Breast Cancer Res Treat (2015) 150:335–346 DOI 10.1007/s10549-015-3335-1
PRECLINICAL STUDY
Clinical and biological significance of glucocorticoid receptor (GR) expression in breast cancer Rezvan Abduljabbar1,2 • Ola H. Negm3,4 • Chun-Fui Lai5 • Dena A. Jerjees1,6 • Methaq Al-Kaabi1,7 • Mohamed R. Hamed3,4 • Patrick J. Tighe3 • Lakjaya Buluwela5 Abhik Mukherjee1 • Andrew R. Green1 • Simak Ali5 • Emad A. Rakha1 • Ian O. Ellis1
•
Received: 3 March 2015 / Accepted: 4 March 2015 / Published online: 12 March 2015 Ó Springer Science+Business Media New York 2015
Abstract The glucocorticoid receptor (GR) is a member of the nuclear receptor superfamily of transcription factors, which exerts anti-proliferative and anti-apoptotic activities. The GR is expressed in a large proportion of breast cancer (BC) although levels generally decrease during cancer progression. This study aimed to determine the clinical and biological significance of GR expression using a large series of early-stage BC with long-term follow-up and BC cell lines. Immunohistochemistry was used to assess the expression of GR in 999 cases of primary invasive BC prepared as tissue microarrays. Reverse phase protein microarray was used to assess the expression of GR in MCF7 and MDA-MB-231 cell lines. Nuclear expression of GR was observed in 61.6 % of breast tumours and was associated with features of good prognosis including smaller
tumour size and lower grade with less pleomorphism and low mitotic count. GR expression was positively correlated with expression of oestrogen (ER) and progesterone receptors. In ER-positive tumours, GR was associated with other features of favourable outcome including FOXA1, GATA3 and BEX1 expression, while low GR expression was associated with high Ki67, p53 and CD71 expression. GR expression is associated with features of good outcome but does not provide prognostic information independent of size, stage and grade. Understanding the receptor and its effects on BC behaviour is essential for avoiding any unwanted effects from the use of glucocorticoids in routine oncology practice. Keywords Glucocorticoid receptor Breast cancer Nuclear receptor Immunohistochemistry
Ola H. Negm: First joint author. & Rezvan Abduljabbar [email protected] 1
Division of Cancer and Stem Cell, Department of Histopathology, The University of Nottingham, City Hospital Campus, Nottingham, UK
2
Department of Oncology, Azadi Teaching Hospital, 1014 AM Duhok, Kurdistan, Iraq
3
Department of Immunology, School of Life Sciences, University of Nottingham, Nottingham, UK
4
Medical Microbiology and Immunology Department, Faculty of Medicine, Mansoura University, Mansoura, Egypt
5
Department of Surgery and Cancer, Imperial College London, London W12 0NN, UK
6
Department of Pathology, Mosul Medical School, University of Mosul, Mosul, Iraq
7
Department of Pathology, College of Medicine, AlMustansiriya University, Baghdad, Iraq
Introduction Several studies have demonstrated that stress is associated with development of breast cancer (BC) [1–3]. Glucocorticoid (GC) synthesis is enhanced following stressful conditions leading and acting mostly through the glucocorticoid receptor (GR) to regulate inflammatory and immune responses, as well as cellular proliferation and apoptosis [4]. The human GR gene, also known as NR3C1 (nuclear receptor subfamily 3, group C, member 1), is located on chromosome 5 (5q31) [5, 6]. Five isoforms of the receptor (GR-a, GR-b, GR-c, GR-P and GR-A) have been described, of which GR-a is the main isoform responsible for GR-mediated transcriptional activities [7] and is the focus of this study. Nuclear receptors consist of highly conserved DNA- and ligand-binding domains, connected by a hinge region [8]. The main ligands of GR are glucocorticoids
123
336
[cortisol and dexamethasone (DEX)], in addition to other agonists such as progesterone and dehydroepiandrosterone (DHEA). In the absence of ligand, the GR resides in the cell cytosol complexed with a variety of proteins such as heat shock protein 90 (HSP90) and HSP70 and translocates to the nucleus upon ligand binding, where it regulates expression of target genes. The receptor is ubiquitously expressed in many tissues and is essential in the regulation of development, metabolism, immune and anti-inflammatory responses [9, 10]. Activated GR modulates the expression of many genes that are involved in cellular metabolism and frequently affect the transcription of other steroid receptors [11]. GR has also been shown to be involved in non-genomic signal transduction. A variety of biological functions of GR in BC have been described which include anti-apoptotic action, through the activation of NF-RB [12], or AP-1 [13, 14]; or anti-proliferative effects through direct inhibitory role or inhibition of insulin or insulin-like growth factor-induced cell growth [15, 16]. Several nuclear receptors have been implicated in cancer development [11, 17–19]. Studies utilising ligandbinding assays and immunohistochemistry show that approximately 50–70 % of invasive breast cancers express GR [20, 21]. Being expressed to this extent in BC, and since it is a stress-induced hormone-activated receptor, it has been suggested that stress may have an association with BC, although there are no strong conclusions [1, 22, 23]. In normal breast tissue, GR is strongly expressed in the nuclei and the cytoplasm of luminal epithelial cells and in the nuclei of myoepithelial cells, and occasionally expressed in the nuclei of stromal and endothelial cells [21, 24]. Despite extensive studies of GR, conflicting results have been reported [25], these range from anti-proliferative and anti-apoptotic activity in BC cell lines to having no effect on prognosis and aggressiveness of BC. This prompts further investigation of the role of this receptor in the pathogenesis and behaviour of BC. The aim of this study was to investigate the expression of GR in a large and wellcharacterised early-stage BC samples using IHC, in order to determine the association between GR, pathological features, tumour phenotypes and clinical outcomes.
Breast Cancer Res Treat (2015) 150:335–346
patients aged 70 years and under who presented with tumours less than 5 cm in diameter and who were uniformly treated and for which we have well-documented clinical and pathological data. The median age of the patients was 54 years (IQR 47–62 years), 53.5 % of the patients had tumour of equal to or larger than 2 cm in size. 86 % were ductal carcinoma of no special type. During the course of the follow-up, 16.9 % of the patients developed local recurrence, 14.4 % developed regional recurrence and 36.7 % developed distant metastases (median of 107 months (IQR: 40–148 months)). A total of 27 % of the patients were of good prognosis based on the Nottingham prognostic index (NPI) at diagnosis. Based on Nottingham Prognostic Index Plus (NPI?) biological classification [26], 159 (27.0 %) were luminal A, 117 (19.9 %) were luminal N, 96 (16.3 %) were luminal B, 44 (7.5 %) were HER2?/ER?, 57 (9.7 %) were HER2?/ER-, 60 (10.2 %) were basal p53 altered and 56 (9.5 %) were basal p53 normal. A summary of patient characteristics is shown in Table 1. Data on a large panel of relevant biomarkers including ER, PgR, HER2, cytokeratins (CK) and markers downstream of ER [27–37] were available, and correlations with those biological features were also investigated. This study was approved by Nottingham Research Ethics Committee 2. Immunohistochemistry
Patient characteristics and selection for TMA construction
A rabbit IgG polyclonal GR antibody (SC-1003; Santa Cruz Biotechnology) was used at an optimised dilution of 1:80. Antibody detection was performed using the polymer detection system (Novolink) as per manufacturer’s instructions. Antigen retrieval was performed using microwave heating (20 min) in citrate buffer solution (pH 6.0). Primary antibody was incubated for 1 h at room temperature. The H-score method was used for evaluation of the immunostaining, by multiplying the intensity of the staining (0: no staining, 1: weak, 2: moderate and 3: strong staining) with the percentage of the tumour stained. The minimum score was 0 and the maximum was 300 [38]. At least a second scorer evaluated [30 % of the cores independently and the intra-class reliability coefficient was 0.8. X-tile software (University of Yale, Yale, USA) [39] was used to classify the tumours with a GR H-score of less than or equal to 10 as negative, and those with an H-score of more than 10 as positive.
Study cohort
Reverse phase protein microarray (RPPA)
The cohort was derived from the Nottingham-Tenovus primary operable breast carcinoma series and includes
To compare the protein expression of GR, ER and PgR, two different cell lines were used in this study; MCF7 (ER
Methods
123
Breast Cancer Res Treat (2015) 150:335–346
337
Table 1 Clinical and pathological features of the study cohort Clinicopathological characteristics
N (%)
Age B50
416 (41.9)
[50
578 (58.1)
Menopausal status Pre-
410 (41.6)
Post-
576 (58.4)
Tumour size (cm) \2.0
474 (47.8)
C2.0
518 (52.2)
Tumour type Ductal (inc. mixed) Lobular
846 (85.5) 80 (8.1)
Medullary-like
20 (2.0)
Special types
44 (4.4)
NPI GPG
284 (28.6)
MPG
535 (53.9)
PPG
173 (17.4)
Stage 1
603 (60.9)
2
306 (30.9)
3
81 (8.2)
Grade 1
153 (15.5)
2
324 (32.7)
3
513 (51.8)
VI Definite
635 (64.3)
Negative/probable
352 (35.7)
Distant metastases No
626 (63.3)
Yes
363 (36.7)
NPI? Biological classes Luminal A
159 (27.0)
Luminal N
117 (19.9)
Luminal B
96 (16.3)
HER2?/ER?
44 (7.5)
HER?/ER-
57 (9.7)
Basal p53 altered
60 (10.2)
Basal p53 normal
56 (9.5)
NPI Nottingham prognostic index, GPG good prognostic group; MPG moderate prognostic group; PPG poor prognostic group; VI vascular invasion
(Genetix, UK), where each sample was two-fold serially diluted three times in 19 SDS buffer. Samples were robotically spotted in duplicate onto nitrocellulose-coated glass slides (Grace Bio-labs, USA) using a microarrayer (MicroGrid 610, Digilab, Marlborough, MA, USA). Slides were incubated overnight in blocking solution (0.2 % I-block (Tropix, Bedford, MA, USA), 0.1 % Tween-20 in PBS) at 4 °C with shaking. After washing three times for 5 min each with TBS Tween, the slides were incubated with the GR antibody diluted 1:100 in antibody diluent with reducing background (DAKO). In addition, mouse bactin (Sigma Aldrich, UK) diluted 1:1000 in the same diluent was used as a housekeeping protein to control for protein loading. Following washing, as described before, the slides were incubated with diluted infrared (1:5000 in washing buffer) secondary antibodies that 800 CW antirabbit antibody and 700 CW anti-mouse antibodies for 30 min at room temperature in dark with shaking. Then slides were washed and dried by centrifugation at 5009g for 5 min and scanned with a Licor Odyssey scanner at 21 lm resolution at 800 nm (green) and 700 nm (red). The images were processed with Axon Genepix Pro6 Microarray Image Analysis software (Molecular Services Inc.) to obtain fluorescence data for each feature and generate gpr files. Protein signals were quantified with background subtraction and normalization to the internal housekeeping target using RPPanalyzer, a module within the R statistical language on the CRAN (http://www.cran.rproject.org/) [44]. Statistical analysis Statistical analysis was performed using SPSS v 21 (Chicago, IL, USA). Association with clinical and biological markers was examined using v2 test. Cut off points for the markers in this study were derived from previous publications [27, 33, 45]. Survival analysis was performed using Kaplan–Meier survival curves with log-rank test to determine significance. Only patients that died from BC were included in the analysis and those who died from other causes were censored. Univariate analysis was performed followed by Cox multivariate analysis whenever GR expression was of prognostic or predictive significance. p values of less than 0.05 were considered significant.
Results positive) and MDA-MB-231 cells (ER negative). RPPA methodology was done as previously described [40–43]. In brief, the lysates prepared in RIPA buffer were solubilised in 4XSDS sample buffer and loaded onto a 384-well plate
Expression of GR in breast cancer The expression of the GR was evident in the nucleus (Fig. 1) with sparse cytoplasmic staining of the malignant
123
338
cells. In the nucleus, the median expression was H-score 30 (IQR 0–80). Minimum H-score was 0 and maximum H-score was 280. Out of 999 cases, 617 (61.7 %) were GR positive (H-score [ 10). A total of 268 (26.8 %) cases showed complete absence of GR expression. In the ERpositive tumours, a similar distribution was observed: 228 (31.4 %) were GR negative, while 497 (68.6 %) were GR positive. Association of GR with clinicopathological and biological features
Breast Cancer Res Treat (2015) 150:335–346
The cohort was stratified based on ER expression. In the ER-positive subgroup, GR maintained its association with grade and its components, with lobular tumour type and expression of FOXA1, BEX1, HER2 and AGTR1 but lost its association with size, NPI and expression of PgR and GATA3. Positive GR was only associated with HER2?/ ER? NPI plus biological class (Tables 2, 3). In the ER-negative cohort, GR positivity was associated with lower pleomorphism (p = 0.006) and FOXA1 but not with any other clinical or biological feature. Outcome analysis
In the whole series, nuclear GR expression showed a negative association with tumour size (p = 0.013), tumour grade and its components, namely pleomorphism and mitosis (p \ 0.001), and lobular histological tumour type (p \ 0.001). Positive GR implies a good prognostic outcome using the NPI (p \ 0.001). There were positive associations between GR and ER, as well as between GR and ERregulated proteins including ER, PgR, FOXA1, GATA3 and BEX1. There was a negative association with HER2 expression, basal and triple negative phenotypes, the proliferation marker Ki67, CD71, AGTR1 and p53 (Tables 2, 3). Correlation with NPI plus biological classes shows that GR positivity correlates with luminal classes N and B while GR negativity in seen in the classes HER?/ER?, basal P53 altered and basal P53 normal (p \ 0.001).
Univariate analysis showed that in the ER-negative group positive nuclear GR is associated with shorter breast cancerspecific survival (BCSS) with median survival in the GRpositive group of 168 months (95 % CI 143–194 months) compared to median survival of 196 months (95 % CI 177–217 months) in the GR-negative group (Log rank = 4.09, p = 0.043) (Fig. 2a). This association was also observed in TN tumours whereby GR positivity conferred a shorter survival with a median of 162 months (95 % CI 128–195 months), while the median survival of the GRnegative group was 200 months (95 % CI 175–225 months) (Log rank = 4.22, p = 0.04) (Fig. 2b). However, Cox multivariate regression (Table 4) demonstrated that GR is not an independent predictor of survival when other classical
Fig. 1 The expression of GR in BC showing strong (a), moderate (b) and negative (c) intensities of GR expression in ductal carcinomas. (X200)
123
209 (36.3)
218 (42.1)
234 (38.8)
116 (37.9)
30 (37.0)
C2.0 Stage
1
2
3
12 (15.0)
12 (60.0)
14 (31.8)
Lobular
Medullary-like
Special-type
93 (26.1)
275 (46.8)
2
3
83 (26.3)
72 (36.7)
217 (47.5)
1
2
3
Mitosis
4 (16.7)
240 (52.5)
124 (63.3)
233 (73.7)
313 (53.2)
263 (73.9)
20 (83.3)
352 (59.2)
243 (40.8)
1
Pleomorphism
3
1 2
269 (52.4)
228 (70.4)
37 (69.8) 208 (64.8)
244 (47.6)
3
113 (73.9)
30 (68.2)
8 (40.0)
68 (85.0)
503 (59.5)
51 (63.0)
190 (62.1)
369 (61.2)
300 (57.9)
311 (65.6)
367 (63.7)
242 (59.0)
16 (30.2) 113 (35.2)
96 (29.6)
2
Tubules
40 (26.1)
1
Grade
343 (40.5)
Ductal
Tumour type
163 (34.4)
\2.0
Tumour size (cm)
168 (41.0)
363 (62.8)
215 (37.2)
Post-
249 (59.9)
Positive N (%)
167 (40.1)
Pre-
Menopausal status
[50 year
B50 year
Age
Negative N (%)
GR expression (whole series)
PRECLINICAL STUDY
Clinical and biological significance of glucocorticoid receptor (GR) expression in breast cancer Rezvan Abduljabbar1,2 • Ola H. Negm3,4 • Chun-Fui Lai5 • Dena A. Jerjees1,6 • Methaq Al-Kaabi1,7 • Mohamed R. Hamed3,4 • Patrick J. Tighe3 • Lakjaya Buluwela5 Abhik Mukherjee1 • Andrew R. Green1 • Simak Ali5 • Emad A. Rakha1 • Ian O. Ellis1
•
Received: 3 March 2015 / Accepted: 4 March 2015 / Published online: 12 March 2015 Ó Springer Science+Business Media New York 2015
Abstract The glucocorticoid receptor (GR) is a member of the nuclear receptor superfamily of transcription factors, which exerts anti-proliferative and anti-apoptotic activities. The GR is expressed in a large proportion of breast cancer (BC) although levels generally decrease during cancer progression. This study aimed to determine the clinical and biological significance of GR expression using a large series of early-stage BC with long-term follow-up and BC cell lines. Immunohistochemistry was used to assess the expression of GR in 999 cases of primary invasive BC prepared as tissue microarrays. Reverse phase protein microarray was used to assess the expression of GR in MCF7 and MDA-MB-231 cell lines. Nuclear expression of GR was observed in 61.6 % of breast tumours and was associated with features of good prognosis including smaller
tumour size and lower grade with less pleomorphism and low mitotic count. GR expression was positively correlated with expression of oestrogen (ER) and progesterone receptors. In ER-positive tumours, GR was associated with other features of favourable outcome including FOXA1, GATA3 and BEX1 expression, while low GR expression was associated with high Ki67, p53 and CD71 expression. GR expression is associated with features of good outcome but does not provide prognostic information independent of size, stage and grade. Understanding the receptor and its effects on BC behaviour is essential for avoiding any unwanted effects from the use of glucocorticoids in routine oncology practice. Keywords Glucocorticoid receptor Breast cancer Nuclear receptor Immunohistochemistry
Ola H. Negm: First joint author. & Rezvan Abduljabbar [email protected] 1
Division of Cancer and Stem Cell, Department of Histopathology, The University of Nottingham, City Hospital Campus, Nottingham, UK
2
Department of Oncology, Azadi Teaching Hospital, 1014 AM Duhok, Kurdistan, Iraq
3
Department of Immunology, School of Life Sciences, University of Nottingham, Nottingham, UK
4
Medical Microbiology and Immunology Department, Faculty of Medicine, Mansoura University, Mansoura, Egypt
5
Department of Surgery and Cancer, Imperial College London, London W12 0NN, UK
6
Department of Pathology, Mosul Medical School, University of Mosul, Mosul, Iraq
7
Department of Pathology, College of Medicine, AlMustansiriya University, Baghdad, Iraq
Introduction Several studies have demonstrated that stress is associated with development of breast cancer (BC) [1–3]. Glucocorticoid (GC) synthesis is enhanced following stressful conditions leading and acting mostly through the glucocorticoid receptor (GR) to regulate inflammatory and immune responses, as well as cellular proliferation and apoptosis [4]. The human GR gene, also known as NR3C1 (nuclear receptor subfamily 3, group C, member 1), is located on chromosome 5 (5q31) [5, 6]. Five isoforms of the receptor (GR-a, GR-b, GR-c, GR-P and GR-A) have been described, of which GR-a is the main isoform responsible for GR-mediated transcriptional activities [7] and is the focus of this study. Nuclear receptors consist of highly conserved DNA- and ligand-binding domains, connected by a hinge region [8]. The main ligands of GR are glucocorticoids
123
336
[cortisol and dexamethasone (DEX)], in addition to other agonists such as progesterone and dehydroepiandrosterone (DHEA). In the absence of ligand, the GR resides in the cell cytosol complexed with a variety of proteins such as heat shock protein 90 (HSP90) and HSP70 and translocates to the nucleus upon ligand binding, where it regulates expression of target genes. The receptor is ubiquitously expressed in many tissues and is essential in the regulation of development, metabolism, immune and anti-inflammatory responses [9, 10]. Activated GR modulates the expression of many genes that are involved in cellular metabolism and frequently affect the transcription of other steroid receptors [11]. GR has also been shown to be involved in non-genomic signal transduction. A variety of biological functions of GR in BC have been described which include anti-apoptotic action, through the activation of NF-RB [12], or AP-1 [13, 14]; or anti-proliferative effects through direct inhibitory role or inhibition of insulin or insulin-like growth factor-induced cell growth [15, 16]. Several nuclear receptors have been implicated in cancer development [11, 17–19]. Studies utilising ligandbinding assays and immunohistochemistry show that approximately 50–70 % of invasive breast cancers express GR [20, 21]. Being expressed to this extent in BC, and since it is a stress-induced hormone-activated receptor, it has been suggested that stress may have an association with BC, although there are no strong conclusions [1, 22, 23]. In normal breast tissue, GR is strongly expressed in the nuclei and the cytoplasm of luminal epithelial cells and in the nuclei of myoepithelial cells, and occasionally expressed in the nuclei of stromal and endothelial cells [21, 24]. Despite extensive studies of GR, conflicting results have been reported [25], these range from anti-proliferative and anti-apoptotic activity in BC cell lines to having no effect on prognosis and aggressiveness of BC. This prompts further investigation of the role of this receptor in the pathogenesis and behaviour of BC. The aim of this study was to investigate the expression of GR in a large and wellcharacterised early-stage BC samples using IHC, in order to determine the association between GR, pathological features, tumour phenotypes and clinical outcomes.
Breast Cancer Res Treat (2015) 150:335–346
patients aged 70 years and under who presented with tumours less than 5 cm in diameter and who were uniformly treated and for which we have well-documented clinical and pathological data. The median age of the patients was 54 years (IQR 47–62 years), 53.5 % of the patients had tumour of equal to or larger than 2 cm in size. 86 % were ductal carcinoma of no special type. During the course of the follow-up, 16.9 % of the patients developed local recurrence, 14.4 % developed regional recurrence and 36.7 % developed distant metastases (median of 107 months (IQR: 40–148 months)). A total of 27 % of the patients were of good prognosis based on the Nottingham prognostic index (NPI) at diagnosis. Based on Nottingham Prognostic Index Plus (NPI?) biological classification [26], 159 (27.0 %) were luminal A, 117 (19.9 %) were luminal N, 96 (16.3 %) were luminal B, 44 (7.5 %) were HER2?/ER?, 57 (9.7 %) were HER2?/ER-, 60 (10.2 %) were basal p53 altered and 56 (9.5 %) were basal p53 normal. A summary of patient characteristics is shown in Table 1. Data on a large panel of relevant biomarkers including ER, PgR, HER2, cytokeratins (CK) and markers downstream of ER [27–37] were available, and correlations with those biological features were also investigated. This study was approved by Nottingham Research Ethics Committee 2. Immunohistochemistry
Patient characteristics and selection for TMA construction
A rabbit IgG polyclonal GR antibody (SC-1003; Santa Cruz Biotechnology) was used at an optimised dilution of 1:80. Antibody detection was performed using the polymer detection system (Novolink) as per manufacturer’s instructions. Antigen retrieval was performed using microwave heating (20 min) in citrate buffer solution (pH 6.0). Primary antibody was incubated for 1 h at room temperature. The H-score method was used for evaluation of the immunostaining, by multiplying the intensity of the staining (0: no staining, 1: weak, 2: moderate and 3: strong staining) with the percentage of the tumour stained. The minimum score was 0 and the maximum was 300 [38]. At least a second scorer evaluated [30 % of the cores independently and the intra-class reliability coefficient was 0.8. X-tile software (University of Yale, Yale, USA) [39] was used to classify the tumours with a GR H-score of less than or equal to 10 as negative, and those with an H-score of more than 10 as positive.
Study cohort
Reverse phase protein microarray (RPPA)
The cohort was derived from the Nottingham-Tenovus primary operable breast carcinoma series and includes
To compare the protein expression of GR, ER and PgR, two different cell lines were used in this study; MCF7 (ER
Methods
123
Breast Cancer Res Treat (2015) 150:335–346
337
Table 1 Clinical and pathological features of the study cohort Clinicopathological characteristics
N (%)
Age B50
416 (41.9)
[50
578 (58.1)
Menopausal status Pre-
410 (41.6)
Post-
576 (58.4)
Tumour size (cm) \2.0
474 (47.8)
C2.0
518 (52.2)
Tumour type Ductal (inc. mixed) Lobular
846 (85.5) 80 (8.1)
Medullary-like
20 (2.0)
Special types
44 (4.4)
NPI GPG
284 (28.6)
MPG
535 (53.9)
PPG
173 (17.4)
Stage 1
603 (60.9)
2
306 (30.9)
3
81 (8.2)
Grade 1
153 (15.5)
2
324 (32.7)
3
513 (51.8)
VI Definite
635 (64.3)
Negative/probable
352 (35.7)
Distant metastases No
626 (63.3)
Yes
363 (36.7)
NPI? Biological classes Luminal A
159 (27.0)
Luminal N
117 (19.9)
Luminal B
96 (16.3)
HER2?/ER?
44 (7.5)
HER?/ER-
57 (9.7)
Basal p53 altered
60 (10.2)
Basal p53 normal
56 (9.5)
NPI Nottingham prognostic index, GPG good prognostic group; MPG moderate prognostic group; PPG poor prognostic group; VI vascular invasion
(Genetix, UK), where each sample was two-fold serially diluted three times in 19 SDS buffer. Samples were robotically spotted in duplicate onto nitrocellulose-coated glass slides (Grace Bio-labs, USA) using a microarrayer (MicroGrid 610, Digilab, Marlborough, MA, USA). Slides were incubated overnight in blocking solution (0.2 % I-block (Tropix, Bedford, MA, USA), 0.1 % Tween-20 in PBS) at 4 °C with shaking. After washing three times for 5 min each with TBS Tween, the slides were incubated with the GR antibody diluted 1:100 in antibody diluent with reducing background (DAKO). In addition, mouse bactin (Sigma Aldrich, UK) diluted 1:1000 in the same diluent was used as a housekeeping protein to control for protein loading. Following washing, as described before, the slides were incubated with diluted infrared (1:5000 in washing buffer) secondary antibodies that 800 CW antirabbit antibody and 700 CW anti-mouse antibodies for 30 min at room temperature in dark with shaking. Then slides were washed and dried by centrifugation at 5009g for 5 min and scanned with a Licor Odyssey scanner at 21 lm resolution at 800 nm (green) and 700 nm (red). The images were processed with Axon Genepix Pro6 Microarray Image Analysis software (Molecular Services Inc.) to obtain fluorescence data for each feature and generate gpr files. Protein signals were quantified with background subtraction and normalization to the internal housekeeping target using RPPanalyzer, a module within the R statistical language on the CRAN (http://www.cran.rproject.org/) [44]. Statistical analysis Statistical analysis was performed using SPSS v 21 (Chicago, IL, USA). Association with clinical and biological markers was examined using v2 test. Cut off points for the markers in this study were derived from previous publications [27, 33, 45]. Survival analysis was performed using Kaplan–Meier survival curves with log-rank test to determine significance. Only patients that died from BC were included in the analysis and those who died from other causes were censored. Univariate analysis was performed followed by Cox multivariate analysis whenever GR expression was of prognostic or predictive significance. p values of less than 0.05 were considered significant.
Results positive) and MDA-MB-231 cells (ER negative). RPPA methodology was done as previously described [40–43]. In brief, the lysates prepared in RIPA buffer were solubilised in 4XSDS sample buffer and loaded onto a 384-well plate
Expression of GR in breast cancer The expression of the GR was evident in the nucleus (Fig. 1) with sparse cytoplasmic staining of the malignant
123
338
cells. In the nucleus, the median expression was H-score 30 (IQR 0–80). Minimum H-score was 0 and maximum H-score was 280. Out of 999 cases, 617 (61.7 %) were GR positive (H-score [ 10). A total of 268 (26.8 %) cases showed complete absence of GR expression. In the ERpositive tumours, a similar distribution was observed: 228 (31.4 %) were GR negative, while 497 (68.6 %) were GR positive. Association of GR with clinicopathological and biological features
Breast Cancer Res Treat (2015) 150:335–346
The cohort was stratified based on ER expression. In the ER-positive subgroup, GR maintained its association with grade and its components, with lobular tumour type and expression of FOXA1, BEX1, HER2 and AGTR1 but lost its association with size, NPI and expression of PgR and GATA3. Positive GR was only associated with HER2?/ ER? NPI plus biological class (Tables 2, 3). In the ER-negative cohort, GR positivity was associated with lower pleomorphism (p = 0.006) and FOXA1 but not with any other clinical or biological feature. Outcome analysis
In the whole series, nuclear GR expression showed a negative association with tumour size (p = 0.013), tumour grade and its components, namely pleomorphism and mitosis (p \ 0.001), and lobular histological tumour type (p \ 0.001). Positive GR implies a good prognostic outcome using the NPI (p \ 0.001). There were positive associations between GR and ER, as well as between GR and ERregulated proteins including ER, PgR, FOXA1, GATA3 and BEX1. There was a negative association with HER2 expression, basal and triple negative phenotypes, the proliferation marker Ki67, CD71, AGTR1 and p53 (Tables 2, 3). Correlation with NPI plus biological classes shows that GR positivity correlates with luminal classes N and B while GR negativity in seen in the classes HER?/ER?, basal P53 altered and basal P53 normal (p \ 0.001).
Univariate analysis showed that in the ER-negative group positive nuclear GR is associated with shorter breast cancerspecific survival (BCSS) with median survival in the GRpositive group of 168 months (95 % CI 143–194 months) compared to median survival of 196 months (95 % CI 177–217 months) in the GR-negative group (Log rank = 4.09, p = 0.043) (Fig. 2a). This association was also observed in TN tumours whereby GR positivity conferred a shorter survival with a median of 162 months (95 % CI 128–195 months), while the median survival of the GRnegative group was 200 months (95 % CI 175–225 months) (Log rank = 4.22, p = 0.04) (Fig. 2b). However, Cox multivariate regression (Table 4) demonstrated that GR is not an independent predictor of survival when other classical
Fig. 1 The expression of GR in BC showing strong (a), moderate (b) and negative (c) intensities of GR expression in ductal carcinomas. (X200)
123
209 (36.3)
218 (42.1)
234 (38.8)
116 (37.9)
30 (37.0)
C2.0 Stage
1
2
3
12 (15.0)
12 (60.0)
14 (31.8)
Lobular
Medullary-like
Special-type
93 (26.1)
275 (46.8)
2
3
83 (26.3)
72 (36.7)
217 (47.5)
1
2
3
Mitosis
4 (16.7)
240 (52.5)
124 (63.3)
233 (73.7)
313 (53.2)
263 (73.9)
20 (83.3)
352 (59.2)
243 (40.8)
1
Pleomorphism
3
1 2
269 (52.4)
228 (70.4)
37 (69.8) 208 (64.8)
244 (47.6)
3
113 (73.9)
30 (68.2)
8 (40.0)
68 (85.0)
503 (59.5)
51 (63.0)
190 (62.1)
369 (61.2)
300 (57.9)
311 (65.6)
367 (63.7)
242 (59.0)
16 (30.2) 113 (35.2)
96 (29.6)
2
Tubules
40 (26.1)
1
Grade
343 (40.5)
Ductal
Tumour type
163 (34.4)
\2.0
Tumour size (cm)
168 (41.0)
363 (62.8)
215 (37.2)
Post-
249 (59.9)
Positive N (%)
167 (40.1)
Pre-
Menopausal status
[50 year
B50 year
Age
Negative N (%)
GR expression (whole series)
