TROP2, androgen receptor, and PD-L1 status in histological subtypes of high-grade metaplastic breast carcinomas

[HER2], programmed death ligand-1 [PD-L1], and trophoblast cell surface antigen 2 [TROP2]). Results : The median age at diagnosis was 59.5 years. Six (9%) patients had metastatic disease at diagnosis. Among the nonmetastatic patients receiving neoadjuvant therapy, 26% (5/19) achieved pathological complete response. Most tumours were pT1/pT2 (77%) and 12% were pN + . Histological subtypes (mixed, squamous, mesenchymal, and spindle cell) were 40%, 35.5%, 15.5%, and 9%, respectively. Tumour-inﬁltrating lymphocytes were low or moderate except when squamous differentiation was present. Most tumours were triple-negative (92%). AR and TROP2 were positive in 34% and 85% of the cases, respectively. PD-L1 was positive in tumour cells in 18% (cut-off: 1% of positive tumour cells) of the cases and in tumour-inﬁltrating immune cells in 40% (cutoff: 1% of tumour area) of the cases. Notably, spindle cell and mesenchymal metaplastic breast carcinomas were mostly PDL1-negative. Lastly, 21 (32.3%) cases were HER2-low, all being HER2 1 + , with no HER2 2 + . Conclusion : Metaplastic breast carcinoma could beneﬁt from tailored therapeutic strategies adapted to the phenotypic speciﬁcities of histological subtypes.


Introduction
Metaplastic breast carcinoma (MBC) is a subtype accounting for 0.2-2% of invasive mammary carcinomas. 1 Based on the current World Health Organization (WHO) classification, this breast carcinoma group was characterised by the presence of epithelial neoplastic cells with squamous, spindle, or mesenchymal differentiation, representing at least 10% of the tumour area. They represent a heterogeneous spectrum of entities displaying histological and prognostic specificities. A small proportion of them, including fibromatosis-like carcinoma and low-grade adenosquamous carcinoma, are typically classified as low-grade MBCs, 2,3 with most of them classified as high-grade MBCs (HG-MBC), as associated with a high risk of visceral metastasis (50% at 5 years). 4 This group encompasses squamous cell carcinomas, mesenchymal MBC, and spindle cell carcinomas, often containing a mixture of different subtypes, at times associated with a nonspecific type (NST)/ductal component.
Our study sought to assess TIL levels and PD-L1, AR, and TROP2 expression in a retrospective series of HG-MBC patients from our tertiary care centre, while comparing our findings with the existing literature.

C O H O R T
We retrospectively included patients initially diagnosed with HG-MBC at Institute Curie's Pathology Department, Paris, between January 2005 and September 2017. MBC diagnosis was confirmed following the last WHO classification. All MBCs that were low-grade, i.e. low-grade adenosquamous cancer and low-grade fibromatosis-like cancer, were excluded. To classify each tumour in an HG-MBC subtype, we considered every component that made up at least 1% of the tumour on the surgical specimen or core needle biopsy sample. Our study was validated and approved by the Breast Cancer Study group and Ethical Committee of the Curie Institute (approval number: DATA190184). All patients provided consent for their tumour material to be used for morphological and phenotypical review.
Clinical, treatment, and follow-up data were collected using the electronic medical record, until September 2018. OS was defined as the time from diagnosis to death from any cause or last follow-up. Breast cancer-specific survival (BCSS) was defined as the time from diagnosis to death from MBC or last follow-up.
Pathological evaluation was performed by three pathologists, including: number, type, and percentage of histological contingents, histological tumour type, and percentage of TILs. For the latter, three categories were defined as follows: low for tumours with 0-10% of TILs, intermediate for tumours with 11-49% of TILs, and high for tumours with ≥50% of TILs. For patients who received neoadjuvant chemotherapy, data from core needle biopsy were used to determine: number, type, and percentage of each neoplastic  Immunohistochemistry (IHC) analyses were performed on 3-lm formalin-fixed paraffin-embedded (FFPE) block sections using the following antibodies: TROP2 (Leica Bond-III automat, Leica Microsystems, Wetzlar, Germany), AR, and PD-L1 (Dako Autostainer Plus automat, DakoCytomation, Glostrup, Denmark). They were assessed on surgical specimens for patients without neoadjuvant chemotherapy and on core needle biopsy samples for those with neoadjuvant chemotherapy. Oestrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2) status was assessed for each tumour by reviewing the original diagnostic slides. Details regarding antibody clones, manufacturers, dilutions, incubation times, and pretreatment buffers are listed in Table S1.
Nuclear staining was assessed for ER, PR, and AR (negative status defined as the presence of <1% positive tumour cells). HER2 status was based on ASCO/ CAP 2018 recommendations, with 2+ positive cases screened by fluorescence in situ hybridisation (FISH) to assess HER2 gene status. Cases were considered HER2-low if they were 1+ or 2+ not amplified. PD-L1 was assessed on tumour cells (positive status: ≥1% membranous tumour cell positivity) and TILs (positive status: ≥1% tumour surface occupied by PD-L1+ immune cells). TROP2 was assessed on tumour cells (negative status: <1% membranous or cytoplasmic tumour cell positivity).

S T A T I S T I C A L A N A L Y S I S
Statistical analyses were performed using R project for Statistical Computing 3.2.1, Rstudio 7.6 (Vienna, Austria), and GraphPad Prism 7 for Windows (San Diego, CA, USA). All tests were two-sided. Statistical significance was set at P < 0.05. Patients with missing information were excluded. For continuous variables, P-values were based on the Wilcoxon-Mann-Whitney test; for categorical variables, on two-sided Fisher's exact test. Survival curves were analysed using the Kaplan-Meier method. Univariate analysis investigated the association between survival and clinicopathological parameters. The only patient that was metastatic from a previous non-MBC at MBC diagnosis was excluded. Local recurrence-free survival and distant recurrence-free survival were studied only in nonmetastatic patients at diagnosis.

C L I N I C O P A T H O L O G I C A L C H A R A C T E R I S A T I O N
Overall, 73 patients were identified through our electronic data search (Table 1). Six patients were excluded as follows: MBC recurrence diagnosed before 2005 (n = 4), treatment started before referral to our centre (n = 1), and no clinical data available (n = 1). Two additional patients were excluded as their slide review excluded any HG-MBC, with 65 patients left for analysis.
Median age at diagnosis was 59 years. Overall, 16 (24.6%) patients had a history of previous breast cancer, 14 (21.5%) of whom had received radiotherapy. Median clinical and radiological tumour size at diagnosis was 25 mm. Twelve (18.5%) patients had clinically or radiologically suspicious lymph nodes.

H I S T O P A T H O L O G I C A L F E A T U R E S
The median tumour size was 21 mm, with more than 80% of tumours classified as pT1/pT2, and lymph node metastases detected in eight (12.3%) patients. Lymphovascular invasion was noticed in 15 (23.1%) tumours. Among the 19 nonmetastatic patients receiving neoadjuvant therapy, five (26%) achieved pCR. Most tumours were Elston-Ellis grade 3 (55/64; 84.6%). The median mitotic index was 9.65 mitoses/ mm 2 .
Considering the HG-MBC subclassification, mixed MBC was the most common subtype (40%), followed by pure squamous cell (35.4%), mesenchymal (15.4%), and spindle cell carcinomas (9.2%). Among MBC components, the most common was squamous cells, in 44 (67.7%) tumours, followed by spindle cells, in 26 (40%) tumours, and mesenchymal cells, in 20 (30.8%) tumours. Four tumours had an NST component, and one an apocrine component. The TIL percentage was available for 64/65 tumours. One core needle biopsy was nonassessable due to its limited size. The median TIL percentage was low (10%), and tumours with high TIL rates (>50% of One formalin/acetic acid/alcohol (FAA)-fixed or FFPE block representative of MBC was available for 59 patients, with additional IHC stainings performed. IHC analysis was completed for 57/59 patients; for the remaining two patients, the material was sufficient to perform the whole analysis except for PD-L1 antibody. IHC positivity is summarised in Table 2, with IHC positivity rates across histological subtypes illustrated in Figure 1.

Discussion
In this 65-patient cohort, MBCs exhibited high TROP2 and low TIL expression levels. MBCs expressed AR, PD-L1, and low HER2 levels, regardless of the MBC subtype, yet with some specificities, such as AR missing in pure spindle cell MBCs and PD-L1 being more frequently expressed in squamous cell carcinomas.
In our study, 40.4% of carcinomas were PD-L1positive when assessing immune cells and 17.5% were PD-L1-positive at the tumour cell level. For assessing PD-L1 positivity in TILs, we followed the criteria defined in the IMPassion130 study for atezolizumab prescription in TNBC, 17 indicating that a significant proportion of our MBC patient cohort could benefit from anti-PD-L1 therapy. Other studies reported on PD-L1 expression in MBCs, focusing on tumour cells, immune cells, or both (Tables S2 and  S3). The percentages of PD-L1-positive MBCs varied from 0 to 73% in tumour cells and from 9.5% to 80% in inflammatory cells. 20,[23][24][25][26][27][28] The criteria of positivity and antibody clone used for IHC varied from one study to another. Notably, assessing PD-L1 antibody expression was highly variable due to preanalytical and analytical factors (biopsy versus surgical specimen; quality of collection), thus limiting the opportunities for comparison. In our cohort, PD-L1 expression was higher in the squamous cell subtype, the latter similarly exhibiting the highest TIL levels. We hypothesize that the variability in PD-L1 rates reported in the literature could possibly be explained by the different MBC subtype percentages included in the series. ICIs, such as anti-PD-L1 immunotherapy, have been assessed in several studies on TNBCs, with none specifically focused on MBCs. It is, thus, crucial to improve our knowledge about PD-L1 antibody expression in MBCs to better identify patients who could benefit from ICI therapy.
AR is expressed in 70-90% of breast cancers and 10-50% of TNBCs. 38 A 2016 meta-analysis on AR expression in TNBCs showed that TNBCs expressing AR have longer recurrence-free survival with no impact on OS. 39 However, its prognostic value remains controversial in TNBC. In our study, 34% of tumours expressed AR; yet we found no correlation between AR expression and survival. The positivity rate found in our study was higher than that (0-11%) from previous studies on MBC that included a smaller patient number (2-34 patients). 22,[32][33][34] AR was expressed in all histological types except spindle cell carcinoma. However, caution should be exercised in drawing conclusions due to the small number of spindle cell carcinomas. In mixed MBCs, AR was expressed in all histological components, including the spindle cell contingent. The AR positivity rate in our study clearly suggested that some patients could potentially be included in trials on antiandrogen hormone therapy. However, the percentage of tumour cells expressing AR ranged from low to moderate. Several antiandrogenic drugs were tested in treating TNBC, 38 with none having yet received approval for managing AR-positive TNBC; to our knowledge, no study has so far focused on the efficacy of antiandrogenic therapies in MBCs.
While breast carcinomas were frequently reported to display high TROP2 expression, 40,41 no study has specifically assessed TROP2 expression in MBCs. Thus, our findings turn out to be quite novel. In our cohort, 85% of MBCs expressed TROP2. The staining was most often membranous, with rare tumours showing cytoplasmic or both membranous and cytoplasmic staining. The percentage of expression was often high in squamous cell carcinomas with squamous cell differentiation, the vast majority of which were TROP2positive. The high frequency of TROP2 expression in MBCs suggests that drugs such as sacituzumab govitecan may prove to be quite interesting in this setting. However, results on using this treatment in breast cancer are very recent; hence, the relevance of using TROP2 expression at the tumour cell level for prescribing this drug must still be confirmed. Besides, in breast cancer, high TROP2 expression has been associated with triple-negative tumours, high histological grade, and high TNM stage, with a higher frequency of locoregional and distant lymph node metastases. 41,42 Moreover, a study investigating the link between TROP2 expression determined by IHC and survival revealed different associations depending on the location of TROP2 expression: TROP2 expression at the membrane level was associated with poorer outcomes, while TROP2 expression at the cytoplasmic level was associated with a better outcome. 40 In our study, membranous expression of TROP2 was not associated with a statistically significant decrease in survival. Cytoplasmic expression was too infrequent to further investigate its association with survival. As a result, further research is needed to better understand the relevance of TROP2 expression at the membranous or cytoplasmic cell levels on clinical outcomes.
Interestingly, HER2-low breast cancer is emerging as a potential candidate for targeted therapy with antibody-drug conjugates. Indeed, we observed that almost 32% of MBCs analysed were HER2-low, with new interesting therapeutic perspectives.
In conclusion, in our cohort a significant proportion of MBCs expressed PD-L1, AR, and TROP2, while being HER2-low. Based on these observations, potential MBC patients may prove to be candidates for using new targeted therapeutic strategies for this rare, aggressive, and poorly documented clinical entity.

Supporting Information
Additional Supporting Information may be found in the online version of this article: Table S1. List of antibodies used. Table S2. Reported positivity of PDL1 immunohistochemistry in tumour cells in MBC. Table S3. Reported positivity of PDL1 immunohistochemistry in immune cells in MBC.