British Journal of Cancer (2008) 99, 1269 – 1275 & 2008 Cancer Research UK All rights reserved 0007 – 0920/08 $32.00 www.bjcancer.com C-Fos expression is a molecular predictor of progression and survival in epithelial ovarian carcinoma *,1,2 1,2 1,2 1 1 1 1 S Mahner , C Baasch , J Schwarz , S Hein ,LWo¨lber ,FJa¨nicke and K Milde-Langosch 1 Department of Gynecology and Gynecologic Oncology, University Medical Center Hamburg-Eppendorf, Martinistrasse 52, 20246 Hamburg, Germany Members of the Fos protein family dimerise with Jun proteins to form the AP-1 transcription factor complex. They have a central function in proliferation and differentiation of normal tissue as well as in oncogenic transformation and tumour progression. We analysed the expression of c-Fos, FosB, Fra-1 and Fra-2 to investigate the function of Fos transcription factors in ovarian cancer. A total of 101 patients were included in the study. Expression of Fos proteins was determined by western blot analysis, quantified by densitometry and verified by immunohistochemistry. Reduced c-Fos expression was independently associated with unfavourable progression-free survival (20.6, 31.6 and 51.2 months for patients with low, moderate and high c-Fos expression; P ¼ 0.003) as well as overall survival (23.8, 46.0 and 55.5 months for low, moderate and high c-Fos levels; P ¼ 0.003). No correlations were observed for FosB, Fra-1 and Fra-2. We conclude that loss of c-Fos expression is associated with tumour progression in ovarian carcinoma and that c-Fos may be a prognostic factor. These results are in contrast to the classic concept of c-Fos as an oncogene, but are supported by the recently discovered tumour-suppressing and proapoptotic function of c-Fos in various cancer types. British Journal of Cancer (2008) 99, 1269 – 1275. doi:10.1038/sj.bjc.6604650 www.bjcancer.com & 2008 Cancer Research UK Keywords: ovarian cancer; prognostic factors; c-Fos; AP-1; survival; progression Ovarian cancer accounts for the highest tumour-related mortality assessed for prognostic significance over the past decades, but among women with gynaecologic malignancies. The American none of them yielded conclusive and reproducible results (Omura Cancer Society estimates about 23 000 new cases of ovarian cancer et al, 1991; Hornung et al, 2004; Winter et al, 2007). each year in the United States and that 70% of the affected women Recent efforts to develop accurate predictors of clinical outcome will die from their disease (American-Cancer-Society, 2007). have mainly focused on assessment of global gene expression by Although aggressive surgical cytoreduction and platinum-based DNA microarrays. This technology provided information on combination chemotherapy have improved outcome for many differential gene expression in a number of tumours including patients, long-term survival could not generally be improved. ovarian cancer (Hartmann et al, 2005) and has identified gene Identification of additional prognostic factors could help to stratify profiles associated with early relapse and decreased survival patients into different biological subgroups. In ovarian cancer, this (Spentzos et al, 2004). Members of the Fos family (c-Fos, FosB, Molecular Diagnostics is especially important for the group of patients with early relapse Fra-1 (Fos-related antigen 1) and Fra-2) are often represented in (within 6 months after first-line treatment) that usually die within these profiles (Meinhold-Heerlein et al, 2005). They dimerise with 6–12 months (du Bois et al, 2003). These patients (approximately the gene products of c-Jun, JunB or JunD to form the transcription 25%) do not benefit from current treatment modalities while factor Activating Protein 1 (AP-1). AP-1 binds to the promoter suffering from the sometimes severe side effects of therapy. region of specific target genes, converting extracellular signals into Subsequent research could then focus on the establishment of changes of gene expression (Milde-Langosch, 2005). As a member more targeted and individual treatment strategies in this subgroup, of AP-1, c-Fos has been implicated mainly in signal transduction, as previously shown for Her2/neu expression and trastuzumab cell differentiation and proliferation (Shaulian and Karin, 2001). treatment in breast cancer (Pegram et al, 2000). Many studies focused on its oncogenic functions and found that Widely accepted prognostic factors in patients with epithelial c-Fos regulated genes important for tumorigenesis, causing the ovarian cancer are International Federation of Gynecology and downregulation of tumour-suppressor genes (Bakin and Curran, Obstetrics (FIGO) stage and residual tumour volume after primary 1999) and leading to invasive growth of cancer cells (Hu et al, surgical cytoreduction (Bristow et al, 2002; Tingulstad et al, 2003). 1994). Furthermore, c-Fos can induce a loss of cell polarity and Several other clinical and biological factors such as age, epithelial-mesenchymal transition, leading to invasive and meta- performance status, tumour histology and grade have been static growth in mammary epithelial cells (Fialka et al, 1996). In addition to these experimental results, several reports investigated the function of c-Fos expression in human tumour *Correspondence: Dr S Mahner; E-mail: [email protected] tissue. In osteosarcoma and endometrial carcinoma, c-Fos over- 2 These authors contributed equally to this study. expression was associated with high-grade lesions and adverse Received 30 May 2008; revised 13 August 2008; accepted 13 August outcome (Gamberi et al, 1998; Bamberger et al, 2001). In a 2008 comparative analysis between precancerous lesion of the c-Fos expression in ovarian carcinoma S Mahner et al 1270 cervix uteri and invasive cervical cancer, c-Fos expression was Table 1 Patient characteristics significantly lower in precancerous lesions (Prusty and Das, 2005). C-Fos has also been identified as independent predictor of No. of patients 101 decreased survival in breast cancer (Bland et al, 1995). However, some more recent studies have raised the idea that Age (years) Mean 59.2 c-Fos may also have tumour-suppressor activity and might have a Median 61 function in apoptosis (Teng, 2000). Overexpression of c-Fos was Range 21 –87 found to inhibit cell cycle progression, stimulated murine hepatocyte cell death and strongly suppressed tumour formation FIGO stage in vivo (Mikula et al, 2003). A functional involvement of c-Fos in I10 apoptosis has been shown by its regulatory involvement during II 8 remodelling and stress response in various tissues in mouse III 60 development (Jochum et al, 2001). IV 23 Besides c-Fos, FosB, Fra-1 and Fra-2 have also been shown to Grading have a function in progression of various tumour types: FosB is 16 downregulated in poorly differentiated mammary carcinomas 229 (Bamberger et al, 1999), whereas Fra-1 and, partly, Fra-2 358 overexpression leads to enhanced tumour cell motility and Not determined/unknown 8 invasion in breast cancer, colorectal cancer and mesothelioma (Milde-Langosch, 2005). Lymph node metastasis This study investigated the potential function of Fos transcrip- N0 31 tion factors in ovarian cancer and analysed the expression and N1 55 prognostic significance of c-Fos, FosB, Fra-1 and Fra-2 in patients NX 15 with invasive epithelial ovarian carcinoma. Residual tumour after surgery Microscopic 67 o0.5 cm 7 MATERIALS AND METHODS 0.5– 1 cm 10 1–2cm 2 Patients 42cm 6 Not determined/unknown 9 Patients with epithelial ovarian carcinoma who presented for primary surgery at the University Medical Center Hamburg- Histologic subtype Eppendorf between 1997 and 2006 were included in this study. Serous 74 Thus, a total of 101 patients were retrospectively analysed. Detailed Mucinous 6 patient characteristics are listed in Table 1. Clinicopathologic Endometrioid 5 Molecular Diagnostics factors were evaluated by reviewing medical charts and pathologic Clear cell 1 records. Tissue slides were reviewed for histological classification Undifferentiated 7 and clinical outcome was followed from the date of surgery to the Mixed differentiation 8 date of death or until the end of 2007. Two cases were lost to Platinum-based first-line chemotherapy follow-up right after surgery and were excluded from the survival Adjuvant 90 analysis. All patients gave written informed consent to access their Neoadjuvant 6 tissue and review their medical records according to our None/unknown 5 investigational review board and ethics committee guidelines. CA 125 before surgery (kU lÀ1) Tissue samples Mean 2178 Median 436 Fresh-frozen samples were obtained intraoperatively and immedi- Range 21 – 47 000 ately stored at À801C. The histological characteristics of each sample were assessed on cryo-cut and haematoxylin-eosin-stained Survival (months), n ¼ 99 Progression-free survival sections and the tissue was trimmed if necessary to obtain at least Mean 19.7 70% tumour cells in the sample used for protein extraction. Median 15.2 Range 0.4–98 Protein extraction Overall survival Samples of approximately 100 mg were cut from the tissue and Mean 24.4 pulverised using a micro-dismembrator (Braun-Melsungen, Mel- Median 20 sungen, Germany) for 2 Â 45 s at 200 r.p.m. Proteins were lysed in Range 0.4–98 ice-cold sample buffer (50 mM Tris pH 6.8, 1% sodium dodecyl À1 FIGO ¼ International Federation of Gynecology and Obstetrics; kU/l ¼ kilo Units per sulphate (SDS), 10% sucrose and 10 mlml protease inhibitor liter. cocktail (Sigma, Taufkirchen, Germany)), and protein concentra- tion was determined following the standard protocols and using bovine serum albumin protein standards as described previously proteins from the ovarian cancer cell lines Ovcar5 and Ovcar8 and (Bamberger et al, 1999; Milde-Langosch et al, 2005). the breast cancer cell line MCF7 were loaded on each gel. After electrophoresis, blotting to polyvinylidene difluoride membranes 1 Western blot analysis and overnight incubation at 4 C in blocking solution, membranes were incubated for 1 h at room temperature with the following Equal amounts of protein (20 mg) of each sample were loaded per primary antibodies (all from Santa Cruz): c-Fos polyclonal well, and equal loading was verified by immunoblotting with actin antibody no.