- Email: [email protected]
Metaplastic carcinomas of the breast. II. Spindle cell carcinoma
Metaplastic Carcinomas of the Breast. II.Spindle Cell Carcinoma ERIC S. WARGOlZ, MD, PHILLIP HoDEOS, MD, AND HENRY J. NORRIS, MD The clinical and pathologic features of 100 examples of spindle cell carcinoma (SpCC) of the breast are reported. Eighty-three neoplasms contained overt carcinoma; 72 had infiltrating ductal or intraductal carcinoma and in 11 the carcinomatous component was purely squamous. Seventeen neoplasms lacked overt carcinoma, but were identified as SpCC by immunoreactivity for keratin and the typical bland spindle cell proliferation forming a variable complex of fibrocollagenous stroma with feathered, myxoid, angioid, and storiform patterns. Areas of epithelium merging imperceptibly with the spindle cell component were commonly observed. Sixty neoplasms were studied by immunohistochemistry for the presence of keratin, epithelial membrane antigen (EMA), vimentin, S-100, and actin. The spindle cell component in 98% of SpCC was immunoreactive for keratin. Most were also immunoreactive for vimentin and actin, and in approximately one half, S-100 immunoreactivity was noted. These findings, in conjunction with histopathologic features, and ultrastructural observations from three cases, support myoepithelium as an integral component of SpCC. The cumulative &year survival rate for SpCC was 64%, better than survival rates usually reported for metaplastic carcinomas. Of 47 patients with axillary dissection, only 6% had metastases to axillary lymph nodes. Development of metastasis was an ominous sign as 29 of the 30 patients who developed metastases died from tumor. Local recurrence was not as ominous as only 29% who had only local recurrence subsequently died from tumor. The difference in size between tumors that recurred (mean, 5.0 cm) and those that did not (mean 3.7 cm), and the presence or absence of complete microscopic circumscription, were both significant prognostic factors. HUM PATHOL 20:732-740. 0 1989 by W.B. Saunders Company.
Spindle cell carcinoma (SpCC) of the breast is a rare neoplasm in which the spindle cell component predominates, resembling a low-grade sarcoma or a reactive process such as fasciitis or granulation tissue. A variety of neoplasms have been included in the designation “metaplastic carcinoma” and the behavior of specific variants is unclear. We define the histopathologic features of 100 spindle cell carcinomas of the breast, correlate the features with the clinical course, and discuss the origin as suggested from the results of immunohistochemistry and ultrastructural observations.
From the Department of Breast and Gynecologic Pathology, Armed Forces Institute of Pathology, Washington, DC. Accepted for publication December 30, 1988. Key words: metaplastic carcinoma, spindle cell carcinoma, breast carcinoma with pseudosarcomatous metaplasiaktroma, carcinosarcoma, squamous carcinoma, myoepithelium, immunohistochemistry, electron microscopy. Address correspondence and reprint requests to Henry J. Norris, MD, Department of Breast and Gynecologic Pathology, Armed Forces Institute of Pathology, Washington, DC 203066000. 0 1989 by W.B. Saunders Company. 0046-8177/89/2008-0003$5.00/O
732
MATERIAL AND METHODS All lesions of the breast coded as metaplastic carcinoma, spindle cell carcinoma, carcinoma with spindle cell metaplasia, pseudosarcoma, carcinosarcoma, squamous carcinoma, low-grade mesenchymal tumor, fasciitis, and sarcoma in the files of the Armed Forces Institute of Pathology (AFIP) accessioned before January 1983 were retrieved and reviewed. Of these, 100 examples of SpCC were identified. Neoplasms were classified as SpCC if intraductal or infiltrating carcinoma was contiguous or subtly merged with a bland, monotonous-appearing spindle cell proliferation composed of bipolar cells with plump nuclei growing as wavy overlapping or interweaving fascicles often forming a storiform pattern with admixed dense collagen, and focal myxoid and angioid stroma. The spindle cell component always comprised more than half of the neoplasm and was more than 80% of it in the majority of cases. Neoplasms lacking overt carcinoma, but having the characteristic spindle cell stroma, were included if they were immunoreactive for keratin. An average of six blocks of tumor were studied per case. Neoplasms with multinucleate osteoclastic giant cells were excluded and will be dealt with in another report. The pleomorphism of the spindle cell component was assessed on a scale of 1 to 3 with grade 1 assigned to relatively uniform, monotonous-appearing cells with uniform nuclei, a fine granular chromatin pattern, and inconspicuous nucleoli. Grade 2 neoplasms had a moderate degree of cellular pleomorphism and nuclei with a single nucleolus, but no multinucleate tumor giant cells. Grade 3 was assigned to neoplasms having areas of marked cellular pleomorphism and prominent nucleoli, often with multinucleate tumor giant cells. The architectural differentiation of the overt carcinoma was based on the degree of gland formation and scored as 1 (well differentiated), 2 (moderately differentiated), and 3 (poorly differentiated). Squamous differentiation, when present, was not graded. Cellularity was inversely proportional to the amount of collagen and was graded 1 + (25% or fewer cells) to 4+ (75% or more cells) following microscopic examination of the tissue sections through a 25 x objective. Neoplasms were designated as fibrocollagenous, fibromyxoid, or angioid depending on the predominant histopathologic features. Areas of maximum cellularity were identified and number of mitotic figures per 10 high power fields (HPFs) (area = 0.159 mm2) was recorded. In 1 + cellular lesions, 20 HPFs were examined with the result divided by two and expressed per 10 HPFs. Chronic inflammatory infiltrates, whether admixed with the neoplasm or predominantly peripheral, were assessed on a scale of 1 + (sparse) to 3 + (marked). Neoplasms with lymphoid follicles either admixed or at the periphery were considered as 3 + . Clinical data were tabulated from the patients’ charts and analyzed in conjunction with the following gross and microscopic characteristics: size, contour (nodular, irregular), microscopic borders (circumscribed, or infiltrative), cellularity, cellular pleomorphism, mitotic activity, inflammation and presence or absence of intraductal or overt infiltrating carcinoma and the degree of differentiation. All
MAMMARY SPINDLE CELL CARCINOMA (Wargotz et al]
Gross and Microscopic Pathology
patients were eligible for 5-year follow-up, and follow-up information was obtained. Life tables were designed according to the method of Berkson and Gage.’ Analysis of variance, (anova) with size as the response variable, and hypothesis testing of differences in proportions were used to test the significance of histologic differences between neoplasms that recurred and those that did not. Analysis of differences in cumulative 5-year survival rates was based on a standard normal test of proportions. All P values were derived from two-tailed tests.
All SDCC were within the mammarv Darenchyma. They ranged in size from 0.6 cm to ‘21 cm, with a mean diameter of 4.4 cm, and a median and mode of 4.0 cm. Forty-five were nodular and apneared well-circumscribed. Twentv had an irregular beriphery, and in 35 the contour bas not descrybed. The neoplasms were firm to hard and gritty, with a dull, white to grey or pink to tan cut surface. They were frequently mottled, and with one or more cysts. Several neonlasms had vellow or chalkv streaks. but central necrosis and hemorrhage were not evident in any. Microscopically, 49 SpCC had infiltrating borders. while 18 were Dredominantlv circumscribed or nodular, and the peiiphery of 3 l’had combinations of nodular and finger-like infiltrating growth patterns (Fig 1). The borders of two neoplasms were not reoresented in the available tissue -sections. None w&e encapsulated. Vascular invasion by the spindle cell component was found in three neoplasms, while neural invasion was not identified. Although infiltrating carcinoma exclusively was evident in 48 neodasms. it was often limited to onlv a few fields. Intriductal carcinoma exclusively ias found in seven cases. Both intraductal and infiltrating carcinoma were observed in 17 cases. Intraductal carcinema. when oresent. was most often identified at the periphery. Infiltrating carcinoma usually blended with the spindle cell component. None of the intraductal carcinomas had spindle cell metaplasia within the ducts. The maioritv of infiltrating carcinomas were moderate or hoorly differentiate;. Squamous differentiation of the carcinoma component was
lmmunohistochemistryand Transmission Electron Microscopy Immunohistochemical studies, performed on paraffinembedded tissue from 60 neoplasms, included cytokeratins (two immunoreagents), epithelial membrane antigen (EMA), S-100 protein, actin, factor VIII, and vimentin. Studies for cyto-keratins were performed on all 60 neoplasms. The polyclonal antiserum to keratin proteins was prepared at the AFIP as described previously.2 The immunohistochemical techniques used, and monospecificities of the commercial polyclonal antibodies to S-100 protein and factor VIII, as well as specificity of commercial monoclonal antibodies to keratins (AE1/3), EMA, actin, and vimentin, were evaluated as before.2 The pattern of cellular immunoreactivity of the tumor was semiquantitated and designated as being diffuse (>30%), intermediate (10 to 30%), focal ( a single focus or multiple discrete foci), or rare (occasional isolated cells). The relative intensity of individual cellular immunoreactivity was not examined. Wet tissue from three cases was available for transmission electron microscopy. Samples were processed and examined as previously described.2
RESULTS Clinical Findings The patients ranged in age from 29 to 95 years, with an average age of 63, median age of 65, and a mode of 72. All but one were women. Of the 82 with known racial extraction, only two were black, two 78 (92%) were were Hispanic, and the remaining white. Of the 99 women, only 11 were premenopausal. All patients presented with a single palpable, firm mass. The tumor was fixed to the skin in seven cases and to the chest wall in one. Skin or nipple ulceration was present in 11 cases, nipple retraction was found in six, and nipple discharge was noted in one. Axillary enlargement was absent in all cases. Fifty-two neoplasms occurred in the right breast, 44 in the left, and four were not designated. In the cases in which the quadrant was given, 46% occurred in the upper outer quadrant. The remainder presented with near equal frequency in the lower outer and the two inner quadrants, or as a subareolar mass. Neoplasms had been present 1 week to 5 years, but most were present for 2 weeks to 3 months. One woman claimed the lump had been present for 45 years. Only six reported pain or tenderness. None of the patients gave a history of trauma before the occurrence of the tumor. Serum carcinoembryonic antigen was not elevated in two patients tested.
FIGURE 1. Despite gross circumscription, the margins of this SpCC are infiliratbe. Both nodular and finger-like growths are evident. (Magnification x 5.1
733
HUMAN PATHOLOGY
Volume 20, No. 8 (August 1989)
present in 35 neoplasms. It was the exclusive form of overt carcinoma in 11 neoplasms. Squamous nests, some with keratin pearls, were often connected to one another by thin strands of epithelium. The cells at the periphery of these nests, and of some infiltrating duct carcinomas, appeared to stream off the main mass and imperceptibly merge with the surrounding spindle cells (Figs 2 through 4). Other ducts had a marked circumferential myoepithelial cell prominence resulting in a cuffed appearance (Fig 5). The spindle cell component was predominantly fibrocollagenous, with the spindle cells arranged in wavy, interlacing and overlapping fascicles often splaying or fanning out at low power, resembling the feathered pattern of fasciitis in some areas (Fig 6) and the storiform pattern of dermatofibrosarcoma protuberans and malignant fibrous histiocytoma (Figs 7 and 8) in others. Some neoplasms resembled cellular examples of fibromatosis or a low-grade fibrosarcoma with finger-like extensions into adjacent fat (Fig 9). Focally, epithelium alone or with spindle cells was interspersed among rounded trabecular bands of hyaline collagen, sometimes indistinguishable from osteoid, and aggregates of basement membrane material (Fig 10). Angioid and/or fibromyxoid-appearing areas were frequent minor components and the predominant component of a few neoplasms. Angioid areas were distinctive, consisting of complex arrangements of intersecting strands of flattened endotheliallike cells in a collagenous or fibromyxoid and edematous stroma mimicking a capillary network (Fig 11). Several neoplasms had areas of broad channels lined by plump endothelioid epithelial cells mimicking angiosarcoma (Fig 12). Nineteen percent of the neoplasms were of 1 + cellularity, 42% were 2+, 33% were 3+, and 6% were 4 + . Some 1 + cellular neoplasms were densely collagenous with only small foci of spindle cells (Fig 10) or a thin rim of spindle and inflammatory cells
FIGURE 2 An interweaving pattern is formed bY the squamous epithelium merging imperceptibly with the spindle cell stroma. (Magnification x 174.)
FIWRE 3. Differentiation zone of a squamous island. (Magnification x 174.)
along the periphery. Seventy percent were without appreciable pleomorphism of the spindle cell component (grade l), and 28% had only a mild to moderate degree of pleomorphism with some cells possessing a nucleolus (grade 2). None of the neoplasms had tumor giant cells, and only two had small foci of marked Spindle-shaped carcinoma cells pleomorphism. tended to have rounded plump nuclei, bipolar processes, and were twisted or contorted between collagen fibers (Fig 10). Many appeared to subtly merge with fibroblasts and myofibroblasts (Figs 2 and 3). In cellular areas the spindle cells were more plump (Figs 4 and 5), while in myxoid areas stellate shapes predominated (Fig 11). Mitotic figures of the spindle cell component ranged from 0 to 11 per 10 HPF with a mean of three and median of one. Abnormal forms were rarely seen. Eighty-four neoplasms had an inflammatory infiltrate predominantly composed of lymphocytes and plasma cells. The intensity of the infiltrate was 1 + in
FIGURE 4. Differentiation zones were also associated wtth duct carcinoma lacldng a squamous component. (Magnkaltcn x 209.)
734
MAMMARY SPINDLE CELL CARCINOMA (Wargotz et al)
FIGURE 5. Malignant myoepithelial proliferation around focus of intraductal carcinoma or atypical intraductal hyperplasia producing a “cuffed” appearance. (Magnification x 104.)
FIGURE 7. A focal storiform pattern is frequently present in SpCC of the breast. (Magnification x 87.)
The spindle cell component of 59 (98%) of the 60 neoplasms was immunoreactive for keratin (Fig 13). Nearly all neoplasms showed an intermediate or dif-
fuse pattern of immunoreactivity in the spindle cell component, while in only three neoplasms were rare cells positive. Immunoreactivity was not limited to the perinuclear zone and was also present in the tapered processes. Of the 47 neoplasms studied by both keratin immunoreagents, there was a concordance in only 77% due to the absence of AE1/3 immunoreactivity in ten polyclonal immunocytokeratin reactive neoplasms. In only one instance was AEU3 immunoreactivity observed in the absence of immunoreactivity to the polyclonal keratin antibody. The spindle cell component of 11 (23%) of 48 neoplasms evaluated for EMA was immunoreactive and the pattern of immunoreactivity was usually rare to focal. Vimentin was detected in the spindle cells of 53 (91%) of 58 neoplasms evaluated, usually in an intermediate or diffuse pattern. In five neoplasms, vimentin was also
FIQRRE 6. Cellular spindle cells areas arranged In irregular bundles with admixed inflammatory cells resembling the feathered pattern of fasciltts. (Magnification x 130.)
FIQRRE 8. This sparsely cellular 3pCC consists of dense collagen with narrow, meandering spindle cell fascicles. (Magnlflcattons: left, x 60; right, x 164.)
21 neoplasms, 2 + in 32, and 3 + in 3 1. In nine neoplasms, the inflammation was limited to the periphery. Eosinophils and mast cells were conspicuous in a few neoplasms. Granulomatous inflammation or acute inflammation was not observed. Necrosis, when present, was characteristic, and was represented by focal ischemic, densely acellular collagenous areas. Five neoplasms had focal osseous metaplasia, one had focal cartilaginous metaplasia, and two others had both focal cartilaginous and osseous metaplasia. Twenty neoplasms tested for mucin were negative.
lmmunohistochemistry
735
HUMAN PATHOLOGY
Volume 20, No. 8 (August 1989)
FIGRRE 11. (Left) Angioid areas consist of a complex arrangement of epithelial strands in a fibromyxoid and edematous, or collagenous stroma simulating a vascular network (Magnification x 38.) (Right) Higher magnification of another field shows focal gland formation. Often only the strands are present. (Magnification x 150.)
FIGRRE 9. Finger-like extension into adjacent mammary stroma is common in SpCC and resembles fibromatosis and fibrosarcoma. (Magnificatton x 35.)
detected in the overt carcinoma and/or foci of squamous differentation. S-100 protein was detected in the spindle cell component of 25 (51%) of 49 neoplasms evaluated and the pattern of immunoreactivity ranged from rare to focal in most, predominantly staining the more rounded cells. An intermediate to diffuse pattern of immunoreactivity for actin was noted in the spindle cells of 48 (96%) of 53 neoplasms evaluated. Of 39 neoplasms evaluated with the entire panel of six antibodies, keratins were detected in 97%, EMA in 21%, vimentin in 95%, S-100 protein in 51%, and actin in 87%. Some of these substances were coexpressed by the same subpopulation of spindle cells. Stromal myofibroblasts and fibroblasts accounted for at least some of the actin and vimentin immunoreactivity. None of the seven neoplasms evaluated for fac-
tor VIII showed component.
immunoreactivity
of the angioid
Ultrastructural Findings The three neoplasms studied by transmission electron microscopy were similar, composed of spindle cells and collagen. Three predominant types of spindle cells could be distinguished on the basis of filament content or composition: (1) spindled epithelial cells having bundles of tonofilaments (prekeratin or cytokeratin); (2) myoepithelial cells, identified by the presence of bundles of tonofilaments and actin filaments; and (3) myofibroblasts, identified by actin filaments, but lacking tonofilaments (Figs 14 and 15).
FIGRRE 10. Trabecular hyaline collagen (left) and areas of epithelium and basement membrane material (right). (Magnificatton x 156.)
FIGURE 12. Hemorrhagic area resembling angiosarcoma. (Magnification x 107.)
736
MAMMARY SPINDLE CELL CARCINOMA CJvargotzet al)
of 27 nodes. With the exception of one patient who chose incisional biopsy with radiation and chemotherapy, surgical margins appeared free of tumor. Receptor Stalus Most of the patients were seen before the availability of routine receptor analysis, but estrogen receptor analysis was performed on 18 neoplasms. In one, a marginally elevated level was found (14 fmol/ mg cytosol protein), while in 17 patients levels were not elevated. Progesterone receptor analysis was performed on 14 of the 18 neoplasms analyzed for estrogen receptors, and none were elevated. Recurrence and Survival Follow-up information was obtained on all 100 patients with a mean and median follow-up interval of 5.4 years. Local recurrence or metastasis occurred in 56 patients. In nine of these, the site and interval were unclear. Seventeen patients developed only local recurrence, 15 had local recurrence and metastasis, and 15 had metastasis only. First recurrence occurred from 3 weeks to 11 years with a mean of 22 months, and a median of 1 year. Of the 56 patients with recurrence, 44 (79%) subsequently died from tumor. Metastasis was an ominous sign as 29 of the 30 patients subsequently died from tumor, whereas following only local recurrence, 7 1% (12 of 17 patients) survived. The difference in mortality rate between these two groups was significant (P = .OlO). The three patients with positive axillary nodes at initial therapy died from widespread metastatic tumor. Four of the eight patients with neoplasms containing cartilage or bone subsequently died from tumor. The most common pattern of progression was local recurrence, metastasis to the pleura and lungs, and then metastasis to bone and other sites. Recurrence rates following each of the five surgical approaches were radical mastectomy 47%, modified radical mastectomy 44%, simple mastectomy 56%, partial mastectomy 38%, and biopsy 78%. Local recurrence was generally treated by surgery, and was most often a form of mastectomy for patients initially treated by biopsy. Of the eight patients who received adjuvant radiation and/or chemotherapy following initial surgery, only two have not died from tumor. One patient was treated by modified radical mastectomy followed by three cycles of doxorubicin hydrochloride and has been free of disease for 7 years. The second patient was initially treated by biopsy and radiation therapy (5,049 rads to the breast and supraclavicular area, and 780 rads to the axilla), developed a recurrence at 26 months and was treated by radical mastectomy. She is free of disease at 5 years after mastectomy. Radiation and/or chemotherapy for recurrence offered no significant advantage over surgery alone, as all 14 recipients died from tumor. Seven (4 1%) of the 17 neoplasms that lacked intraductal carcinoma or overt infiltrating carcinoma had been diagnosed originally as fasciitis, fibromatosis, or low-grade mesenchymal tumor and had re-
FIGURE 13. lmmunoreactivity for cytokeratin is nearly always present in SpCC and identifies them when an overt form of carclnoma is not apparent. [Left, hematoxylin and eosin. magnification x 205; right, polyclonal immunocytokeratin PAP, magnification x 164.)
No one cell type predominated and all were intermingled with one another. Most cells were rich in organelles. Well-formed desmosomes were infrequent, but poorly developed junctions were common. Microvilli projecting into slit-like intercellular lumina were identified in one case (Figs 14 and 15). Nuclear contour was variable, and nearly all cells had a prominent nucleolus. The curvilinear bundles of tonoftlaments were of the prekeratin or keratin type (10 nm), and were distributed in the perinuclear zone and within the tapered cytoplasmic processes (Figs 14 and 15). Actin filaments (6 nm) were arranged parallel to the long axis of the cell (Fig 15). Some cells had numerous dense bodies. Initial Therapy and Node Status Mastectomy was the form of initial therapy in 66 patients. Radical mastectomy was performed on 19, modified radical mastectomy on 29, and simple mastectomy on 18. Excisional biopsy was the initial therapy for 25 patients, and partial mastectomy (lumpectomy, tylectomy, or wide local excision) was performed on eight. Postoperative radiation was received by three patients following simple mastectomy and one patient following biopsy. Two patients received chemotherapy following modified radical mastectomy. Two other patients had both postoperative radiation and chemotherapy following partial mastectomy and incisional biopsy. Of the 48 patients with axillary dissection performed at initial presentation, the lymph node status is known for 47. Only three (6%) of the 47 were found to have metastatic carcinoma. In one woman, the spindle cell component was present exclusively in three of nine nodes, while in the two others, overt carcinoma was present exclusively in one of 15 and 16 737
HUMAN PATHOLOGY
Volume 20, No. 8 (August 1989)
FIGURE 14. Portions of three spindle cells, one with bundles of prekeratin and keratin tonofilaments [IQ and all three contributing to the formation of intercellular slit lumina with microvilli (L). (Magnification x 3,784.)
cinema or presence of squamous epithelium, and inflammation did not differ significantly between the two groups (Table 1). Despite the increased frequency of local recurrence in patients who were treated by biopsy, the delay in more extensive surgery did not result in increased death from tumor as 44% died from tumor, versus 48% of patients treated by a form of mastectomy. Only two of eight patients (25%) died from tumor following partial mastectomy.
ceived only excisional biopsy. Five (71%) of these recurred locally and more extensive surgical therapy was performed, but two of the patients subsequently died from tumor. Four of the other ten patients in this group who were treated by a form of mastectomy also eventually died from tumor. Twelve patients died of causes unrelated to their breast cancer. The cumulative 5-year survival rate for 100 patients with SpCC, adjusting for patients who died from other causes, was 64%. Histologic material and/or pathology reports of local recurrences and metadtases were available in 14 cases. Of these, seven were the spindle component only, three were overt carcinoma, and four showed an admixture of both. In four of these 14 cases, autopsies revealed widespread metastases and involved hilar and/or periaortic lymph nodes.
DISCUSSION SpCC is a distinct form of metaplastic carcinoma with a cumulative 5-year survival rate of 64%. This contrasts with 5-year survival rates of 44% to 55% for compilations of various forms of metaplastic carcinoma, none of which exceeds 33 patients with follow-up. 3-5 These reports provide important observations, but sound conclusions on the behavior of SpCC and other variants are lacking. The designation “spindle cell carcinoma” has not been applied consistently in the literature. At times, it has been used synonymously with squamous carcinoma with spindle cell metaplasia, pseudosarcoma, sarcomatoid carcinoma, and carcinosarcoma. The criteria for a diagnosis of SpCC are two-fold. There must be a dominant spindle cell proliferation composed of bland-appearing bipolar cells with plump
Analysis of Recurrences There was little difference in age between patients with recurrence (mean age, 64 years) and patients without (mean age, 65 years). Tumors that recurred were significantly larger than those that did not progress (P = .033). Although only 20% of the nonrecurrent neoplasms were circumscribed microscopically, when compared with the 7% in the recurrent group this difference was significant at a level of P = .053. Other histologic features such as grade, cellularity, mitotic activity, differentiation of the car738
MAMMARYSPINDLECELL CARCINOMA(Wargotzet al)
FIGURE 15. Portions of two spindle cells, one with actinfilaments (A) and bundles of prekerattnand kerattntonofilaments (K) identified as a myoepithelial cell (ME), and the other cell has tonofilaments only. An intercellular slit lumen (L] wtth microvilli Is also present. [Magnification x 4,300.)
nuclei, insignificant pleomorphism, and low mitotic activity, growing as feathered, overlapping or interweaving wavy fascicles, often in a storiform pattern. Collagen production is characteristic as are myxoid and angioid areas. Second, there must be in situ carcinoma or ductal, lobular, squamous, or mixed infiltrating carcinoma that is contiguous or subtly merges with the spindle cell component. If overt carcinoma is absent, the epithelial nature of the neoplasm must be proved through immunohistochemistry or electron microscopy. With these features in mind, some examples of squamous carcinoma with spindle metaplasia and carcinoma with pseudosarcomatous stroma,3,4*6*7 spcc,5.s-t0 acantholytic variant of squamous carcinoma,” low-grade adenosquamous metaplastic carci-
TABLE 1. Selected Pathologic Features of Spindle Cell Carcinomas: Recurrent v Nonrecurrent Neoplasms Recurrent Size* Microscopic circumscription Cellularity* Mitotic figures* Pleomorphism* Inflammatory infiltrate
* Mean
values.
5.0 cm 7% 2.2+ 3/10 HPF grade 1.3 82%
Non-Recurrent 3.7 cm 20% 2.0+ 2.5110 HPF grade 1.0 82%
(P = ,033) (P = ,053)
noma,‘* and infiltrating or malignant myoepitheliomar3 correspond to the spindle cell carcinomas included in this report. This study finds a relationship between neoplasm size and the likelihood of progression. The three patients with metastases to axillary lymph nodes at initial surgery (6%) had neoplasms 3.0 cm or greater in diameter. When present, complete microscopic circumscription was significant at a level of P = .053. Other histologic differences between neoplasms that recurred and those that did not were not significant. Immunohistologic studies for cytokeratin were useful in confirming the epithelial nature in 59 of 60 cases tested. EMA was not as useful, as only 2 1% were immunoreactive. The dual expression of keratin and vimentin by epithelial cells, and occasionally by mesenchymal cells, has been documented previously.14-lg The demonstration of S- 100 (5 1%) and actin (96%) in SpCC suggests that myoepithelial cells are a component of these neoplasms. Stromal cells no doubt account for some of the actin reactivity; however, our ultrastructural findings also identified cells containing bundles of tonoftlaments and actin filaments (myoepithelial cells), as well as myofibroblasts and spindle-shaped cells containing tonofilaments. This indicates that SpCC is at least in part derived from or
HUMAN PATHOLOGYVolume20, No.8 [August1989) REFERENCES
differentiating toward, myoepithelium. Disordered proliferation and differentiation by myoepithelial or progenitor cells could result in aberrant intraductal growth, neoplastic spindle cell proliferation, and an overt infiltrating carcinoma. Such disordered growth could explain many of the peculiarities of SpCC such as transitional or differentiation zones, cuffing, and adenomyoepithelioma-like areas consisting of pericellular and periductal accumulations of basement membrane material (Fig 10). In addition to our immunohistochemical findings, myoepithelial participation is supported by actin and/or S-100 immunoreactivity reported for sarcomatoid mammary carcinoma.20*21 Also, there is a report of infiltrating myoepithelioma that by light microscopy is a typical SpCC, and ultrastructurally had myoepithelial and squamous features.r3 SpCC might be difficult to distinguish from malignant fibrous histiocytoma (MFH) because it may have areas of storiform pattern. SpCC differs from MFH in that it usually has associated intraductal or infiltrating carcinoma or a squamous component and usually has angioid stroma. It is immunoreactive for cytokeratin, and lacks tumor giant cells present in most examples of MFH. Also, MFH is extremely rare in the breast and should have a pattern of immunoreactivity similar to MFH occurring elsewhere. Transmission electron microscopy might also prove helpful in distinguishing the two in rare instances where immunohistochemical results are equivocal. Carcinosarcoma of the breast has been used as a generic term for biphasic neoplasms composed of both carcinoma and a sarcoma. Although the stroma of carcinosarcoma may show some immunoreactivity for keratin, the high degrees of cellularity, nuclear pleomorphism, and mitotic activity in the spindle cell component distinguish it from SpCC, which has relatively bland stroma and seldom has marked pleomorphism. When an overt carcinoma is absent, SpCC might easily be confused with fasciitis, fibromatosis, or a low-grade stromal sarcoma of the breast. These are very rare lesions and should only be diagnosed after thorough sampling of the lesion fails to reveal any carcinoma or the mixture of patterns present in SpCC. Any further suspicion can be laid to rest by immunohistochemical study for keratin. SpCC of the breast is a distinct form of metaplastic carcinoma with a cumulative 5-year survival rate of 64% that contrasts with lower survival rates for metaplastic carcinoma overall. Careful definition of other variants of metaplastic carcinoma will be necessary in order to assess the importance of this difference.
1. Berkson J, Gage RP: Calculation of survival rates for cancer. Proc Mavo Clin 25:270. 1950 2. Wargbtz ES, Norris HJ: Metaplastic carcinomas of the breast. I. Matrix-producing carcinoma. HUM PATHOL 20:628-635, 1989 3. Huvos AC, Lucas JC, Foote FW: Metaplastic breast carcinoma. NY State J Med 73:1078, 1973 4. Kaufman MW, Marti JR, Gallager HS, et al: Carcinoma of the breast with nseudosarcomatous metaplasia. Cancer 53: 1908, 1984 5. Oberman HA: Metaplastic carcinoma of the breast. A cliniconatholoeic studv of 29 uatients. Am 1 Sure Path01 11:918, 1987 ’ 6. Jo&s EL: Primar; squamous-&l ca;cinoma of the breast with pseudosarcomatous stroma. J Path01 97:383, 1969 7. Willis RA: Squamous-cell mammary carcinoma of predominantly fibrosarcoma-like structure. J Path01 Bacterial 76:5 11, 1958 8. Gersell DJ, Katzenstein AL: Spindle cell carcinoma of the breast: A clinicopathologic and ultrastructural study. HUM PATHOL 12:550, 1981 9. Bauer TW, Restock RA, Eggleston JC, et al: Spindle cell carcinoma of the breast: Four cases and review of the literature. HUM PATHOL 15: 147, 1984 10. Ellis IO, Bell J, Ronan JE, et al: Immunocytochemical investigation of intermediate filament proteins and epithelial membrane antigen in spindle cell tumours of the breast. J Path01 154:157, 1988 11. Eusebi V, Lamovec J, Cattani MC, et al: Acantholytic variant of squamous-cell carcinoma of the breast. Am J Surg Path01 10:855, 1986 12. Rosen PP, Ernsberger D: Low-grade adenosquamous carcinoma. A variant of metaplastic mammary carcinoma. Am J Surg Path01 11:351, 1987 13. Erlandson RA, Rosen PP: Infiltrating myoepithelioma of the breast. Am J Surg Path01 6:785, 1982 14. Ellis CL, Langloss JM, Heffner DK, et al: Spindle-cell carcinoma of the aerodigestive tract. An immunohistochemical analysis of 21 cases. Am J Surg Path01 11:335, 1987 15. Zarbo RJ, Crissman JD, Venkat H, et al: Spindle-cell carcinoma of the upper aerodigestive tract mucosa. An immunohistologic and ultrastructural study of 18 biphasic tumors and comparison with seven monophasic spindle-cell tumors. Am J Surg Path01 10:741, 1986 16. Azumi N, Battifora H: The distribution of vimentin and keratin in epithelial and nonepithelial neoplasms. A comprehensive immunohistochemical study on formalinand alcohol-fixed tumors. Am J Clin Path01 88:286, 1987 17. Van Muijen GNP, Ruiter DJ, Warnaar SO: Coexpression of intermediate filament polypeptides in human fetal and adult tissues. Lab Invest 57:359, 1987 18. Brown DC, Theaker JM, Banks PM, et al: Cytokeratin expression in smooth muscle and smooth muscle tumours. Histopathology 11:477, 1987 19. Wick MR, Brown BA, Young RH, et al: Spindle-cell proliferations of the urinary tract. An immunohistochemcial study. Am J Surg Path01 12:379, 1988 20. Cattani MC, Lamovec J, Ceccarelli C, et al: Sarcomatoid carcinomas of the breast: An immunohistochemical study. Path Res Pratt 182:475, 1987 (abstr) 21. Meis JM, Ordonez NC, Gallager S: Sarcomatoid carcinoma of the breast: An immunohistochemical study of six cases. Virchows Arch (A) 40:415, 1987 L
740
Spindle cell carcinoma (SpCC) of the breast is a rare neoplasm in which the spindle cell component predominates, resembling a low-grade sarcoma or a reactive process such as fasciitis or granulation tissue. A variety of neoplasms have been included in the designation “metaplastic carcinoma” and the behavior of specific variants is unclear. We define the histopathologic features of 100 spindle cell carcinomas of the breast, correlate the features with the clinical course, and discuss the origin as suggested from the results of immunohistochemistry and ultrastructural observations.
From the Department of Breast and Gynecologic Pathology, Armed Forces Institute of Pathology, Washington, DC. Accepted for publication December 30, 1988. Key words: metaplastic carcinoma, spindle cell carcinoma, breast carcinoma with pseudosarcomatous metaplasiaktroma, carcinosarcoma, squamous carcinoma, myoepithelium, immunohistochemistry, electron microscopy. Address correspondence and reprint requests to Henry J. Norris, MD, Department of Breast and Gynecologic Pathology, Armed Forces Institute of Pathology, Washington, DC 203066000. 0 1989 by W.B. Saunders Company. 0046-8177/89/2008-0003$5.00/O
732
MATERIAL AND METHODS All lesions of the breast coded as metaplastic carcinoma, spindle cell carcinoma, carcinoma with spindle cell metaplasia, pseudosarcoma, carcinosarcoma, squamous carcinoma, low-grade mesenchymal tumor, fasciitis, and sarcoma in the files of the Armed Forces Institute of Pathology (AFIP) accessioned before January 1983 were retrieved and reviewed. Of these, 100 examples of SpCC were identified. Neoplasms were classified as SpCC if intraductal or infiltrating carcinoma was contiguous or subtly merged with a bland, monotonous-appearing spindle cell proliferation composed of bipolar cells with plump nuclei growing as wavy overlapping or interweaving fascicles often forming a storiform pattern with admixed dense collagen, and focal myxoid and angioid stroma. The spindle cell component always comprised more than half of the neoplasm and was more than 80% of it in the majority of cases. Neoplasms lacking overt carcinoma, but having the characteristic spindle cell stroma, were included if they were immunoreactive for keratin. An average of six blocks of tumor were studied per case. Neoplasms with multinucleate osteoclastic giant cells were excluded and will be dealt with in another report. The pleomorphism of the spindle cell component was assessed on a scale of 1 to 3 with grade 1 assigned to relatively uniform, monotonous-appearing cells with uniform nuclei, a fine granular chromatin pattern, and inconspicuous nucleoli. Grade 2 neoplasms had a moderate degree of cellular pleomorphism and nuclei with a single nucleolus, but no multinucleate tumor giant cells. Grade 3 was assigned to neoplasms having areas of marked cellular pleomorphism and prominent nucleoli, often with multinucleate tumor giant cells. The architectural differentiation of the overt carcinoma was based on the degree of gland formation and scored as 1 (well differentiated), 2 (moderately differentiated), and 3 (poorly differentiated). Squamous differentiation, when present, was not graded. Cellularity was inversely proportional to the amount of collagen and was graded 1 + (25% or fewer cells) to 4+ (75% or more cells) following microscopic examination of the tissue sections through a 25 x objective. Neoplasms were designated as fibrocollagenous, fibromyxoid, or angioid depending on the predominant histopathologic features. Areas of maximum cellularity were identified and number of mitotic figures per 10 high power fields (HPFs) (area = 0.159 mm2) was recorded. In 1 + cellular lesions, 20 HPFs were examined with the result divided by two and expressed per 10 HPFs. Chronic inflammatory infiltrates, whether admixed with the neoplasm or predominantly peripheral, were assessed on a scale of 1 + (sparse) to 3 + (marked). Neoplasms with lymphoid follicles either admixed or at the periphery were considered as 3 + . Clinical data were tabulated from the patients’ charts and analyzed in conjunction with the following gross and microscopic characteristics: size, contour (nodular, irregular), microscopic borders (circumscribed, or infiltrative), cellularity, cellular pleomorphism, mitotic activity, inflammation and presence or absence of intraductal or overt infiltrating carcinoma and the degree of differentiation. All
MAMMARY SPINDLE CELL CARCINOMA (Wargotz et al]
Gross and Microscopic Pathology
patients were eligible for 5-year follow-up, and follow-up information was obtained. Life tables were designed according to the method of Berkson and Gage.’ Analysis of variance, (anova) with size as the response variable, and hypothesis testing of differences in proportions were used to test the significance of histologic differences between neoplasms that recurred and those that did not. Analysis of differences in cumulative 5-year survival rates was based on a standard normal test of proportions. All P values were derived from two-tailed tests.
All SDCC were within the mammarv Darenchyma. They ranged in size from 0.6 cm to ‘21 cm, with a mean diameter of 4.4 cm, and a median and mode of 4.0 cm. Forty-five were nodular and apneared well-circumscribed. Twentv had an irregular beriphery, and in 35 the contour bas not descrybed. The neoplasms were firm to hard and gritty, with a dull, white to grey or pink to tan cut surface. They were frequently mottled, and with one or more cysts. Several neonlasms had vellow or chalkv streaks. but central necrosis and hemorrhage were not evident in any. Microscopically, 49 SpCC had infiltrating borders. while 18 were Dredominantlv circumscribed or nodular, and the peiiphery of 3 l’had combinations of nodular and finger-like infiltrating growth patterns (Fig 1). The borders of two neoplasms were not reoresented in the available tissue -sections. None w&e encapsulated. Vascular invasion by the spindle cell component was found in three neoplasms, while neural invasion was not identified. Although infiltrating carcinoma exclusively was evident in 48 neodasms. it was often limited to onlv a few fields. Intriductal carcinoma exclusively ias found in seven cases. Both intraductal and infiltrating carcinoma were observed in 17 cases. Intraductal carcinema. when oresent. was most often identified at the periphery. Infiltrating carcinoma usually blended with the spindle cell component. None of the intraductal carcinomas had spindle cell metaplasia within the ducts. The maioritv of infiltrating carcinomas were moderate or hoorly differentiate;. Squamous differentiation of the carcinoma component was
lmmunohistochemistryand Transmission Electron Microscopy Immunohistochemical studies, performed on paraffinembedded tissue from 60 neoplasms, included cytokeratins (two immunoreagents), epithelial membrane antigen (EMA), S-100 protein, actin, factor VIII, and vimentin. Studies for cyto-keratins were performed on all 60 neoplasms. The polyclonal antiserum to keratin proteins was prepared at the AFIP as described previously.2 The immunohistochemical techniques used, and monospecificities of the commercial polyclonal antibodies to S-100 protein and factor VIII, as well as specificity of commercial monoclonal antibodies to keratins (AE1/3), EMA, actin, and vimentin, were evaluated as before.2 The pattern of cellular immunoreactivity of the tumor was semiquantitated and designated as being diffuse (>30%), intermediate (10 to 30%), focal ( a single focus or multiple discrete foci), or rare (occasional isolated cells). The relative intensity of individual cellular immunoreactivity was not examined. Wet tissue from three cases was available for transmission electron microscopy. Samples were processed and examined as previously described.2
RESULTS Clinical Findings The patients ranged in age from 29 to 95 years, with an average age of 63, median age of 65, and a mode of 72. All but one were women. Of the 82 with known racial extraction, only two were black, two 78 (92%) were were Hispanic, and the remaining white. Of the 99 women, only 11 were premenopausal. All patients presented with a single palpable, firm mass. The tumor was fixed to the skin in seven cases and to the chest wall in one. Skin or nipple ulceration was present in 11 cases, nipple retraction was found in six, and nipple discharge was noted in one. Axillary enlargement was absent in all cases. Fifty-two neoplasms occurred in the right breast, 44 in the left, and four were not designated. In the cases in which the quadrant was given, 46% occurred in the upper outer quadrant. The remainder presented with near equal frequency in the lower outer and the two inner quadrants, or as a subareolar mass. Neoplasms had been present 1 week to 5 years, but most were present for 2 weeks to 3 months. One woman claimed the lump had been present for 45 years. Only six reported pain or tenderness. None of the patients gave a history of trauma before the occurrence of the tumor. Serum carcinoembryonic antigen was not elevated in two patients tested.
FIGURE 1. Despite gross circumscription, the margins of this SpCC are infiliratbe. Both nodular and finger-like growths are evident. (Magnification x 5.1
733
HUMAN PATHOLOGY
Volume 20, No. 8 (August 1989)
present in 35 neoplasms. It was the exclusive form of overt carcinoma in 11 neoplasms. Squamous nests, some with keratin pearls, were often connected to one another by thin strands of epithelium. The cells at the periphery of these nests, and of some infiltrating duct carcinomas, appeared to stream off the main mass and imperceptibly merge with the surrounding spindle cells (Figs 2 through 4). Other ducts had a marked circumferential myoepithelial cell prominence resulting in a cuffed appearance (Fig 5). The spindle cell component was predominantly fibrocollagenous, with the spindle cells arranged in wavy, interlacing and overlapping fascicles often splaying or fanning out at low power, resembling the feathered pattern of fasciitis in some areas (Fig 6) and the storiform pattern of dermatofibrosarcoma protuberans and malignant fibrous histiocytoma (Figs 7 and 8) in others. Some neoplasms resembled cellular examples of fibromatosis or a low-grade fibrosarcoma with finger-like extensions into adjacent fat (Fig 9). Focally, epithelium alone or with spindle cells was interspersed among rounded trabecular bands of hyaline collagen, sometimes indistinguishable from osteoid, and aggregates of basement membrane material (Fig 10). Angioid and/or fibromyxoid-appearing areas were frequent minor components and the predominant component of a few neoplasms. Angioid areas were distinctive, consisting of complex arrangements of intersecting strands of flattened endotheliallike cells in a collagenous or fibromyxoid and edematous stroma mimicking a capillary network (Fig 11). Several neoplasms had areas of broad channels lined by plump endothelioid epithelial cells mimicking angiosarcoma (Fig 12). Nineteen percent of the neoplasms were of 1 + cellularity, 42% were 2+, 33% were 3+, and 6% were 4 + . Some 1 + cellular neoplasms were densely collagenous with only small foci of spindle cells (Fig 10) or a thin rim of spindle and inflammatory cells
FIGURE 2 An interweaving pattern is formed bY the squamous epithelium merging imperceptibly with the spindle cell stroma. (Magnification x 174.)
FIWRE 3. Differentiation zone of a squamous island. (Magnification x 174.)
along the periphery. Seventy percent were without appreciable pleomorphism of the spindle cell component (grade l), and 28% had only a mild to moderate degree of pleomorphism with some cells possessing a nucleolus (grade 2). None of the neoplasms had tumor giant cells, and only two had small foci of marked Spindle-shaped carcinoma cells pleomorphism. tended to have rounded plump nuclei, bipolar processes, and were twisted or contorted between collagen fibers (Fig 10). Many appeared to subtly merge with fibroblasts and myofibroblasts (Figs 2 and 3). In cellular areas the spindle cells were more plump (Figs 4 and 5), while in myxoid areas stellate shapes predominated (Fig 11). Mitotic figures of the spindle cell component ranged from 0 to 11 per 10 HPF with a mean of three and median of one. Abnormal forms were rarely seen. Eighty-four neoplasms had an inflammatory infiltrate predominantly composed of lymphocytes and plasma cells. The intensity of the infiltrate was 1 + in
FIGURE 4. Differentiation zones were also associated wtth duct carcinoma lacldng a squamous component. (Magnkaltcn x 209.)
734
MAMMARY SPINDLE CELL CARCINOMA (Wargotz et al)
FIGURE 5. Malignant myoepithelial proliferation around focus of intraductal carcinoma or atypical intraductal hyperplasia producing a “cuffed” appearance. (Magnification x 104.)
FIGURE 7. A focal storiform pattern is frequently present in SpCC of the breast. (Magnification x 87.)
The spindle cell component of 59 (98%) of the 60 neoplasms was immunoreactive for keratin (Fig 13). Nearly all neoplasms showed an intermediate or dif-
fuse pattern of immunoreactivity in the spindle cell component, while in only three neoplasms were rare cells positive. Immunoreactivity was not limited to the perinuclear zone and was also present in the tapered processes. Of the 47 neoplasms studied by both keratin immunoreagents, there was a concordance in only 77% due to the absence of AE1/3 immunoreactivity in ten polyclonal immunocytokeratin reactive neoplasms. In only one instance was AEU3 immunoreactivity observed in the absence of immunoreactivity to the polyclonal keratin antibody. The spindle cell component of 11 (23%) of 48 neoplasms evaluated for EMA was immunoreactive and the pattern of immunoreactivity was usually rare to focal. Vimentin was detected in the spindle cells of 53 (91%) of 58 neoplasms evaluated, usually in an intermediate or diffuse pattern. In five neoplasms, vimentin was also
FIQRRE 6. Cellular spindle cells areas arranged In irregular bundles with admixed inflammatory cells resembling the feathered pattern of fasciltts. (Magnification x 130.)
FIQRRE 8. This sparsely cellular 3pCC consists of dense collagen with narrow, meandering spindle cell fascicles. (Magnlflcattons: left, x 60; right, x 164.)
21 neoplasms, 2 + in 32, and 3 + in 3 1. In nine neoplasms, the inflammation was limited to the periphery. Eosinophils and mast cells were conspicuous in a few neoplasms. Granulomatous inflammation or acute inflammation was not observed. Necrosis, when present, was characteristic, and was represented by focal ischemic, densely acellular collagenous areas. Five neoplasms had focal osseous metaplasia, one had focal cartilaginous metaplasia, and two others had both focal cartilaginous and osseous metaplasia. Twenty neoplasms tested for mucin were negative.
lmmunohistochemistry
735
HUMAN PATHOLOGY
Volume 20, No. 8 (August 1989)
FIGRRE 11. (Left) Angioid areas consist of a complex arrangement of epithelial strands in a fibromyxoid and edematous, or collagenous stroma simulating a vascular network (Magnification x 38.) (Right) Higher magnification of another field shows focal gland formation. Often only the strands are present. (Magnification x 150.)
FIGRRE 9. Finger-like extension into adjacent mammary stroma is common in SpCC and resembles fibromatosis and fibrosarcoma. (Magnificatton x 35.)
detected in the overt carcinoma and/or foci of squamous differentation. S-100 protein was detected in the spindle cell component of 25 (51%) of 49 neoplasms evaluated and the pattern of immunoreactivity ranged from rare to focal in most, predominantly staining the more rounded cells. An intermediate to diffuse pattern of immunoreactivity for actin was noted in the spindle cells of 48 (96%) of 53 neoplasms evaluated. Of 39 neoplasms evaluated with the entire panel of six antibodies, keratins were detected in 97%, EMA in 21%, vimentin in 95%, S-100 protein in 51%, and actin in 87%. Some of these substances were coexpressed by the same subpopulation of spindle cells. Stromal myofibroblasts and fibroblasts accounted for at least some of the actin and vimentin immunoreactivity. None of the seven neoplasms evaluated for fac-
tor VIII showed component.
immunoreactivity
of the angioid
Ultrastructural Findings The three neoplasms studied by transmission electron microscopy were similar, composed of spindle cells and collagen. Three predominant types of spindle cells could be distinguished on the basis of filament content or composition: (1) spindled epithelial cells having bundles of tonofilaments (prekeratin or cytokeratin); (2) myoepithelial cells, identified by the presence of bundles of tonofilaments and actin filaments; and (3) myofibroblasts, identified by actin filaments, but lacking tonofilaments (Figs 14 and 15).
FIGRRE 10. Trabecular hyaline collagen (left) and areas of epithelium and basement membrane material (right). (Magnificatton x 156.)
FIGURE 12. Hemorrhagic area resembling angiosarcoma. (Magnification x 107.)
736
MAMMARY SPINDLE CELL CARCINOMA CJvargotzet al)
of 27 nodes. With the exception of one patient who chose incisional biopsy with radiation and chemotherapy, surgical margins appeared free of tumor. Receptor Stalus Most of the patients were seen before the availability of routine receptor analysis, but estrogen receptor analysis was performed on 18 neoplasms. In one, a marginally elevated level was found (14 fmol/ mg cytosol protein), while in 17 patients levels were not elevated. Progesterone receptor analysis was performed on 14 of the 18 neoplasms analyzed for estrogen receptors, and none were elevated. Recurrence and Survival Follow-up information was obtained on all 100 patients with a mean and median follow-up interval of 5.4 years. Local recurrence or metastasis occurred in 56 patients. In nine of these, the site and interval were unclear. Seventeen patients developed only local recurrence, 15 had local recurrence and metastasis, and 15 had metastasis only. First recurrence occurred from 3 weeks to 11 years with a mean of 22 months, and a median of 1 year. Of the 56 patients with recurrence, 44 (79%) subsequently died from tumor. Metastasis was an ominous sign as 29 of the 30 patients subsequently died from tumor, whereas following only local recurrence, 7 1% (12 of 17 patients) survived. The difference in mortality rate between these two groups was significant (P = .OlO). The three patients with positive axillary nodes at initial therapy died from widespread metastatic tumor. Four of the eight patients with neoplasms containing cartilage or bone subsequently died from tumor. The most common pattern of progression was local recurrence, metastasis to the pleura and lungs, and then metastasis to bone and other sites. Recurrence rates following each of the five surgical approaches were radical mastectomy 47%, modified radical mastectomy 44%, simple mastectomy 56%, partial mastectomy 38%, and biopsy 78%. Local recurrence was generally treated by surgery, and was most often a form of mastectomy for patients initially treated by biopsy. Of the eight patients who received adjuvant radiation and/or chemotherapy following initial surgery, only two have not died from tumor. One patient was treated by modified radical mastectomy followed by three cycles of doxorubicin hydrochloride and has been free of disease for 7 years. The second patient was initially treated by biopsy and radiation therapy (5,049 rads to the breast and supraclavicular area, and 780 rads to the axilla), developed a recurrence at 26 months and was treated by radical mastectomy. She is free of disease at 5 years after mastectomy. Radiation and/or chemotherapy for recurrence offered no significant advantage over surgery alone, as all 14 recipients died from tumor. Seven (4 1%) of the 17 neoplasms that lacked intraductal carcinoma or overt infiltrating carcinoma had been diagnosed originally as fasciitis, fibromatosis, or low-grade mesenchymal tumor and had re-
FIGURE 13. lmmunoreactivity for cytokeratin is nearly always present in SpCC and identifies them when an overt form of carclnoma is not apparent. [Left, hematoxylin and eosin. magnification x 205; right, polyclonal immunocytokeratin PAP, magnification x 164.)
No one cell type predominated and all were intermingled with one another. Most cells were rich in organelles. Well-formed desmosomes were infrequent, but poorly developed junctions were common. Microvilli projecting into slit-like intercellular lumina were identified in one case (Figs 14 and 15). Nuclear contour was variable, and nearly all cells had a prominent nucleolus. The curvilinear bundles of tonoftlaments were of the prekeratin or keratin type (10 nm), and were distributed in the perinuclear zone and within the tapered cytoplasmic processes (Figs 14 and 15). Actin filaments (6 nm) were arranged parallel to the long axis of the cell (Fig 15). Some cells had numerous dense bodies. Initial Therapy and Node Status Mastectomy was the form of initial therapy in 66 patients. Radical mastectomy was performed on 19, modified radical mastectomy on 29, and simple mastectomy on 18. Excisional biopsy was the initial therapy for 25 patients, and partial mastectomy (lumpectomy, tylectomy, or wide local excision) was performed on eight. Postoperative radiation was received by three patients following simple mastectomy and one patient following biopsy. Two patients received chemotherapy following modified radical mastectomy. Two other patients had both postoperative radiation and chemotherapy following partial mastectomy and incisional biopsy. Of the 48 patients with axillary dissection performed at initial presentation, the lymph node status is known for 47. Only three (6%) of the 47 were found to have metastatic carcinoma. In one woman, the spindle cell component was present exclusively in three of nine nodes, while in the two others, overt carcinoma was present exclusively in one of 15 and 16 737
HUMAN PATHOLOGY
Volume 20, No. 8 (August 1989)
FIGURE 14. Portions of three spindle cells, one with bundles of prekeratin and keratin tonofilaments [IQ and all three contributing to the formation of intercellular slit lumina with microvilli (L). (Magnification x 3,784.)
cinema or presence of squamous epithelium, and inflammation did not differ significantly between the two groups (Table 1). Despite the increased frequency of local recurrence in patients who were treated by biopsy, the delay in more extensive surgery did not result in increased death from tumor as 44% died from tumor, versus 48% of patients treated by a form of mastectomy. Only two of eight patients (25%) died from tumor following partial mastectomy.
ceived only excisional biopsy. Five (71%) of these recurred locally and more extensive surgical therapy was performed, but two of the patients subsequently died from tumor. Four of the other ten patients in this group who were treated by a form of mastectomy also eventually died from tumor. Twelve patients died of causes unrelated to their breast cancer. The cumulative 5-year survival rate for 100 patients with SpCC, adjusting for patients who died from other causes, was 64%. Histologic material and/or pathology reports of local recurrences and metadtases were available in 14 cases. Of these, seven were the spindle component only, three were overt carcinoma, and four showed an admixture of both. In four of these 14 cases, autopsies revealed widespread metastases and involved hilar and/or periaortic lymph nodes.
DISCUSSION SpCC is a distinct form of metaplastic carcinoma with a cumulative 5-year survival rate of 64%. This contrasts with 5-year survival rates of 44% to 55% for compilations of various forms of metaplastic carcinoma, none of which exceeds 33 patients with follow-up. 3-5 These reports provide important observations, but sound conclusions on the behavior of SpCC and other variants are lacking. The designation “spindle cell carcinoma” has not been applied consistently in the literature. At times, it has been used synonymously with squamous carcinoma with spindle cell metaplasia, pseudosarcoma, sarcomatoid carcinoma, and carcinosarcoma. The criteria for a diagnosis of SpCC are two-fold. There must be a dominant spindle cell proliferation composed of bland-appearing bipolar cells with plump
Analysis of Recurrences There was little difference in age between patients with recurrence (mean age, 64 years) and patients without (mean age, 65 years). Tumors that recurred were significantly larger than those that did not progress (P = .033). Although only 20% of the nonrecurrent neoplasms were circumscribed microscopically, when compared with the 7% in the recurrent group this difference was significant at a level of P = .053. Other histologic features such as grade, cellularity, mitotic activity, differentiation of the car738
MAMMARYSPINDLECELL CARCINOMA(Wargotzet al)
FIGURE 15. Portions of two spindle cells, one with actinfilaments (A) and bundles of prekerattnand kerattntonofilaments (K) identified as a myoepithelial cell (ME), and the other cell has tonofilaments only. An intercellular slit lumen (L] wtth microvilli Is also present. [Magnification x 4,300.)
nuclei, insignificant pleomorphism, and low mitotic activity, growing as feathered, overlapping or interweaving wavy fascicles, often in a storiform pattern. Collagen production is characteristic as are myxoid and angioid areas. Second, there must be in situ carcinoma or ductal, lobular, squamous, or mixed infiltrating carcinoma that is contiguous or subtly merges with the spindle cell component. If overt carcinoma is absent, the epithelial nature of the neoplasm must be proved through immunohistochemistry or electron microscopy. With these features in mind, some examples of squamous carcinoma with spindle metaplasia and carcinoma with pseudosarcomatous stroma,3,4*6*7 spcc,5.s-t0 acantholytic variant of squamous carcinoma,” low-grade adenosquamous metaplastic carci-
TABLE 1. Selected Pathologic Features of Spindle Cell Carcinomas: Recurrent v Nonrecurrent Neoplasms Recurrent Size* Microscopic circumscription Cellularity* Mitotic figures* Pleomorphism* Inflammatory infiltrate
* Mean
values.
5.0 cm 7% 2.2+ 3/10 HPF grade 1.3 82%
Non-Recurrent 3.7 cm 20% 2.0+ 2.5110 HPF grade 1.0 82%
(P = ,033) (P = ,053)
noma,‘* and infiltrating or malignant myoepitheliomar3 correspond to the spindle cell carcinomas included in this report. This study finds a relationship between neoplasm size and the likelihood of progression. The three patients with metastases to axillary lymph nodes at initial surgery (6%) had neoplasms 3.0 cm or greater in diameter. When present, complete microscopic circumscription was significant at a level of P = .053. Other histologic differences between neoplasms that recurred and those that did not were not significant. Immunohistologic studies for cytokeratin were useful in confirming the epithelial nature in 59 of 60 cases tested. EMA was not as useful, as only 2 1% were immunoreactive. The dual expression of keratin and vimentin by epithelial cells, and occasionally by mesenchymal cells, has been documented previously.14-lg The demonstration of S- 100 (5 1%) and actin (96%) in SpCC suggests that myoepithelial cells are a component of these neoplasms. Stromal cells no doubt account for some of the actin reactivity; however, our ultrastructural findings also identified cells containing bundles of tonoftlaments and actin filaments (myoepithelial cells), as well as myofibroblasts and spindle-shaped cells containing tonofilaments. This indicates that SpCC is at least in part derived from or
HUMAN PATHOLOGYVolume20, No.8 [August1989) REFERENCES
differentiating toward, myoepithelium. Disordered proliferation and differentiation by myoepithelial or progenitor cells could result in aberrant intraductal growth, neoplastic spindle cell proliferation, and an overt infiltrating carcinoma. Such disordered growth could explain many of the peculiarities of SpCC such as transitional or differentiation zones, cuffing, and adenomyoepithelioma-like areas consisting of pericellular and periductal accumulations of basement membrane material (Fig 10). In addition to our immunohistochemical findings, myoepithelial participation is supported by actin and/or S-100 immunoreactivity reported for sarcomatoid mammary carcinoma.20*21 Also, there is a report of infiltrating myoepithelioma that by light microscopy is a typical SpCC, and ultrastructurally had myoepithelial and squamous features.r3 SpCC might be difficult to distinguish from malignant fibrous histiocytoma (MFH) because it may have areas of storiform pattern. SpCC differs from MFH in that it usually has associated intraductal or infiltrating carcinoma or a squamous component and usually has angioid stroma. It is immunoreactive for cytokeratin, and lacks tumor giant cells present in most examples of MFH. Also, MFH is extremely rare in the breast and should have a pattern of immunoreactivity similar to MFH occurring elsewhere. Transmission electron microscopy might also prove helpful in distinguishing the two in rare instances where immunohistochemical results are equivocal. Carcinosarcoma of the breast has been used as a generic term for biphasic neoplasms composed of both carcinoma and a sarcoma. Although the stroma of carcinosarcoma may show some immunoreactivity for keratin, the high degrees of cellularity, nuclear pleomorphism, and mitotic activity in the spindle cell component distinguish it from SpCC, which has relatively bland stroma and seldom has marked pleomorphism. When an overt carcinoma is absent, SpCC might easily be confused with fasciitis, fibromatosis, or a low-grade stromal sarcoma of the breast. These are very rare lesions and should only be diagnosed after thorough sampling of the lesion fails to reveal any carcinoma or the mixture of patterns present in SpCC. Any further suspicion can be laid to rest by immunohistochemical study for keratin. SpCC of the breast is a distinct form of metaplastic carcinoma with a cumulative 5-year survival rate of 64% that contrasts with lower survival rates for metaplastic carcinoma overall. Careful definition of other variants of metaplastic carcinoma will be necessary in order to assess the importance of this difference.
1. Berkson J, Gage RP: Calculation of survival rates for cancer. Proc Mavo Clin 25:270. 1950 2. Wargbtz ES, Norris HJ: Metaplastic carcinomas of the breast. I. Matrix-producing carcinoma. HUM PATHOL 20:628-635, 1989 3. Huvos AC, Lucas JC, Foote FW: Metaplastic breast carcinoma. NY State J Med 73:1078, 1973 4. Kaufman MW, Marti JR, Gallager HS, et al: Carcinoma of the breast with nseudosarcomatous metaplasia. Cancer 53: 1908, 1984 5. Oberman HA: Metaplastic carcinoma of the breast. A cliniconatholoeic studv of 29 uatients. Am 1 Sure Path01 11:918, 1987 ’ 6. Jo&s EL: Primar; squamous-&l ca;cinoma of the breast with pseudosarcomatous stroma. J Path01 97:383, 1969 7. Willis RA: Squamous-cell mammary carcinoma of predominantly fibrosarcoma-like structure. J Path01 Bacterial 76:5 11, 1958 8. Gersell DJ, Katzenstein AL: Spindle cell carcinoma of the breast: A clinicopathologic and ultrastructural study. HUM PATHOL 12:550, 1981 9. Bauer TW, Restock RA, Eggleston JC, et al: Spindle cell carcinoma of the breast: Four cases and review of the literature. HUM PATHOL 15: 147, 1984 10. Ellis IO, Bell J, Ronan JE, et al: Immunocytochemical investigation of intermediate filament proteins and epithelial membrane antigen in spindle cell tumours of the breast. J Path01 154:157, 1988 11. Eusebi V, Lamovec J, Cattani MC, et al: Acantholytic variant of squamous-cell carcinoma of the breast. Am J Surg Path01 10:855, 1986 12. Rosen PP, Ernsberger D: Low-grade adenosquamous carcinoma. A variant of metaplastic mammary carcinoma. Am J Surg Path01 11:351, 1987 13. Erlandson RA, Rosen PP: Infiltrating myoepithelioma of the breast. Am J Surg Path01 6:785, 1982 14. Ellis CL, Langloss JM, Heffner DK, et al: Spindle-cell carcinoma of the aerodigestive tract. An immunohistochemical analysis of 21 cases. Am J Surg Path01 11:335, 1987 15. Zarbo RJ, Crissman JD, Venkat H, et al: Spindle-cell carcinoma of the upper aerodigestive tract mucosa. An immunohistologic and ultrastructural study of 18 biphasic tumors and comparison with seven monophasic spindle-cell tumors. Am J Surg Path01 10:741, 1986 16. Azumi N, Battifora H: The distribution of vimentin and keratin in epithelial and nonepithelial neoplasms. A comprehensive immunohistochemical study on formalinand alcohol-fixed tumors. Am J Clin Path01 88:286, 1987 17. Van Muijen GNP, Ruiter DJ, Warnaar SO: Coexpression of intermediate filament polypeptides in human fetal and adult tissues. Lab Invest 57:359, 1987 18. Brown DC, Theaker JM, Banks PM, et al: Cytokeratin expression in smooth muscle and smooth muscle tumours. Histopathology 11:477, 1987 19. Wick MR, Brown BA, Young RH, et al: Spindle-cell proliferations of the urinary tract. An immunohistochemcial study. Am J Surg Path01 12:379, 1988 20. Cattani MC, Lamovec J, Ceccarelli C, et al: Sarcomatoid carcinomas of the breast: An immunohistochemical study. Path Res Pratt 182:475, 1987 (abstr) 21. Meis JM, Ordonez NC, Gallager S: Sarcomatoid carcinoma of the breast: An immunohistochemical study of six cases. Virchows Arch (A) 40:415, 1987 L
740