Prognostic value of androgen receptor and FOXA1 co-expression in non-metastatic triple negative breast cancer and correlation with other biomarkers

Background In luminal androgen receptor (AR) tumours, FOXA1 may direct AR to sites occupied by ER in luminal tumours, thus stimulating proliferation. Methods AR and FOXA1 expression were evaluated by immunohistochemistry in 333 non-metastatic triple-negative breast cancers (TNBC). Positivity threshold was set at ≥ 1% staining. Lymphocytic infiltration, PD-L1expression, PIK3CA mutations, PTEN defects and BRCA1 promoter methylation were assessed. Results AR + /FOXA1 + tumours (42.4%) were more frequently: found in older patients, lobular, of lower nuclear grade, with more frequently PIK3CA mutations; exhibited less frequently BRCA1 promoter methylation, defects of PTEN and PD-L1 expression than others. Recurrence-free and overall survivals were significantly lower for AR + /FOXA1 + TNBC (median follow-up: 7.8 years). Conclusions AR + /FOXA1 + expression defines a luminal-like TNBC subgroup affected with a worse outcome compared to other TNBC and a higher risk of late recurrences. This subgroup appears enriched in PIK3CA mutations, suggesting a role for PI3K inhibitors in this subgroup.


INTRODUCTION
Triple negative breast cancers (TNBCs) represent 15% of all breast cancer (BC) and are defined by the lack of oestrogen receptor (ER), progesterone receptor (PR) and HER2 expression/amplification. This subgroup of BC is extremely heterogeneous: Lehmann et al. distinguished seven TNBC subtypes displaying unique gene expression profiles, including the luminal androgen receptor (LAR) subtype. 1 These tumours are ER-, CK5/6-, but express genes usually expressed in ER + luminal tumours such as the androgen receptor (AR) and FOXA1, are distinct of basal-like tumours and represent 8-12% of all BC. LAR cell lines are sensitive to AR antagonists. 1 Using immunohistochemistry (IHC), AR is expressed in 8-53% of TNBC 2 and its prognostic value is controversial. 3,4 The phosphatidylinositol 3-kinase (PI3K) signalling pathway is crucial for cell growth and survival. PIK3CA activating mutations and PTEN loss of expression may contribute to BC therapeutic resistance. In the study of Lehmann et al., PIK3CA mutations were more frequent in AR + TNBC than in AR-TNBC. 5 Programmed cell death (PD-1) is an immune checkpoint receptor. Its ligand, PD-L1, is expressed by immune cells, BC cells, and tumour-infiltrating lymphocytes (TILs). PD-1 binding to PD-L1 specifically inhibits T-cell activation and represents one mechanism of immune tumour escape. 6 TNBCs are thought to be more immunogenic than other BC and AR + TNBC show a higher frequency of PD-L1 expression. 7 In LAR tumours, AR functionality is dependent on FOXA1: FOXA1 is required for AR binding chromatin, AR transcriptional activity and cell growth, directing AR to sites normally occupied by ER in luminal tumours, inducing an oestrogen-like programme stimulating proliferation. 8 Our team has shown that AR + /FOXA1 + TNBC even tend to behave like luminal tumours. 9 These two biomarkers could be useful to identify a particular subgroup of TNBC.
This study aimed to evaluate in a large series of non-metastatic TNBC with a long follow-up both the profiles and the prognostic value of AR/FOXA1 co-expression and its correlation with other biomarkers like PD-L1 and PIK3CA status.

MATERIALS AND METHODS
This study included 349 patients with unifocal, unilateral, untreated, non-metastatic TNBC operated in our institution between 2002 and 2012. ER and PR negativity was defined as < 10% by IHC and HER2 negativity was defined as IHC 0/1 + or 2 + and negative fluorescent/chromogenic hybridisation in situ.

DISCUSSION
In our study, with cut-offs of 10% and 1% for ER/PR and AR positivity, respectively, we showed a AR + TNBC rate of 58.6%. There is currently no standardised assay to assess AR expression but recent data suggested that AR-targeted therapies may enhance the efficacy of chemotherapy even in TNBC with low AR expression by targeting cancer stem cell-like cells. 10 In a metaanalysis of TNBC based on retrospective studies and population data, AR positivity was significantly associated with prolonged RFS but had no significant impact on OS. 4 In our study, patients with AR + tumours had a poorer prognosis but AR was not an independent prognostic factor in multivariate analysis. Prospective data are needed to conclude on the independent prognostic value of this biomarker. basal-like tumours by gene expression profiling. 1 In a previous study, we found that AR + /FOXA1 + TNBC seemed to behave like luminal tumours with a morphological profile distinct from other TNBC, including AR + /FOXA1-tumours. 9 In this independent series, we confirmed these data and added the notion of a lower frequency of basal-like phenotype and BRCA1 promoter methylation in this population. The association of these two biomarkers easily identifies, in IHC, a particular subgroup of TNBC. In our study, patients with AR + /FOXA1 + tumours had a shorter RFS and OS than other TNBC and AR/ FOXA1 co-expression was an independent prognostic factor for OS. Usually, in TNBC, relapses occur in the first 3 years of followup. These data are confirmed in our population for patients with AR-or AR + /FOXA1-tumours. Interestingly, in our study, relapses could occur even after a long follow-up in patients with AR + /FOXA1 + tumours, like in patients with ER + /HER2tumours. Anti-androgen therapies are under development in AR + TNBC. For AR + /FOXA1 + TNBC with a risk of late relapse, an adjuvant anti-androgen therapy could be considered, like in ER + tumours.
anti-androgen therapy and PIK3 inhibitor. FOXA1 will allow the identification of a TNBC subgroup that could retrieve more benefit from this therapeutic strategy.
Stromal TILs constitute a robust and an independent prognostic marker in TNBC treated with chemotherapy. In our series, TILs density was also a strong and independent prognostic factor for both RFS and OS. To the best of our knowledge, no study has yet evaluated the correlation between TILs and AR expression. In our study, there was no difference between AR + and AR-tumours on this point, including in the subgroup analyses according to FOXA1 status.
Cancer cell could escape from immune surveillance by upregulating PD-L1 expression. The rate of PD-L1 expression in TNBC is extremely variable across studies, due to the lack of standard assay and its prognostic value remains controversial. In our study, 56.1% of TNBC expressed PD-L1 and it was not an independent prognostic marker in multivariate analysis. The 1% threshold for positivity was selected based on data demonstrating a clinical response to PD-L1 inhibition at this expression level in some cancers. There was no difference regarding PD-L1 expression and AR status in our series, contrary to the results of Tung et al. 7 reporting a 2.6-fold higher PD-L1 rate in their AR + TNBC population. 7 A higher rate of PD-L1 expression was seen in the AR + /FOXA1-subgroup (76.7% vs 49.6% in AR + /FOXA1 + and 55.2% in AR-). Among patients with a PD-L1 + tumour, we found significantly poorer RFS and OS in case of AR/FOXA1 coexpression. It could be interesting to specifically evaluate the benefit of anti-PD1 or PD-L1 targeted therapies in association with an anti-androgen in this subgroup.
In this large homogeneous series with a long follow-up, 42% of non-metastatic TNBC presented an AR/FOXA1 co-expression. AR + /FOXA1 + expression defines a luminal-like TNBC subgroup with a worse outcome compared to other TNBC and a higher risk of late recurrences mimicking luminal tumours behaviour and suggesting a putative role for anti-androgen therapies. This subgroup appears enriched in PIK3CA mutations, advocating for PIK3 inhibitors evaluation, alone or in association with antiandrogens, in this specific subgroup. All these data suggest the need for a concomitant evaluation of AR/FOXA1 to identify a specific subgroup of patients with TNBC who could benefit from targeted therapies.
Note: This work is published under the standard license to publish agreement. After 12 months the work will become freely available and the license terms will switch to a Creative Commons Attribution 4.0 International (CC BY 4.0).