Main

Tyrosine kinases, proteins that phosphorylate substrate proteins on tyrosine residues, regulate key processes in tumorigenesis, including proliferation, angiogenesis, and cellular survival. For this reason, tyrosine kinases have the potential to predict tumor progression and serve as targets for therapeutic intervention.1 While the expression patterns of tyrosine kinases such as HER-2/neu and epidermal growth factor receptor have been extensively characterized in tumors,2 the functions of other tyrosine kinases in cancer are only beginning to be understood.

Focal adhesion kinase (FAK) is a 125 kDa protein tyrosine kinase that localizes to sites of cellular adhesion in cultured cells. FAK functions in integrin-signaling pathways,3, 4, 5, 6 cellular motility,7 and apoptosis.8, 9, 10 The avian FAK homologue was originally identified through its association with the v-Src oncogene protein,11 suggesting a role for FAK in tumorigenesis. In our lab, human FAK was subsequently identified in high-grade sarcomas12 and breast tumors.13, 14, 15 FAK is overexpressed in a variety of human tumors in comparison with adjacent normal tissue, including breast,13, 14, 15, 16 colon,13, 14, 17, 18 thyroid,19 ovarian,20 cervix,16 head and neck,21 and esophagus.22

Early work suggested that FAK might function in the later stages of tumor progression, such as invasion and metastasis, promoting adhesion in invading cells. However, FAK expression is detected in early stage tumors, such as ductal carcinoma in situ,15 suggesting that FAK might have functions in cancer progression that precede invasion and metastasis. Work from this lab and others has demonstrated that FAK regulates antiapoptotic survival signaling, even in cells maintained in the absence of adhesion.23, 24 This suggests that FAK could function in the earlier stages of tumor progression as a regulator of survival, suppressing apoptotic responses associated with uncontrolled proliferation and signaling.

In the present study, we have analyzed the expression of FAK in a large population-based study of breast tumor samples. Although we had previously shown FAK overexpression in breast tumors,13, 14, 15 the previous studies lacked sufficient numbers of patient samples to derive relationships between FAK expression and other important clinical markers.

Materials and methods

Study Population

The Carolina Breast Cancer Study is a population-based, case–control study whose study participants included women 20–74 years of age from 24 contiguous counties of central and eastern North Carolina.25, 26 Using the North Carolina Central Cancer Registry, 1153 women with a first diagnosis of invasive breast cancer between 1993 and 1996 were identified.27 African-American and women under the age of 50 years were over-sampled using randomized recruitment. Informed consent was obtained and 861 invasive breast cancer cases were eligible for participation in Phase 1 of Carolina Breast Cancer Study, overall response rate=75%. Medical records or direct histopathological review of tumor tissue provided clinical data and information about tumor characteristics. The University of North Carolina School of Medicine Institutional Review Board approved all aspects of this research.

Tissue Specimens

Formalin-fixed, paraffin-embedded tumor blocks were obtained from participating hospitals' pathology departments and tumors were sectioned by the Specialized Programs of Research Excellence (SPORE) Immunohistochemistry Core facility as described previously.28 All sections were stored at 4°C. Of the 861 formalin-fixed, paraffin embedded specimens received for the Carolina Breast Cancer Study, 629 were available and contained sufficient invasive cancer for FAK immunohistochemistry.

Immunohistochemistry

FAK immunohistochemistry was performed using the monoclonal antibody 4.47 (1:250; Upstate Biotechnology, Lake Placid, NY, USA) to the amino-terminal region of FAK.14 This region of FAK is preferred as an antigenic target since anti-FAK antibodies directed at the carboxy terminus of the FAK molecule do not distinguish between FAK and FAK-related nonkinase (FRNK) proteins.14 The 4.47 antibody has been previously characterized and recognizes FAK in formalin-fixed, paraffin-embedded breast tumor sections, with minimal crossreactivity.14

FAK immunohistochemical analyses were performed as described previously.18 Briefly, slides were incubated overnight with anti-FAK 4.47 monoclonal antibody (1:250; Upstate Biotechnology) after heat-induced epitope recovery and blocking with normal horse serum (Vectastatin Elite Kit; Vector Laboratories, Burlingame, CA, USA). The DAKO Autostainer (DAKO Corporation) was utilized for application and incubation of the biotinylated horse antimouse IgG (Vectastain Elite Kit, Vector Laboratories), avidin–biotin complex (Vectastain Elite Kit, Vector Laboratories), and slate gray chromagen (Vector Laboratories).

HER-2/neu immunohistochemistry had been previously performed in the SPORE Immunohistochemistry laboratory using clone CB11 (1:100; BioGenex, San Ramon, CA, USA).29 These assays were scored previously (LG and MI) and results maintained in the SPORE Immunohistochemistry database. Immunohistochemistry was performed using an avidin–biotin immunoperoxidase system with 3′,3′ diaminobinzidine as chromagen. Briefly, slides were incubated overnight with clone CB11 after quenching endogenous peroxidase using a methanol-peroxide bath and blocking with normal horse serum (Vectastain Elite Kit, Vector Laboratories). During day 2, slides were incubated with biotinylated secondary antibody (Vectastain Elite Kit, Vector Laboratories), avidin–biotin complex (Vectastain Elite Kit, Vector Laboratories), and 3′,3′ diaminobinzidine (Sigma, St Louis, MO, USA). Cases were considered positive for HER-2/neu if there was unambiguous membrane staining of a weak, moderate, or strong intensity in at least 10% of the invasive cancer cells.

Immunohistochemistry Scoring for FAK

A single board-certified pathologist (CAL) scored each tissue section for FAK expression based on a scoring system that measured intensity (0, none; 1, borderline; 2, weak; 3, moderate; 4, strong), percentage positive cells (0–100), and cellular localization (nucleus, cytoplasm, membrane, or combination). In order to better define FAK expression in a large study population, we dichotomized FAK expression as high (3+ or 4+ intensity and ≥90% positive cells) and not high (0, 1, or 2 intensity and/or <90% positive cells) (Figure 1).

Figure 1
figure 1

Immunohistochemical analysis of FAK expression in invasive breast cancer. Anti-FAK 4.47 monoclonal antibody (1:250) was used to detect FAK expression with slate gray chromagen. Tissue sections for FAK expression is based on a scoring system that measured intensity (0, none; 1, borderline; 2, weak; 3, moderate; 4, strong), percentage positive cells (0–100), and cellular localization (nucleus, cytoplasm, membrane, or combination). FAK expression was dichotomized as high (3+ or 4+ intensity and ⩾90% positive cells) or not high. (a) FAK expression in invasive breast cancer demonstrating strong cytoplasmic expression ( × 400 magnification). (b) Weak FAK expression demonstrated by faint cytoplasmic staining within invasive breast cancer cells ( × 400 magnification). (c) No FAK expression demonstrated in breast cancer cells ( × 400 magnification).

Full size image

Statistical Methods

Fisher's exact test was used for data categorized into two by two contingency tables. For the larger contingency tables (ie the two by three and two by four tables of mitotic index, nuclear grade, histological grade, and stage at diagnosis by dichotomized FAK expression), we used the nonparametric Jonckheere–Terpstra method to test for ordered differences. With this test, the null hypothesis is that the distribution of the response does not differ across ordered categories. Statistical analyses were performed with SAS statistical software, Version 8.2, SAS Institute Inc., Cary, NC, USA. Logistic regression was implemented using SAS version 6.11 (SAS Institute Inc.) to estimate odds ratios for the association of high FAK expression with the following potential predictive variables: race, menopausal status, stage at diagnosis, lymph node status, nuclear grade, histologic grade, mitotic index, HER-2/neu expression, and estrogen receptor (ER) status. Models were adjusted for age.

Results

Formalin-fixed, paraffin-embedded tumor blocks were obtained for 629 (73%) of 861 tumors of breast cancer patients who consented to participate in Phase 1 of Carolina Breast Cancer Study. Among the 629 Carolina Breast Cancer Study breast cancer patients, 41% were African-American and 52% were premenopausal30 (Table 1). The mean age was 48 years (minimum age 23 years, maximum age 74 years; Table 1). The majority of these women presented with stage 1 or 2 disease (89%)31 and negative lymph nodes (60%, Table 1). The majority of the breast cancers were poorly differentiated (65%) with moderate to marked nuclear pleomorphism (88%, Table 1), ER (59%) and progesterone receptor (PR; 57%) positive, and negative for HER-2/neu overexpression 77% (Table 2). In all, 25% of the breast tumors demonstrated high FAK expression (Table 2). In this study population, recurrence and survival data have not been compiled to date.

Table 1 Characteristics of breast cancer patients in the CBCS with FAK data
Full size table
Table 2 Tumor marker expression in CBCS breast cancer patients
Full size table

Tumors that expressed high levels of FAK (Table 3) were compared with other clinicopathological features. High FAK expression was highly associated with a mitotic index of >10 mitoses per 10 consecutive high-power fields (P<0.0001) and nuclear grade 3 breast tumors (P<0.0001). Breast tumors with high FAK expression were also more likely to be architectural grade 3 tumors (P=0.038, Table 3).

Table 3 Correlation between FAK immunohistochemical expression and clinicopathological features
Full size table

FAK expression was also associated with several markers of poor prognosis. Breast tumors expressing high levels of FAK were associated with ER negative (P=0.002) and PR negative (P=0.0004, Table 3) phenotype. In addition, high FAK expression was associated with overexpression of the tyrosine kinase receptor HER-2/neu (P=0.004, Table 3). After multivariable adjustment, the only significant predictors of breast tumors with high FAK expression were HER-2/neu positivity (2.2 (1.4–3.4), P=0.001) and nuclear grade 3 tumors (2.3 (1.4–4.0), P=0.002).

In the Carolina Breast Cancer Study population, patients presenting at diagnosis with stage 3 or 4 breast cancers (65/584) (American Joint Committee on Cancer, 5th Edition, Cancer Staging Classification for Breast Cancer Stages IIIA, IIIB, or IV) were more likely to have tumors with high FAK expression (P=0.08). Patients presenting with positive lymph nodes were also more likely to have breast tumors with high FAK expression (P=0.066). Although not statistically significant, high FAK expression is common in advanced stage.

Discussion

We have examined the characteristics of a population-based, case–control study of breast cancer. High FAK expression in the Carolina Breast Cancer Study breast tumors is associated with a poor prognostic phenotype. Histologically, these tumors have a high mitotic index (<10 mitoses per high power field), grade 3 nuclear and architectural features, lack ER and PR, and overexpress HER-2/neu. To date, recurrence and survival data for this Carolina Breast Cancer Study population is not available in order to evaluate FAK as a prognostic marker.

Overexpression of HER-2/neu is usually associated with a worse prognosis in primary breast cancer and is observed in 15–30% of breast cancers.32, 33, 34 In the current study of Carolina Breast Cancer Study patients, 23% of breast cancers overexpressed HER-2/neu. The association of high FAK expression with HER-2/neu overexpression suggests that high FAK expression may also be a poor prognostic marker. Even though the molecular interactions between FAK and HER-2/neu signaling are just beginning to be understood, work by Vadlamudi et al35 suggest that HER-2/neu and heregulin are involved in FAK signaling and selectively upregulate FAK tyrosine phosphorylation.

Work from this lab and others has indicated a role for FAK in survival signaling, which is the propagation of signals that are required to maintain viability and suppress apoptosis. Attenuation of FAK expression with antisense oligonucleotides leads to apoptosis in tumor cells but not in nontransformed cell lines.8 Subsequent work has utilized overexpression of a fragment of the FAK carboxy-terminal domain (FRNK or FAK-CD for FAK-carboxy-terminal domain) that acts as a dominant-negative protein. FRNK/FAK-CD leads to the inhibition of endogenous FAK phosphorylation36 and the degradation of FAK in tumor cells.23, 37 As a result, FRNK/FAK-CD leads to apoptotic cell death in colon37 and breast cancer cells,23 but not in normal nontumor cells.23, 37 Furthermore, FRNK/FAK-CD inhibits apoptosis even in cells maintained in suspension, indicating a role for FAK in suppression of apoptosis even in the absence of adhesive signaling. Expression of the N-terminus of FAK in breast carcinoma cell lines resulted in rounding, detachment and apoptosis.38 Recent work has shown FAK association with receptor-interacting protein, a serine/threonine kinase that contains a death domain.39 When FAK binds to receptor-interacting protein, the proapoptotic signals supplied by receptor-interacting protein are suppressed. This further supports the role of FAK as a survival signal.39

In this study, high FAK expression had a borderline association with positive lymph node status and stage at diagnosis. Owing to the low number of patients presenting with stage III and IV disease we cannot confirm this association in this study. However, since high FAK was associated with other poor prognostic indicators, examining a population with higher numbers of stage III and IV disease would be worthwhile. Miyazaki et al22 reported an association between FAK overexpression (intense staining of >40% of carcinoma cells) and lymph node metastasis in esophageal squamous cell carcinoma. Indeed, the mechanistic framework of FAK as a survival signal would suggest a role for FAK in metastatic spread to lymph nodes and such a role has been observed in esophageal carcinoma.22

These results also emphasize the potential of FAK as a molecular target for cancer therapeutics. Although we dichotomized FAK expression as high or not high, our cutoff for high FAK was expression in at least 90% of cells, with an intensity of 3+ or 4+. Moreover, normal breast cells express negligible levels of FAK, demonstrating a ‘therapeutic window’ between normal and tumor cells. Taken together, our results demonstrate the aggressive phenotype of high FAK expression and offer the potential for therapeutics designed to abrogate its function.