Metaplastic Breast Cancer: Characteristics and Survival Outcomes

Objectives Metaplastic breast cancer (MBC) is a rare neoplasm accounting for <1% of all breast cancer. We evaluated the clinical characteristics and survival outcomes of MBC. Methods Patients diagnosed with pathologically proven MBC were reviewed from the institutional breast cancer database from 2000 to 2017. Results A total of 136 patients diagnosed with MBC were included in the study. The median age of the diagnosis was 60 years, and 60% of patients were stage II at diagnosis, and 22% were stage III. About two-thirds of the patients were triple-negative; 93% had nuclear grade III, and 25% had a lymphovascular invasion. Squamous differentiation (29%) was the most common histologic subtype, followed by the spindle subtype (21%). The most common distant metastases were lung (22%), followed by bone (13%). Moreover, 60% had a mastectomy, 19% had endocrine therapy, 58% had radiation, 51% received anthracycline-based chemotherapy, 26% had non-anthracycline chemotherapy, and 22% received no chemotherapy. In the entire cohort, the two-year overall survival (OS) and five-year OS were 79% and 69%, respectively, and the two-year progression-free survival (PFS) and five-year PFS were 72% and 61%, respectively. On multivariable analysis, the stage of MBC (stage III: hazard ratio (HR), 5.065 (95% confidence interval (CI), 1.02-25.27) (p=0.048)), poor functional status (Eastern Cooperative Oncology Group (ECOG) score, 2; HR, 24.736 (95% CI, 1.92-318.73) (p=0.014)), and distant metastasis to the brain (HR, 8.453 (95% CI, 1.88-38.04) (p=0.005)) and lung (HR, 42.102 (95% CI, 7.20-246.36) (p<0.001)) were significant predictors of decreased OS. Conclusions MBC demonstrated early disease progression and poor overall survival. The stage of MBC, decreased performance status, and metastasis to the lung and brain were independent poor prognostic factors.


Introduction
Metaplastic breast cancer (MBC) is a rare neoplasm that accounts for less than 1% of all breast cancers, and it is characterized by histologic and molecular heterogeneity. MBC is histologically characterized by the differentiation of the neoplastic epithelium into squamous cells and/or mesenchymal-looking elements [1]. Evidence also suggests that MBC has an aggressive nature and tends to have a worse prognosis of MBC as compared to other breast cancers. This is possibly due to its rarity, tumor heterogeneity, and lack of targeted treatment [2,3]. Various clinical and immunohistochemical factors along with genetic markers have been described in the literature; however, no validated prognostic markers have been identified. To date, MBC remains a clinical challenge for physicians regarding pathogenesis, clinicopathological features, and its management [2]. Most MBCs are triple-negative; hence, they are managed in a similar way to triple-negative breast cancer (TNBC) with anthracycline, taxane, and platinum-based therapy. There is no standard therapeutic approach available for this breast cancer subtype.
In this study, we evaluated the clinical, histopathologic, and molecular characteristics across all locations of our health system. We also sought to identify potential factors attributing to the progression of the disease and survival.

Patients
The complete list of the study cohort from January 2000 to December 2017 was identified from a report generated from the institutional database. The study was approved by the Cleveland Clinic Institutional Review Board (IRB# . The database for pathologically proven MBC cases was maintained in the breast tumor registry from three centers in the Cleveland Clinic health system. The inclusion criteria for the study include age > 18 years, female gender, and all pathologically diagnosed and proven cases of MBC from the database. Patients aged <18 years and males were excluded. A total of 136 patients fulfilled the inclusion criteria and were included in the study for analysis. The retrospective chart review was done using an electronic medical record for data abstraction. Study variables for data collection include age at the diagnosis, demographics, Eastern Cooperative Oncology Group (ECOG) performance status, histopathology, staging, hormone receptor status, lymphovascular invasion, molecular markers, treatment received (surgery, radiation therapy, chemotherapy, and hormonal therapy), date of local and distant progression, and date of death or last follow-up. The stage of MBC was documented based on the breast cancer staging system by the American Joint Committee on Cancer (AJCC).

Pathology
The histopathology of breast tissue was evaluated by an expert pathologist in breast cancer and documented in the electronic medical record. MBC was defined based on the World Health Organization (WHO) classification for breast tumors [1]. It is a unique group of invasive ductal carcinoma, which is characterized by the differentiation of tumor cells into purely epithelial or mixed epithelial and mesenchymal components. The epithelial group includes squamous, adenocarcinoma with spindle cell differentiation, and adenocarcinoma, including mucoepidermoid. Mixed epithelial and mesenchymal components are comprised of carcinoma with chondroid metaplasia, carcinoma with osseous metaplasia, and carcinosarcoma. We broadly categorized histology into four groups for analysis as follows: squamous subtype, spindle cell subtype, mixed epithelial plus mesenchymal differentiation, and other types of metaplastic carcinoma with no special type. Other histology of MBC with no special type includes ductal carcinoma in situ micropapillary pattern, heterologous type with matrix production, high-grade metaplastic carcinoma, large cells with intermingling mature plasma cells, matrix-producing carcinoma, and sarcomatoid features, and metaplastic carcinoma.

Statistical analysis
Categorical data are summarized as frequencies and percentages, whereas continuous data as medians and ranges. Overall survival (OS) and progression-free survival (PFS) were the primary outcomes. OS was calculated from the date of the diagnosis of metaplastic breast cancer to the date of death or last follow-up. PFS was calculated from the date of the diagnosis of MBC to the date of distant progression or local progression of disease or death or follow-up, whichever comes first. Time to event variable was summarized using the Kaplan-Meier method; the log-rank test and univariate cox regression model were used to estimate the association between outcomes and patient clinical and pathological characteristics, including tumor-node-metastasis (TNM) stage (I-IV), distance metastasis, hormone receptor status, histologic subtypes, type of therapy, and performance status. The variables that were selected through stepwise selection were included in the multivariable Cox model. All statistical analyses were performed using R version 3.5.0. p-value < 0.05 was considered statistically significant.
On histologic evaluation, spindle cell tumor was observed in 29 (21%) patients, and 40 (29%) patients showed squamous subtype. Fifty-five (40%) patients had mixed epithelial plus mesenchymal carcinoma, and 12 (9%) were found to be other types of metaplastic carcinoma with no specific type. Breast pathologists documented the diagnosis of MBC in the chart. Histology documentation in the pathology report was utilized to divide MBCs into different subtypes. Seven patients had BRCA mutations (five had BRCA1 and two had BRCA2). Forty-three (31.6%) patients expressed p63. The most common distant metastases were observed in the lungs (n=30 (22%)), followed by the bone (n=18 (13%)) and brain (n=9 (7%)). The adrenal and spleen were rare sites of metastasis, each occurring in one patient. Fifty (40%) patients had a lumpectomy, 76 (60%) had a mastectomy, 25 (20%) received hormonal therapy, 76 (58%) received radiation, 51% received anthracycline-based chemotherapy, 26% had non-anthracycline chemotherapy, and 22% received no chemotherapy. Unfortunately, not all clinical and pathological data were available for patients included in the study during the chart review. Complete demographics, clinical, and pathological variables are outlined in Table 1.

Variables Level Overall
Age at diagnosis (median (range)) 60 (27,   #Other histology in our study includes ductal carcinoma in situ micropapillary pattern, heterologous type with matrix production, high-grade metaplastic carcinoma, large cells with intermingling mature plasma cells, matrix-producing carcinoma, and sarcomatoid features, and metaplastic carcinoma.

Survival analysis
At the time of analysis, a total of 44 patients had died, and 36 patients were found to have progression (local or distant). Four patients were excluded from survival analysis due to missing information. Two patients had missing dates of diagnosis, and two patients had missing dates of the last follow-up or date of death. The two-year and five-year OS rates in the entire cohort were 79% and 69%, respectively. We observed a two- year PFS rate of 72% and a five-year PFS rate of 61%.

Progression-free survival
Univariable analysis for PFS was statistically significant for the stage of the MBC and higher functional status (ECOG 0 and 1 ). Lumpectomy and mastectomy were the types of surgery included as a variable for analysis. The five-year PFS rates for stages I, II, and III were 68%, 72%, and 27%, respectively ( Figure 1). Stage IV had worse survival with only a one-year PFS rate of 14%. Metastases to the lung, brain, bone, and liver were found to have worse PFS. The complete univariable analysis is summarized in Table 2.    (Figure 2). Better functional or performance status (ECOG 0 and 1) demonstrated better five-year OS, and poor functional or performance status (ECOG 3 and 4) was correlated with increased mortality. Lymphovascular invasion, p63 positivity, hormonal therapy, radiation therapy, and histologic subtype did not show better outcomes.

OS, overall survival; Ref: reference
Metastasis to the lung and brain at any time during follow-up was associated with poor survival. In patients with lung metastasis, the two-year and five-year OS rates were 53% and 27%, respectively, with an HR of 7.85 (95% CI, 4.18-14.72) (p<0.001). The complete analysis is summarized in Table 4.  Four patients were excluded from survival analysis due to missing information such as the date of diagnosis, last follow-up, or date of death.

Variable
*Progression was treated as an event and death as a competing risk event.
#Other histology in our study includes ductal carcinoma in situ micropapillary pattern, heterologous type with matrix production, high-grade metaplastic carcinoma, large cells with intermingling mature plasma cells, matrix-producing carcinoma, and sarcomatoid features, and metaplastic carcinoma.

Discussion
The identification of MBC has evolved over the past two decades since the official recognition of this subtype by the WHO [3]. Clinically, most MBC patients present with a well-circumscribed palpable mass, and the presentation resembles an invasive ductal carcinoma. However, previous studies reported larger tumor sizes, greater than 5 cm, which is associated with worse outcomes [4][5][6]. A large national database study by Pezzi et al. reported an increased proportion of MBC in older patients with a mean age of 61.1 years and a higher prevalence in African-American and Hispanic patients [6]. In our study cohort, the median age of diagnosis was 60 years (27-92 years), and most tumors were diagnosed at AJCC stage II, followed by stage I. The white population (76%) represented the dominant cohort of the study, which could possibly be due to the large referral center with better access to care and regional demographics of the MBC population.
Currently, MBC is classified into purely epithelial and mixed epithelial and mesenchymal components based on the updated fourth edition of the WHO classification for breast tumors [7]. It is a heterogeneous tumor with diverse histologic subtypes. Previous studies demonstrated variation in the prevalence of histologic subtypes in different ethnic populations. The spindle cell subtype resembles a low-grade sarcoma, which is the most common histologic subtype found in western patients and is reported to be associated with poor prognosis [8,9]. The squamous subtype was found to be more common in the Asian population [4,8]. However, Zhang et al. found the most common histology subtype of spindle cell carcinoma (34%) followed by the squamous subtype (31%) in the Chinese population [10]. In our study cohort, we identified the squamous subtype (29%) more commonly in our study population, followed by the spindle subtype (21%).
Tumor protein p53 is a gene that functions as a tumor suppressor. The p53 mutation has been commonly reported in MBC with a high frequency ranging from 53% to 64% [11,12]. Tumor protein p63, which is a member of the p53 family, is expressed in MBC tumor cells and is also a myoepithelial marker [13]. Evidence suggests that p63 can be utilized as a diagnostic marker for MBC but with no prognostic value [13,14]. We found 43 (31%) patients positive for p63 expression in the entire cohort. No difference in outcome was found between patients positive for p63 versus negative for p63 expression.
MBC is known for its rapid tumor growth and an increased tendency for recurrence [3,15]. Typically, tumors are of higher nuclear grade and characterized by the absence of ER, PR, and HER2/neu expression [7,16,17]. Moreover, the prognosis of metaplastic triple-negative breast cancer is worse when compared to nonmetaplastic triple-negative breast cancer [18]. In our study cohort, approximately two-thirds of the patients had metaplastic triple-negative breast cancer. The cohort with any hormone receptor positivity and treatment with hormonal therapy did not show any improvement in survival when compared with the metaplastic triple-negative cohort. We also observed the heightened potential of tumor metastasis to the lung (22%), followed by the bone (13%), brain (6%), and liver (4%). In addition, there was evidence of rare metastasis to the spleen and adrenal gland in the study cohort. Evidence suggests the metastatic spread of tumors to the lungs, bone, and brain via vasculature instead of lymphatics [19]. Patients with metastasis to the lung and brain in our study cohort demonstrated significantly poor survival outcomes.
Rakha et al. reported histologic subtype as an independent prognostic factor in MBC; the spindle subtype tumor had aggressive behavior with a worse prognosis as compared to the matrix-producing tumor and squamous subtype [8]. However, in other studies, the histologic subtype did not show any statistical significance as a prognostic factor [10,20]. In our study cohort, histologic subtypes did not prove to be a predictor of outcomes ( Tables 6, 7). Nevertheless, young age (<40 years), skin invasion, and squamous carcinoma subtype with nodal involvement have been identified as independent predictors of outcome in MBC patients [15].   In one of the large US population-based studies, 1,011 MBC patients were compared with 253,818 infiltrating ductal carcinoma (IDC) [21]. The study noted significantly worse five-year survival in MBC patients than in IDC patients (78% versus 93% (p<0.0001)). Besides, MBC was associated with higher tumor grades and larger tumor sizes. Similar findings were demonstrated in another study; authors reported decreased five-year OS when compared with invasive ductal carcinoma and triple-negative invasive ductal carcinoma (54.5% versus 85.1% versus 73.3% (p<0.001)) [5]. El Zein et al. also observed a worse prognosis in MBC than in triplenegative breast cancer (TNBC) [22]. The study documented that MBC patients had almost double the risk of local recurrence than TNBC patients (95% CI, 1.01-3.83 (p=0.05)). The authors also noted worse disease-free survival (DFS) and OS in MBC patients when compared with matched TNBC patients (p<0.001, p=0.033). Further review of the literature revealed a five-year OS ranging from 50% to 83% in various retrospective studies [10,17,19,[23][24][25][26][27][28][29]. Most of the studies demonstrated a five-year DFS of 41%-65% [10,19,23,[26][27][28][29][30], except for one study that demonstrated a five-year DFS of 84% [24].
In our study, we observed a five-year OS of 69% and PFS of 61%, which correlates with previous studies. On multivariable analysis, the stage of the disease, performance status (ECOG score), and distant metastasis to the lung, bone, brain, and liver proved to be significant predictors of outcomes in the MBC patient population. More importantly, on analysis for prognostication of MBC, we observed that patients with bone metastasis have a lower risk of death. On the other hand, a lower risk of progression was identified in the cohort of patients with liver metastasis. However, larger study samples would be needed to identify if these factors are truly associated with positive outcomes.
Our study is limited by its retrospective design with the possibility of selection bias and a small sample size. Few patients in the study cohort had missing variables; therefore, the total number of patients included for analysis for some of the specified variables was less than 136. We intended to do a descriptive analysis because of the above reasons. We utilized the pathology report documented by breast pathologists for MBC in our retrospective cohort study and did not review slides. Patients were not divided based on the type of presentation, such as primary first-time diagnosed MBC and concurrent malignancy in the contralateral breast. Another weakness of the study is that progression included both local and distant progression; no analysis was explicitly conducted for local recurrence. As far as treatment is concerned, we did not describe whether the patient received neoadjuvant or adjuvant chemotherapy, the combination of surgery with radiation therapy or hormonal therapy, or chemotherapy.
Despite the above limitations, the strength of the study is a sizable cohort of MBC patients over a long period of time depicting clinical and pathological characteristics.