eISSN: 2084-9869
ISSN: 1233-9687
Polish Journal of Pathology
Current issue Archive Manuscripts accepted About the journal Supplements Editorial board Abstracting and indexing Subscription Contact Instructions for authors Ethical standards and procedures
SCImago Journal & Country Rank
vol. 70
Original paper

Clinicopathologic characteristics of mixed epithelial/mesenchymal metaplastic breast carcinoma (carcinosarcoma): a meta-analysis of Chinese patients

Dequan Xu
Limin Hou

Pol J Pathol 2019; 70 (3): 174-182
Online publish date: 2019/12/07
Article file
- 03-PJP-01467.pdf  [0.83 MB]
Get citation
JabRef, Mendeley
Papers, Reference Manager, RefWorks, Zotero


Breast carcinosarcoma, also known as metaplastic carcinoma of the breast with mesenchymal differentiation (MCMD), is an uncommon mixed tumour composed of both malignant epithelial and mesenchymal cells, with an incidence of less than 0.2% [1]. Although the exact histogenesis has not been clearly defined, MCMD has been suggested to derive from malignant transformation of myoepithelial cells or myofibroblastic metaplasia of malignant epithelial cells. Recently, MCMD has been classified by the World Health Organisation (WHO) as a subtype of metaplastic breast carcinoma (MBC) [2, 3, 4, 5, 6]. To date, because of its rarity, only relatively small series have attempted to delineate several pathological and clinical aspects that make breast carcinosarcoma different from common malignant breast cancer. For the purposes of the current study, we analysed 215 breast carcinosarcoma patients identified through Chinese databases to describe clinical and pathologic characteristics.

Material and methods

Patients’ characteristics

Studies were identified from the following electronic databases: China Biology and Medicine data disc (CBMDISK), China National Knowledge Infrastructure (CNKI), MEDLINE, and Wanfang medicine database. We conducted a comprehensive review on reported cases of breast carcinosarcoma from January 1990 to March 2015 in China. The titles and abstracts of the search findings were screened for potentially eligible studies. In order to meet the analysis requirements and reduce deviation, included studies fulfilled the following criteria: I the study was based on population samples rather than volunteers; I case collection was based on a field survey; II there were accurate study dates and validated diagnostic criteria; and III if there were many articles based on the same sample, only the one that reported the most detailed data was included. The following combination of key words was used to search the databases: breast carcinosarcoma, breast cancer, carcinosarcoma, or metaplastic carcinoma. No attempt was made to retrieve unpublished studies. The search yielded 34 articles, including 215 cases. All patients in this study underwent surgical treatment with or without adjuvant oncologic treatment. The patients’ medical records were retrospectively reviewed to obtain demographic, clinicopathologic, treatment, and prognostic information as well as the immunohistochemistry of biologic factors such as ER, PR, HER2, CK, EMA, S100, and vimentin.

Statistical analysis

Descriptive statistics were used to characterise the patient population. Variables were presented as absolute numbers and percentage or median values and range. The chi-square test was used to analyse the significance of difference in the proportion of variables, and Kaplan-Meier analysis was used to assess the five-year survival. Statistical analysis was performed by SPSS 19.0 software, and the meta-analysis was conducted using Stata 12.0 statistical software. Heterogeneity was assessed with I-squared statistics. An I-squared value of more than 50% was considered to indicate high statistical heterogeneity.


Figure 1 summarises the process of identifying eligible breast carcinosarcoma studies. There were 34 studies left after the quality assessment (Table I). Table II shows the characteristics of the studies, which covered the location of lesions, tumour size, and axillary lymph node metastasis. The clinical and pathological features were calculated separately for studies, and the incidence can be found.

Patient and tumour demographics for the cohort

The mean age of the cohort of 215 patients was 53 years (range: 25-82 years). Although male breast cancer was quite rare, three male breast carcinosarcoma cases were diagnosed in the current cohort. There were no significant differences in left or right breast or in the quadrant of the breast. The tumour size ranged from 2.5 cm to 18 cm. The incidence of pathologically confirmed lymph node metastases was 30.81% (Table II). Very few cases were at pT1 stage (15%); most of them were either pT2 or pT3.

Pathological elements of breast carcinosarcoma

Mixed epithelial/mesenchymal metaplastic carcinomas are often regarded as “matrix producing carcinomas” and show carcinoma mixed with heterologous mesenchymal elements ranging from areas of bland chondroid and osseous differentiation to frank sarcoma. When the mesenchymal component is malignant, the designation of carcinosarcoma is used. Among the 215 breast carcinosarcoma patients, 76 with available pathological and IHC data were included in our studies (Table III). The epithelial component in a tumour may be composed of invasive ductal carcinoma (84.21%), squamous cell carcinoma (7.89%), lipid-rich carcinoma (6.58%), or adenocarcinoma (1.31%). Mesenchymal components may contain different elements ranging from fibrosarcoma (63.16%) to chondrosarcoma (19.73%), osteosarcoma (9.21%), liposarcoma (3.95%), or leiomyosarcoma (3.95%).
Of these 76 patients, 11 (14.4%) were HER2+, and 28 (36.8%) were triple-negative breast cancer (TNBC) (ER– and PR– and HER2–). The result of immunohistochemistry showed that the positive rates of CK (cytokeratin), EMA (epithelial membrane antigen), ER, and PR in breast epithelial components were significantly higher than those in mesenchymal components, while the positive rates of vimentin and S100 in mesenchymal components were higher than those in epithelial components. Detailed immunohistochemistry characteristics of the studied cases are presented in Table IV.

Five-year survival of breast carcinosarcoma

Thirty-four articles were identified by the search terms, but data on five-year survival could be obtained only from two studies including 119 patients. For the rest, the following data were independently extracted from each identified study: first author, study period, sample size, pathology reports, and duration of follow-up.
Overall, 30 patients with duration of follow-up were identified and assigned to group A. For those patients, survival curves were performed by Kaplan-Meier method. Survivals were expressed as median five-year percentages. Figure 2 shows the overall percentage survival in the 30 patients who belonged to group A.
The overall five-year survival rate was 56.7% with a median survival time of 71 months. Meta-analysis including three studies with comparative data revealed the five-year survival of the breast carcinosarcoma to be 62.6% (D + L pooled ES = 0.626; 95% CI: 0.549-0.703; p = 0.606; I2=0.0%). Figure 3 shows meta-analysis forest plots of five-year survival of breast carcinosarcoma in 149 patients from three studies.
Owing to the lack of controlled, clinical trials, the present analysis exclusively included retrospective, uncontrolled studies. Some important details like immunohistochemistry and adjuvant treatment after surgical operation were not adequately found in the databases. Yang et al. reported on retrospectively collected patients with breast carcinosarcoma, who were resected from January 1995 to January 2012 [7]. Several key points were made by Yang et al. as follows (Table V): in univariate analysis tumour size, axillary lymph node metastasis, treatment, and clinical stage were significant factors for local relapse; the five-year OS (overall survival) rate of patients with breast carcinosarcoma, who received adjuvant treatment after surgical operation was higher than those who only received surgical operation.


To our knowledge, this is to date one of the largest analyses systematically summarising the available evidence on the feature and survival of patients with breast carcinosarcoma. True carcinosarcoma of the breast is extremely rare. Since its first description by Pasternak and Wirth in 1936 as sarcomatous adenoacanthoma, these admixed breast tumours have been crowned with a plethora of names and have long been referred to as carcinosarcoma (CS) of the breast [8]. The strict definition of this tumour requires both a carcinomatous component and a malignant non-epithelial component of mesenchymal origin, without evidence of a transition zone between the two elements [9].
Classification of metaplastic carcinoma was proposed by the World Health Organisation in 2003 as: 1) squamous cell carcinoma; 2) adenocarcinoma with spindle cell proliferation; 3) adenosquamous, including mucoepidermoid; and 4) mixed epithelial and mesenchymal. It should be noted that upon extensive sampling, a large proportion of metaplastic breast cancers display a mixture of different elements. The wide variety of mixed epithelial/mesenchymal metaplastic carcinomas, some of which are also regarded as “matrix producing carcinoma”, show infiltrating carcinoma often mixed with heterologous mesenchymal elements ranging from areas of bland chondroid and osseous differentiation to frank sarcoma (chondrosarcoma, osteosarcoma, rhabdomyosarcoma, liposarcoma, fibrosarcoma) [10, 11]. When the mesenchymal component is malignant, the designation of carcinosarcoma is used (Figure 4) [12]. Undifferentiated spindle cell elements may form part of the tumour. In our study, the epithelial component in a tumour may be composed of invasive ductal carcinoma (84.21%), squamous cell carcinoma (7.89%), lipid-rich carcinoma (6.58%), or adenocarcinoma (1.31%). Mesenchymal components may contain different elements ranging from fibrosarcoma (63.16%) to chondrosarcoma of bone (19.73%), osteosarcoma (9.21%), liposarcoma (3.95%), or leiomyosarcoma (3.95%).
Given the multiple histological appearances exhibited by metaplastic breast cancers, the fourth edition of the WHO Classification of Tumours of the Breast stated that a plethora of terms have been coined to refer to subgroups of metaplastic breast cancers, including carcinosarcoma, sarcomatoid carcinoma, carcinoma with pseudosarcomatous metaplasia, etc. [6]. However, the definitions of the different types of mixed metaplastic carcinomas are confusing in the fourth edition of the WHO Classification of Tumours of the Breast. It was stated that a large proportion of metaplastic breast cancers display a mixture of different elements. Different types of metaplastic carcinomas have been described arising in association with complex sclerosing lesions and papillomas [13]. Microarray-based gene-expression profiling has demonstrated that metaplastic breast tumours are preferentially classified as of basal-like subtype [14]. However, specific genetic aberrations may be restricted to specific components within a cancer.
Mammography, sonography, and magnetic resonance mammography are frequently used breast imaging techniques in the diagnosis of breast neoplasms. However, all three techniques have unsatisfactory specificity in the diagnosis of breast carcinosarcoma [15]. Frozen section diagnosis of breast lumps is of high accuracy, with a sensitivity and specificity of more than 90 and 99%, respectively [16]. However, there may be some limitations to frozen section diagnosis because the small selected piece of tissue for frozen section diagnosis is unable to represent the overall profile of the tumour.
Lymph node invasion frequently occurs in breast carcinosarcoma. Tokudome et al. [17] reported a case of recurrence in a patient with breast carcinosarcoma. In their study, the recurrent tumour showed proliferation of carcinoma cell, resembling the epithelial component of the breast tumour without any sarcoma. Some studies concluded that the tissue types of recurrent breast carcinosarcoma might consist of carcinomatous or sarcomatous components or an admixture of both [18]. In our studies, the incidence of pathological confirmed lymph node metastases was 30.81%. However, the histological type of invaded lymph nodes is often complicated. It may consist of carcinomatous or sarcomatous components or an admixture of both.
The findings clearly show that five-year survival of breast carcinosarcoma is 62.6% (CI: 54.9%~0.703%). The mean age of the patients was 53 years (range, 25-82 years). Hennessy et al. [19] reported on 100 patients with biphasic metaplastic sarcomatoid carcinoma and 98 patients with carcinosarcoma identified through the SEER database. The authors identified five-year overall survival at stage I, II, III, and IV as 0.73, 0.59, 0.44, and 0.00, respectively. However, data from our study showed five-year overall survival at stage I, II, III, and IV as 1.00, 0.889, 0.661, and 0.111, respectively. Adjuvant treatment after surgical operation may improve the five-year OS of patients with breast carcinosarcoma.
Expertise and evidence-based information on optimal treatment is very limited due to the low incidence and inconsistent classification. A multidisciplinary approach is required for treatment. Mastectomy was often performed with or without axillary dissection [20, 21]. Anthracycline/taxane-based chemotherapy is recommended for chemotherapy [22]. Wargotz et al. [8] reported that in breast carcinosarcomas, 55% of the sarcomatous component is immunoreactive for keratin and 98% for vimentin. The aggressiveness of this tumour is attributed to its high grade and negativity for oestrogen and progesterone receptors. These tumours also do not overexpress HER-2/neu oncogene, ER, and PR [23]. Jia et al. reported that most cases of metaplastic breast cancer did not express ER and PR (89.4%) and showed negative expression of HER-2 (78.9%) [24]. In our study, unlike others, positive-expression rates of cytokeratin and vimentin in sarcomatous tissue were 19.74% and 82.89%, respectively. Eleven patients (14.4%) were HER-2+, and 19 patients (25%) expressed ER and PR.
The prognosis of breast carcinosarcoma remains controversial. Some studies reported that, compared to patients with invasive ductal carcinoma (IDC), those with breast carcinosarcoma have a worse prognosis.
Their overall survival and disease-free survival are both lower despite presenting more commonly as node-negative disease; however, almost all MCDC recurrences happened during the first five years, whereas recurrence curves for IDC continued to fall over time, suggesting the possibility that MCMD may show an earlier recurrence than IDC [25].
Because neoplastic cells often extend into the perivascular tissue and beyond the tumour capsule, local recurrence is common. Data from our study showed that the recurrence rate is around 50% (64/133) and tends to be locoregional in approximately 1/3 of the cases. Hematogenous spread is the common route of metastasis, and lung and pleura are the most common locations of distant metastasis. Accordingly, careful periodic follow-up after the initial treatment is strongly recommended to detect metastasis and recurrence early.
Despite the previous findings, this study has some limitations. The studies cannot capture every subtle factor, some of which may be critical for clinicians. In addition, other important details like immunohistochemistry and adjuvant treatment after surgical operation were not recorded enough in the databases. We did not perform a further analysis of factors that might affect survival, given that there are no demonstrated data. Moreover, out study cannot assess the gene modules associated with these basal like tumours. Owing to the lack of controlled, clinical trials, the present analysis exclusively included retrospective, uncontrolled studies with an inherent risk of bias.
Although the present study represents the most comprehensive analysis of the clinicopathological features and survival status of patients with breast carcinosarcoma, one should consider that the aggregate study population of patients was rather small and heterogenous with respect to the surgical procedures and adjuvant treatment after surgical operation. In conclusion, this is a very rare, malignant tumour, the diagnosis of which is made easier with detailed histological investigation to differentiate it from other, similar tumours.
There are not enough data and standard guidelines for its optimal treatment, and information about management is based on small retrospective studies rather than randomised trials. Further research is required in order to fully evaluate the potential of such therapy in patients with carcinosarcoma of the breast.
The authors declare no conflict of interest.


1. Esses KM, Hagmaier RM, Blanchard SA, et al. Carcinosarcoma of the breast: two case reports and review of the literature. Cases J 2009; 2: 1-5.
2. Pechoux C, Gudjonsson T, Ronnov-Jessen L, et al. Human mammary luminal epithelial cells contain progenitors to myoepithelial cells. Dev Biol 1999; 206: 88-99.
3. Salemis NS. Metaplastic carcinoma of the breast with mesenchymal differentiation (carcinosarcoma). A unique presentation of an aggressive malignancy and literature review. J Breast Dis 2018; 37: 169-175.
4. Wick MR, Swanson PE. Carcinosarcomas: current perspectives and an historical review of nosological concepts. Semin Diagn Pathol 1993; 10: 118-127.
5. Zhang Y, Cao MZ. Clinical characteristic of breast carcinosarcoma in pathology and biology. Chin Arch Gen Surg 2009, 3: 500-502.
6. Gobbi H, Simpson JF, Jensen RA, et al. Metaplastic spindle cell breast tumors arising within papillomas, complex sclerosing lesions, and nipple adenomas. Mod Pathol 2003; 16: 893-901.
7. Yang YX, Zhang LP, Guo GC. Analysis of clinicopathological features and prognosis of 94 patients with breast carcinosarcoma. Tumor 2013; 33: 63-69.
8. Pasternak JG, Wirth JE. Adeno-acanthoma Sarcomatodes of the mammary gland. Am J Pathol 1936; 12: 423.
9. Beatty JD, Atwood M, Tickman R, Reiner M. Metaplastic breast cancer: clinical significance. Am J Surg 2006; 191: 657-664.
10. Pope TL, Fechner RE, Brenbridge AN. Carcinosarcoma of the breast: radiologic, ultrasonographic, and pathologic correlation. Can Assoc Radiol J 1987; 38: 50-51.
11. Tavassoli FA, Devilee P. Pathology and Genetics of Tumours of the Breast and Female Genital Organs. World Health Organization Classification of Tumours. IARC Press, Lyon 2003; 37-41.
12. Lakhani SR, Ellis IO, Schnitt SJ, et al. WHO classification of tumours of the breast. IARC Press, Lyon 2012.
13. Geyer FC, Weigelt B, Natrajan R, et al. Molecular analysis reveals a genetic basis for the phenotypic diversity of metaplastic breast carcinomas. J Pathol 2010; 220: 562-73.
14. Tian Wei, Xu Dong. Diagnosis and management of multiple carcinosarcoma of the breast in a young chinese patient. Breast Care 2012; 7: 147-149.
15. Sultana N, Kayani N. Validity of frozen section in the diagnosis of breast lumps: 5 years experience at the Aga Khan University Hospital. J Pak Med Assoc 2005; 55: 533-536.
16. Tokudome N, Sakamoto G, Sakai T, et al. A case of carcinosarcoma of the breast. Breast Cancer 2005; 12: 149-153.
17. Foschini MP, Dina RE, Eusebi V. Sarcomatoid neoplasms of the breast: proposed definitions for biphasic and monophasic sarcomatoid mammary carcinomas. Semin Diagn Pathol 1993; 10: 128-136.
18. Hennessy BT, Giordano S, Broglio K, et al. Biphasic metaplastic sarcomatoid carcinoma of the breast. Ann Oncol 2006; 17: 605-613.
19. Buzdar AU, Valero V, Theriault RL. Pathological complete response to chemotherapy is related to hormone receptor status. San Antonio Breast Cancer Symposium 2003; Abstr 302.
20. Tse GM, Tan PH, Putti TC, et al. Metaplastic carcinoma of the breast: a clinicopathological review. J Clin Pathol 2006; 59: 1079-1083.
21. Jia Y, He C, Liu L, et al. A Retrospective Study of the Imaging and Pathological Features of Metaplastic Breast Carcinoma and Review of the Literature. J Med Sci Monit 2019; 25: 248-258.
22. Park HS, Park S, Kim JH, et al. Clinicopathologic features and outcomes of metaplastic breast carcinoma: comparison with invasive ductal carcinoma of the breast. Yonsei Med J 2010; 51: 864-869.
Copyright: © 2019 Polish Association of Pathologists and the Polish Branch of the International Academy of Pathology This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0) License (http://creativecommons.org/licenses/by-nc-sa/4.0/), allowing third parties to copy and redistribute the material in any medium or format and to remix, transform, and build upon the material, provided the original work is properly cited and states its license.
Quick links
© 2020 Termedia Sp. z o.o. All rights reserved.
Developed by Bentus.
PayU - płatności internetowe