Expression of c-ABL, c-KIT, and platelet-derived growth factor receptor-β in ovarian serous carcinoma and normal ovarian surface epithelium

Abstract

BACKGROUND

Tyrosine kinases, such as c-KIT, c-ABL, and platelet-derived growth factor-beta (PDGFR-β), are important regulators of cell growth. Highly potent and selective inhibitors of tyrosine kinases are being investigated as alternatives to standard chemotherapy. One such inhibitor, imatinib mesylate, is being used to treat gastrointestinal stromal tumors and chronic myelogenous leukemia. Ovarian carcinomas frequently develop resistance to conventional chemotherapeutic agents. Immunohistochemical expression of c-ABL, PDGFR-β, and c-KIT was evaluated in ovarian carcinomas to determine whether treatment with imatinib mesylate might be feasible.

METHODS

The expression of c-ABL, c-KIT, and PDGFR-β in tumors was evaluated by immunohistochemical analysis of 52 ovarian serous carcinomas, including 21 low-grade (well differentiated) and 31 high-grade (poorly differentiated) tumors. Fourteen normal ovaries were also evaluated.

RESULTS

In normal ovarian surface epithelium, c-ABL was expressed universally. PDGFR-β was expressed in the majority (93%) of samples of normal ovarian epithelium, whereas the c-KIT protein was undetectable in normal ovarian surface epithelium. Overall, c-ABL was expressed in 71% of serous carcinomas. c-ABL was expressed more frequently in the low-grade serous carcinomas (81%) compared with the high-grade serous carcinomas (65%). PDGFR-β expression was observed in 81% of serous carcinomas overall and was observed more frequently in higher-grade tumors. c-KIT immunohistochemical staining was absent in low-grade tumors but was present in 26% of high-grade serous carcinomas.

CONCLUSIONS

The majority of ovarian serous carcinomas express one or more of the kinases targeted by the tyrosine kinase inhibitor, imatinib mesylate, suggesting the potential usefulness of this drug in the treatment of ovarian carcinoma. Cancer 2003;98:758–64. © 2003 American Cancer Society.

DOI 10.1002/cncr.11561

Ovarian carcinoma is the fifth leading cause of cancer death among women in the United States and is the most common cause of cancer death from a gynecologic malignancy.1, 2 Unfortunately, ovarian carcinoma rarely is diagnosed in its early, most curable stages. The cornerstone of ovarian carcinoma management is surgery, which includes hysterectomy, bilateral salpingo-oophorectomy, omentectomy, surgical staging, and the removal of all visible tumor. After surgery, ovarian carcinoma is treated with combination chemotherapy using taxanes and platinum-based drugs.3–8 Greater than 70% of patients initially respond to chemotherapy, but greater than 50% of these patients will experience disease recurrence within 2–5 years. Attempts to refine and optimize existing therapies and to improve the outcome of patients with advanced disease have had limited success. Recurrent tumors become increasingly chemoresistant to currently available therapies.5–7 The development of novel, more effective treatment strategies against advanced disease is necessary to improve patient survival.

Protein tyrosine kinases (PTKs) play important roles in the regulation of cellular proliferation and differentiation and represent potential drug targets for anticancer therapies.9–14 Highly potent and selective inhibitors of PTKs are being developed and investigated as drugs for the treatment of a variety of cancers, including ovarian carcinoma.15–17 Imatinib mesylate (Gleevec, Novartis Pharma AG, Basle, Switzerland) is a selective inhibitor of c-ABL, platelet-derived growth factor receptor-beta (PDGFR-β), and c-KIT tyrosine kinase activity.18–20 It is used to treat chronic myelogenous leukemia (CML) because these leukemic cells characteristically express an activating mutation of the ABL tyrosine kinase.18–21 In addition, imatinib mesylate is used to treat gastrointestinal stromal tumors (GIST), which express a constitutively activated mutant c-KIT.18–20, 22, 23 Several well-characterized fusion genes of PDGFR-β have been described and treated successfully with imatinib mesylate in chronic myelomonocytic leukemia. These fusions, resulting in constitutively active PDGFR-β kinase, include tel(ETV6)-PDGFRB, RAB5EP-PDGFRB, Huntington interacting protein1/PDGFRB, and others.24 COL1A1-PDGFRB fusions have been identified in the skin tumor dermatofibrosarcoma protuberans.25 In addition, PDGFR-β expression is associated with tumor neoangiogenesis.26, 27 It has been hypothesized that inhibition of PDGFR-β may be both antitumorigenic and antiangiogenic. Clinical trials also are underway to investigate the utility of imatinib mesylate in the treatment of brain tumors that overexpress PDGFRs and other solid tumors that express imatinib mesylate-sensitive kinases. To determine the possible use of imatinib mesylate and related tyrosine kinase inhibitors as treatments for epithelial ovarian carcinoma, we characterized the immunohistochemical expression of c-ABL, PDGFR-β, and c-KIT in ovarian serous carcinoma, which is the most common epithelial malignancy of the ovary.

MATERIALS AND METHODS

Patients and Tissue Samples

Formalin-fixed, paraffin-embedded sections from 14 normal ovaries, 31 ovarian high-grade (poorly differentiated) papillary serous carcinomas, and 21 low-grade (well-differentiated) papillary serous carcinomas were obtained from the pathology files at the University of Texas M. D. Anderson Cancer Center (MDACC; Houston, TX). All slides were reviewed by two gynecologic pathologists (R. B. and A. M.) to confirm the diagnosis.

The ovarian serous carcinomas were graded on a two-tiered system (either low- or high-grade), based on the degree of cytologic atypia and the mitotic index per 10 microscopic high power fields (HPF) as is the usual practice at MDACC.28 Low-grade carcinomas were characterized by mild to moderate cytologic atypia and up to 10 mitoses per 10 HPF. High-grade carcinomas were characterized by severe cytologic atypia and more than 10 mitoses per HPF. Low-grade carcinomas were generally comparable to FIGO (International Federation of Obstetrics and Gynecology) Grade 1 tumors and high-grade carcinomas were comparable to FIGO Grade 2 and 3 tumors. The 14 normal ovary specimens were chosen based on the presence of at least 1 epithelial compartment (surface epithelium, inclusion cysts, follicles).

Immunohistochemistry

Immunohistochemical staining was performed on 5-μm paraffin-embedded sections using the universal Dako labeled streptavidin-biotin 2 system (Dako LSAB2 System, horseradish peroxidase [HRP], Dako, Carpinteria, CA), as per the manufacturer's instructions. Briefly, sections were deparaffinized in xylene and rehydrated in a decreasing gradient of ethanol in water. Antigen retrieval was performed subsequently in a pressure cooker with 0.01 M citrate buffer, pH 6.0, for 20 minutes. Hydrogen peroxide (0.3%) was applied to quench the endogenous peroxidase activity. The slides were then incubated in protein blocking agent to reduce nonspecific binding. The sections were incubated with primary antibodies for 1 hour. The sections were then washed in phosphate-buffered saline (PBS) to remove unbound primary antibody, after which they were incubated with a biotinylated secondary antibody. The sections were washed in PBS and staining was completed by incubation with streptavidin-HRP and 3,3′-diaminobenzidine colorimetric reagents. Finally, sections were counterstained with hematoxylin.

The antibodies used for immunohistochemical analysis included c-ABL AB-1 monoclonal (Neomarkers, LabVision, Fremont, CA), c-KIT CD117 polyclonal (Dako), and PDGFR-β monoclonal (BD-Transduction Labs, Lexington KY).

The intensity of the immunostaining was graded as negative (no staining), weak (1+), moderate (2+), or strong (3+). Tumors with 2+ or 3+ staining in greater than 10% of the tumor cells were considered to be positive. In parallel with all of the experiments, positive controls (including breast carcinoma for PDGFR-β, placenta for c-ABL, and GIST for c-KIT) were also performed to confirm antibody staining. c-KIT and PDGFR-β typically demonstrated membrane and cytoplasmic staining, whereas c-ABL demonstrated cytoplasmic and nuclear staining with some membrane association.

RESULTS

Expression of c-ABL, c-KIT, and Platelet-Derived Growth Factor Receptor-β in Normal Ovarian Epithelium

c-ABL, PDGFR-β, and c-KIT each demonstrates unique patterns of expression in different epithelial compartments of the normal ovary (Table 1; Fig. 1). c-ABL was expressed highly in all 13 samples of normal ovarian surface epithelium (Fig. 1A). The epithelial lining of ovarian inclusion cysts, however, demonstrated inconsistent immunohistochemical staining for c-ABL. c-ABL was not detected in the follicular epithelium of two normal ovary samples obtained from premenopausal patients. No c-ABL staining was observed in ovarian stromal tissue.

Table 1. Kinase Expression in the Ovary

Tissue	c-ABL	PDGFR-β	c-KIT
PDGFR-β: platelet-derived growth factor-beta.
Normal ovary (%)
Stroma	0 (0/14)	14 (2/14)	7 (1/14)
Surface epithelium	100 (13/13)	92 (12/13)	0 (0/13)
Inclusion cyst epithelium	0 (0/3)	100 (3/3)	33 (1/3)
Follicular epithelium	0 (0/2)	0 (0/2)	100 (2/2)
Low-grade serous carcinoma (%)	81 (17/21)	67 (14/21)	0 (0/21)
High-grade serous carcinoma (%)	65 (20/31)	90 (28/31)	26 (8/31)
All serous carcinoma (%)	71 (37/52)	81 (42/52)	15 (8/52)

Figure 1. Immunohistochemical analysis of c-ABL (A), platelet-derived growth factor receptor-beta (PDGFR-β, B) and c-KIT (C) in normal ovary. Normal ovarian surface epithelium immunohistochemically expressed c-ABL and PDGFR-β. c-KIT was absent in normal ovarian epithelium, but was detectable in normal ovarian primordial follicles. (Magnification: A, × 400; B,C, × 200.)

PDGFR-β was expressed intensely in 12 of 13 (92%) samples of ovarian surface epithelium, as well as in the epithelial lining of ovarian inclusion cysts (Fig. 1B). Like c-ABL, PDGFR-β was undetectable in developing follicles (Table 1). Faint immunoreactivity was observed throughout the ovarian stromal tissue, with intense staining of single stromal cells occasionally observed.

c-KIT was undetectable in normal ovarian surface epithelium and was inconsistently detectable in ovarian inclusion cysts (Table 1). Only 1 of 14 normal ovaries showed positive stromal staining for c-KIT. However, c-KIT was expressed strongly in the primordial follicles of two specimens of normal ovary containing maturing oocytes (Fig. 1C).

Expression of ABL, c-KIT, and Platelet-Derived Growth Factor Receptor-β in Ovarian Serous Carcinoma

The expression of c-ABL, c-KIT, and PDGFR-β proteins was evaluated by immunohistochemical analysis of 52 specimens of serous carcinoma of the ovary, which is the most common histology of ovarian carcinoma. To characterize the grade-specific expression of these kinases, tumors were subdivided into two categories: low-grade (Grade 1; Fig. 2A–C) and high-grade (Grades 2 and 3; Fig. 2D–F).

Figure 2. Immunohistochemical analysis of c-ABL, platelet-derived growth factor receptor-beta (PDGFR-β), and c-KIT in low-grade and high-grade ovarian serous carcinoma. c-ABL shows positive immunohistochemical staining in both (A) low-grade and (D) high-grade tumors. Similarly, PDGFR-β is expressed in the majority of both (B) low-grade and (E) high-grade ovarian carcinomas. (C) Low-grade ovarian serous carcinomas do not demonstrate c-KIT immunohistochemical expression, but (F) c-KIT is detectable in approximately 26% of high-grade serous carcinomas. Magnification: A–F, ×200.

c-ABL was expressed in 71% (37 of 52) of serous carcinomas (Table 1). It was expressed more highly in the low-grade tumors, although the difference between groups was not statistically significant (P = 0.2). Positive staining was observed in 81% (17 of 21) of the low-grade tumors (Fig. 2A) and in 65% (20 of 31) of high-grade serous carcinomas (Fig. 2D).

PDGFR-β immunoreactivity was observed in 81% (42 of 52) of the serous carcinomas. Converse to c-ABL expression, PDGFR-β expression was more common in the high-grade tumors. Ninety percent (28 of 31) of high-grade tumors expressed PDGFR-β (Fig. 2E), whereas 67% (14 of 21) of low-grade tumors stained positively for PDGFR-β (Fig. 2B). Chi-square analysis indicated that PDGFR-β expression was associated significantly with the grade of the tumor (P = 0.03).

c-KIT expression was observed in 15% (8 of 52) of serous ovarian carcinomas overall (Table 1). None of the low-grade tumors expressed c-KIT (Fig. 2C), but c-KIT was expressed in 26% (8 of 31) of the high-grade serous carcinomas (Fig. 2F). In addition, c-KIT expression was associated significantly with high-grade serous carcinoma (P = 0.01).

DISCUSSION

Many novel targeted therapeutics are being developed by the pharmaceutical industry and are in clinical trials for the treatment of a variety of malignancies.13–17 The profiling of tyrosine kinase expression in epithelial ovarian carcinoma may provide oncologists with potential therapeutic targets for the treatment of this malignancy. In the current immunohistochemical study, we evaluated the expression of c-ABL, PDGFR-β, and c-KIT in normal ovaries and in high and low-grade ovarian serous carcinoma. These kinases are the therapeutic targets of imatinib mesylate, a relatively nontoxic tyrosine kinase inhibitor currently being used with considerable success to treat CML and GIST.

The c-ABL tyrosine kinase is a 140-kilodalton (kD) nonreceptor tyrosine kinase that belongs to the src family of tyrosine kinases.29, 30 The effects of c-ABL are mediated by numerous protein-protein and protein-DNA interaction domains in addition to its tyrosine kinase domain. The amino-terminal region of c-ABL contains SH3 and SH2 domains required for protein-protein interactions, as well as the c-ABL tyrosine kinase domain. The carboxy-terminal region of c-ABL contains a DNA-binding domain. A role for c-ABL in transcription and DNA recombination and repair has been postulated. The c-ABL carboxy terminus also contains an actin-binding domain In normal tissue specimens, c-ABL is localized to both the cytoplasm and nucleus, where it is believed to participate in signaling pathways that govern cellular proliferation, differentiation, and apoptosis.

The expression pattern of the c-ABL tyrosine kinase has not been described extensively in normal ovary and ovarian carcainoma. One small study characterizing c-ABL expression in a wide variety of human fetal and adult tissue specimens and tumors reported the expression of c-ABL in four of four ovarian serous carcinomas, but c-ABL was not expressed in the one ovarian small cell carcinoma specimen.31 In the current study, c-ABL was expressed in all samples of normal ovarian surface epithelium and was undetectable in ovarian stroma and ovarian follicles. This pattern of staining is consistent with previous reports of the ubiquitous expression of c-ABL in the ciliated epithelium of the fallopian tube, epididymis, bronchi, and vas deferens.31 When 52 ovarian serous carcinomas were evaluated, c-ABL was expressed in the majority (71%) of tumors. The subcellular localization of c-ABL in both normal ovarian epithelium and in the serous carcinomas was primarily membranous and cytoplasmic. In the majority of tumors, c-ABL is excluded from the nucleus.

The platelet-derived growth factors (PDGF-A, B, C, and D) comprise a family of homodimeric and heterodimeric polypeptides. PDGFs stimulate proliferative and migratory responses in a variety of cells by activating specific receptor tyrosine kinases (e.g., PDGF α- and β receptors).32–34 Studies using knockout mice have demonstrated the critical role of PDGFR signaling in embryonic and postnatal development. The presence of PDGFRs in normal ovary and ovarian tumors has been debated in the research literature. Versnel et al.35 reported the absence of PDGFRs in several ovarian tumor cell lines as well as in six primary epithelial and granulosa cell tumors. However, Dabrow et al.36 demonstrated that in culture, PDGF enhanced the growth of cultured normal ovarian surface epithelial cells, suggesting the presence of PDGFRs on these cells. Dabrow et al. evaluated 21 Stage III ovarian carcinomas for the immunohistochemical presence of PDGFR-α and PDGFR-β. Of these, only one tumor stained positively for PDGFR-α, whereas eight tumors stained positively for PDGFR-β. The authors hypothesized that loss of PDGFR-α and PDGFR-β was related to short median survival times and suggested that growth factor independence was characteristic of more aggressive tumors. Henriksen et al.37 described the absence of PDGFR-β in 45 of 45 epithelial ovarian tumors of various histologies and stages, including 17 serous ovarian carcinomas, and reported the observation of PDGFR-α in 16 of 45 malignant ovarian neoplasms. Both PDGFR-α and PDGFR-β are absent in normal ovarian epithelium. All studies, however, demonstrated that epithelial tumors produce PDGFs.

In the current study, we demonstrated that the surface epithelium of normal ovaries expressed PDGFR−β. In contrast to previously published studies, we also found that 67% of low-grade and 90% of high-grade serous carcinomas express PDGFR-β. In addition, expression of PDGFR-β was associated significantly with high-grade tumors. The previous studies and the current article suggest an autocrine or paracrine mechanism of ovarian carcinoma growth. We hypothesize that the tumors themselves produce PDGFs, which can bind to and activate PDGFRs also expressed by surrounding tumor cells, thereby driving cellular proliferation.

c-KIT is a 145 kD transmembrane receptor tyrosine kinase. It is closely related to PDGFRs, which bind Steel factor, also known as the c-KIT ligand or stem cell factor. c-KIT signaling plays a role in the regulation of hematopoiesis and melanocyte proliferation as well as the proliferation and survival of germ cells.38 The requirement of c-KIT and the c-KIT ligand for oocyte migration during embryonic development and for follicular development in the adult ovary has been described in the research literature.39–41 Consistent with these data, we have confirmed the expression of c-KIT in the primordial follicles of normal human ovaries (Fig. 1C).

c-KIT expression and activation are well documented in a variety of human tumors including germ cell tumors, GIST, acute myelogenous leukemia and other leukemias, neuroblastoma, and breast carcinoma.22, 23, 38, 42 The expression of c-KIT in epithelial ovarian tumors has been described by numerous investigators, but the data vary considerably among groups. Several groups have reported elevated levels of c-KIT in epithelial ovarian tumors. Parrott et al.43 described the expression of KIT in four of four Stage I ovarian tumors and four of four Stage III ovarian tumors. They also described c-KIT immunoreactivity in normal human and bovine ovarian surface epithelium. In addition, these authors demonstrated a growth response to c-KIT ligand in cultured normal ovarian epithelium, suggesting the presence of the receptor.44 Less intense staining of ovarian stromal tissue also was reported. Tonary et al.45 observed no detectable c-KIT immunoreactivity in normal ovarian surface epithelium but found that 76% (19 of 25) of serous carcinomas expressed detectable levels of c-KIT. Loss of c-KIT expression in advanced-stage, malignant histologies of ovarian malignancies also was documented. These data are in agreement with Arber et al.46 who reported c-KIT immunoreactivity in 87% of epithelial ovarian carcinomas. Other groups reported very low levels of c-KIT expression in ovarian carcinoma. Wrigley et al.47 did not detect c-KIT expression in any of 38 ovarian carcinoma specimens. Inoue et al.48 observed c-KIT immunoreactivity in only 2 of 9 (22%) serous carcinomas of the ovary. Disparities with respect to staining for KIT in the ovary may reflect antibody variability, as each group used different antibodies for its immunohistochemistry experiments.44–47

The current study was conducted using the c-KIT CD117 polyclonal antibody (Dako). This antibody was chosen because it is widely used to assess c-KIT expression in formalin-fixed, paraffin-embedded sections for patients with GIST who enroll in imatinib mesylate clinical trials. In agreement with both Tonary et al.45 and Inoue et al.,48 we observed no detectable staining of normal ovarian surface epithelium. Our data most closely resemble those of Inoue et al., as we detected c-KIT in 26% of high-grade serous carcinomas of the ovary. c-KIT was undetectable in low-grade ovarian tumors.

Standardized methods of immunohistochemical staining for c-KIT should be carried out to more accurately characterize c-KIT levels in ovarian tumors. The Dako CD117 antibofy is recommended by Novartis to screen GIST specimens for c-KIT expression and to determine patient eligibility for treatment with imatinib mesylate. This antibody has been identified by Novartis as the most reliable of the commercial anti-c-KIT antibodies available for immunohistochemistry. If patients with ovarian carcinoma are to be considered in clinical trials for imatinib mesylate, we suggest the continued use of the Dako CD117 antibody for c-KIT immunohistochemistry.

The current study indicates that the majority of serous epithelial ovarian tumors express one or more of the kinases targeted by the tyrosine kinase inhibitor, imatinib mesylate. Although positive immunostaining for c-ABL, PDGFR, and c-KIT do not suggest specific functional activities, their presence does suggest that these kinases may play a role in the growth regulation of normal ovarian epithelium and in the development and growth of ovarian serous carcinomas. Activating mutations of c-ABL and c-KIT contribute to the pathogenesis of CML and GISTs. The exact contribution of c-ABL, PDGFR, and c-KIT to the pathogenesis of ovarian carcinoma has yet to be determined. We are currently conducting a Phase II clinical trial of imatinib mesylate in patients with recurrent platinum-resistant, taxane-resistant epithelial ovarian carcinoma, primary peritoneal carcinoma, or fallopian tube carcinoma. Primary tumor tissue must express any or all of c-ABL, PDGFR, or c-KIT for patients to be eligible for these trials. Expression levels of c-ABL, PDGFR, and c-KIT will be tested for correlations with patient outcome. We anticipate that these studies will determine whether these kinases contribute to disease progression and whether imatinib mesylate and related tyrosine kinase inhibitors will be effective in the treatment of ovarian carcinoma.