Eukaryotic Initiation Factor 4E (eIF4E) and angiogenesis: prognostic markers for breast cancer

Tissue samples were studied from 122 breast carcinoma patients ranging from 41 to 77 years of age (median age, 56.2 years), who had not received chemotherapy, radiotherapy, or hormone therapy prior to surgery. All patients were treated at the Department of Surgery in Tongji Hospital (Huazhong University of Science and Technology, Wuhan, China) and were sequential cases seen from October 1996 to July 1999. These select untreated breast cancer patients underwent axillary node excision combined with wide local excision with margin clearance or mastectomy. Study samples were collected at this time. Adjuvant axillary irradiation, systemic chemotherapy, and antiestrogen therapy were then offered and administered as indicated per current post-surgical standard of care. Patient follow-up ranged from 1–72 months (median, 44 months): Clinical records show that 77 patients were found to be disease-free, while 45 breast cancer patients relapsed, of whom 30 subsequently died. The 5-year survival rate was 76.4%. Mean survival time was 41.89 ± 8.38 months and median survival time was 39 months. Overall patient survival was calculated as the period from surgery until the date of death, while disease-free survival was the period from surgery to date of metastasis.

The research protocol was approved by our Institutional Review Board (IRB) before all specimens were examined by experienced pathologists. Histological examination was carried out on paraffin-embedded sections stained with hematoxylin & eosin (H&E). These specimens had a spectrum of breast carcinomas represented, including 80 invasive ductal carcinomas (IDC), 15 ductal carcinomas in situ (DCIS), and 5 lobular carcinomas in situ (LCIS). In addition, special invasive carcinoma types included: 14 invasive lobular carcinomas (ILC), 6 medullary carcinomas (MC), and 2 tubular carcinomas (TC). The 102 invasive carcinoma cases seen included 27 welldifferentiated (grade I), 41 moderately differentiated (grade II), and 34 poorly differentiated (grade III) carcinomas. Similarly, there were 5 low-grade, 3 intermediate-grade, and 12 high-grade cases in the non-invasive carcinomas (DCIS and LCIS). Tumor size varied from 0.5 to 6.8 cm (mean, 3.1 cm). At the time of surgery, it was found that a total of 74 patients had experienced lymph node metastases. Other clinical and pathologic parameters were obtained from the hospital pathology reports, including estrogen receptor (ER) and progesterone receptor (PR) status, Her2/neu status, and the TNM stage (Table 1).

The paraffin-embedded, formalin-fixed archival breast tissue was cut into four-micrometer sections and fixed on charged slides using heat immobilization, by baking at 60°C for 2 hours. The slides were then deparaffinized in xylene and subsequently rehydrated through a graded ethanol series ending in deionized water. Any endogenous peroxidase was blocked by treatment with 3% hydrogen peroxide for 15 minutes, followed by 3 rinses of 5 minutes each in deionized water. Antigen retrieval was performed by placing slides in 1× citrate buffer for 15 minutes at 100°C (in a microwave). The eIF4E primary antibody (Santa Cruz, California, USA), diluted to 1:100 in phosphate-buffered saline (PBS) and 1% bovine serum albumin, was left on tissue overnight at 4°C, in a humidified chamber. As a negative control, preimmune serum was substituted for the primary antibody. All sections were rinsed in PBS 3 times, 5 minutes each. Secondary antibody (Envison, Anti-Mouse/Rabbit-HRP, DAKO) was applied to all sections and incubated for 30 minutes at room temperature (RT), followed by 3 PBS rinses of 5 minutes each. Finally, the sections were developed with 3,3'-diaminobenzidine (DAB) for 5 minutes at RT, then counterstained with hematoxylin. After staining, sections were dehydrated through a graded series of ethanol baths followed by xylene and cover-slipped. Using microscopy, the section's tumor cell immunoreactivity for eIF4E was scored according to the cytoplasmic staining present: Both the percentage of positively stained tumor cells and the stain intensity were taken into account to determine eIF4E expression values. The percentage of positive cells was rated as follows: 1 point, <5% positive tumor cells; 2 points, 5–25% positive cells; 3 points, 26–75% positive cells; and 4 points, >75% positive cells. Stain intensity was rated as follows: 1 point, weak intensity; 2 points, moderate intensity; and 3 points, strong intensity. The breast cancer specimens were attributed to four groups, according to their overall score: Absent expression, when <5% of cells stained positive, regardless of intensity; weak expression, a total of 3 points; moderate expression, 4–5 points; and strong expression, 6–7 points. For statistical reasons, tumors were then scored according to a two-scale system: Low reactivity denoting tumors with absent or weak expression, and high reactivity denoting tumors with moderate to strong expression. The association of eIF4E with other parameters was assessed using eIF4E as either a categorical variable (low reactivity vs. high reactivity) or a continuous variable (the percentage of eIF4E-positive cells within a sample).

VEGF and IL-8 expression were assessed using purified mouse anti-human monoclonal antibodies (purchased from Santa Cruz, California, USA), diluted 1:100 and 1:50, respectively, in an overnight incubation, following standard immunohistochemistry procedures. The percentage of cancer cells having cytoplasmic staining for VEGF or IL-8 reactivity was recorded. Staining was graded in a four-grade classification as follows: -, for those where reactivity was not detected; 1+, those with positive stain in less than 5% of tumor cells; 2+, a positive stain between 5% and 50%; and 3+, when greater than 50% positive. The median value was used as a cut-off point to define those tumors having either high or low reactivity for VEGF and IL-8.

Tumor angiogenesis was assessed by calculating standard microvessel density (MVD). Microvessels were identified by immunostaining endothelial cells with the mouse anti-human monoclonal antibody CD105, which targets the receptor for the transforming growth factor beta 1 (TGFβ-1). Although the TGFβ-1 receptor has weak or negative expression in the vascular endothelium of normal human tissues, it is more specific and sensitive in the vascular endothelium of cancer tissues [16]. First, each slide was scanned at low magnification (×100) to identify four areas with the highest density of microvessels ("hotspots"). Each hotspot was then evaluated at high power magnification (×400) for the number of stained microvessels per field. Any endothelial cell having a brown stain was considered to be a single countable microvessel, and the presence of a clearly defined vessel lumen was not required to verify the structure. The final microvessel score was the average of vessel counts from four fields. MVD was quoted as a continuous variable.

One-way ANOVA testing was used to assess correlations between the continuous variables. Contingency tables were analyzed for trends with the chi-square test, in order to evaluate differences in clinicopathologic factors between the patient groups. The Spearman rank correlation coefficient test was used to examine the correlations between different variables. Linear regression analysis was also used to assess the correlation between continuous variables. Survival curves were plotted using the method of Kaplan and Meier and the significance of the observed differences was assessed using the log-rank test. Survival analysis was carried out using either cancer recurrence or death from disease as the endpoints for disease-free survival (DFS) and overall survival (OS), respectively. The influence of each variable on survival was assessed with the Cox proportional hazards regression model. The level of statistical significance was defined as P < 0.05. All analyses were carried out using SPSS 11.0 for Windows 2000 software.

In breast carcinoma, localization of eIF4E protein overexpression was mainly cytoplasmic in the cancer cells, but not in the adjacent normal-looking ductal epithelial cells and stromal elements. Of 122 tumors, 59 (48.4%) showed low reactivity (absent or weak expression) and 63 (51.6%) showed high reactivity (moderate to strong expression) (Fig. 1A,B, and 1C). Immunopositive reactions for VEGF (Fig. 1D,E and 1F) and IL-8 (Fig. 1G,H and 1I) were confined to the cytoplasm of cancer cells. Expression of VEGF was found to be negative, weak, moderate, and strong in 29 (23.8%), 38 (31.1%), 34 (27.9%), and 21 (17.2%) tumor tissues, respectively. Also, the IL-8 protein showed negative, weak, moderate, and strong staining in 21 (17.2%), 40 (32.8%), 37 (30.3%) and 24 (19.7%) of 122 tumors, respectively. In addition, it was evident that both VEGF and IL-8 had higher immunoreactivity in the IDC than in the DCIS. In a few specimens, the most intense reaction was demonstrated distal to the blood vessels and surrounding necrotic areas, but in the vast majority of positive cancer cells we observed a diffuse expression pattern that was independent of vessel proximity.

Vascular endothelial cell or endothelial cell clustering around the tumors was evident using the anti-CD105 antibody (Fig. 1J,K and 1L). The mean value of MVD was found to be 15.22 ± 6.93. The microvessels around the solid tumor nest were much denser than those in the adjacent normal tissues.

A significant positive association between eIF4E immunoreactivity and the histological grade of the invasive tumor was found (p = 0.016). In addition, we found that invasive carcinoma was different from non-invasive carcinoma with respect to eIF4E immunostaining (p = 0.034). No significant correlations occurred between eIF4E expression and patients' age (p = 0.49), histological type in invasive carcinoma (p = 0.72), tumor size (p = 0.25), node metastasis (p = 0.23), TNM stage (p = 0.54), or the ER, PR, and HER-2 status provided (p = 0.44, p = 0.38, and p = 0.45, respectively) (Table 2). In addition, we found that VEGF and IL-8 were closely related to both the tumor grade (p = 0.003 and p = 0.022, respectively) and the clinical stage (p = 0.007 and p = 0.048, respectively) (Table 3). Increasing MVD was also linked with increasing tumor grade (p < 0.001) and stage (p < 0.001) (Table 4).

Using the Spearman rank correlation, significant correlations of eIF4E with both VEGF (p = 0.007, r = 0.316) and IL-8 (p = 0.007, r = 0.317) expression in breast cancer were noted (Fig. 2A,2B). Using linear regression analysis, we studied the association of the percentage of tumor cells staining positive for eIF4E and MVD. We found a significant correlation between the proportion of eIF4E-positive cells and MVD (p = 0.006, r = 0.34) in these breast cancer samples (Fig. 3).

The Kaplan-Meier survival curves for eIF4E, VEGF, and IL-8 (Fig. 4) demonstrated that when the survival of breast cancer patients having moderate to strong expression of eIF4E is compared with the survival of patients with absent to weak expression of eIF4E through log-rank testing, there was a significantly unfavorable influence for eIF4E expression on both the overall survival (OS) (p = 0.01) and disease-free survival (DFS) (p = 0.006) of the patients (Fig. 4A,B). Univariate analysis revealed that overexpression of VEGF and IL-8 were also associated with a worse prognosis, with both unfavorable OS (p = 0.02 and p = 0.03, respectively) and DFS (p = 0.01 and p = 0.02, respectively) linked to these two pro-angiogenic growth factors as well (Fig. 4C–F).

To examine the independent prognostic significance of eIF4E in relation to the classical clinicopathological parameters and angiogenic variables, multivariate analysis of OS was performed including eIF4E and all three angiogenic markers, as well as other risk factors such as tumor grade, stage and the ER, PR, and Her-2 status (Table 5). In a multivariate model that included tumor grade, stage and the ER, PR, and Her-2 status, as well as levels of VEGF, IL-8, MVD, and eIF4E found in the tissues; eIF4E emerged as an independent prognostic factor for OS (p = 0.001), along with stage (P = 0.005), node status (P = 0.046), and the MVD present (P = 0.004). Table 6 shows the results of the multivariate analysis of DFS including the previously mentioned parameters. The eIF4E, MVD, and tumor stage still emerged as independent prognostic factors (p = 0.02, p = 0.011, and p = 0.048, respectively), but node metastasis was found to be of borderline significance (p = 0.062).