Tumor and patient characteristics
Fresh material obtained during surgery was kept on ice and examined by a pathologist. Representative specimens with more than 60% tumour cells were sent to the Stiftung Tumorbank Basel (STB), immediately shock frozen and cryopreserved (-80°C). All activities of the STB are in accordance with an official Swiss permit, which guarantees patient confidentiality and respects ethical issues. For the present study, 156 samples of primary breast tumours were selected. Those samples overexpressing ErbB-2 (>500 U/mg total protein) were selected, based on ErbB-2 protein expression levels routinely detected using EIAs at the time of surgery by the STB [
23]. EIA ErbB-2 positive samples correlate strongly with DAKO 3+ and with ErbB-2 amplification detected by fluorescent
in situ hybridization (FISH; data not shown).
All patients underwent primary surgery before January 1996. Sixty-seven patients (43%) experienced disease recurrence within the median follow-up time of 57 months (range 27–88 months). Sixty-six patients (42%) were node negative, and 90 (58%) were node positive. Forty tumours (26%) were oestrogen receptor (ER)-α negative. Ninety-five patients (61%) had ErbB-2-negative (<500 U/mg total protein) and 61 patients (39%) had ErbB-2 positive tumours [
23]. None of the patients received neoadjuvant therapy. Patient and tumour characteristics are summarized in Table
1.
Table 1
Clinicopathological characteristics of the patients
Patients enrolled | 156 |
Age (years): | |
<40 | 12 (8) |
40–60 | 85 (54) |
>60 | 59 (38) |
Histology type: | |
Ductal | 109 (70) |
Lobular | 17 (11) |
Other | 30 (19) |
Tumour size: | |
T1 | 49 (31) |
T2 | 90 (58) |
T3-T4 | 17 (11) |
Lymph-node status | |
Node negative | 66 (42) |
Node positive | 90 (58) |
Histopathological grade | |
I + II | 57 (37) |
III | 86 (55) |
Not analyzed | 13 (8) |
Oestrogen receptor | |
Positive (>20 fmol/mg) | 116 (74) |
Negative (≤ 20 fmol/mg) | 40 (26) |
Progesterone receptor | |
Positive (>20 fmol/mg) | 85 (54) |
Negative (≤ 20 fmol/mg) | 71 (46) |
Cell lines and tissue culture
MCF-7 breast cancer cells were cultured in IMEM-ZO (improved minimal essential medium with zinc option) supplemented with 5% foetal bovine serum, l-glutamine and antibiotics (penicillin/streptomycin) at 37°C in a 5% carbon dioxide incubator. For the phospho-standard preparation, subconfluent MCF-7 cells were serum starved for 48 hours in serum-free media, and were treated with NaF and Na3VO4 for 1 hour, and then with 10% foetal bovine serum for 10 min. Cells were lysed for 5 min on ice in EB lysis buffer (20 mmol/l Tris-HCl [pH 7.4], 0.5 mol/l NaCl, 10 mmol/l EDTA, 1% Triton X100, 20 mmol/l NaF, 20 mmol/l glycerophosphate, 2 mmol/l Na3VO4, proteinase inhibitor cocktail [Roche, Indianapolis, IN, USA]), centrifuged at 20,000 g for 5 min and supernatant was stored at -80°C.
Measurement of oestrogen receptor, progesterone receptor and ErbB-2 protein levels in tumour extracts by enzyme immunoassay
Tissue homogenates were prepared in accordance with standard procedures for tumour marker measurement using EIAs, as previously described [
23]. In brief, the frozen tissues were pulverized in liquid nitrogen using a Micro-Dismembrator U (B Braun Melsungen AG, Melsungen, Germany). The powder was homogenized using a tissue homogenizer (Ultra-Turrax; Janke & Kunkel, IKA-Werke, Staufen, Germany) for 20 s in three volumes of ice-cold extraction buffer. The homogenate was centrifuged at 800
g for 30 min at 2°C, and the resulting supernatant re-centrifuged in an ultracentrifuge (Beckman Instruments, Fullerton, CA, USA) at 100,000
g. The resulting supernatants (cytosols) were used for measurement of the hormone receptors (ER, progesterone receptor [PgR]), and the membrane fractions were used for EIA measurement of membrane-associated ErbB-2. ER and PgR concentrations were measured from tumour cytosolic extracts by commercial quantitative ER and PgR EIA kits (Abbott Laboratories, Abbott Park, IL, USA) using a Quantum II photometer (Abbott Laboratories, Abbott Park, IL, USA). Quality control of ER and PgR measurements was carried out in collaboration with the Receptor Biomarker Group of the European Organization for Research and Treatment of Cancer. ErbB-2 receptor levels were determined on the particulate membrane fractions of tumour extracts using a commercial monoclonal antibody EIA kit, described by Eppenberger-Castori and coworkers [
23].
Immunoassay of phosphorylated Akt level
Neither antibody used in the CLISA discriminates between Akt isoforms. The catching antibody (anti-Akt/PKB, PH domain, clone SKB1; Upstate Biotechnology, Lake Placid, NY, USA) recognizes Akt1/PKBα, Akt2/PKBβ and Akt3/PKBγ (weak to none) based on immunoblot analysis using 100 ng recombinant fusion protein for each isoform, as reported by the manufacturer. The detecting phospho-specific (S473) Akt monoclonal antibody (4E2) detects endogenous levels of Akt1 only when phosphorylated at serine-473. This antibody also recognizes Akt2 (S472) and Akt3 (S474) if they are phosphorylated at the corresponding residues, according to the information obtained from the manufacturer (Cell Signaling Technology, Inc., Beverly, MA, USA). However, 4E2 does not recognize other Akt phosphorylation sites.
S473 phosphorylated Akt levels were measured using a novel two-site CLISA. Black 96-well microtitre plates (Nunc Black MaxiSorp Surface; Nalgen Nunc International, Rochester, NY, USA) were coated with coating antibody at a concentration of 3 mg/ml of coating buffer (phosphate-buffered saline with 0.6 mmol/l EDTA) in a volume of 100 μl/well and kept at 4°C overnight. To measure P-Akt, respective tumour extracts were prepared as described above in the presence of NaF and Na3VO4. Before sample applications, the coated microtitre plates were washed five times with 200 μl/well washing buffer (25 mmol/l HEPES [pH 7.4], 300 mmol/l NaCl, 0.05% Tween-20) and then blocked for 2 hours at room temperature with 250 μl blocking buffer (25 mmol/l HEPES [pH 7.4], 300 mmol/l NaCl, 0.05% Tween-20, 3% TopBlock [Juro AG, Lucerne, Switzerland]). Blocked wells were washed five times with 200 μl washing buffer, and then 100 μl diluted tumour membrane extracts or reference material was added to the wells and incubated overnight at 4°C.
As a reference for each assay, an extract of MCF-7 cells, prepared as described above, was used. For use in the assay, MCF-7 cell extracts were sequentially diluted with sample dilution buffer (blocking buffer, proteinase inhibitor cocktail, NaF and Na3VO4) at ratios of 1×, 0.75×, 0.5×, 0.25×, 0.125× and 0.025×, and then 100 μl aliquots were incubated on each microtitre plate, together with tumour tissue extracts and negative controls (containing only dilution buffer). After incubation of the samples and reference material, wells were washed five times with 200 μl washing buffer at room temperature to eliminate unbound particles. Biotinylated detection antibody was added, followed by incubation for 2 hours at room temperature. Complexes were detected with horseradish peroxidase-conjugated streptavidin, diluted in conjugate diluents for 1 hour at room temperature. Horseradish peroxidase activity was detected using SuperSignal WestPico substrate (Pierce, Rockford, IL, USA) in a glow luminometer. The response data for diluted reference material was fitted, and the respective curve was used for the quantification of tumour extracts. The value of undiluted MCF-7 extracts was denominated as 1 U/ml.
Quantitative real-time RT-PCR for the detection of proliferation markers
RNA was extracted using RNeasy kit (Qiagen, Hilden, Germany). Quality and quantity were checked using a Bioanalyzer 2100 (Agilent, Palo Alto, CA, USA). All genes were examined using SYBR Green I methods with Taqman 7000 (Applied-Biosystems, Foster City, CA, USA). Relative quantification (ΔΔCt) was obtained by normalization with ribosomal 18S and a standardization step with Human Universal Standard RNA (Stratagene, La Jolla, CA, USA). Quantitative real-time RT-PCR results were expressed in arbitrary units of reverse transcribed RNA (U/μg rt-RNA).
Statistical methods
The statistical significance of the association between P-Akt and other dichotomous variables (e.g. node status) was assessed using Mann–Whitney U-test. Spearman rank correlation (r
s) was calculated to assess associations between continuous markers (e.g. ErbB-2 or tumour size and P-Akt protein expression levels). The continuous variable function of CLISA-determined P-Akt values was first tested for prognostic significance by univariate Cox regression. A cutoff or prognostic threshold value with respect to relapse-free survival was sought by means of classification and regression tree analysis [
24,
25]. Survival probabilities were calculated using the Kaplan–Meier method and compared by means of log-rank analysis [
26]. The Cox proportional hazards regression model was also applied over multivariate analyses, with the associated likelihood ratio test used to assess test-of-trend differences. The results of multivariate Cox regression analysis were summarized in a table and expressed as relative risk for relapse.