Previous attempts at subclassifying pure mucinous carcinomas of the breast based on architectural patterns and other associated features are on record. A distinctive micropapillary variant with most tumor cells arranged in micropapillae/pseudoacini has not been described previously.
The author reviewed the histologic slides from all 556 patients who underwent wide excision/mastectomy for mammary ductal carcinoma, either in situ or invasive, at Pamela Youde Nethersole Eastern Hospital, Hong Kong, during an 8-year period from early 1994 to the end of 2001. Five patients with pure mucinous carcinoma with diffuse micropapillary architecture were noted. The cytologic features (if available) were correlated with clinical, radiologic, and pathologic findings.
Among the five patients with pure mammary mucinous carcinoma (micropapillary variant), three patients underwent preoperative fine-needle aspiration biopsies and had specimens available for review. The direct smears and cytospin preparations were of moderate cellularity and showed cohesive clusters and micropapillae of mildly pleomorphic tumor cells among a mucoid background. True tumor papillae with fibrovascular cores were not present. Nuclear hobnailing was observed commonly, and occasional psammoma bodies were found. There were also scanty isolated tumor cells scattered around. The pseudoacinar pattern was appreciated more readily in the cell block sections. Histologic examination of the surgical specimens showed features of pure mucinous carcinoma with diffuse micropapillary architecture. The micropapillary arrangement was confirmed further by the demonstration of a reverse polarity immunostaining pattern for epithelial membrane antigen and the identification of microvilli rimming the periphery of tumor cell clusters under the electron microscope. Peritumoral lymphovascular permeation and ipsilateral axillary lymph node metastasis was found in one of the patients.
Mammary mucinous carcinoma is defined as ductal carcinoma of the breast associated with abundant extracellular mucin, which is often discernible macroscopically.1 It can be subdivided into pure and mixed types according to the different proportions of mucinous and nonmucinous elements. The precise definition of pure mucinous carcinoma of the breast is still controversial and has not been agreed on universally. In general, if the mucinous component constitutes > 90% of the tumor area, a diagnosis of pure mucinous carcinoma is warranted.2 The distinction between pure and mixed mucinous carcinoma is important, because patients with the former type have a much better prognosis, with a low propensity for lymph node and distant metastasis.3–8 Conversely, the prognosis for patients with the mixed type of mucinous carcinoma does not differ from the prognosis for patients with conventional invasive ductal carcinoma.9 The metastatic foci often consist of nonmucinous elements. With regard to pure mucinous carcinoma in general, the tumor cells may be arranged in ribbon-like cell cords, trabeculae, cribiform structures, and solid lobules. They sometimes show neuroendocrine differentiation,10–13 especially if there is a prominent, organoid, cellular arrangement. In general, mucinous carcinoma with neuroendocrine differentiation tends to occur among patients in older age groups. Previous attempts to subclassify mammary mucinous carcinoma based on architectural growth patterns and other associated features are record.10, 12 Mucinous carcinoma with tumor cells that have a diffuse micropapillary arrangement, morphologically equivalent to a mucinous counterpart of invasive micropapillary carcinoma, have not been described in the literature. The author reviewed five patients with the micropapillary variant of pure mucinous breast carcinoma who were seen at Pamela Youde Nethersole Eastern Hospital, Hong Kong, during an 8-year period from early 1994 to the end of 2001, with immunohistochemical and ultrastructural correlations. The fine-needle aspiration cytology findings from three of those patients were analyzed.
MATERIALS AND METHODS
During the 8-year period from early 1994 to the end of 2001, a total of 556 wide excision/mastectomy specimens of mammary ductal carcinoma (either ductal carcinoma in situ [DCIS] or invasive ductal carcinoma) were obtained from patients who were seen at Pamela Youde Nethersole Eastern Hospital, Hong Kong. After reviewing all histology slides, 21 samples (3.8%) of pure mucinous carcinoma and 15 samples (2.7%) of mixed mucinous carcinoma were noted. Among the 21 patients who had pure mucinous carcinoma, 5 patients had tumors with diffuse micropapillary architecture. Four of the five patients (Patients 1, 2, 3, and 5) underwent fine-needle aspiration biopsy before surgical resection. The cytology slides from Patient 5, who was a consultation patient, were no longer available for this retrospective study. The preoperative diagnosis for Patient 4 was made with histologic examination of the core biopsy.
The clinical, radiologic, cytologic (if any), and histologic features of all five patients were reviewed. The various parameters are summarized in Table 1. Chest X-ray, ultrasound examination, computerized tomography scan of the abdomen, and bone scan (if indicated) were carried out to rule out any distant metastasis. The duration of follow-up ranged from 3 months to 3.5 years.
Table 1. Summary of Clinical, Radiologic, and Pathologic Features of the Five Reported Patients with the Micropapillary Variant of Mammary Mucinous Carcinoma
LN: lymph node.
Left breast mass (1.8 cm)
Ductal carcinoma with mucinous features
Wide local excision, sentinel LN sampling, and adjuvant chemoradiotherapy
Left breast mass (6.0 cm)
Ductal carcinoma with mucinous features
Mastectomy, axillary LN dissection, oral tamoxifen, and adjuvant radiotherapy
Left breast mass (2.0 cm)
Ductal carcinoma with mucinous features
Wide local excision, axillary LN dissection, oral tamoxifen, and adjuvant radiotherapy
Right breast mass (3.0 cm)
Mastectomy, axillary LN dissection, and adjuvant chemotherapy
Right breast mass (3.7 cm)
Ductal carcinoma with mucinous features
Mastectomy, axillary LN dissection, and oral tamoxifen
The fine-needle aspiration biopsies for Patients 1–3 were performed using 21-gauge or 22-gauge needles, either free-hand or under ultrasound guidance. If the material was sufficient, several direct smears were prepared and promptly fixed in 95% ethanol overnight. The needles and syringes were rinsed in an equal volume of 50% ethanol and normal saline, which was centrifuged later at a rate of 1500 rpm for 20 minutes. The sediment obtained, if any, was processed using the same method that was used for routine paraffin blocks. Cytospin preparations (Shandon, Cheshire, United Kingdom) were also produced from the supernatant. All smears and cytospin preparations were stained with Papanicolaou stain (EA 65) or hematoxylin and eosin. Four-micron-thick paraffin sections were cut from the cell blocks and stained with hematoxylin and eosin. Immunohistochemical studies using the Streptavidin-biotin complex technique for epithelial membrane antigen (EMA; Dako, Glostrup, Denmark) also were performed on the cell block sections.
The surgical specimens were submitted in 10% neutral buffered formalin. Representative tissue blocks were taken from the tumors. The samples were then processed using the method used for routine paraffin blocks. Four-micron-thick sections were stained with hematoxylin and eosin as well as mucicarmine stain. Immunohistochemical studies (Table 2) using the Streptavidin-biotin complex technique for estrogen receptor (Ventana, Tucson, AZ), progesterone receptor (Zymed, San Francisco, CA), c-erbB2 (Zymed), chromogranin A (Dako), synaptophysin (Dako), and EMA were performed on the paraffin sections either manually or with the help of a Ventana automated immunohistochemistry slide-staining device. Ultrastructural studies were also carried out for Patients 1 and 4 using paraffin embedded tissue and formalin fixed tumor samples, respectively. For both studies, 1-mm-thick tissue blocks were trimmed, dewaxed, rehydrated (if necessary), postfixed in 1% osmium tetroxide, embedded in epoxy resin, sectioned, stained with uranyl acetate-lead citrate, and examined under a transmission electron microscope (CM100; Philips, Eindhoven, The Netherlands).
Table 2. Monoclonal Antibodies Used for Immunohistochemical Study
A Ventana automated immunohistochemistry slide-staining device was used.
All patients were Chinese women, and their ages ranged from 36 years to 72 years. Except for Patient 5, they were premenopausal. The sizes of tumors varied from 1.8 cm to 6.0 cm in greatest dimension. All patients underwent definitive surgery that was done after receiving an initial cytologic/core biopsy diagnosis of ductal carcinoma with mucinous features. Ipsilateral axillary dissection was carried out in all but Patient 1 to look for possible lymph node metastasis. Postoperatively, the patients were given adjuvant treatment, including oral tamoxifen, local irradiation, and/or systemic chemotherapy, depending on the hormone receptor status of their tumors as well as other clinical parameters. All five patients were still alive and remained disease free after a follow-up of 3 months to 3.5 years. There was no clinical or radiologic evidence of distant metastasis.
Histologic examination of the surgical specimens showed the features of pure mucinous carcinoma with diffuse micropapillary architecture (Fig. 1). Nonmucinous, solid elements, including invasive micropapillary carcinoma, were not evident. The histology recapitulated that seen in the cytology cell block sections. Many of the tumor micropapillae resembled small acini with central lumen. However, these were not true acini, as verified by the demonstration of reverse polarity immunostaining pattern for EMA (Fig. 2) and the identification of microvilli rimming the periphery of tumor clusters under the electron microscope (Fig. 3). The tumor cells showed a mild degree of nuclear pleomorphism. The mitotic count measured much less than 1 per 10 high-power fields (×40x objective; Nikon, Tokyo, Japan). At most, intracytoplasmic mucin was demonstrated occasionally. Immunohistochemical studies of neuroendocrine markers (chromogranin A and synaptophysin) were negative, except for focal cytoplasmic staining in some tumor cells from Patient 5. The surrounding stroma commonly showed sclerotic change and mild chronic inflammation. Psammoma bodies were identified in Patients 2 and 3. Small foci of associated, low-grade, cribiform DCIS were noted only in Patient 1. Peritumoral lymphovascular permeation and ipsilateral axillary lymph node metastasis were found in Patient 4 (Fig. 4). The mucinous and micropapillary features, as well as the EMA staining pattern, were maintained in the lymphatic tumor emboli and metastatic foci. The hormone receptor status and c-erbB2 expression for each patient are summarized in Table 3.
Table 3. Immunohistochemical Studies of Hormone Receptors and c-erbB2 Performed on Surgical Specimens from the Five Reported Patients
Fine-Needle Aspiration Cytology Findings
The fine-needle aspirates from all the three patients showed similar features. The direct smears and cytospin preparations were of moderate cellularity and contained cohesive clusters and micropapillae of mildly pleomorphic tumor cells with low mitotic activity along with a mucoid background (Fig. 5). Some of the micropapillae showed abortive branching (Fig. 6). True arborizing tumor papillae with fibrovascular cores were not seen. Nuclear hobnailing was not uncommon (Fig. 7). Occasional tumor cells contained intracytoplasmic vacuoles. Scanty psammoma bodies were found in aspirates from Patients 2 and 3. There also were small numbers of isolated tumor cells scattered around. Some tumor cell sheets had a monolayered, honeycomb appearance (Fig. 8), superficially mimicking benign proliferative breast lesions. Naked nuclei and foamy macrophages were not seen in the background. The micropapillary/pseudoacinar pattern was appreciated better in the cell block sections (Fig. 9). Immunohistochemical studies performed on the cell block sections (available only for Patients 1 and 2) showed mainly cytoplasmic positivity for EMA. However, the reverse polarity pattern (with EMA staining outlining the periphery of tumor micropapillae rather than the central pseudolumen) that is observed characteristically in invasive micropapillary carcinoma of the breast was not seen readily.
Classic examples of ductal carcinoma with micropapillary architecture include micropapillary DCIS and invasive micropapillary carcinoma. The nature of these two entities is different and probably is not related. Micropapillary DCIS, which is a low-grade in situ malignancy that is commonly associated with multicentricity,14 may undergo cystic changes and may show histologic overlaps with cystic hypersecretory DCIS.15 Conversely, invasive micropapillary carcinoma carries a high propensity for lymphatic permeation and regional lymph node metastasis.16–20 Histologically, it is characterized by micropapillary clusters of tumor cells separated by empty spaces and sclerotic fibrous septae. Immunohistochemical studies with EMA often show a reverse polarity staining pattern, confirming its true micropapillary nature. Several small series on fine-needle aspiration cytology findings in patients with invasive micropapillary carcinoma of the breast are on record.21–24 Cytologically, there are multiple, tightly cohesive clusters of mildly to moderately pleomorphic tumor cells with focal, abortive branching. True tumor papillae with fibrovascular cores are not found. Some of the micropapillary clusters tend to mold into one another, recapitulating what seen in histologic sections. Some authors have described the focal presence of mucoid stroma in fine-needle aspirates.25 However, the features of mucinous carcinoma, including an abundance of background mucin, are not evident. In view of its high propensity for peritumoral lymphovascular permeation and local recurrences, mastectomy, rather than other conservative surgical management, is usually advocated.26 Except for the abundance of extracellular mucin, the current examples do show an architectural pattern equivalent to the pattern seen in invasive micropapillary carcinoma. In addition, other associated stromal features (namely, sclerotic change, focal chronic inflammation, and small numbers of psammoma bodies) are also similar. At least morphologically, the entity described here may represent a mucinous counterpart of invasive micropapillary carcinoma.
The micropapillary variant constitutes 24% of all pure mammary mucinous carcinomas that were encountered during the study period. It is apparent that this peculiar morphologic subtype has been under-recognized. In fact, some of the corresponding illustrations in textbooks on breast pathology, including those showing lymphatic permeation by pure mucinous carcinoma, show a conspicuous micropapillary architectural pattern.2 The glandular pattern seen in some mucinous carcinoma may be due simply to pseudoacinar configurations of the tumor micropapillae. Further morphologic categorization, together with EMA immunostaining, however, has not been investigated thoroughly. Given the distinctive cytohistologic appearance of this micropapillary variant and its possible histogenetic links with invasive micropapillary carcinoma, it should be included in the list of mammary mucinous carcinoma subcategories. Focal expression of neuroendocrine markers was also noted in Patient 5, suggesting some immunophenotypic overlaps with the neuroendocrine subtype of mucinous carcinoma.
Cytologically, the micropapillary variant of mucinous carcinoma has many features that recapitulate the features seen in histologic sections. In addition to the presence of abundant, thick, mucoid substance in the background, there are multiple, cohesive clusters of low-grade tumor cells with obvious micropapillary arrangement. Nuclear hobnailing and loosely dispersed, malignant cells are found on occasion. Psammoma bodies, which are not uncommon in tumors with papillary/micropapillary growth patterns (such as papillary thyroid carcinoma and ovarian serous adenocarcinoma), are also observed in some patients. The cell block sections, if available, allow better appreciation of the micropapillary architecture. Superficially, many of the tumor micropapillae in histologic sections resemble tubules and acini and contain a central pseudolumen. Unlike immunohistochemical studies performed on surgical specimens, EMA staining using the cell block materials often fails to demonstrate the reverse polarity pattern. Instead, there is mostly cytoplasmic positivity. The main reason is probably related to distortion of membranous antigen during specimen processing, which involves a step of vigorous centrifugation. Similar experience of EMA immunostaining has also been noted in fine-needle aspiration cytology specimens from patients with invasive micropapillary carcinoma.
With regard to the cytologic differential diagnosis, this entity must be distinguished from other subtypes of mucinous carcinoma, invasive micropapillary carcinoma, micropapillary/cystic hypersecretory DCIS, other mucocele-like lesions, and even myxoid fibroadenoma. In contrast to the micropapillary variant, the neuroendocrine subtype of mucinous carcinoma tends to yield loosely dispersed tumor cells on fine-needle aspiration biopsies.27, 28 The tumor cells show neuroendocrine cytology, including stippled chromatin, a plasmacytoid appearance, and granular eosinophilic cytoplasm. Conversely, fine-needle aspirates of mucinous carcinoma with conventional ribbon-like architecture often are hypocellular and are comprised of nondescript cell clusters and sheets. A striking micropapillary pattern is often not identified. The micropapillary variant of mucinous carcinoma does sometimes mimic invasive micropapillary carcinoma, because both entities show an essentially similar cellular arrangement. However, mucinous features in invasive micropapillary carcinoma, if any, are just occasional. The tumor cells are often of higher grade and show more florid mitotic activity compared with the tumor cells in mucinous carcinoma. Nuclear hobnailing and cellular dispersion in invasive micropapillary carcinoma is less conspicuous. With regard to micropapillary DCIS, the micropapillary configurations are not as obvious as what was shown in the current study.29 In general, fine-needle aspirates of DCIS also do not contain discernible amounts of background mucin. In patients with cystic hypersecretory DCIS, thick, colloid-like substances are seen instead.30–35 The tumor cells in cystic hypersecretory DCIS also demonstrate high secretory activity, with frequent intracytoplasmic vacuoles noted. Mucocele-like lesions, which are considered by some authors in a continuum with mucinous carcinoma,36–38 tend to produce paucicellular cytologic preparations.39–41 Three-dimensional, cohesive clusters of ductal cells with discernible nuclear pleomorphism are not found. Finally, the abortive branching pattern of the micropapillae may mimic the staghorn configurations of the ductal cell clusters seen in fine-needle aspirates of fibroadenoma. The background mucin, when admixed with isolated tumor cells of low cytologic grade, may also resemble myxoid stromal tissue fragments of myxoid fibroadenoma. The mimicry of benign proliferative breast diseases is enhanced further by the focal, monolayered, honeycomb arrangement of the mucinous carcinoma cells. The controversy can be resolved by paying attention to the monotonous cell population, the presence of cytoplasm (although scanty) in the dispersed cells, and the true mucinous nature (as confirmed by mucicarmine stain) of the background substances in patients with mucinous carcinoma. The predominance of micropapillary structures and the absence of admixed, actin positive myoepithelial cells in the cell block sections is also helpful in making the distinction.
In one of the five reported patients (i.e., 20% of all patients with pure mucinous carcinoma of the micropapillary variant described here), peritumoral lymphovascular permeation and regional lymph node metastasis, consisting of pure mucinous element, was found. Although lymph node metastasis of mixed mucinous carcinoma is not uncommon, metastatic pure mucinous carcinoma is considered a rarity in breast pathology.2 Moreover, the tumor cells in Patient 4 showed strong and diffuse membranous staining for c-erbB2. Is this propensity for lymph node metastasis related to micropapillary growth pattern, which is analogous to that in invasive micropapillary carcinoma, or, is this related to tumor cell overexpression of c-erbB2? Further analysis of larger series of patients will be required for better clarification. In general, despite the occasional occurrence of lymph node secondary disease, the clinical course of micropapillary variant of mucinous carcinoma appears relatively indolent compared with the clinical course of invasive micropapillary carcinoma. The abundant extracellular mucin that was present here may serve as an obstacle to lymphovascular permeation, explaining the relatively good prognosis of patients with this peculiar morphologic entity.