The study was conducted on healthy women and on patients with breast cancer. Most of the patients were recruited from individuals attending the Screening Program run by the Prevention Unit of the Medical Oncology Department of Pierantoni Hospital, and all of the women were older than 50 years of age (range, 51–92 years; median, 67 years). All patients had histologically confirmed breast cancer.

The control group comprised women who were free from any disease associated with an increased angiogenic activity such as diabetic retinopathy, heart disease or lung disease, which could affect VEGF levels and, possibly, bFGF levels. Healthy women ranged in age from 51 to 77 years, with a median age of 60 years.

The study was examined and approved by the Ethics Committee of the Local Health and Social Services (Azienda USL, Forlì) in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki. Written informed consent was obtained from all the women participating in the study.

Growth factors were determined in the tumor and in the serum. Specifically, VEGF and bFGF expression was determined in 62 and 63 tumor samples, respectively. Serum VEGF and bFGF levels were determined in 54 and 65 healthy women, and in 69 and 73 breast cancer patients, respectively.

Tumor tissue removed during surgery was fixed in neutral buffered 10% formalin and embedded in paraffin. Four-micrometer sections of the histologically confirmed tumor samples were mounted on positive-charged slides (BioOptica, Milan, Italy), deparaffinized with xylene, rehydrated, and the endogenous peroxidase activity blocked by 3% hydrogen peroxide solution.

VEGF expression was determined using a polyclonal antibody that specifically reacts with isoforms 121, 165, 189, and 206 (Biogenex, San Ramon, CA, USA). bFGF expression was determined using a monoclonal antibody that reacts with the bFGF 18–24 kDa isoforms (Transduction Laboratories, Lexington, KY, USA). Expression of the estrogen receptor (ER) and of the progesterone receptor (PgR) was determined using 1D5 and 1A6 monoclonal antibodies, respectively (Biogenex).

VEGF, bFGF, ER, and PgR antigen retrieval was performed by microwaving at 75 W for 15 min in 10 mM citrate buffer (pH 6.0) followed by cooling at room temperature for at least 20 min. The sections were then treated for non-specific binding with 3% bovine serum albumin in PBS for 20 min, after which they were incubated for 1 hour at room temperature with prediluted polyclonal anti-VEGF antibody or monoclonal anti-ER and anti-PgR antibodies or anti-bFGF monoclonal antibody diluted 1:100 in PBS. Positive control slides (HL-60, K562, and HeLa tumor cell lines for VEGF and bFGF, and normal breast tissue for ER and PgR) were stained in parallel with the antibodies. The sections were washed with PBS, incubated with universal biotinylated secondary antibody, rinsed in PBS, and incubated with streptavidin–peroxidase conjugate (LSAB+kit; DAKO Corporation, Carpinteria, CA, USA) for 15 min. Sections were rinsed again in PBS, and antibody binding was detected by staining with diaminobenzidine/hydrogen peroxidase chromogen solution (DAB+liquid substrate–chromogen solution; DAKO Corporation). Finally, sections were rinsed in deionized water, were counterstained blue by Mayer's Hemalum, and were mounted in Eukitt (Bio-Optica).

VEGF and bFGF expression was evaluated at a light microscope (200 ×) by two independent observers (AMG and LM). Immunoreactivity was expressed as the percentage of the immunopositive area in relation to the total area of invasive neoplastic tissue of the whole section. ER and PgR analysis was performed by an image analyzer system (CAS 200; Becton Dickinson, San Jose, CA, USA) and expressed as already described.

Peripheral venous blood was collected in sterile test tubes (Vacutainer System; Becton Dickinson), left to coagulate at 4°C for 30 min, centrifuged at 2000 × g for 10 min, and then stored in aliquots at -70°C for a maximum of 1 month.

VEGF and bFGF levels were determined using a quantitative sandwich enzyme immunoassay technique (Quantikine; R&D System, Minneapolis, MN, USA) and all samples were tested in duplicate. Briefly, 100 μl assay diluent and 100 or 150 μl serum sample for VEGF and bFGF, respectively, or scalar concentrations of VEGF (15.6, 31.2, 62.5, 125, 250, 500, 1000 and 2000 pg/ml) and bFGF (0.5, 1, 2, 4, 8, 16, 32 and 64 pg/ml) were added to each well of a microtiter plate precoated with mouse antihuman VEGF or mouse antihuman bFGF monoclonal antibody. Plates were incubated at room temperature for 2 or 3 hours for VEGF and bFGF, respectively. The wells were then washed to remove unbound VEGF or bFGF, supplemented with 200 μl enzyme-linked anti-VEGF or anti-bFGF polyclonal antibodies and incubated for 2 hours at room temperature. The wells were washed again, supplemented with antibody-linked enzyme substrates, and left for 25 min at room temperature. For bFGF assay only, an amplifier solution was added and the reaction was prolonged for 25 min.

After the addition of 50 μl stop reagent (2 M sulphuric acid), the color intensity in each well was measured at 450 nm for VEGF and at 490 nm for bFGF using a spectrophotometer (Medical System, Genoa, Italy). The optical density found in the serum samples was compared with a standard curve of VEGF and bFGF concentrations and was quantified. The coefficient of linearity, which shows the linear correlation between measured absorbance and known amounts of standards, was at least 0.9 for both bFGF and VEGF. The minimum detectable concentration ranged from 0.05 to 0.56 pg/ml for bFGF and was less than 9 pg/ml for VEGF, as quoted by the manufacturer.

With regard to intra-assay reproducibility, the determinations of growth factor serum levels were performed in duplicate and were repeated when the coefficient of variation (CV) exceeded 15%. The highest disagreement between the pair values was observed for low levels of the growth factors. Overall, the CV was less than 15% in 85% of cases for either VEGF (105 out of 123 samples) or bFGF (117 out of 138 samples). The CV was less than 10% in 76% of cases for VEGF and in 68% of cases for bFGF.

Assessment of interassay reproducibility was made by calculating the CV for growth factor levels of serum samples from five individuals run in quintuplicate. The CV for VEGF determinations ranged from 4.9% to 9.5%, and that for bFGF from 3% to 13.3%. In consideration of the low serum concentrations of bFGF and of the fact that the tests were carried out manually, we considered a CV higher than 10% but lower than 15% acceptable.

Nonparametric ranking statistics (median test) were used to analyze the relationship between the percentage of positive tumor cells and the serum levels of VEGF and bFGF, considered as continuous variables, and patient characteristics. Spearman's correlation coefficient was used to investigate the relationship between the two markers in the serum or in the tumor. The relation between VEGF and bFGF serum levels and the presence of breast cancer in the case–control study was analyzed using the median test.

The most accurate cut-off value capable of discriminating between healthy women and cancer patients, in the absence of internationally accepted cut-off values for serum bFGF and VEGF concentrations, was identified using a receiver operating characteristic (ROC) curve analysis. In the ROC curve the true positive rates (sensitivity) were plotted against the false positive rates (specificity) for all classification points. Ninety-five percent confidence intervals were calculated for sensitivity and specificity values.

VEGF serum levels in healthy women ranged from 0 to 707.6 pg/ml, with a median value of 145.7 pg/ml. bFGF serum levels in healthy women varied from 0 to 5.7 pg/ml, with a median value of 0.4 pg/ml.

A considerable overlapping of serum VEGF levels between healthy women and breast cancer patients was observed, and median values were not significantly different (P = 0.055). Values in cancer patients exceeded the highest value (707.6 pg/ml) detected in healthy women in only two (2.8%) cases (Fig. 1a).

A partial overlapping of bFGF serum levels in healthy women and in breast cancer patients was also observed, but the difference between the median values of the two subgroups was significantly different (P < 0.001) (Fig. 1b). Moreover, all values equal to or higher than 6.0 pg/ml (8/8) referred to cancer patients, while the total absence of bFGF in the serum (27/27) was only observed in healthy women.

The ROC curve did not highlight any acceptable concentration of serum VEGF in terms of either sensitivity or specificity (Fig. 2a). Conversely, when calculated using serum bFGF values of healthy women and of breast cancer patients (Fig. 2b), the ROC curve identified a concentration of 1.0 pg/ml, with an 84.9% sensitivity and a 63.1% specificity, as the best discriminant (Table 4). For patients older than 70 years of age, a similar sensitivity (77.8%) and a higher specificity (78.6%) were observed.