Breast Cancer Research and Treatment

, Volume 129, Issue 1, pp 11–21

FOXO3a nuclear localisation is associated with good prognosis in luminal-like breast cancer

Authors

  • Hany Onsy Habashy
    • Division of Pathology, School of Molecular Medical SciencesUniversity of Nottingham
    • Department of Pathology, Faculty of MedicineMansoura University
  • Emad A. Rakha
    • Department of PathologyNottingham University Hospitals NHS Trust and University of Nottingham
  • Mohammed Aleskandarany
    • Division of Pathology, School of Molecular Medical SciencesUniversity of Nottingham
    • Department of PathologyMenoufia University
  • Mohamed AH Ahmed
    • Division of Pathology, School of Molecular Medical SciencesUniversity of Nottingham
    • Department of Pathology, Faculty of MedicineSuez Canal University
  • Andrew R. Green
    • Division of Pathology, School of Molecular Medical SciencesUniversity of Nottingham
    • Division of Pathology, School of Molecular Medical SciencesUniversity of Nottingham
    • Department of HistopathologyNottingham University Hospitals NHS Trust and University of Nottingham
  • Desmond G. Powe
    • Department of PathologyNottingham University Hospitals NHS Trust and University of Nottingham
Preclinical study

DOI: 10.1007/s10549-010-1161-z

Cite this article as:
Habashy, H.O., Rakha, E.A., Aleskandarany, M. et al. Breast Cancer Res Treat (2011) 129: 11. doi:10.1007/s10549-010-1161-z

Abstract

Oestrogen receptor (ER)-positive breast cancer (BC) constitutes a heterogeneous group of tumours with regard to outcome and response to therapy. Accurate stratification of ER-positive BC according to risk of relapse and response to therapy will be achieved through an improved understanding of ER and ER-related biological pathways. Recent studies have identified Forkhead box O3a (FOXO3a) transcription factor as an intracellular mediator of ERα expression and as an important downstream target of the Akt/PI3K pathway indicating a biological and potential clinical role for FOXO3a in ER-positive BC. In this study, we investigated the clinical relevance and biological associations of FOXO3a protein expression, using tissue microarrays and immunohistochemistry, in a large series of patients with invasive breast cancer. FOXO3a protein expression showed both nuclear and/or cytoplasmic staining patterns. FOXO3a predominant nuclear expression was positively associated with biomarkers of good prognosis including PgR, FOXA1 and p27 expression. There was an inverse association with mitotic counts, MIB1 growth fraction, C-MYC and PIK3CA expression. With respect to patient outcome, FOXO3a nuclear localisation was associated with longer BC specific survival (P < 0.001) and longer distant metastasis free interval (P = 0.001), independently of the well-established breast cancer prognostic factors. In conclusion, our results demonstrate the biological and prognostic role of FOXO3a protein expression and its subcellular localisation in ER-positive/luminal-like BC possibly through its involvement in controlling cell proliferation.

Keywords

Breast carcinomaFOXO3aOestrogen receptorPrognosisImmunohistochemistryLuminal-like

Introduction

Oestrogen receptor (ER)-positive tumours comprise the majority of breast cancers, accounting for 60–70% of cases and are generally expected to show good response to hormonal treatment and favourable clinical outcome [1]. However, ER-positive tumours are heterogeneous with respect to their clinical behaviour and biology. By studying novel biomarkers with strong association with ER signalling it should be possible to gain a better understanding of the biology of ER-positive disease.

FOXO3a (FKHRL1) gene belongs to the forkhead family of transcription factors [2] and their activity is regulated by several post-translational modifications, including phosphorylation and acetylation [3]. Zou et al. [4] have reported that FOXO3a can suppress ERα-dependent breast cancer cell proliferation and tumourigenesis in the MCF-7 breast cancer cell line, suggesting a crosstalk between FOXO3a and ER signalling pathways in ER-dependent breast cancer [4]. Other studies demonstrated FOXO3a has an important intracellular mediator role in ERα expression, suggesting possible therapeutic intervention [5]. Importantly, FOXO3a is a downstream target in the Akt/PI3K pathway and when phosphorylated, is prevented from translocating to the nucleus resulting in its loss of functional activity. In contrast, FOXO3a dephosphorylation leads to nuclear localisation and subsequent target gene activation [610]. Therefore, as a target within the Akt/PI3K signalling pathway FOXO3a regulates the expression of proapoptotic genes, cell cycle-regulated genes, and genes that control cellular homeostasis [6, 11]. Alvarez et al. [12] suggested that an efficient mitotic programme depends on downregulation of Akt/PI3K and consequent induction of FOXO3a transcriptional activity. However, there is also evidence that an Akt-independent mechanism of FOXO3a regulation exists. In vitro co-transfection of FOXO3a and IKK resulted in strong inhibition of FOXO3a activity independent of Akt [13].

In breast cancer, FOXO3a activity has been shown to elevate p27 resulting in cell cycle arrest [14]. Furthermore, nuclear FOXO3a can induce cellular apoptosis through upregulation of Fas ligand (Fas-L) [6] and Bim [15] and has been implicated in resistance to oxidative stress and longevity [16]. Other studies have highlighted the importance of FOXO3a for maintenance of the hematopoietic stem cell pool [17, 18]. It has been reported that activation of FOXO3a could induce p53-dependent apoptosis even in cells with a transcriptionally inactive p53 [19].

FOXO3a may have therapeutic implications because some clinical anticancer treatments target FOXO3a through three main oncogenic kinases (Akt, IKK and ERK) [2, 20]. For instance, nuclear localisation of FOXO3a could potentially improve the effectiveness of anti-EGFR agents such as gefitinib by mediating proliferative arrest [21]. Gefitinib treatment causes cell cycle arrest and induces apoptosis due to the effects of FOXO3a dephosphorylation and nuclear translocation at Akt sites. In contrast, resistant cells show phosphorylated FOXO3a are restricted to the cytoplasm [21]. Furthermore, up-regulation of FOXO3a by paclitaxel has been reported to increase Bim mRNA and protein level with subsequent induction of apoptosis in breast cancer cells [22, 23].

The value of FOXO3a as a prognostic biomarker for ER-positive luminal-like breast cancer remains unclear. Therefore, in this study, we have investigated the clinical relevance and biological associations of FOXO3a protein expression in a large series of patients with luminal-like ER-positive invasive breast cancers using high-throughput tissue microarrays (TMAs) and immunohistochemistry.

Materials and methods

Patient selection and TMA construction

This study was approved by the Nottingham Research Ethics Committee 2 under the title “Development of a molecular genetics classification of breast cancer”.

Formalin fixed paraffin embedded (FFPE) TMAs were prepared from a series of cases of primary operable (stage I and II) breast carcinoma cases from patients age <70 presenting consecutively to the Nottingham Breast Unit with tumours of less than 5 cm diameter between 1988 and 1998 as previously reported [24]. This consecutive patient series is well characterised and contains patients’ clinical and pathological data including patients’ age, histologic tumour type, primary tumour size, lymph node status, mitotic count and histologic grade [25], Nottingham prognostic index (NPI) [26], vascular invasion (VI), therapy and follow-up data. In addition, data on a large panel of biomarkers with strong relevance to breast cancer were available including oestrogen receptor-α (ER), ER-related genes (progesterone receptor (PgR), androgen receptor (AR), FHIT, FOXA1, CARM1 and PELP1), proliferation, apoptosis and cell cycle-related genes (BCL2, p27, p53, C-MYC and MIB1), and biomarkers related to PI3K pathway (PIK3CA) [24, 27, 28].

Patient management was based on the Nottingham Prognostic Index (NPI) score and ER status as previously described [27, 29], patients within the good prognostic NPI group (≤3.4) did not receive adjuvant systemic therapy. Hormonal therapy (Tamoxifen ± Zoladex if premenopausal) was given to patients with ER-positive tumours and NPI scores of >3.4. Pre-menopausal patients with moderate and poor prognostic NPI groups were given chemotherapy (cyclophosphamide, methotrexate, and 5-fluorouracil). ER-positive postmenopausal patients with moderate or poor NPI were offered Hormonal therapy, while ER-negative patients received CMF if fit to receive these cytotoxic agents with no concurrent disease that were considered as potential contraindication to the use of chemotherapy.

Data are maintained on prospective basis for breast cancer specific survival (BCSS), development of distant metastases (DM) and/or locoregional tumour recurrence.

BCSS was defined as the time (in months) from the date of the primary surgical treatment to the time of death from breast cancer. Distant metastasis free interval DMFI was defined as the interval (in months) from the date of the primary surgical treatment to the date of development of the first distant metastasis. FOXO3a expression was assessed in the whole patient series (n = 934), and a cohort of ER-positive/luminal-like patients (n = 633).

Immunohistochemistry

Rabbit polyclonal antibody to FOXO3a (Antibody 9467, Cell Signalling Technology, Danvers, MA) was optimised at a working dilution of 1:50 using FFPE TMA sections and full-face sections of breast cancer tissue to assess the staining distribution. Immunohistochemical optimisation and staining of FOXO3a was performed using a DakoCytomation Techmate 500 plus (DakoCytomation, Cambridge, UK) immunostainer with a linked streptavidin biotin (LSAB) technique in accordance with the manufacturer’s instructions after microwave antigen retrieval in 0.01 M citrated buffer pH 6 and as previously described [24]. Negative controls were performed by omitting the primary antibody and substitution with a diluent. Sections were counterstained in haematoxylin and cover-slipped using DPX mounting medium.

Western blotting was performed on breast cancer cell lysates of the human breast cancer cell line MCF-7 to confirm the specificity of the FOXO3a antibody used in immunohistochemistry. MCF-7 cells were obtained from the American Type Culture Collection (Rockville, MD, USA) and cultured in RPMI 1640 medium in T75 flasks supplemented with 10% foetal calf serum (FCS), penicillin (100 IU/ml) and streptomycin (100 μg/ml). The sub-confluent cells were washed with PBS, then 30 μl of protease inhibitors (Sigma–Aldrich) were added to 470 μl of ice-cold lysis buffer (20 mM Tris, pH 7.5, 150 mM NaCl, 1% TritonX-100, 0.5% sodium deoxycholate, 1 mM EDTA, 0.1% SDS). Western blotting was done on the cell lysates to confirm the specificity of the antibody used in immunohistochemistry. Lysates (20 μg) were added to 4× SDS loading buffer with 5% β-mercaptoethanol (Sigma–Aldrich, UK) and denatured by heating at 100°C for 10 min prior to loading then added for 5 min into ice. Samples were subjected to sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) using a 10% resolving polyacrylamide gel and transferred onto a Hybond-P PVDF membrane (Amersham Bioscience, Buckinghamshire, UK). After blocking with 5% milk powder 0.1% TPBS (Tween 20 in PBS solution) for 60 min, the membrane was then incubated with 1:1000 dilution of the FOXO3a rabbit polyclonal antibody (9467) at 4°C overnight.

The membrane was washed with 0.1% PBS/Tween 20 3 times for 5 min each then incubated for 1 h at room temperature with a horseradish peroxidase-linked secondary antibody (Sigma–Aldrich) (1:4000, anti-rabbit) diluted with 5% milk powder PBS containing 0.1% Tween 20. After a further three washes, the membrane was visualised with enhanced chemiluminescence reagents (Amersham Bioscience, Buckinghamshire, UK). The monoclonal Anti-β-actin antibody (Sigma–Aldrich) in a dilution of 1:2000 against the ubiquitous β-actin protein was used.

Assessment of IHC staining results

Staining was initially evaluated on full face sections to assess the expression and the distribution of staining. Firstly, FOXO3a TMAs IHC staining results were categorised into negative and positive expression, regardless of FOXO3a localisation. Examination of the TMAs has shown that some cases showed nuclear pattern and others were mainly cytoplasmic. Since the expression pattern and localisation of FOXO3a protein expression show variable biological functions, we categorised the positive cases according to whether they showed predominant nuclear (N) or predominant cytoplasmic (C) localisation. Both patterns scored separately using the percentage of the positive cells in each TMA core. Cases were categorised as showing a nuclear or cytoplasmic pattern in 50% or more of the informative TMA core provided that there is more than 20% variation between both patterns. We have defined the cases with obvious overlap (less than 20% variation, n = 31) and were excluded to ensure a clear separation in their patterns of expression. The cases were scored without the knowledge of patient outcome.

The histochemical score (H-score) was used to assess the intensity of staining and the percentage of stained cells for FOXA1 [27], PELP1 [30], CARM1 and PIK3CA [31]. The H-score was used to assess staining intensity and percentage of stained tumour cells following immunohistochemistry. Staining intensity was scored 0, 1, 2 or 3 (negative to strong) and the percentage of positive cells at each intensity subjectively estimated to produce a final score in the range 0–300. For other biomarkers we used the percentage of the positive cells. Intensity of C-MYC staining was scored as negative, low, moderate or strong. MIB1 staining was done on full face breast cancer sections. Suppliers, dilutions, antigen retrieval methods, cutoffs and methods of staining assessment used for the biomarkers included in this study is summarised in Table 1.
Table 1

Suppliers, dilutions, antigen retrieval methods, cutoffs and methods of staining assessment used for the biomarkers included in this study

Antibody (clone)

Dilution

Source

Pretreatment

Cutoff

Hormone receptors and ER-related markers

 ER (clone 1D5)

1:80

DakoCytomation

Microwave

10%

 PR (clone PgR 636)

1:100

DakoCytomation

Microwave

10%

 AR (clone F39.4.1)

1:30

Biogenex

Microwave

10%

 FOXA1(clone2F83)

1:2000

Abcam, UK

Microwave

10a

 CARM1

1:300

Novus Biological

Microwave

30,150a

 PELP1

1:100

Novus Biological

No

5,170a

Tumour suppressor genes

 p53 (clone DO7)

1:50

Novocastra

Microwave

10%

 FHIT (cloneZR44)

1:600

Zymed Laboratories

5%

Cell cycle associated, proliferation and Akt/PIK3 pathway associated markers

 p27

1:40

Dako, Denmark

Microwave

50%

 Ki67 (MIB1)

1:100

Dako, UK

10%

 PIK3CA

1:50

Sigma–Aldrich

100a

 C-MYC(clone 9E100)

1:100

Abcam, UK

b

 BCL2 (clone 124)

1:100

Dako, UK

30%

aH-score

bScored as (0, absent; 1, weak; 2, moderate; 3, strong expression)

Statistical analysis

We have reported our results according to the REMARK criteria for reporting tumour marker prognostic studies [32, 33]. Statistical analysis was performed using SPSS 16 statistical software (SPSS Inc., Chicago, IL, USA). Association between FOXO3a staining patterns and different clinicopathological parameters was evaluated using Chi-square test or Fishers exact test. Survival analysis was estimated by the Kaplan–Meier plots and Log Rank test to assess significance. Multivariate Cox proportional hazard regression models were used to evaluate any independent prognostic effect of the variables with 95% confidence interval. A P value of <0.05 was considered to reflect a significance.

Results

The median age of the patients was 56 years (range 27–70). At the time of the primary diagnosis, Forty-seven percent of patients had tumours less than 2 cm in size and 31.5% had grade 2 tumours. During follow-up, 30.6% of the patients developed metastatic disease.

The specificity of the FOXO3a antibody was confirmed using Western blotting (Fig. 1). In normal breast tissue, FOXO3a expression was detected mainly in the nuclei of mammary epithelial cells (Fig. 2a). In the malignant tissues, FOXO3a showed both nuclear and cytoplasmic staining patterns but one pattern were obviously dominant.
https://static-content.springer.com/image/art%3A10.1007%2Fs10549-010-1161-z/MediaObjects/10549_2010_1161_Fig1_HTML.gif
Fig. 1

Western blotting analysis of MCF-7 cell lysates using the FOXO3a rabbit polyclonal antibody. Enhanced chemiluminescence was used to visualise the membrane. The expected band size ranges from 82 to 97 kD. Lane a FOXO3a, lane b β-actin

https://static-content.springer.com/image/art%3A10.1007%2Fs10549-010-1161-z/MediaObjects/10549_2010_1161_Fig2_HTML.jpg
Fig. 2

a FOXO3a expression in normal tissue (mainly nuclear). b Predominant FOXO3a nuclear expression. c Predominant FOXO3a cytoplasmic expression

In whole patient, about 23% showed predominant nuclear expression pattern and 34% cytoplasmic pattern while 3% showed both with less 20% difference in the predominant pattern. In ER-positive patient cohort, 24% showed nuclear expression pattern (Fig. 2b) and 31% showed cytoplasmic pattern (Fig. 2c).

Correlation of FOXO3a protein expression and other clinicopathological variables

The tumour-specific FOXO3a IHC staining characteristics were initially categorised into negative and positive expression (showing either nuclear or cytoplasmic expression), regardless of FOXO3a localisation. FOXO3a expression status did not show significant associations with the other clinicopathological variables including tumour grade, size, stage, NPI and vascular invasion (P > 0.05).

Subsequently, the cases were categorised according to the pattern of expression into nuclear and non-nuclear, the latter including cytoplasmic predominant localisation and negative expression. Analysis of whole patient series revealed that FOXO3a nuclear localisation is positively associated with low mitotic counts, lower grade tumour, less frequent development of distant metastasis (P < 0.001) and tumour recurrence (P = 0.001) (Table 2). Its expression was significantly associated with markers of good prognosis (Table 3).
Table 2

Associations between FOXO3a immunostaining patterns and various clinicopathological parameters in the whole series

Variable

Non-nuclear localisation

Predominant nuclear localisation

Total

χ2

P value

Age

 <40

59 (85.5)

10 (14.5)

69

5.276

0.153

 40–50

215 (78.2)

60 (21.8)

275

 51–60

234 (79.2)

61 (20.6)

295

 >60

194 (73.8)

69 (26.6)

263

Tumour size

 ≤2 cm

335 (78.1)

94 (21.9)

429

0.032

0.873

 >2 cm

367 (77.6)

106 (22.4)

473

Lymph node stage

 1 (Negative)

413 (76.6)

126 (23.4)

539

1.169

0.557

 2 (1–3 LN)

219 (79.1)

58 (20.9)

277

 3 (>3 LN)

68 (81)

16 (19)

84

Tumour grade

 1

97 (71.3)

39 (28.7)

136

9.115

0.010

 2

214 (74.6)

73 (25.4)

287

 3

391 (81.6)

88 (18.4)

479

Vascular invasion

 No/Probable

457 (77.9)

130 (22.10

587

0.003

1.000

 Definite

244 (77.7)

70 (22.3)

314

NPI

 Good

164 (73.5)

59 (26.5)

223

4.364

0.113

 Moderate

408 (78.3)

113 (21.7)

521

 Poor

131 (82.4)

28 (17.6)

159

Mitotic counts

 1

204 (73.6)

73 (26.4)

277

20.308

<0.001

 2

109 (67.7)

52 (32.2)

161

 3

365 (83.5)

72 (16.5)

437

DM

 No

451 (73.5)

163 (26.5)

614

21.375

<0.001

 Positive

242 (87.4)

35 (12.6)

277

Recurrence

 No

374 (73.9)

132 (26.1)

506

10.000

0.002

 Positive

314 (82.8)

65 (17.2)

379

Table 3

Associations between FOXO3a immunostaining patterns and other biomarkers in the whole series

Variable

Non-nuclear localisation N (%)

Predominant nuclear localisation N (%)

Total

χ2

P value

ER

 Negative

202 (83.8)

39 (16.20)

241

6.356

0.013

 Positive

466 (75.9)

148 (24.1)

614

PgR

 Negative

319 (84.4)

59 (15.6)

378

17.014

<0.001

 Positive

344 (72.6)

130 (27.4)

474

AR

 Negative

257 (82.5)

55 (17.5)

312

4.364

0.035

 Positive

378 (75.9)

120 (24.1)

498

FOXA1

 Negative

319 (85.1)

56 (14.9)

375

15.145

<0.001

 Positive

234 (73.1)

86 (26.9)

320

CARM1

 Negative/low

159 (80.7)

38 (19.30)

197

1.806

0.405

 Moderate

281 (81.2)

65 (18.8)

346

 Strong

97 (75.8)

31 (24.20

128

PELP1

 Negative/low

76 (74.5)

26 (25.5)

102

2.612

0.271

 Moderated

412 (80.7)

99 (19.3)

511

 Strong

69 (75.8)

22 (24.2)

91

p53

 Negative

474 (77.8)

135 (22.2)

609

0.080

0.856

 Positive

196 (78.8)

53 (21.3)

249

MIB1

 Low

126 (71.2)

51 (28.8)

177

5.180

0.028

 High

407 (79.5)

105 (20.5)

512

PIK3CA

 Negative

175 (70.9)

72 (29.1)

247

11.224

0.001

 Positive

444 (81.5)

101 (18.5)

545

p27

 Negative

318 (83.9)

61 (16.1)

379

11.229

0.001

 Positive

272 (73.9)

96 (26.1)

368

C-MYC

 Negative

87 (88.8)

11 (11.2)

98

8.113

0.044

 Low

203 (80.6)

49 (19.4)

252

 Moderate

203 (75.5)

66 (24.5)

269

 Strong

113 (80.10)

28 (19.9)

141

FHIT

 Negative

118 (88.7)

15 (11.3)

133

11.092

0.001

 Positive

475 (75.5)

154 (24.5)

629

BCL2

 Negative

236 (80.8)

56 (19.2)

292

1.896

0.193

 Positive

312 (76.5)

96 (23.5)

408

In the luminal-like cohort (ER cutoff 10%), FOXO3a nuclear expression pattern was associated with low mitotic counts (P = 0.008), and less frequent development of distant metastasis (P < 0.001) and tumour recurrence (P = 0.035) (Table 4). The nuclear pattern showed significant positive associations with molecular biomarkers associated with good prognosis including PgR (P = 0.004), FHIT (P = 0.008), FOXA1 (P < 0.001) and p27 ( = 0.004) expression. It also showed an inverse correlation with the expression of cell proliferation-related markers, MIB1 expression (P = 0.039) and with PIK3CA (P = 0.006) (Table 5).
Table 4

Associations between FOXO3a immunostaining patterns and various clinicopathological parameters in the luminal-like cohort

Variable

Non-nuclear localisation N (%)

Predominant nuclear localisation N (%)

Total

χ2

P value

Age

 <40

24 (72.7)

9 (27.3)

33

1.535

0.674

 40–50

135 (73.8)

48 (26.2)

183

 51–60

159 (78.7)

43 (21.3)

202

 >60

147 (75.4)

48 (24.6)

195

Tumour size

 ≤2 cm

231 (75.2)

76 (24.8)

307

0.126

0.777

 >2 cm

234 (76.5)

72 (23.5)

306

Lymph node stage

 1 (Negative)

269 (73.9)

95 (26.1)

364

2.135

0.344

 2 (1–3 LN)

153 (77.7)

44 (22.3)

197

 3 (>3 LN)

41 (82)

9 (18)

50

Tumour grade

 1

88 (71.5)

35 (28.5)

123

4.905

0.086

 2

186 (73.5)

67 (26.5)

253

 3

191 (80.6)

46 (19.34)

237

Vascular invasion

 No/Probable

302 (76.1)

95 (23.9)

397

0.040

0.844

 Definite

162 (75.3)

53 (24.70

215

NPI

 Good

142 (72.4)

54 (27.6)

196

5.078

0.079

 Moderate

247 (75.5)

80 (24.5)

327

 Poor

77 (84.6)

14 (15.40

91

Mitotic counts

 1

184 (73.3)

67 (26.7)

251

9.643

0.008

 2

90 (68.2)

42 (31.8)

132

 3

172 (82.3)

37 (17.7)

209

DM

 No

306 (71.5)

122 (28.5)

428

13.825

<0.001

 Positive

155 (85.6)

26 (14.4)

181

Recurrence

 No

251 (72.3)

96 (27.7)

347

4.627

0.035

 Positive

207 (79.9)

52 (20.1)

259

Table 5

Associations between FOXO3a immunostaining patterns and other biomarkers in the luminal-like cohort

Variable

Non-nuclear localisation N (%)

Predominant nuclear localisation N (%)

Total

χ2

P value

PgR

 Negative

123 (84.2)

23 (15.8)

146

7.968

0.004

 Positive

333 (72.7)

125 (27.3)

458

AR

 Negative

110 (79.1)

29 (20.9)

139

0.860

0.423

 Positive

326 (75.3)

107 (24.7)

433

FOXA1

 Negative

183 (84.7)

33 (15.3)

216

14.177

<0.001

 Positive

180 (70)

77 (30)

257

CARM1

 Negative/low

121 (78.1)

34 (21.9)

155

2.128

0.345

 Moderate

191 (79.6)

49 (20.4)

240

 Strong

44 (71)

18 (29)

62

PELP1

 Negative/low

57 (74)

20 (26)

77

2.136

0.344

 Moderated

278 (79)

74 (21)

352

 Strong

40 (71.4)

16 (28.6)

56

p53

 Negative

374 (76.8)

113 (23.2)

487

0.0366

0.545

 Positive

86 (74.1)

30 (25.9)

116

MIB1

 Low

102 (68.9)

46 (31.1)

148

4.580

0.039

 High

253 (78.1)

71 (21.9)

324

PIK3CA

 Negative

136 (69.4)

60 (30.6)

196

8.120

0.006

 Positive

276 (80.2)

68 (19.8)

344

p27

 Negative

162 (83.9)

31 (16.1)

193

8.438

0.004

 Positive

221 (72.7)

83 (27.3)

304

C-MYC

 Negative

56 (88.9)

7 (11.1)

63

6.756

0.080

 Low

137 (78.3)

38 (21.7)

175

 Moderate

138 (73.4)

50 (26.60)

188

 Strong

69 (79.3)

18 (20.7)

87

FHIT

 Negative

64 (87.7)

9 (12.3)

73

6.573

0.008

 Positive

353 (73.8)

125 (26.2)

478

BCL2

 Negative

92 (78.6)

25 (21.40)

117

0.351

0.616

 Positive

275 (76)

87 (24)

362

Patients’ outcome

In whole patient series, initial univariate analysis of FOXO3a expression status (as positive vs. negative) was not associated with BCSS (Log Rank (LR) = 0.005, P = 0.942) nor DMFI (LR = 0.015, P = 0.904) but when the localisation of expression was considered FOXO3a nuclear expression was associated with better outcome in terms of longer BCSS (LR = 24.079, P < 0.001) and longer DMFI (LR = 15.996, P < 0.001).

Univariate analysis of survival showed no associations between FOXO3a expression status (as positive vs. negative) and patient outcome in terms of BCSS (LR = 0.234, P = 0.628) (Fig. 3a) or DMFI (LR = 0.198, P = 0.656). In the luminal-like ER-positive cohort of patient (n = 633) (median follow up time = 126 months), FOXO3a nuclear localisation showed a significant association with both longer BCSS (LR = 15.813, P < 0.001) (Fig. 3b) and longer DMFI (LR = 11.836, P = 0.001) (Fig. 3c).
https://static-content.springer.com/image/art%3A10.1007%2Fs10549-010-1161-z/MediaObjects/10549_2010_1161_Fig3_HTML.gif
Fig. 3

a Kaplan–Meier plot of FOXO3a expression status regardless of its subcellular localisation shows no significant difference in BCSS. b Kaplan–Meier plot of FOXO3a nuclear verse non-nuclear protein expression with respect to BCSS. Nuclear expression was significantly associated with improved survival. c Kaplan–Meier plot of FOXO3a expression with respect of DMFI shows a significant longer DMFI in patients with nuclear FOXOa3 expression. d Kaplan–Meier plot of FOXO3a expression in non-treated cohort with respect of DMFI. Nuclear expression was significantly associated with improved survival in untreated patients. e Kaplan–Meier plot of FOXO3a expression patterns with respect of BCSS in the whole series. f Kaplan–Meier plot of FOXO3a expression patterns with respect of DMFI in the whole series

When we analysed patient survival using categorisation of the cohort into three groups: predominant nuclear, predominant cytoplasmic and negative, our results showed that subcellular localisation differences of FOXO3a are associated with survival differences. Specifically there is a contrast between nuclear and cytoplasmic expression localisation where nuclear pattern showed the most favourable BCSS (LR = 18.279, P < 0.001) (Fig. 3e) and longer DMFI (LR = 14.775, P = 0.001) (Fig. 3f).

According to systemic therapy groups

When systemic therapy was considered, similar associations of longer survival were found in the subgroup of ER-positive patients who did not receive adjuvant systemic therapy (n = 222) with regard DMFI (LR = 10.110, P = 0.001) (Fig. 3d) and in the subgroup of patients who received tamoxifen monotherapy (n = 221) with regards BCSS (LR = 5.201, P = 0.023).

Multivariate analyses

Since many potential prognostic factors may interact with specific therapies and therefore are compounded by the effect of adjuvant hormone therapy and chemotherapy, we have included the systemic therapy groups (given vs. not given) in the multivariate analysis together with the other well-established prognostic variables such as MIB1, PgR, tumour size, stage, grade to assess the prognostic independence of nuclear FOXO3a expression in the ER+ patient cohort.

FOXO3a nuclear expression was an independent prognostic factor for predicting better outcome in terms of longer BCSS (Hazard ratio (HR) = 0.392, P = 0.006, 95% CI = 0.202–0.760) (Table 6) and longer DMFI (HR = 0.530, P = 0.020, 95% CI = 0.310–0.906) (Table 7) in ER-positive/luminal-like breast cancer.
Table 6

Cox model of predictors of BCSS in the luminal-like breast cancer

Variable

P value

HR

95% CI

Lower

Upper

FOXO3a nuclear localisation

0.006

0.392

0.202

0.760

PgR expression

0.049

0.642

0.413

0.997

MIB1 expression

0.011

2.105

1.184

3.742

Tumour size

0.001

2.228

1.411

3.520

LN stage

<0.001

1.746

1.290

2.363

Tumour grade

0.006

1.629

1.152

2.302

Endocrine therapy

0.502

0.846

0.519

1.379

Chemotherapy

0.305

0.715

0.377

1.358

Table 7

Cox model of predictors of DM in the luminal-like breast cancer

Variable

P value

HR

95% CI

Lower

Upper

FOXO3a nuclear localisation

0.020

0.530

0.310

0.906

PgR expression

0.035

0.641

0.424

0.970

MIB1 expression

0.008

1.989

1.194

3.311

Tumour size

0.001

2.094

1.380

3.177

LN stage

<0.001

1.836

1.380

2.442

Tumour grade

0.048

1.375

1.003

1.887

Endocrine therapy

0.889

0.967

0.604

1.548

Chemotherapy

0.938

0.977

0.536

1.779

Discussion

Oestrogen receptor (ER) plays an important role in breast cancer progression and response to therapy and is typically expressed in the most frequent biological class of breast cancers. However, ER-positive breast cancer does not appear to be a homogenous group; some tumours respond to therapy and others do not. There is therefore a requirement for improved understanding of the biology of ER-positive breast cancer which could be achieved through studying ER-related pathways and their downstream targets. One of the important pathways that has been implicated in the pathogenesis of the poor prognosis ER-positive (luminal B) subtype is the Akt/PI3K pathway and its related genes [34]. FOXO3a is a key downstream target in this pathway which prompted us to study its expression and associations in breast cancer particularly in ER-positive/luminal-like subtype.

The Akt/PI3K pathway regulates the sub-cellular localisation of FOXO3a by phosphorylation which prevents the protein from translocating to the nucleus to regulate transcription [6]. This indicates that absence of nuclear FOXO3a expression in a tumour, with either complete absence or cytoplasmic localisation, should indicate phosphorylation by Akt. This may represent an important biological mechanism responsible in part for poor prognosis in ER-positive breast cancer. This proposal is supported by our finding that absence of nuclear expression of FOXO3a is associated with poorer outcome and shows a significant association with PIK3CA, a marker strongly related to Akt function. Other breast cancer studies have also shown an association between Akt/PI3K activation and cytoplasmic FOXO3a expression pattern with decreased patient survival, in agreement with our findings [13].

In this study, we did not find a significant association with survival when patients were categorised into either negative or positive FOXO3a expressers per se. Instead, we found that subcellular localisation indicates functional relevance as evidenced here by more favourable outcome in patients with predominant nuclear expression. Supporting these findings, previous studies have shown that nuclear FOXO3a induces the expression of genes that inhibit cell cycle progression such as the CDK inhibitors [4, 6]. Subsequently, we found a significant positive association between nuclear FOXO3a and the expression of the cell cycle inhibitor p27 implying a role in the induction of cell cycle arrest.

In this patient series including ER-positive/luminal-like subtype, nuclear localisation of FOXO3a was associated with markers of good prognosis such as FHIT, PgR [35] and FOXA1 expression which is required for the expression of 50% of ER-regulated genes [36]. Furthermore, we have also shown that nuclear FOXO3a expression is significantly associated with longer BCSS and DMFI which implies its role in stratification of ER-positive groups into prognostic subgroups, possibly explained by a tumour suppressor function associated with cell cycle arrest.

Previous studies have shown that loss of FOXO3a function indicated by its absence or by cytoplasmic localisation is positively associated with markers of increased proliferation [14]. We have found that BC, especially luminal-like cases, expressing nuclear FOXO3a are characterised by low proliferation, specifically low mitotic frequency and low MIB1 expression. In addition, we found an inverse association between FOXO3a nuclear expression and C-MYC protein supporting the proposal that the Akt/PI3K/FOXO pathway modulates MYC function by inhibition of MYC-dependent transcription resulting in decreased cellular proliferation [37]. Taken together, our findings support the interaction of FOXO3a as a downstream target of Akt/PI3K pathway with markers related to proliferation and cell cycle, a role which is independent of the systemic therapy as shown here by our multivariate analysis results.

In conclusion, our results demonstrate the biological and prognostic role of FOXO3a protein expression, as a downstream target of AKT pathway, and its subcellular localisation in BC. Loss of nuclear translocation of FOXO3a could tilt the balance in favour of increased proliferation and more aggressive behaviour in luminal-like breast cancers.

Acknowledgments

We thank the Ministry of Higher Education (Egypt) for funding H. O. Habashy, M. Ahmed, M. Aleskandarany and E. Rakha.

Conflict of interest

None.

Copyright information

© Springer Science+Business Media, LLC. 2011