Outcomes of Metaplastic Breast Cancer Versus Triple-Negative Breast Cancer: A Propensity Score Matching Analysis

This study aims to compare the survival outcomes of metaplastic breast cancer (MPBC) with triple-negative breast cancer (TNBC) and identify prognostic factors that influence the survival outcomes of MPBC patients and TNBC patients. Patients with nonmetastatic MPBC or TNBC were reviewed from our database. Patients’ clinicopathologic and molecular features were analyzed with respect to outcomes, including disease-free survival (DFS) and overall survival (OS). Propensity score matching (PSM) with a one-to-three matching between MPBC patients and TNBC patients was performed. A total of 857 female patients (76 MPBC patients and 781 TNBC patients) were included in this study. A subgroup of triple-negative MPBC (n  =  60) was matched with TNBC (n  =  180) cases based on patient characteristics and treatments. Kaplan–Meier analysis indicated that the MPBC group was associated with worse OS and DFS before (P  =  0.0046 both) and after (P  =  0.011 and P  =  0.0046, respectively) PSM. Multivariable analysis revealed that a higher T stage (T > 2) (P  =  0.032) and higher lymph node stage (N3 vs. N0-2, P  =  0.012) were associated with worse OS after PSM. For DFS, the MPBC group (P  =  0.012), higher T stage (T > 2) (P  =  0.032), and higher lymph node stage (N3 vs. N0-2, P  =  0.045) were associated with worse DFS. Among the 76 MPBC patients, a higher T stage and mesenchymal differentiation were associated with worse OS (pT1/2 vs. pT3/4 and mesenchymal differentiation vs. other subtypes, P  =  0.007 and P  =  0.011, respectively). Compared with TNBC, MPBC was associated with worse OS and DFS. Mesenchymal differentiation has a worse DFS than other subtypes of MPBC.


Introduction
Metaplastic breast cancer (MPBC) is a rare histological subtype of breast carcinoma with substantial heterogeneity, containing mixed epithelial and metaplastic components composed of glandular and non-glandular tissues, accounting for only 0.25-1% of all breast cancer [1].At present, it is generally recognized that MPBC has the characteristics of older age, larger primary tumor size, lower axillary lymph node metastasis rate, lower estrogen receptor (ER) positive rates, lower progesterone receptor (PR) positive rates, and lower human epidermal growth factor receptor-2 (HER2) positive rates [2][3][4].
Because most MPBC have a triple-negative phenotype, MPBC was regularly compared with triple-negative breast cancer (TNBC).Whether the prognosis of MPBC is worse than that of TNBC is controversial.Some studies revealed that the prognosis of MPBC seems to be worse than that of TNBC [5][6][7][8][9][10].Other studies showed that MPBC has no significant difference in clinical outcomes with TNBC [11,12].The controversy may come from the fact that limited data are available due to the rarity of this disease.Previous studies with a small sample size had performed propensity score matching analysis with few variables to reduce selection bias [5,11,13].However, the sample sizes from these studies have mainly been small.Therefore, a more comprehensive propensity score matching analysis needed to be performed to reduce the effect of selection bias and potential confounding.
Hence, this study aims to compare survival outcomes between nonmetastatic MPBC and nonmetastatic TNBC by propensity score matching analysis.Furthermore, the association of survival outcomes with neo-adjuvant chemotherapy, chemotherapy, radiotherapy, surgery, histopathological and molecular characteristics among MPBC patients was analyzed.

Study design
We queried the database of the Clinical Research Center for Breast, West China Hospital, Sichuan University, a prospectively maintained database, for MPBC and TNBC cases with stages I to III. Between January 2002 and August 2020, a total of 84 nonmetastatic patients were diagnosed with MPBC in our institution.The pathological diagnosis was confirmed by experienced pathologists.Furthermore, 890 patients without metastasis were diagnosed with TNBC between January 2010 and December 2019.Among these patients, 117 were excluded due to failure to follow-up or no record of essential information.Finally, 76 MPBC patients and 781 TNBC patients were included in this study (Figure 1).Five patients with MPBC, diagnosed during 2002-2009, lacking information of HER-2 expression were not included in the PSM but were included in the section of the subtype analysis of MPBC.
Clinical and laboratory data were extracted from a prospectively maintained database.To compare the survival outcomes between patients with triple-negative MPBC and TNBC, we performed propensity score matching (PSM) with one-to-three matching.Patients were matched on characteristics, which included age, T-stage, lymph node status, Ki-67 index, Nottingham grade, type of surgery, radiotherapy, neo-adjuvant chemotherapy, and chemotherapy.

Inclusion and exclusion criteria
The inclusion criteria were diagnosis with MPBC or invasive ductal carcinoma with ER-/PR-/HER2-, as well as pathology and treatment information were available.The exclusion criteria were as follows: failed to follow up or not treated at our institution; metastatic disease at the time of diagnosis; and other preexisting malignancies.

Evaluation criterion and follow-up
The efficacy evaluation of treatment was performed according to the Response Evaluation Criteria in Solid Tumors objective efficacy evaluation standards (RECIST, version 1.1) [14].Disease progression-free survival (DFS) is the time from diagnosis to the appearance of any local recurrence, distant metastasis, or death.Overall survival (OS) was defined as the time from diagnosis to death from any cause.Data were collected electronically from the Clinical Research Center for Breast, West China Hospital, Sichuan University database.Follow-up was performed by clinic visits, medical record reviews, and telephone contact.

Statistical analysis
Propensity score matching was performed by using R 4.2.0.Each of 60 metaplastic breast cancer cases was matched with three controls for a total of 180 matched controls.Multivariable logistic regression models were used to estimate propensity scores.The matching parameters including age, T-stage, lymph node status, Ki-67 index ( \ 30% or C30%), Nottingham grade, neo-adjuvant chemotherapy, chemotherapy, type of surgery, and radiotherapy.
Statistical analysis was performed by using SPSS (IBM SPSS 26.0, SPSS Inc.) and R 4.2.0.Kaplan-Meier survival analysis was performed to estimate the survival curve.Kaplan-Meier survival curves were plotted using R 4.2.0.
The independent prognostic factors of risk of breast tumor mortality were determined using univariate Cox regression and multivariate Cox regression.P values less than 0.05 were considered to be significant.

Survival analysis before and after matching
Before matching, the median follow-up time ranged from 2 to 216 months, with an average follow-up time of 50 94.1%)and 5-year DFS (80.9% vs. 92.7%).OS and DFS were significantly worse in the MPBC group before and after propensity score matching, and survival curves are shown for groups (Figure 2, P = 0.005 both).The average follow-up time of these two matching groups was 45 months (median 45 months, range, 2 to 216 months).During the follow-up time, disease recurrence after surgery occurred in 11 (18.3%)patients with MPBC and 9 (5%) patients with TNBC (P = 0.001).Seven (11.7%) patients  2, P = 0.011 and 0.005, respectively).

Survival results for MPBC subgroups
A total of 76 MPBC patients were included in the current study.Higher T stage and mesenchymal differentiation were associated with worse overall survival (Figure 3, pT1/ 2 vs. pT3/4 and mesenchymal differentiation vs. other types, P = 0.007 and 0.011, respectively).When compared with other subtypes, mesenchymal differentiation cases tended to have a worse OS and DFS, but this difference did not reach statistical significance (Figure 3, P = 0.079 and 0.052, respectively).

Discussion
We identified 76 MPBC patients and 781 TNBC patients diagnosed and treated at our institution to compare the outcome of MPBC and TNBC.Compared with TNBC patients, a higher T stage and less advanced or similar N stage were observed in MPBC patients in our study, which is concordant with prior researches [1,7,9,15].In addition, the two groups showed a difference in Ki-67 index, Nottingham grade, surgery, radiotherapy, and neo-adjuvant chemotherapy.These differences may cause bias in disease outcome assessment.Before matching, our results showed that MPBC patients had a worse OS, which was consistent with previous studies [2,5,7,8,10,15,16].Propensity score matching was performed with fewer variables in previous studies [5,11,13].We performed propensity matching to identify comparator triple-negative MPBC and TNBC groups that were similar in age, tumor size, lymph node status, Ki-67 index, Nottingham grade, type of surgery, radiotherapy, neo-adjuvant chemotherapy, and chemotherapy.Our matching is a more comprehensive PSM, which includes more variables when compared with the previous matched cohorts [5,[7][8][9].After matching, the difference in clinical features and treatment between the propensity-matched cohorts was not statistically significant.MPBCs still tend to have a worse DFS and OS when comparing the propensity-matched cohorts.Qualitatively similar results were also found after propensity score matching, suggesting that the results are consistent.These results are in line with previous studies that compared outcomes for patients with MPBC and TNBC [8,9,15].Squamous cell differentiation has been shown to be associated with worse DFS than other metaplastic components in a small retrospective study [8].While another study found the opposite: patients benefited from squamous cell differentiation [13].Patients with heterologous mesenchymal had a better 5-year overall survival and breast cancer-specific survival in previous reports [13,17].In our center, the histologic subtype has no significant impact on survival, which is consistent with previous studies [9,18].However, our research indicates that the mesenchymal subtype tends to have worse DFS than other subtypes.A potential reason for this is that compared with the other subtypes, the mesenchymal group has a higher T stage (T[2) in our study (42.9% vs. 14.0%).Further research is needed to determine whether the pathological subtype of MPBC is associated with a better or worse prognosis.Similar to the majority of previous studies, our research indicated that a lower lymph node stage was a predictor of better DFS in MPBC, and smaller tumor size or lower T stages were good prognostic factors for death [19][20][21][22].
In localized MPBC, surgery remains the standard of treatment [3].An analysis of surveillance, epidemiology, and end results (SEER) data from 1988 to 2006 did not suggest an association of surgery type with survival [23].However, another SEER analysis evaluating 2010-2014 data showed that among MPBC patients who underwent surgery, mastectomy was associated with inferior 3-year OS (hazard ratio = 2.00, 95% CI 1.31-3.06)compared with breast-conserving surgery independent of the stage at presentation [24].Lumpectomy as primary surgery was associated with improved recurrence-free and distant recurrence-free survival.In a prior study, the type of surgery did not relate to DFS [18].This is consistent with our research.
Tseng and Martinez used the SEER database to identify MPBC patients diagnosed from 1988 to 2006.A total of 1501 MPBC patients were included, and radiation therapy (RT) provided an OS (HR = 0.64; 95% CI, 0.51-0.82;P \ 0.001) and disease-specific survival (DSS) (HR = 0.74; CI, 0.56-0.96;P \ 0.03) benefit.When patients were stratified according to the type of surgery, RT provided an OS but not a DSS benefit to lumpectomy and mastectomy patients [23].Consistent with prior reports, C. Marc Leyrer reported that when radiation therapy was stratified by surgery type, receipt of adjuvant radiation after breast-conserving surgery (BCS) was significant for improved locoregional recurrence (LRR) (P \ 0.001) and OS (P = 0.009) over those who did not, and the use of radiation therapy was not significantly associated with LRR, distant metastasis (DM), or OS for mastectomy [18].Postoperative adjuvant RT was given to 20 (26.3%) MPBC patients in our study and compared with those who did not receive RT, and there were no significant differences in OS and DFS.The difference in treatment may be due to selection bias.More detailed and rigorous research is needed to assess the relationship between RT and survival in MPBC.
MPBC is poorly response to neo-adjuvant chemotherapy in prior reports, with a PCR rate in the range of 0-18% [2,19,[25][26][27][28].A previous study, including 44 MPBC patients, showed that 49% of cases had no clinical response or clinical radiological progression to neo-adjuvant chemotherapy.Only one (2%) patient had a pathologic complete response (pCR) [29].In our study, the result of neo-adjuvant chemotherapy was also poor.According to previous studies, with a large number of MPBC patients, adjuvant chemotherapy was associated with better outcomes [2,30,31].Current chemotherapy treatment protocols for the management of MPBC are not very effective, but they remain an important part of the treatment regimen for MPBC.This may be changed by immunotherapy in the future because few previous studies found that PD-L1 was overexpressed in MPBC, and immune checkpoint inhibitors showed good efficacy in MPBC [32,33].
In the current study, we found that MPBC is a more aggressive subtype of breast cancer than TNBC.Our current options for MPBC are far from ideal.Fortunately, immunotherapy and targeted therapy may hold promise as a new therapeutic strategy for MPBC, which has shown efficacy in previous researches [33][34][35].These potential treatment approaches are worth exploring and studying in detail in the future.

Limitations
Several limitations need to be noted regarding the present study.One limitation is that this retrospective study only included 76 MPBC patients, and the small sample size may cause bias.A more comprehensive PSM was performed to minimize the imbalance between the arms, but there may be unmeasured factors that contribute to the outcomes.Some patients who received treatment in our institution in earlier years lacked information on clinical features and histopathological features.

Fig. 1
Fig. 1 Flow diagram of the study cohort

Fig. 2
Fig. 2 Kaplan-Meier survival curves of overall survival and disease-free survival of MPBC vs. TNBC.a OS between MPBC and TNBC before propensity score matching; b OS between MPBC and TNBC after propensity score matching.Abbreviations: OS, overall survival; MPBC, metaplastic breast cancer; TNBC, triple-negative breast cancer; c DFS between MPBC and TNBC before propensity score matching; d DFS between MPBC and TNBC after propensity score matching.Abbreviations: DFS, disease-free survival; MPBC, metaplastic breast cancer; TNBC, triple-negative breast cancer

Fig. 3
Fig. 3 Kaplan-Meier survival curves of MPBC patients stratified by pathological subtype and T stage.a OS between mesenchymal differentiation and other subtypes; b OS between tumors T1-2 and T3-4; c OS of MPBC stratified by pathological subtype.d DFS of MPBC stratified by pathological subtype.Abbreviations: MPBC, metaplastic breast cancer; OS, overall survival; DFS, disease-free survival

Table 1
Baseline patient characteristics before and after propensity score matching

Table 2
Baseline patient characteristics of 76 MPBC cases ER estrogen receptor, PR progesterone receptor, HER2 human epidermal growth factor receptor 2, LN lymph node status, AC-T doxorubicin, and cyclophosphamide followed by taxane, FEC fluorouracil, epirubicin, and cyclophosphamide, TC docetaxel, and cyclophosphamide, X capecitabine, CMF cyclophosphamide, methotrexate, fluorouracil died from developing breast cancer in the MPBC group, and 8 of 180 (4.4%) patients with TNBC died due to the disease.A poorer DFS and OS were found in the MPBC population (Figure

Table 3
Univariate and multivariate Cox regression analysis of factors predictive of overall survival for 240 matched patients MPBC metaplastic breast cancer, TNBC triple-negative breast cancer, LN lymph node status

Table 4
Univariate and multivariate Cox regression analysis of factors predictive of disease-free survival for 240 matched patients MPBC metaplastic breast cancer, TNBC triple-negative breast cancer, LN lymph node status