Epithelial Displacement in Breast Lesions: A Papillary Phenomenon

Displacement of epithelial cells (DE) in the breast is a recognized phenomenon that may occur after needling procedures such as fine-needle aspiration biopsy, core needle biopsy, needle localization, suture placement, and infiltration with local anesthetic.1,2 Both benign and malignant epithelial cells can be displaced into the biopsy site, needle tract, lymphatic channels, and axillary lymph nodes.1–3 Despite the fact that the biologic significance of these findings is currently unknown, it is important to recognize this iatrogenic artifact, so that these findings are not misinterpreted as stromal or lymphatic invasion, especially in cases of benign breast lesions and pure ductal carcinoma in situ (DCIS).

In the few series in the literature that describe DE in the breast, the reported associated entities have included papillary duct hyperplasia, duct hyperplasia, benign and malignant intracystic papillary lesions, atypical duct hyperplasia, lobular carcinoma in situ, DCIS, and invasive carcinoma (IC).1–6 The purpose of this study was to analyze which lesions in the breast, if any, are prone to displacement.

Using the pathology computer database at the Mount Sinai Medical Center, we identified 53 cases with DE in the breast from January 1994 to June 2004. We achieved this by limiting the search to all breast excision specimens (13 334 cases) for the words displacement or displaced. Our institution has 3 dedicated breast pathologists, and thus practically all breast cases are seen by at least one of the authors. Thus, any case with DE within the institution is labeled as such in a consistent manner.

Information regarding prior needling procedures and date of procedure was obtained from the database, requisition sheet, medical chart, surgeon, radiologist, or outside pathologist. All slides from the initial needling procedure and subsequent excision were reviewed to determine the pathology and site of displacement. Alterations secondary to needling procedures, such as the presence of granulation tissue, fat necrosis, inflammation, hemosiderin-laden macrophages, stromal or intraductal hemorrhage, and foreign-body giant cell reaction, were sought in all cases (Figure 1). Displacement of epithelial cells was defined as the presence of epithelial cells, benign or malignant, identical to those of the primary lesion and present in artifactual spaces away from the target lesion.7 These cell clusters were not bound by any basement membrane or specialized breast stroma.1 The needle tract was defined as a linear area of granulation tissue with or without neutrophils. Clinical follow-up information was available in only a few cases. In most cases, the time interval from the biopsy to the subsequent excision containing DE was short. Thus, unfortunately, our study was limited by lack of clinical follow-up information.

The patients consisted of 51 women and 2 men with ages ranging from 22 to 84 years (mean, 55.1 years). Excision of the lesion was performed for a mass (34 cases), calcifications (15), both (3), or bloody nipple discharge (1). All patients had undergone prior needling procedures that consisted of 1 of the following: ultrasound-guided core biopsy with 14-gauge needles (24), 11-gauge mammotome core biopsy (16), anesthetic injection (3), suture placement (5), and needle localization biopsy (1). In 8 cases, fine-needle aspiration was done, 5 in conjunction with a second biopsy (3 mammotome biopsies and 2 ultrasound-guided biopsies), either at the same time (2 cases) or a few days later (3), but prior to the surgical excision that contained DE. In 1 case, the type of needling procedure was unknown, because it was a consultation case from a different institution with no clinical information currently available. However, changes consistent with previous biopsy site changes were present at the site of DE. The time interval from initial needling to excision ranged from a few minutes to a few hours (in the cases of needle localization, suture placement, and local anesthetic infiltration) to 47 days for the other biopsies.

Displacement of epithelial cells was noted in 1 or more of the following sites: breast stroma (2 cases) (Figure 2), previous biopsy site (42 cases) (Figure 1), lymphatic channels (5 cases), or both biopsy tract and lymphatic channels (4 cases) (Figure 3). The presence of a lymphatic space was determined by the presence of histologically clear-cut lining endothelial cells. These areas usually did not occur in isolation, being accompanied by adjacent blood vessels and other lymphatic spaces. Histologic evidence of a previous needling procedure was present in all cases in which there was sufficient time for the development of biopsy site changes. These changes included the presence of 1 or more of the following: granulation tissue, fat necrosis, inflammation, hemosiderin-laden macrophages, stromal and intraductal hemorrhage, and foreign-body giant cell reaction (Figures 1 and 3). Not surprisingly, these changes were absent in some of the cases (3) in which the time interval from the initial needling procedure to the excision ranged from only minutes to a few hours (in the cases of needle localization, suture placement, and local anesthetic infiltration). Instead, we found marginating neutrophils in the path of the needle tract due to the needle localization or suture placement (Figure 2). In these cases, the displaced epithelial cells were present in breast stroma (2 cases) (Figure 2) or lymphatic channels (1). In 4 cases, patients had had an initial lumpectomy that included a needling procedure (1 suture placement and 3 infiltration of local anesthetic) followed subsequently by an excision. Even though the time interval from the needling procedure to the subsequent excision in these cases was brief, previous biopsy site changes were present because of the lumpectomies originally performed for palpable lesions. This then provided the right milieu for the presence of the displaced epithelial cells in previous biopsy site changes in the second specimens.

In most cases (45/53, or 84.9%), a diagnosis of DCIS was rendered. The nuclear grades of the DCIS were as follows: low, 4; intermediate, 24; and high, 17. Cases with mixed features were assigned the higher nuclear grade categorization. The morphologic patterns of DCIS included 1 or more of the following: cribriform (30), micropapillary (20), comedo (12), solid (13), and papillary (6). Half of the papillary DCIS cases also had micropapillary features. Ductal carcinoma in situ involved an intraductal papilloma in 32 cases, and by pattern could be recognized as 1 or more of the following: cribriform (22), solid (13), micropapillary (10), comedo (8), and papillary (3). One of the 3 papillary DCIS cases also had micropapillary features. Six of the intraductal papillomas involved by DCIS also had an associated cystic component. The pattern of DCIS in these lesions was partially micropapillary or papillary types in 4 cases, and the remaining 2 cases were solid and cribriform types.

The next most common diagnosis was IC (17/53 cases, or 32.1%), ranging in size from microinvasive (<0.1 cm) to 1.9 cm (mean, 1.26 cm). The numbers of patients with each grade of these carcinomas were as follows: well differentiated, 1; moderately differentiated, 9; and poorly differentiated, 7. Additional features noted focally to diffusely in 6 cases included: papillary (1), micropapillary (1), mucinous (colloid) (2), and tubulolobular (2) differentiation. Only 2 of these cases were pure, 1 of the colloid carcinomas and the papillary carcinoma, neither of which was associated with an intraductal component. The pattern of DCIS in the IC (15/17 cases) consisted of 1 or more of the following: cribriform (8), micropapillary (6), comedo (3), solid (3), and papillary (3). One of the papillary DCIS cases also had micropapillary features. In 8 cases, the associated DCIS involved an intraductal papilloma; 2 of these had both micropapillary and papillary features. One of the 8 cribriform DCIS cases was pure in type and involved a cyst. The IC with micropapillary features also had micropapillary DCIS.

In the remaining 6 cases, displacement occurred in benign intraductal papillomata, with no carcinoma identified in any of the patients' specimens. The displaced epithelial cells in only these 6 cases were benign (Figure 3), the remainder being cytologically malignant (Figure 1). Morphologically, in all but 1 case, the displaced epithelial cells had a micropapillary architecture (Figures 1 and 3). The exception was a case of high-grade DCIS (comedo type) in which the displaced epithelial cells formed clusters with no specific architecture (Figure 4).

Displacement of epithelial cells has been described in sites such as the thyroid gland (3.3%–43%)5 and lymph nodes (1.4%–20.3%)8 after needling procedures. The variability in incidence of DE is most likely due to differences in the type of needle used, the actual technique, or the nature of the underlying lesion. In the breast, estimates of the incidence of DE have ranged from 7% to 36%. However, a true incidence of DE due to core biopsies in any given series is impossible to ascertain, because the majority of such specimens are benign and are not followed by excision. Displacement is thought to occur less with directional vacuum-assisted biopsy than with automated core biopsy.6 Perhaps this is because of the mechanics of the procedure. In contrast to automated core biopsy, with vacuum-assisted biopsy, the needle is fired once, adjacent to rather than through the lesion; a larger cellular volume is obtained by suction and is pulled into the probe rather than being initially pushed, fragmented, and possibly displaced along the needle tract or left behind at the biopsy site.

It is important to be aware of DE in order to distinguish it from truly infiltrative or potentially metastatic malignancies, especially in cases in which the index lesions are benign or purely DCIS. For this purpose, we required that DE cells be present in stroma or in an area with changes due to previous biopsy or excision (Figures 1 and 3) and lacking a desmoplastic reaction. Similar criteria have been employed in the thyroid gland and lymph nodes to differentiate DE from infiltrative or metastatic malignancies.9 

In the breast, DE has been described in benign lesions such as papillary hyperplasia and intraductal papilloma as well as malignancies such as DCIS and invasive carcinomas.1,2 In our series, with the exception of 3 cases, DE was predominantly (94.3%, 50/53 cases) associated with and possibly a function of 1 or more underlying papillary lesions (Table). There was a spectrum of papillary lesions ranging from intraductal papilloma without any atypia (12.0%, 6/50 cases) to invasive papillary carcinoma (2.0%, 1/50 cases). Given that the most common diagnosis in our series was DCIS (84.9%, 45/53 cases), it was not surprising that the predominant papillary lesions in the remaining cases consisted of DCIS involving intraductal papilloma (64.0%, 32/50 cases), micropapillary or papillary DCIS (46.0%, 23/50 cases), and some with features of both (24.0%, 12/50 cases). One of the cases of micropapillary DCIS also had invasive micropapillary carcinoma.

The displaced cells in most cases were malignant appearing (Figure 1), the exceptions being the intraductal papillomas (Figure 3). With the exception of 1 case of comedo DCIS, the displaced cells in both the benign and malignant papillary lesions consistently had a micropapillary appearance, supporting their origin from papillary lesions. In addition to the case of comedo DCIS, there were 2 more cases that also lacked an underlying papillary lesion but that nevertheless had micropapillary-appearing displaced cells. They included 2 cases of IC, a case of colloid carcinoma, and an IC in which DCIS involved a cyst. The appearance of the cells in these cases can be ascribed to the fluid media in which they were suspended, that is, the cyst fluid and mucin, both of which can cause cells to cluster into micropapillary-like acini.5 

The frequent association of DE with papillary lesions may be explained by their inherently friable nature. In papillary lesions, whether benign or malignant, despite the presence of a supporting fibrovascular core, the arborizing, exophytic, frondlike structures are loose and may break off with or without trauma. Similarly, in micropapillary DCIS and IC of micropapillary type, the delicate micropapillae seem to float in luminal spaces with minimal if any structural support. Papillary and micropapillary lesions may also be admixed, as seen in 24.0% (12/50) of our cases.

Conceivably, with manipulation from needling procedures, these vulnerable, delicate papillary lesions could be amputated relatively easily and also be retained in a healing biopsy site. In fact, the morphology of the displaced cells in cases originating from papillary lesions may be described as “micropapilloma”-like (Figures 1 and 3).3 Thus, we believe it is the inherently fragile structure of papillary lesions that puts them at high risk for DE.

Another contributing finding for DE is the presence of a cystic component. Theoretically, one could explain DE in cysts by the rupture of the cystic area during the needling procedure, leading to spillage of the tumor cells into the needle tract, perhaps being carried by the cystic fluid. This was evident in 1 case of IC in which the DCIS (pure cribriform type) involved a cyst, with no papillary lesion evident (Figure 5). Cystic areas are also often a secondary component of papillary lesions. The dual effect of the friable papillary lesion superimposed on the release of fluid pressure from puncture of the cyst4 may enable the exfoliated epithelial cells to escape into the adjacent tissue and implant into the developing granulation tissue of the needle tract. In fact, in 6 cases, DCIS involved both an intraductal papilloma and a cyst. In addition, with the exception of 2 cases, the pattern of DCIS in all these cases was papillary or micropapillary.

Another previously described association with DE, seen in 2 cases, is mucinous carcinoma.10 The lack of “fibrous reaction or supporting stroma” is attributed as the cause for DE in mucinous carcinomas. Others have described the presence of multifocal hemorrhage within the mucin pools of a colloid carcinoma after fine-needle aspiration,5 possibly due to the soft mucoid nature of the tumor, which provides poor mechanical support for the vasculature. As with the cyst fluid, the mucinous material and hemorrhage may also serve as a transport medium for the DE. There were 2 cases of mucinous carcinoma in our series. The first was a pure colloid carcinoma that lacked DCIS. In the second case, the mucinous differentiation was focal and thus probably a secondary cause for DE, the primary lesion being micropapillary DCIS. Lastly, there was 1 case of high-grade DCIS in which none of the reasons for DE described here were present.

The biologic significance of DE is currently unknown. A few preliminary studies in the literature show that the cells do not survive displacement.8 One explanation is that the displaced epithelial cells die as time progresses or are killed by the actual trauma of the needling procedure. However, it is uncertain whether these displaced cells would have any effect before dying or whether they can actually implant elsewhere and survive. Even assuming that the cells can implant, the order of magnitude needed for successful tumor autotransplantation is probably much greater than that seen in postneedling surgical excision specimens.11,12 In addition, in the absence of the right surrounding milieu, these cells may lack any infiltrative or metastatic potential.

In summary, our series demonstrates that DE occurs predominantly in papillary lesions (94.3%), regardless of the type of needling procedure (encompassing a range from small needles used for anesthesia administration to suture needles to core and mammotome biopsies). These cases included 1 or more of the following: pure intraductal papilloma or intraductal papilloma admixed with DCIS, micropapillary or papillary DCIS, or invasive carcinoma. Other predisposing factors included cysts and colloid carcinoma.