Fax: (301) 480-5142
Granulin-epithelin precursor is a novel prognostic marker in epithelial ovarian carcinoma†
Article first published online: 19 APR 2004
Published 2004 American Cancer Society
Volume 100, Issue 10, pages 2139–2147, 15 May 2004
How to Cite
Davidson, B., Alejandro, E., Flørenes, V. A., Goderstad, J.-M., Risberg, B., Kristensen, G. B., Trope, C. G. and Kohn, E. C. (2004), Granulin-epithelin precursor is a novel prognostic marker in epithelial ovarian carcinoma. Cancer, 100: 2139–2147. doi: 10.1002/cncr.20219
This article is a US Government work and, as such, is in the public domain in the United States of America.
- Issue published online: 29 APR 2004
- Article first published online: 19 APR 2004
- Manuscript Revised: 24 FEB 2004
- Manuscript Accepted: 24 FEB 2004
- Manuscript Received: 12 FEB 2004
- Norwegian Cancer Society
- tumor microenvironment
The granulin-epithelin precursor (GEP) was preferentially expressed in invasive ovarian tumor epithelium specimens compared with specimens of borderline ovarian tumors. The objective of the current study was to evaluate the anatomic site-related and cellular expression of GEP and its association with clinicopathologic parameters and survival in patients with advanced-stage ovarian carcinoma.
Effusions (n = 190), corresponding primary tumor specimens (n = 64), and specimens of metastatic lesions (n = 125) were analyzed using immunohistochemistry with a specific polyclonal antipeptide antibody. In addition, 36 effusions were analyzed using immunoblotting.
GEP was detected in tumor cells in 171 of 190 (90%) effusions and demonstrated both focal membrane and cytoplasmic localization. Mesothelial cells were often GEP positive (81%). GEP was found in carcinoma cells in 180 of 189 (95%) tumor biopsy specimens, with stromal and endothelial cell expression in 93 of 180 (52%) and 124 of 185 (67%) specimens, respectively. Lower GEP expression in stromal cells was observed in metastases sampled during or after chemotherapy (P = 0.034). The presence of GEP-positive stromal cells in untreated primary tumor specimens correlated with worse overall survival (P = 0.014). Significantly more frequent GEP expression was observed in tumor cells of both primary (P = 0.002) and metastatic (P < 0.001) tissue specimens compared with malignant effusions.
GEP expression was observed in primary and metastatic epithelial ovarian carcinoma specimens, with down-regulated expression in tumor cells of malignant effusions. The poor outcome associated with stromal GEP expression suggests a prognostic role for this growth factor in ovarian carcinoma. Cancer 2004. Published 2004 American Cancer Society.
Ovarian carcinoma is the leading cause of death from gynecologic carcinoma among women in industrialized countries.1 Despite improved chemotherapy regimens, only modest improvement has been achieved in patient long-term survival. Metastases to serosal surfaces and associated peritoneal and/or pleural effusions are found in most patients diagnosed with epithelial ovarian or primary peritoneal carcinoma. This is most prominent in the serous or clear cell type, histologic types that are known for their aggressive behavior. Studies of effusions and corresponding solid tumor specimens have demonstrated that cells in effusions are different from their counterparts in solid primary and metastatic lesions.2 Understanding these differences will yield insight into events in tumor progression of ovarian carcinoma.
The granulin-epithelin precursor (GEP/ progranulin/ PC-cell-derived growth factor) is a 68-kilodalton (kD) secreted protein.3, 4 Glycosylation of GEP leads to the formation of multiple higher molecular weight forms, the most common of which is the 88-kD protein.5, 6 GEP also is cleaved into its component granulins (epithelins), small proteins of 6 kD.7, 8 Granulins have been shown to have inhibitory function, opposing that of GEP.7 Expression of the progranulin gene (pgrn) has been documented in cell lines of epithelial, hematopoietic, and mesenchymal origin, as well as in epithelial cells undergoing rapid turnover.9, 10 GEP has been shown to have a role in both physiologic and pathologic processes. It is expressed in various stages of development, including in gametes, blastocysts, implantation, the mature placenta, and embryonic tissues.11, 12 Expression in physiologic processes, such as wound repair,13 has been documented. We previously identified GEP as a growth factor for ovarian carcinoma through its up-regulation in tumor cells of invasive ovarian carcinomas compared with tumors of low malignant potential (LMP). Its broad expression in stromal cells, including endothelial cells, also was shown.14 Transfection of OVCAR-3 ovarian carcinoma cells with antisense GEP reduced cell growth and S-phase fraction, and resulted in loss of density-independent growth, supporting a role for GEP in ovarian carcinoma growth and tumorigenicity.14 The objective of the current study was to analyze the expression and cellular distribution of GEP in primary ovarian carcinomas, their solid metastases, and effusions. Expression was scored in carcinoma cells, reactive mesothelial cells, and macrophages in effusions, and in carcinoma cells, stromal cells, and endothelium of solid tumor specimens to provide comprehensive evaluation of GEP expression in the tumor and its microenvironment. In the current study, we found that GEP is expressed by all cell types in the tumor and its microenvironment. Expression by stromal cells portends a worse survival outcome for patients with ovarian carcinoma. GEP may be an important new molecular therapeutic target.
MATERIALS AND METHODS
Specimens and Patient Data
All specimens and relevant clinical data were obtained from the Department of Gynecologic Oncology, Norwegian Radium Hospital (Oslo, Norway). Table 1 shows the clinicopathologic data of the study cohort.
|Parameters||No. of specimens|
|Median age (range)||61 (38–88 yrs)|
The cytologic material was comprised of 190 fresh nonfixed malignant peritoneal and pleural effusions submitted to the Section of Cytology, Department of Pathology, The Norwegian Radium Hospital, during the period between January 1998–December 2002. Informed consent was obtained according to national Norwegian and institutional guidelines. Effusion specimens (150 peritoneal and 40 pleural specimens) were obtained from 156 patients; 176 effusion specimens were obtained from 145 patients diagnosed with epithelial (predominantly serous) ovarian carcinoma), 6 effusion specimens were obtained from 5 patients with serous carcinoma of the fallopian tube, and 8 effusion specimens were obtained from 6 patients diagnosed with primary peritoneal carcinoma (PPC). Specimens arrived within minutes of removal from the patient and were processed immediately. Cells were suspended and frozen in equal volumes of RPMI supplemented with 20% fetal calf serum and 20% dimethylsulfoxide. Smears and cell block sections from formalin-fixed, paraffin-embedded pellets underwent diagnostic morphologic evaluation by three experienced cytopathologists, and then were characterized further using immunocytochemistry with broad antibody panels against carcinoma, mesothelial, and leukocyte epitopes, as previously detailed.15–17 Fixation and block preparation procedures were identical to those employed for biopsy specimens.
One hundred eighty-nine surgical specimens from 78 patients (74 patients with ovarian carcinoma, 3 with PPC, and 1 with carcinoma of the fallopian tube) were studied. These specimens were comprised of 64 ovarian and 125 extraovarian lesions (58 omental, 32 peritoneal, 19 intestinal, and 16 biopsies from other sites). Treatment status was known for 168 specimens. Of these, 113 specimens were obtained before the administration of chemotherapy, whereas the remaining 55 were specimens resected at interval debulking or were retrieved from patients with recurrent disease. Formalin-fixed, paraffin-embedded tissue blocks were obtained from archival material at the Department of Pathology. All tissue specimens underwent microscopic confirmation of diagnosis, tumor type, and histologic grade using established criteria.18
Production of the Anti–GEP Antibody
Generation and peptide affinity purification of GEP polyclonal antibodies were previously described.14 The antibody used in this series of studies recognizes the 68-kD unglycosylated GEP as well as various glycosylated forms of higher molecular weight, including the predominant 88-kD glycoprotein.5, 6 It has very low affinity for the small molecular weight granulins. A second antibody (no. 737), which is more selective for granulins, directed against amino acids 187–198 (sequence: LAKKLPAQRTNRA) was raised and affinity purified. The specificity of these anti–GEP antibodies was confirmed by competition assays using cognate peptide to demonstrate the selectivity of the GEP antibody used for immunohistochemistry (IHC) for the precursor molecule (Fig. 1).
Sections from paraffin-embedded blocks were deparaffinized by xylene and serial ethanol dilutions. Staining for GEP was optimized and specifically confirmed by peptide competition. Microwave oven antigen retrieval was performed 4 times for 5 minutes in standard citrate buffer (pH = 6). Staining was accomplished with an overnight incubation with a 1:1000 dilution of the antipeptide antibody. Visualization was achieved using the Envision peroxidase system (Dakocytomation, Glostrup, Denmark). Positive controls for these antibodies consisted of an ovarian carcinoma case shown to be positive and specific in a pilot study. Negative controls were stained with an antibody for mouse myeloma protein.
Staining in effusion specimens was scored in carcinoma and reactive cells (mesothelial and macrophages) and in solid tumors in the epithelial, stromal, and endothelial cell compartments. The staining extent in all compartments except the endothelium was scored on a scale of 0–4, corresponding to the percentage of stained cells of 0%, 1–5%, 6–25%, 26–50%, and 51–100%, respectively. At least 500 cells were scored, when present (> 90% of specimens). No specimen contained < 100 tumor, stromal, or reactive cells. Endothelial cell expression was scored as negative (0) or positive (1). All slides were evaluated blinded by two of the authors (B.D. and B.R.). Intensity was not scored because of the difficulty in ruling out fixation-related variation.
Frozen cells available from 32 malignant and 4 reactive effusions were subjected to lysis in standard radioimmunoprecipitation (RIPA) buffer followed by immunoblot for GEP. Twenty-one malignant effusions contained a tumor cell population > 50% of cells. Twenty-five micrograms of lysate was loaded from each effusion sample. Lysis and immunoblotting procedures were as previously described.19 A mouse monoclonal antibody against α-tubulin (clone 57; Oncogene, Cambridge, MA) was evaluated as a loading control. Densitometric analysis of expression intensity was measured using the National Institutes of Health (NIH; Bethesda, MD) Image 1.63 program.
Statistical analysis was performed applying the SPSS-PC package (version 10.1; SPSS Inc., Chicago, IL). P < 0.05 was considered statistically significant. Complete clinicopathologic data were available for 149 patients. Studies of the association between GEP expression in effusions and solid tumor specimens and clinicopathologic parameters were undertaken using the two-sided chi-square test. Comparative analyses of staining results in tumor cells in effusions versus specimens of primary tumors and solid metastases were executed using the Wilcoxon signed rank test. Univariate survival analyses were executed using the Kaplan–Meier method and log-rank test. Expression categories in these tests were clustered as described in the text to allow for a sufficient number of specimens to be included in each category.
GEP is Expressed in Both Ovarian Carcinoma Cells and Reactive Cells in Effusions
Solid tumors, metastases, and malignant and reactive effusions are common in ovarian carcinoma.20 They are an early event in ovarian carcinoma progression and are attributed to tumor shedding. How these cells differ from the tumor mass cells remains unknown. We previously found that GEP is up-regulated in primary invasive ovarian carcinoma.14 We evaluated the expression of GEP at different anatomic sites that are frequently involved in the metastatic spread of ovarian carcinoma. We therefore analyzed protein expression of this growth factor in both tumor and reactive cells in pleural and peritoneal effusions. GEP was detected in the tumor cells in 171 of 190 (90%) effusions (Fig. 2A–C). Cytoplasmic staining was observed in > 25% of tumor cells in 94 (49%) specimens (Table 2). Membrane expression for GEP on carcinoma cells most often was dot-like in pattern, was observed in 43 effusion specimens, and was limited to < 5% of cells in 41 of 43 (95%) specimens. GEP expression was not significantly different in carcinoma cells in peritoneal (ascites) compared with pleural effusions. Mesothelial cells and/or macrophages were present in significant numbers (> 100 cells) in 109 effusions. GEP expression was observed in reactive cells in 88 of these specimens (81%; Fig. 2C). Staining in reactive cells in effusions generally was observed in a higher percentage of cells compared with carcinoma cells. Membrane expression was less frequent, observed in only 13 of 109 (12%) specimens. Immunoblot of effusion cell lysates revealed expression of at least 1 GEP isoform in all 32 malignant and in 4 reactive effusions tested. Specimens showed multiple previously documented GEP isoforms, including the native 68-kD and primary 50-kD forms, as well as the 78, 88, and 100-kD forms (Fig. 3).
|Specimens||Staining extent (% of cells)||Total|
GEP is Expressed by all Tissue Compartments in Solid Tumors
GEP expression was found in malignant, stromal, and tumor-associated endothelial cells in a previous report.14 Cytoplasmic GEP protein expression was found in carcinoma cells in 180 of 189 (95%) specimens and was observed in > 25% of cells in 155 (82%) specimens (Table 2). Focal (< 5%) membrane expression was observed in malignant cells in 97 specimens (with a dot-like pattern similar to that observed in effusions). Stromal fibroblasts expressed GEP in 93 of 180 (52%) specimens (Table 2; Fig. 2D–I). Endothelial cells stained for GEP in 124 of 185 (67%) specimens (37 primary, 87 metastatic). Staining in tumor cells was generally comparable at all sites for a given patient, with greater variation in stromal and endothelial expression. Statistical analysis failed to demonstrate any significant site-related differences in expression.
GEP Expression is Reduced in Carcinoma Cells in Effusion Samples Compared with Corresponding Solid Tumor Specimens
Previous studies of this patient cohort have demonstrated significant differences in expression of metastasis-related molecules in effusions compared with both solid primary and metastatic tumor specimens.2 Despite the frequent expression of GEP in carcinoma cells at all anatomic sites, it appeared that expression in effusions was more limited. A comparative analysis of primary tumor specimens and their respective effusions showed reduced cytoplasmic (P = 0.009) and membrane (P = 0.002) expression in effusions. A similar analysis of solid metastatic specimens and corresponding effusions documented a still greater difference (P < 0.001 for both cytoplasmic and membrane expression).
GEP Expression is Reduced by Chemotherapy
If GEP is an important growth factor, then its expression may be reduced in patients who have had some response to chemotherapy. We examined whether GEP expression correlates with any of the established parameters affecting outcome among patients with in ovarian carcinoma. Significantly lower GEP expression was found to be present in stromal cells of residual disease in metastases sampled after chemotherapy (P = 0.034), with loss of GEP expression in patients who received platinum compounds (P = 0.034; Table 3). No correlation of GEP expression with grade, International Federation of Gynecology and Obstetrics (FIGO) stage of disease, or the extent of residual disease after primary debulking was observed. Expression in effusions was similarly independent of clinicopathologic parameters.
|Treatment||Staining extent (% of cells)||P value|
Inverse Relation between GEP Expression in Different Cellular Compartments and Survival
The mean and median follow-up periods were 24 months and 21 months, respectively (range, 1–83 months). At the last follow-up, 12 patients showed no evidence of disease, 46 were alive with disease, and 91 were dead of disease. FIGO stage was the only predictor of survival in this cohort (P = 0.011) using univariate survival analysis. GEP protein expression by IHC in carcinoma cells in effusions demonstrated no correlation with survival. Immunoblotting results indicated that malignant effusions expressing a high level of the 88-kD form correlated with significantly better overall survival (OS; 49 months vs. 29 months, P = 0.031; Fig. 4), with a trend toward improved disease-free survival (DFS 18 months vs. 6 months, P = 0.08). Trends for poor survival were observed in tumor specimens with GEP-positive stromal cells when the entire cohort was analyzed (P = 0.08 for primary tumor specimens; P = 0.09 for metastatic tissue specimens). Analysis of this expression in untreated tumor specimens allowed examination of its potential as a prognostic marker. The presence of GEP-positive stromal cells in untreated primary tumor specimens correlated with worse outcome (P = 0.014; OS of 23 months vs. 36 months; Fig. 5), with a similar trend in metastatic tissue specimens (P = 0.057; OS of 28 months vs. 35 months). Expression in endothelial and tumor cell populations in primary and metastatic tissue specimens was associated with a trend for better survival in the entire cohort. Patients with metastatic tumors in which endothelial cells were GEP-positive had an OS of 39 months compared with 30 months for GEP-negative lesions (P = 0.07).
Our identification of GEP as a putative autocrine growth factor for epithelial ovarian carcinoma with expression also in the stroma of LMP and invasive malignant tumors14 led to the current hypothesis that it may be of biologic significance in this malignancy. GEP may self-regulate through its degradation to component granulins, which have proinflammatory and antiproliferative function.13 GEP also has a newly demonstrated role in angiogenesis.21 We evaluated a series of advanced ovarian carcinoma solid tumor and effusions with a GEP-selective antibody. GEP was expressed, as expected, in most tumors and its expression was not limited to carcinoma cells but was frequently found in reactive mesothelial cells and macrophages in effusions, as well as in stromal and endothelial cells in solid tumor specimens. GEP expression was found to be more prominent in the tumor cells in solid tumor specimens than in effusions. Production of the 88-kD form in effusions was associated with improved outcome, whereas stromal cell expression was significantly associated with worse OS. Multivariate analysis of a limited cohort for which frozen material was available for Western blot analysis did not demonstrate independence of GEP expression (P = 0.1; n = 30). This may be because of the small numbers or a uniform expression of this growth factor in more advanced cancers.14
Consonant with the findings that stromal production of GEP was associated with worse prognosis is the known ability of stromal cells activated by the presence of tumor to synthesize a wide array of growth and angiogenic factors and metastasis-associated molecules.22 The expression of growth factors and metastasis-associated molecules has been demonstrated both in solid tumor specimens and effusions and has been linked to poor outcome.23–26 A possible reason for the apparently discrepant correlation with prognosis for tumor cells in effusions and stromal cells in solid tumor specimes may reside in the fact that tumor cells at the former site differ from those in solid tumors.2 In addition, stromal cell expression may trigger different signaling pathays than those observed in tumor cells.
Stromal growth and angiogenic factor expression often coincides with endothelial expression of proinvasive and prometastatic factors. Our finding of tumor-associated endothelial cell expression of GEP follows this trend. Recent data demonstrate a proangiogenic activity of GEP in wound healing.20 Mesothelial cells in effusions have been demonstrated to secrete factors that can promote the local malignant milieu, such as angiogenic and growth factors, cytokines, and metalloproteinases (MMP).2, 27 The current study results demonstrate that GEP is produced in both solid and effusion tumor microenvironments and is a characteristic of all cell types present. Whether stromal and vascular expression is a response to the carcinoma, or a stimulus to which carcinoma cells respond in a paracrine fashion, cannot be assessed from the current study. This is being modeled currently in vitro. The findings that stromal GEP expression is associated with poor prognosis in univariate analysis and that it has broad expression in advanced disease make it an ideal target for molecular therapeutic targeting.
Despite the ubiquitous presence of GEP, the percentage of carcinoma cells expressing this growth factor was found to be lower in the malignant effusion cells compared with tumor cells in primary and metastatic sites. We have previously reported on the extensive molecular changes that carcinoma cells in effusions undergo compared with corresponding solid tumor specimens.2 These include reduced expression of both cytokines and growth factors, such as interleukin-8 and vascular endothelial growth factor,28 and growth factor receptors such as the high-affinity nerve growth factor receptor, TrkA.29 Up-regulation of proteases, such as MMP-230 also was observed concomitantly. These observations suggest a difference in the interplay between carcinoma cells and host-related cell populations in the different local environments. The down-regulation in GEP expression in carcinoma cells in effusions is in agreement with these previous findings. One mechanism to explain this is the prominent expression of GEP by activated mesothelial cells in the malignant effusions. Stromal GEP production and secretion may supplement the local milieu and reduce the necessity for tumor cell expression.
To our knowledge, the identity of the downstream targets of GEP signaling is currently unknown and its receptor remains undescribed. Mechanisms through which GEP production is regulated in ovarian carcinoma currently are under investigation. Preliminary data from our laboratory show that activation of the ERK signaling pathway downstream of known ovarian carcinoma growth factor exposure is associated with elevated GEP message and protein in ovarian carcinoma cell lines (unpublished data). We recently reported that expression and phosphorylation of ERK, SAPK/JNK, and p38 MAPK was correlated with improved outcome in 64 ovarian carcinoma effusions.31 Forty-two specimens from this case cohort were also studied for GEP expression. In agreement with the in vitro data, we found a trend for the greater expression of GEP in tumor specimens with activated ERK (P = 0.13). Furthermore, both ERK phosphorylation31 and the expression of the 88-kD GEP were found to be correlated separately with better survival. This supports in vitro results indicating that this pathway may be important in mediating GEP events in ovarian carcinoma. GEP production also has been shown to be increased by exposure to estrogens in breast carcinoma.32 Inhibition through ERK and these other pathways may synergize to reduce stromal and tumor production of GEP.
The finding of common expression of GEP in all tissue specimens of primary and metastatic epithelial ovarian carcinoma suggests that GEP is involved in the tumor process. Modeling in vitro will be necessary to dissect these events to determine whether stromal GEP is a response to the tumor or whether tumor GEP production is regulated by the stroma. The poor prognosis afforded by stromal cell GEP expression underscores the importance of understanding the role and focusing therapy to this aspect of the microenvironment. That GEP is present in the solid and liquid tumor specimens has potential significance with regard to its role as a proliferation and survival factor,14 making GEP a logical new target for molecular stromal therapy of ovarian carcinoma.
The authors gratefully acknowledge the competent technical help of Mrs. Inger-Liv Nordli, Mrs. Mai Nguyen, Mrs. Erika Thorbjørnsen, Mrs. Ann Larsen (immunohistochemistry), and Mrs. Martina Skrede (immunoblotting) at the Department of Pathology, The Norwegian Radium Hospital.
- 16Carbohydrate antigen expression in primary tumors, metastatic lesions, and serous effusions from patients diagnosed with epithelial ovarian carcinoma—evidence of up-regulated Tn and Sialyl Tn antigens expression in effusions. Hum Pathol. 2000; 31: 1081–1087., , , et al.
- 18Surface epithelial-stromal tumors. In: SternbergSS, AntonioliDA, CarterD, MillsSE, ObermanHA, editors. Diagnostic surgical pathology. Philadelphia: Lippincott Williams & Wilkins, Inc., 1999; 2319–2382., , .