (2% ipsilateral, 2% contralateral, 0.3% bilateral). Therefore, if we consider our population of papillomas with various types and patterns of epithelial proliferation as a whole, notwithstanding the degree or extent of atypical features within them, there appears to be an increased risk of subsequent develop-

615 Table 1 Characteristics of papillary lesions that recurred. ADH Atypical ductal hyperplasia, DCIS ductal carcinoma in situ Case number

Age/side

Size

Type of papilloma

Necrosis

Surrounding breast

Recurrence

Location

1

59 years/left

19 mm

None

Nothing

Papilloma

Other breast

2

63 years/left

10 mm

Hyperplasia

3

61 years/right

6 mm

4

64 years/left

32 mm

Invasive carcinoma Invasive carcinoma Papilloma

Other area, same 104 months breast Same area, same 35 months breast Other breast 38 months

5

34 years/right

11 mm

Infarction

Nothing

79 years/left

20 mm

Infarction

ADH

Invasive carcinoma DCIS

7

62 years/right

12 mm

Atypical papilloma

Infarction

Hyperplasia

Papilloma

8

55 years/right

Atypical papilloma

Hyperplasia

DCIS

9

73 years/left

Atypical papilloma

Hyperplasia

10

40 years/right

Atypical papilloma

Hyperplasia

Invasive carcinoma Papilloma

11

81 years/right

20 mm

Nothing

Papilloma

12

47 years/left

8 mm

Hyperplasia

DCIS

13

62 years/left

Tumoral

Hyperplasia

14

44 years/left

12 mm

Infarction

ADH

Carcinoma arising in papilloma DCIS

Other area, same breast Same area, same breast Other area, same breast Same area, same breast Same area, same breast Same area, same breast Same area, same breast Same area, same breast Same area, same breast

162 months

6

Papilloma with florid hyperplasia Papilloma with florid hyperplasia Papilloma with focal atypia Papilloma with focal atypia Papilloma with focal atypia Atypical papilloma

55 months

15

48 years/right

15 mm

Infarction

ADH

Other area, same breast Other breast

16

61 years

15 mm

6 mm

Table 2 Different categories of solitary papillomas among the patients with follow-up and the corresponding carcinomatous recurrences. DCIS ductal carcinoma in situ

Carcinoma arising in papilloma Carcinoma arising in papilloma Carcinoma arising in papilloma Carcinoma arising in papilloma Carcinoma arising in papilloma Carcinoma arising in papilloma

Infarction

Nothing Nothing

Tumoral

Hyperplasia

Invasive carcinoma Carcimona arising in papilloma

Interval

Same area, same breast

96 months

48 months 217 months 145 months 28 months 87 months 18 months 34 months 108 months

44 months 105 months

Category of papilloma

Number of patients

Recurrences

Interval of time

Papilloma with florid hyperplasia Papilloma with focal atypia Atypical papilloma

22

1 invasive carcinoma

104 months

40

2 invasive carcinomas

24

2 DCIS 1 invasive carcinoma 2 DCIS 2 carcinomas in papilloma 1 invasive carcinoma (contralat)

Carcinoma arising in a papilloma

33

ment of invasive carcinoma among them when compared with papillomas in general. Raju and Vertes [10] compared 12 papillomas with ADH with 60 papillomas without ADH. Two contralateral invasive carcinomas subsequently occurred in the papilloma with ADH group compared with one ipsilateral and two contralateral

35 months, 162 months 48 28 34 105 44

months, 145 months months months, 55 months months, 108 months months

invasive carcinomas in the non-atypical papilloma group. The authors concluded from these results that papillomas with atypia are more of a risk factor for subsequent development of invasive carcinoma in the contralateral breast. Page et al. [8] performed a nested case control study on 368 papillomas. Among these 368 benign

616 Table 3 Impact of proliferative breast disease (PBD) (hyperplasia and/or atypical ductal hyperplasia or lobular neoplasia) in surrounding breast on outcome of papillomas. DCIS ductal carcinoma in situ Category of papilloma

Proliferative breast disease

Number of patients

Recurrences

Interval

Papilloma with florid hyperplasia

+PBD PBD +PBD PBD

8 12 15 22

Invasive carcinoma Papilloma

104 months 96 months

Atypical papilloma

+PBD

17

Invasive carcinoma Invasive carcinoma Papilloma DCIS DCIS Invasive carcinoma Papilloma Papilloma

35 162 38 48 145 28 88 217

months months months months months months months months

Carcinoma arising in papilloma

PBD +PBD

7 17

PBD

15

1 1 0 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1

Carcinoma in papilloma Carcinoma in papilloma DCIS DCIS Invasive carcinoma (contralat) Papilloma

105 108 55 34 48 18

months months months months months months

Papilloma with focal atypia

Table 4 Impact of infarction on outcome of papillomas. DCIS ductal carcinoma in situ

Category of papilloma

Infarction

Number of patients

Recurrences

Interval

Papilloma with florid hyperplasia

Infarction

20

104 months 96 months

+Infarction Infarction

2 35

+Infarction Infarction

5 22

+Infarction Infarction

2 30

+Infarction

2

1 Invasive carcinoma 1 Papilloma 0 1 Invasive carcinoma 1 Papilloma 1 Invasive carcinoma 1 Papilloma 1 Papilloma 1 DCIS 1 Invasive carcinoma 1 DCIS 1 Carcinoma in papilloma 1 Carcinoma in papilloma 1 DCIS 1 Papilloma 1 DCIS 1 Invasive carcinoma (contralat)

Papilloma with focal atypia Atypical papilloma

Carcinoma arising in papilloma

papilloma patients, 31 subsequently developed invasive carcinoma (8.4%). They found that the relative risk for invasive carcinoma for patients with papillomas containing atypical hyperplasia (AH) was nine and a half times that of patients without AH within their papillomas. Most of the invasive carcinomas in their study occurred in the same breast as the original papilloma, leading them to conclude that the risk represented by AH in papillomas was mainly local. Interestingly, the increased relative risk for subsequent development of invasive carcinoma observed in papillomas with AH appeared to be strongly linked to the presence of AH in the surrounding breast. The risk of developing invasive carcinoma in the two previous studies was higher than that found in our own series [8, 10]. It is noteworthy that the criteria used to identify the atypical cell population observed in the

162 34 35 217 88 145 28 48 108 105 34 18 55 48

months months months months months months months months months months months months months months

papillomas of the three studies were similar. Judging from the information available in the three studies, the main reason for this discrepancy appears to be the number of cases and type of papillomas studied. While in rare cases the distinction was difficult, a vast majority of our cases were central, solitary papillomas of larger ducts of either simple (papillary projection into basically one distended space) or complex (multiple side branches involved) type. Both the mean and median size of our lesions exceeded 1 cm approaching 1.5 cm; the largest in the study of Page et al. [8] was 1.3 cm. More than half of the papillomas in that study were micropapillomas that measured 3 mm or less. Furthermore, in Raju and Verdes’ study [10], of the 20 papillomas with atypical hyperplasia, 11 were of the solitary type and 9 were of the multiple type. Therefore, we believe that in the above-mentioned studies, there was

617

a significant proportion of multiple peripheral-type papillomas. Surgical margins were not truly assessed in these two studies, as in our own study, due to the absence of specific guidelines in the management of such lesions at the time of their excision. Incomplete excision of foci of atypical epithelial proliferation, especially for multiple peripheral-type papillomas, may partially explain the high rate of recurrences and subsequent development of cancer observed. From both qualitative and quantitative standpoints, the atypical areas in a good proportion of our cases would have easily qualified as DCIS had they been found within the breast proper and out of the confines of a papilloma. The uneventful outcome for these patients can be explained by the complete excision of the atypical cell population localized within a confined lesion. Recurrences associated with carcinoma were usually developed within the same breast and were significantly associated with the presence of ductal cell hyperplasia and/or ADH, or LN in the surrounding breast. The study on the outcome of DCIS by the Van Nuys group [12], where no invasive recurrences occurred within the low-grade DCIS group of patients treated solely by complete excision with adequate margins, supports this explanation. The findings in the study of Carter et al. [3] on intracystic papillary carcinomas parallel ours. In the group of patients treated by excision alone in Carter et al.’s study, subsequent DCIS or invasive carcinomas were observed only in the subgroup of patients with associated DCIS in the surrounding breast while no recurrences were observed in the subgroup of patients presenting with fully developed intracystic papillary carcinoma alone but no evidence of DCIS in the surrounding breast tissues. Surprisingly, a strongly significant morphological feature associated with recurrence in our series was the presence of areas of infarction within the original papilloma. We defined such areas as foci of coagulative necrosis involving the epithelial and myoepithelial cells as well as the underlying fibrovascular stalks. These areas retained their original architecture and “ghost” cells corresponding to the original epithelial cells were clearly visible. Such areas could be easily differentiated from tumoral necrosis, where only the epithelial cell compartment was involved and destroyed. Infarction tended to occur more often in older patients with larger papillomas. We have no clear hypothesis to explain this phenomenon. It could be the result of torsion secondary to dyssynchronous rapid growth of the epithelial, myoepithelial and stromal compartments of a papilloma, underlining its aggressive potential, or it may just reflect the physiological status of the patient. The present study has demonstrated that, when confined to a central papilloma (simple or complex), the extent and quantity of epithelial atypia comparable to that of low-grade DCIS has no prognostic significance or impact on outcome for the patient. Recurrences, when they develop, appear to be related to the presence of proliferative breast lesions in the surrounding breast tissue. When atypia is encountered in a papilloma on an

excisional biopsy, a rim of breast tissue surrounding the papilloma should be excised to allow assessment of the surrounding tissue for any evidence of atypical epithelial proliferations. Decisions on management of these lesions should be made in the context of the surrounding breast tissue and should be more conservative considering the findings of the present study. Assessment of the surrounding breast tissue for presence of an atypical epithelial proliferation is essential since follow-up would be particularly important in this group. Acknowledgement We wish to thank Dr Carol Bodian for helpful and constructive discussion, Vronique Picot for statistical analysis and Isabelle Le Polls for secretarial assistance. Gatan MacGrogan was supported by a “bourse d’aide  la mobilit de la Fondation De France et de la Fdration Nationale des Centres de Lutte Contre le Cancer.”

References 1. Azzopardi J (1979) Papillary tumors. In: Azzopardi J (ed) Problems in breast pathology. Saunders, London, pp 150–166 2. Carter D (1977) Intraductal papillary tumors of the breast. A study of 78 cases. Cancer 39:1689–1692 3. Carter D, Orr SL, Merino MJ (1983) Intracystic papillary carcinoma of the breast. After mastectomy, radiotherapy or excisional biopsy alone. Cancer 52:14–19 4. Haagensen CD (1986) Solitary intraductal papilloma. In: Haagensen CD (ed) Diseases of the breast. Saunders, Philadelphia, pp 181–185 5. Ohuchi N, Abe R, Takahashi T, Tezuka F (1984) Origin and extension of intraductal papillomas of the breast: a threedimensional reconstruction study. Br Cancer Res Treat 4:117– 128 6. Ohuchi N, Abe Rikiya, Kasai M (1984) Possible cancerous change of intraductal papillomas of the breast. Cancer 54:605– 611 7. Page DL, Dupont W, Rogers L, Rados M (1985) Atypical hyperplastic lesions of the female breast: a long-term follow-up study. Cancer 55:2698–2708 8. Page DL, Salhany KE, Jensen RA, Dupont WD (1996) Subsequent breast carcinoma risk after biopsy with atypia in a breast papilloma. Cancer 78:258–266 9. Papotti M, Gugliotta P, Eusebi V, Bussolati G (1983) Immunohistochemical analysis of benign and malignant papillary lesions of the breast. Am J Surg Pathol 7:451–461 10. Raju U, Vertes D (1996) Breast papillomas with atypical ductal hyperplasia: a clinicopathologic study. Hum Pathol 27:1231– 1238 11. Rosen PP (1986) Arthur Purdy Stout and papilloma of the breast. Comments on the occasion of his 100th birthday. Am J Surg Pathol 10:100–107 12. Silverstein MJ, Lagios MD, Craig PH, Waisman JR, Lewinsky BS, Colburn W, Poller DN (1996) A prognostic index for ductal carcinoma in situ of the breast. Cancer 77:2267–2274 13. Tavassoli FA (1992) Benign lesions. In: Tavassoli FA (ed) Pathology of the breast. Appleton and Lange, Norwalk, pp 85 14. Tavassoli FA (1992) Papillary lesions. In Tavassoli FA (ed) Pathology of the breast. Appleton and Lange, Norwalk, pp 203– 209 15. Tavassoli FA, Norris HJ (1990) A comparison of the results of long-term follow-up for atypical intraductal hyperplasia and intraductal hyperplasia of the breast. Cancer 65:518–529 16. Tavassoli FA, Norris HJ (1994) Intraductal apocrine carcinoma: a clinicopathologic study of 37 cases. Modern Pathol 7:813– 818