SEARCH

SEARCH BY CITATION

Keywords:

  • ovarian cancer;
  • interferon regulatory factors;
  • cancer immunology;
  • prognosis

Abstract

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. References

IRF-1 and IRF-2 expression was determined by real-time PCR in 138 ovarian cancer samples and 30 healthy ovarian biopsies and was correlated with the expression of other relevant immunologic parameters and common clinicopathologic variables. Regulation of IRF-1 and IRF-2 was evaluated by cytokine treatment of various ovarian cancer cell lines, human peritoneal mesothelial cells and ovarian surface epithelium. IRF-1 but not IRF-2 was constitutively over-expressed in 5 of 7 ovarian cancer cell lines. Both IRFs were inducible with IFN-γ and to a lesser extent with IL-1 or TNF-α, but not with IL-6. Epidermal growth factor (EGF) treatment down-regulated both IRFs. In ovarian cancer samples only IRF-1, but not IRF-2 mRNA, was up-regulated when compared with healthy ovarian tissue. IRF-1 but not IRF-2 expression was significantly associated with interferon (IFN)-γ and forkhead box P3 (FoxP3). In univariate survival analysis, strong expression of IRF-1 and IRF-2 predicted improved disease-free survival (DFS) and overall survival (OS). In Cox regression analyses, IRF-1 retained independent prognostic significance for DFS and OS and IFN-γ for OS. In contrast to other solid tumors, IRF-2 expression cannot be regarded as a classic oncoprotein associated with poor prognosis in ovarian cancer. Of the immunologic parameters investigated, intratumoral IRF-1 expression is the most powerful independent predictor of a favorable clinical outcome. © 2008 Wiley-Liss, Inc.

Epithelial ovarian cancer is the second-most common gynecologic cancer, with an incidence of about 15 cases per 100,000 women in Western countries and ∼205,000 new cases and 125,000 deaths worldwide, annually.1 Most of these tumors are diagnosed at an advanced stage; more than half of such patients achieve remission after surgical debulking and primary chemotherapy, but overall 5-year survival remains poor at less than 40%. Although tumor stage, residual disease after primary debulking surgery and responsiveness to chemotherapy decisively affect outcome, there is still considerable variability in progression-free and overall survival among patients with similar clinical and pathological characteristics. One explanation for these differences in patient outcome could be that ovarian cancer is a tumor entity in which malignant progression has been shown to be particularly affected by the interplay between the host's immune system and the tumor. Recently, Zhang et al.2 reported that the presence of intratumoral CD3+ T cells independently correlated with delayed recurrence or delayed death in multivariate analysis and was associated with increased expression of interferon-gamma (IFN-γ), interleukin-2 and lymphocyte-attracting chemokines in the tumor. The absence of intratumoral T cells was associated with increased levels of vascular endothelial growth factor. Furthermore, Curiel et al.3 demonstrated that in the cellular interplay of local anti-tumor defense CD4+CD25+ FoxP3-expressing regulatory T cells (Tregs), which are key mediators of immune tolerance, play a crucial role. The number of these tumor-infiltrating Tregs revealed to be inversely and independently related to patient survival in ovarian cancer. We recently found that IFN-γ mRNA expression in ovarian cancer specimens is an independent prognosticator with regard to progression-free and overall survival.4 These results might indicate that IFN-γ expression in ovarian cancer-infiltrating T-lymphocytes directly inhibits tumor growth and furthermore represents a surrogate for high activity of cytotoxic T-lymphocytes (CTL) and stimulation of other immune competent cells. Moreover, clinical trials evidenced significant anti-tumor activity of IFN-γ administered via either the intraperitoneal or the subcutaneous route in the first line treatment of advanced ovarian cancer.5, 6 Although these results, at least for low-dose subcutaneous IFN-γ application, have not been confirmed in a larger randomized trial,7 the above-mentioned findings nonetheless indicate an important role of IFN-γ in the biology of ovarian cancer.

Since the discovery of interferons, a great deal of progress has been made in understanding how these pleiotropic cytokines work. The family of interferon regulatory factors (IRFs), which is composed of 9 individual members, emerged as the most prominent group of transcription factors in the regulation of interferon-related cellular responses in antiviral defense, immune modulation or in control of cell growth and survival.

The most relevant members of this family are IRF-1 and IRF-2. Although IRF-1 activates transcription of many IFN-γ-inducible genes thus causing inhibition of cell proliferation and stimulation of apoptosis,8, 9 IRF-2 binds to the same DNA sequences as IRF-1 and blocks the transcription of IRF-1 target genes.10 Generally, IRF-2 is synthesized after IRF-1 on IFN-γ stimulation and appears to play a key role in inhibiting the cellular effects of IFN-γ mediated by IRF-1. There is accumulating evidence to suggest that both factors are critically involved in tumorigenesis and tumor progression, in that IRF-1 exhibits anti-oncogenic properties whereas IRF-2 acts as an oncoprotein. Over-expression of IRF-2 in NIH3T3 fibroblasts was shown to promote their malignant transformation and tumorigenicity in a nude mouse model. Furthermore, these oncogenic effects were successfully reversed by forced expression of IRF-1.10 In human melanoma and breast cancer, loss of IRF-1 and gain of IRF-2 expression was associated with a more malignant phenotype.11, 12

In various ovarian cancer cell lines growth suppression and apoptotic index upon IFN-γ exposure were highly related to the inducibility of IRF-1 in the respective cell lines. Moreover, in PA-1 cells, ligand-independent apoptosis was strongly induced when IRF-1 was transiently over-expressed. Interestingly, coexpression of IRF-1 together with IRF-2 abrogated the proapoptotic effect of IRF-1.13 However, the clinical significance of both molecules, with regard to prognosis and possible associations with relevant clinicopathologic parameters, has not been investigated thus far in ovarian cancer. Therefore, we quantitatively assessed the constitutive expression and the inducibility of both regulatory factors on IFN-γ IL-1, IL-6, TNF-α and EGF treatment in various ovarian cancer cell lines. In addition, we determined the expression pattern of IRF-1 and IRF-2 by means of quantitative RT-PCR in normal ovarian tissue and in 138 ovarian cancer specimens and correlated these findings with the most relevant clinicopathological features of the disease and with the intratumoral presence of mRNA of other immunological parameters.

Material and methods

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. References

Patients

Tissue samples (n = 138) from patients with ovarian cancer were collected during primary debulking surgery at the Department of Obstetrics and Gynecology in Innsbruck between 1998 and 2006. Selection of cancer specimens was performed on a random basis using the registration code from the gynecopathologic unit. The collective represents 60% of patients (n = 232) undergoing primary surgery for ovarian cancer at our department during the study period. Only epithelial ovarian cancers were included in this study and tumors with borderline malignancy were excluded. Ovarian tissue samples obtained from postmenopausal patients during surgery for other than inflammatory or malignant conditions served as control (n = 30). Use of tissue samples for research purposes was accorded by the patients' written informed consent and approved by the local Institutional Review Board.

The median observation period of the included patients was 44.52 (20.76–102.00) months. Clinicopathologic characteristics are summarized in Table I. Primary debulking surgery was performed in all but 3 patients, who received carboplatin-based chemotherapy due to impaired performance status. With the exception of 15 patients who presented in FIGO stage I a and b with low-grade tumors, all patients received 6 cycles of a platinum-based chemotherapy (89.1%; n = 123), 55.1% (n = 76) of the study patients died during follow-up.

Table I. Clinicopathologic Characteristics of Included Patients with Primary Ovarian Cancer Stage I–IV
Characteristicsn 
  • 1

    Variables are given as median and interquartile range.

  • 2

    Histopathologic grading of two patients was not available.

  • 3

    15 patients with low-grade, stage Ia and b tumors did not receive chemotherapy.

  • y, year; m, month; DFS, disease-free survival; OS, overall survival; FIGO, Fédération Internationale de Gynécologie et d'Obstétrique.

Age (y)113862 (51–72)
Median DFS (m)1 24 (9.9–89.2)
Median OS (m)1 45 (20.7–102.0)
 n%
FIGO stage
 I28(20)
 II12(9)
 III78(56)
 IV20(15)
Histologic subtype
 Serous64(46)
 Mucinous42(30)
 Endometrioid23(17)
 Undifferentiated9(7)
Histopathologic grading2
 15(4)
 276(56)
 355(40)
Residual disease
 None52(38)
 Yes86(62)
Debulking surgery
 Yes138(100)
 No0(0)
Chemotherapy3
 Yes123(89)
 No15(11)

Cell lines

Healthy ovarian surface epithelium cells (OSE) and human peritoneal mesothelial cells (HPMC) were derived from normal ovaries and samples of omental tissue, respectively. Specimens were removed during surgery for other than inflammatory or malignant conditions, after patients' written informed consent had been obtained. Cells were isolated and cultured using methods previously described elsewhere.14, 15 OSE and HPMC were used as controls in all subsequent in vitro experiments. Various established ovarian cancer cell lines, HTB-77, HOC-7, OVCAR-3, SKOV-6, A2780, AG-6000, A2774, and the control cells were cultured and passaged in Dulbecco's modified minimum essential medium (Biochrom KGF, Schöller Pharma, Vienna) containing 10% fetal bovine serum, 1% l-glutamine, 0.2% penicillin, 0.2% streptomycin and 1% nonessential amino acids (all from PAA Laboratories GmbH Linz, Austria). Cells were plated in 25-cm2 culture flasks (Falkon; Becton Dickinson, Franklin Lakes, NJ) and allowed to attach overnight. Approximately 75–80% confluent cells were washed with phosphate-buffered saline solution and treated with Imukin® (Boehringer Ingelheim, Vienna, Austria), rh-IL-1, rh-IL-6, rhTNF-α (ImmunoTools, Friesoythe, Germany) and rhEGF (Roche Molecular Biochemicals, Mannheim, Germany) over 1, 3, 6 and 12 hr. The final concentration of the reagents was 10 ng/ml. At the end of incubation cells were counted with an electronic particle counter (Coulter, Dunstable, UK) and collected for subsequent total RNA isolation.

RNA extraction and RT reaction

Total RNA was isolated from patient samples using the guanidium thiocyanate-phenol-chloroform method according to the manufacturer's protocol (RNAgents® Total RNA Isolation System, Promega, Madison, WI). Integrity was evaluated by assessing the 18 S and 28 S ribosomal RNA bands in 1% ethidium bromide-stained agarose gels. To remove any contaminating genomic DNA, DNAse treatment of typically 4 μg total RNA was performed according to the manufacturer's instructions (Roche Molecular Biochemicals, Mannheim, Germany). Total RNA was stored at −80°C or immediately used for subsequent reverse transcription.

Reverse transcription of typically 2 μg total RNA was performed in a final volume of 25 μl containing 1× RT buffer (50 mM Tris HCl, pH 8.3, 75 mM KCl, 5 mM MgCl2), 40 U of rRNAsin RNAse Inhibitor® (Promega, Madison, WI), 10 mM dithiothreitol, 250 nM random hexamers and 200 U of M-MLV reverse transcriptase (Invitrogen). Incubation periods were 10 min at 25°C and 50 min at 37°C, followed by heating at 70°C for 15 min to inactivate the reverse transcriptase enzyme.

Primers and probes

Gene expression Assay Mixes for CD3E and IFN-γ were purchased from Applied Biosystems (Applied Biosystems Assay ID: Hs00167894_m1 [CD3E], Hs00174143_m1 [IFN-γ]). Specific primers and probes for IRF-1, IRF-2, FoxP3, iNOS, SOCS1 and for the TATA box-binding protein (TBP, a component of the DNA-binding protein complex TFIID as an endogenous RNA control) were determined with the computer program “Primer Express” (Applied Biosystems, Foster City, CA). To prevent amplification of contaminating genomic DNA, the forward primer or the probe was placed at a junction between 2 exons. Sequences of primers and probes are shown in Table II.

Table II. Primers and Probes Used for Real-Time PCR
GeneDirectionSequence
  1. Gene expression Assay Mixes for CD3E (Hs00167894_m1) and IFN-γ (Hs00174143_m1) were purchased from Applied Biosystems.

IRF-1Forward Primer5′-TTT GTA TCG GCC TGT GTG AAT G-3′
Reverse Primer5′-AAG CAT GGC TGG GAC ATC A-3′
Probe5′-FAM-CAG CTC CGG AAC AAA CAG GCA TCC TT-TAMRA-3′
IRF-2Forward Primer5′-CGC CCC TCG GCA CTC T-3′
Reverse Primer5′-TCT TCC TAT GCA GAA AGC GAA AC-3′
Probe5′-FAM-TTC ATC GCT GGG CAC ACT ATC AGT-TAMRA-3′
iNOSForward Primer5′-CCA ACA ATG GCA ACA TCA GG-3′
Reverse Primer5′-TCG TGC TTG CCA TCA CTC C-3′
Probe5′-FAM-CGG CCA TCA CCG TGT TCC CC-TAMRA-3′
FoxP3Forward Primer5′-TGG CTA GGA AAA TGG CA-3′
Reverse Primer5′-GCA GGA GCC CTT GTC GG-3′
Probe5′-FAM-TGA CCA AGG CTT CAT CTG TGG CAT CA-TAMRA-3′
SOCS-1Forward Primer5′-TTT TCG CCC TTA GCG TGA AG-3′
Reverse Primer5′-CAT CCA GGT GAA AGC GGC-3′
Probe5′-FAM-CCT CGG GAC CCA CGA GCA TCC-TAMRA-3′

Real-time PCR amplification

RT-PCRs were performed using an ABI Prism 7900 Detection System (Applied Biosystems, Foster City, CA) in a total volume of 25 μl reaction mixture containing 5 μl of each appropriately diluted RT sample (standard curve points and test samples), 12.5 μl TaqMan Universal PCR Master Mix (Applied Biosystems, Foster City, CA), 900 nM of each primer and 250 nM of the probe. Cycling conditions were an initial step at 50°C for 2 min, a denaturing step at 95°C for 10 min, and 45 cycles at 95°C for 15 sec and 65°C for 1 min. Serially diluted cDNA isolated from the ovarian cancer cell line HTB-77 was used to calculate the standard curve. Real-time PCR assays were conducted in triplicate and the mean value was used for calculation. Levels of IRF transcripts detected in patient samples were normalized to the TATA box-binding protein (TBP).

Immunohistochemistry

IRF-1 and IRF-2 expression was determined in a subset of 25 paraffin-embedded 4 μm sections of ovarian cancer specimens. Sections were pretreated with microwave (80°C, 30 min) for antigen retrieval, and the primary antibody human anti-IRF-1 (sc-497) and human anti-IRF-2 (sc-498) (Santa Cruz) was applied at a dilution of 1:500. Standard ABC technique (ABC-Elite-Kit, Vector Laboratories, Burlingham, CA) was followed for immunostaining. Slides were then put into a diaminobenzidine solution as chromogen and were finally counterstained with Mayer's Hemalum solution (Merck, Darmstadt, Germany).

Statistical analysis

Differences in expression levels of IRF-1 and IRF-2 mRNAs between normal and malignant tissues were assessed with the Mann-Whitney U Test. Because IRF-1, IRF-2 and IFN-γ expression showed neither a clear negative value nor a biphasic distribution and no clinically relevant cut-off levels have so far been defined for IRF-1 and IRF-2 expression in ovarian cancer, the optimal cut-off level had to be determined in our training set of 138 patients. For this purpose, data were divided at each percentile from 10 to 80, and differences in overall survival were assessed with the log-rank test. The greatest difference between the 2 groups was defined as the optimal clinical cut-off value (Table III).

Table III. Optimal Clinical Cut-Off Values and Respective Percentiles and p Values for Disease-Free (DFS) and Overall Survival (OS) of Interferon Regulatory Factor (IRF)-1, IRF-2, Interferon (IFN)-γ and Forkhead Box P3 (FoxP3)
 Cut-off valuePercentilep value DFSp value OS
IRF-13.429225th0.0160.0001
IRF-22.258840th0.0010.0001
IFNγ0.180430th0.1020.048
FoxP30.003582th0.0420.010

Differences between groups characterized by weak or strong IRF-1, IRF-2, FoxP3 or IFN-γ expression in clinicopathologic characteristics were evaluated by the Mann-Whitney U Test. Spearman's correlation coefficient was calculated to show inter-relationship between the expression of IRF-1, IRF-2, IFN-γ, FoxP3, iNOS, SOCS1 and CD-3. Survival analyses were performed for progression-free survival and overall survival with the Kaplan-Meier method, and differences between groups were determined in a univariate analysis using the log-rank test. To assess the independence of the predictive value of the investigated parameters, the Cox proportional hazard model with a stepwise backward method was used with adjustment for confounding variables. The final model included age at diagnosis, histopathologic grading, histologic subtype, FIGO stage, residual disease after primary debulking surgery, and levels of IRF-1, IRF-2, FoxP3 and IFN-γ expression. SOCS1, CD-3 and iNOS were excluded from the final calculations because of the lack of any relevance to patient outcome.

Statistical significance was defined as p < 0.05. SPSS for Windows 15.0 software (SPSS, Chicago, IL) was used for all analyses.

Results

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. References

Expression of IRF-1 and IRF-2 mRNA in ovarian cancer cell lines

In established ovarian cancer cell lines, the constitutive expression level of IRF-1-specific m-RNA was found to be either significantly higher (HTB-77, HOC-7) or lower (A2780, AG-6000, SKOV-6, OVCAR-3, A2774) than in primary cultured normal human mesothelial cells (HMC) or ovarian surface epithelium (OSE). IRF-2 expression did not differ in HTB-77 or HOC-7 when compared with that in HMC and OSE, but was significantly weaker in all the other cell lines investigated (p < 0.05). Of special note is the fact that constitutive IRF-1 expression was higher than that of IRF-2 in all the ovarian cancer cell lines studied (p < 0.05), with the exception of the cdk-deficient cell line AG-6000 as well as its parental cell line A2780, in which, by contrast, constitutive IRF-2 templates were higher than those of IRF-1 (p = 0.043 and p = 0.038, respectively).

With IFN-γ treatment both IRF-1 and IRF-2 were inducible, but the magnitude of up-regulation was more pronounced for IRF-1 (range: 10-fold to 80-fold) than for IRF-2 (range: 2-fold to 4.5-fold). Furthermore, IRF-2 peaks were delayed for 3 hr when compared with those of IRF-1, which were already observed after 3 hr of treatment. Immunhistochemically, IFN-γ treatment resulted in a considerable enhancement of IRF-1 cytoplasmatic staining, which was by far lesser for IRF-2 staining. Interestingly, no increase in nuclear immunoreactivity was observed for both IRFs after IFN-γ exposure. In addition, treatment with IL-1 also caused a significant enhancement of IRF-1 mRNA expression in all the cell lines tested (range: 1.8-fold to 7.8-fold) with peaks after 3 hr of treatment (p < 0.05). Expression of IRF-2 mRNA was stimulated to a lesser extent by IL-1 (range: 1.8-fold to 3-fold), (p < 0.05). This was true for all cell lines with the exception of HTB-77 and HOC-7 where levels of IRF-2 mRNA remained unaffected following IL-1 exposure. In all ovarian cancer cell lines, 3 hr of treatment with TNF-α caused a 2.3-fold to 35-fold and a 1.9-fold to 5-fold stronger expression of IRF-1 (p < 0.01) and IRF-2 (p < 0.05), respectively (Fig. 1).

thumbnail image

Figure 1. Expression of interferon regulatory factor-1 and interferon regulatory factor-2 mRNA after in vitro treatment of the various ovarian cancer cell lines with pro-inflammatory cytokines and epidermal growth factor. Results are given as fold-expression of untreated controls. Each symbol represents a cell line as follows (♦ = Hoc-7, ▪ = HTB-77, ▴= SKOV-6, • = AG-6000, ⋄ = A2780, □ = ZL2774; ▵ = OVCAR-3). Assays were performed in triplicate.

Download figure to PowerPoint

It is noteworthy that when the absolute numbers of induced transcripts of both regulatory factors were considered at their respective peaks, induction by either IFN-γ, IL-1 or TNF-α resulted in significantly higher levels for IRF-1 than for IRF-2 in all responsive ovarian cancer cell lines (p < 0.01) and consequently the relationship between IRF-1 and IRF-2 moved toward a considerable excess in favor of IRF-1. IL-6 exposure however, did not result in changes in the expression of IRF-1 or IRF-2 in the ovarian cancer cell lines investigated.

In contrast to the tested pro-inflammatory cytokines, treatment with the tumor-promoting EGF over 12 hr, caused a stepwise down-regulation of both IRF-1 and IRF-2 mRNAs. IRF-1 transcripts ranged from 40 to 20% (p < 0.03) and IRF-2 transcripts from 65 to 45% (p < 0.05) of that measured in untreated controls (Fig. 1). In all cell lines except AG6000 and A2780, the decline in the transcripts was more pronounced for IRF-1 than for IRF-2, resulting in a shift in the IRF-1/IRF-2 ratio with a median value of 2.1 (range: 1.45 to 4) in untreated cell lines toward 1 (range: 0.95 to 1.24) after 12 hr of EGF treatment.

Relationships between clinicopathologic features of ovarian cancers and intratumoral expression of IRF-1 and IRF-2

Although no consistent association between IRF-1 expression and FIGO stage was revealed, IRF-2 expression was negatively associated with the stage of disease (ρ = −0.257; p = 0.002). IRF-2 transcripts were found to be highest in stage I with a median value of 2.97 [Q1: 2.32; Q3: 4.10] and lowest in stage IV (median value: 2.09; [Q1: 1.67; Q3: 2.52]); (p = 0.012). In line with this observation, intratumoral levels of IRF-2 mRNA were found to be higher in cases with no residual disease after primary debulking surgery than in those where complete tumor clearance was impossible (p = 0.043). In contrast, no statistical relationship between either histopathological grading or histological subtype and both IRFs was demonstrated. Of special note is the fact that high expression of IFN-γ mRNA (above the 30th percentile) was frequently found in grade 3 tumors and that low levels of IFN-γ transcripts were predominantly associated with grade 1 and 2 tumors (p = 0.013).

Intratumoral expression of IRF-1 and IRF-2 and other immunological parameters

IRF-1 m-RNA transcripts were found to be significantly up-regulated in ovarian cancers (median value: 5.21 [Q1: 3.54; Q3: 8.49]) when compared with normal ovarian tissue (median value: 2.20 [Q1: 1.71; Q3: 3.06]); (p = 0.0001). However, this was not the case for IRF-2 transcripts with median values of 2.46 [Q1: 1.91; Q3: 3.24] and 2.71 [Q1: 2.15; Q3: 3.23] for ovarian cancers and normal ovaries, respectively (p = 0.235). In ovarian cancer samples, IRF-1 expression showed significantly higher levels than did IRF-2 expression (p = 0.0001), but neither IRF differed significantly in normal ovarian tissue (p = 0.606). Nonetheless, statistical analysis showed a significant positive correlation between IRF-1 and IRF-2 expression in ovarian cancer samples (ρ = 0.590; p = 0.0001). As indicated in Table IV, IRF-1 expression was strongly correlated with the intratumoral mRNA of IFN-γ, the STAT-1-dependent gene SOCS-1, the T-cell marker CD-3 and FoxP3, the specific marker for Tregs. However, IRF-2 was positively associated only with CD-3 transcripts and interestingly not with IFN-γ or FoxP3. Both IRFs failed to show any statistical relationship to the IFN-γ-dependent STAT-1-induced iNOS gene transcription (data not shown).

Table IV. Correlations Between Interferon Regulatory Factor (IRF)-1 or IRF-2 and Other Immunologic Variables
  IRF-1IRF-2IFNγFoxP3CD3
  1. Spearman Rank Correlation Coefficients (ρ) and levels of significance are indicated.

IRF-1ρ 0.5900.4650.2980.587
p value 0.00010.00010.0060.0001
n 138138138138
IRF-2ρ0.590 0.138−0.0110.271
p value0.0001 0.2230.9260.010
n138 138138138

Immunohistochemical identification of cells expressing IRF-1 and IRF-2

Immunohistochemical determination of IRF-1 and IRF-2 was performed to elucidate the nature of cells expressing these regulatory factors in ovarian cancer. As shown in Figures 2a and 2b, both IRF-1 and IRF-2 were found to be localized mainly in the cytoplasm and to a lesser extent in the nuclei of cancer cells. Although specific immunostaining for IRF-1 was occasionally detected predominantly in the nucleus of infiltrating immune cells (Fig. 2a), IRF-2 was undetectable in immune-competent cells. Neither IRF-1 nor IRF-2 was demonstrated in the fibroblasts of the stromal tissue of the investigated ovarian cancer specimens.

thumbnail image

Figure 2. (a) Cytoplasmatic interferon regulatory factor (IRF)-1 immunostaining in ovarian cancer cells with IRF-1 negativity in the infiltrating immune cells. Note the inserted picture with a higher magnification showing a different tumor with a small group of tumor-infiltrating immune cells exhibiting nuclear IRF-1 immunoreactivity. (b) strong IRF-2-specific immunostaining confined to the cytoplasm of ovarian cancer cells. Notably, immunohistochemistry detected no IRF-2 expression in tumor-infiltrating immune cells. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Download figure to PowerPoint

Survival in a training set of 138 ovarian cancer patients according to IRF-1 and IRF-2 expression

As depicted in Figures 3a3d univariate survival analyses showed high-level intratumoral expression of both IRF-1 and IRF-2 as predictors of favorable disease-free and overall survival in the studied training set of 138 ovarian cancer patients. Interestingly, in those patients classified as belonging to the high-level IRF-1 expressing group, disease-free and overall survival were significantly poorer when intratumoral FoxP3 expression was found to be above the threshold of the 82nd percentile (Figs. 4a and 4b). In contrast, expression of FoxP3 was irrelevant for survival in the subgroup of patients with cancers showing high expression of IRF-2 mRNA. Expression of IFN-γ within the tumor samples was associated with favorable prognosis regarding overall survival (p = 0.048), but failed to show significant prognostic value with regard to disease-free survival. However, in multivariate analyses only the expression of IRF-1, but not that of IRF-2, retained independent prognostic significance as well for disease-free as for overall survival and high-level expression of IFN-γ revealed to be an independent prognosticator of improved overall survival (Table V).

thumbnail image

Figure 3. Kaplan-Meier curves based on high- and low-level IRF-1 expression (dichotomized along the 25th percentile) with regard to disease-free survival (a) and overall survival (b); and IRF-2 expression (dichotomized along the 40th percentile) for disease-free survival (c) and overall survival (d).

Download figure to PowerPoint

thumbnail image

Figure 4. Subgroup survival analyses in patients with high-level IRF-1-expressing cancers, dichotomized for low and high intratumoral forkhead box P3 (FoxP3) mRNA expression; (threshold level for FoxP3: 82nd percentile). Kaplan-Meier curves for disease-free survival (a) and overall survival (b).

Download figure to PowerPoint

Table V. Cox Regression Hazard Model (Stepwise Backward)
VariablesDisease-free survivalOverall survival
Hazard ratio95% CIp valueHazard ratio95% CIp value
  1. The final model included age at diagnosis, residual disease, FIGO stage, histopathologic grading, histologic subtype, IFNγ, FoxP3, IRF-1 and IRF-2 expression.

Age at diagnosis      
 < median1.000  1.000  
 > median1.420(0.772–2.609)0.2591.603(0.843–3.048)0.150
Residual disease      
 Optimally debulked1.000  1.000  
 Tumor residual > 2cm4.553(2.467–8.403)0.00015.055(2.572–9.935)0.0001
IFNγ      
 Low1.000  1.000  
 High0.629(0.326–1.214)0.1670.456(0.222–0.936)0.032
IRF-1      
 Low1.000  1.000  
 High0.402(0.213–0.759)0.0050.341(0.169–0.690)0.003
IRF-2      
 Low1.000  1.000  
 High0.780(0.387–1.569)0.4851.033(0.479–2.227)0.935

Discussion

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. References

Recently, the presence of tumor-infiltrating CD3+ T-cells was reported to be independently associated with improved disease-free and overall survival in ovarian cancer.2 In accordance, high intratumoral expression of IFN-γ mRNA, has also been shown to have independent impact in predicting favorable outcome in cancers of the ovary.4 Although demonstrated intratumoral levels of IFN-γ mRNA were interpreted to originate from tumor-infiltrating lymphocytes (i.e. CTL, T-Helper cells and NK cells), intratumoral IRF-1 and IRF-2 studied here are considered to be derived mainly from tumor cells and to a lesser extent from tumor-infiltrating immune cells. This was corroborated by the immunohistochemical determination of both IRFs in a subset of ovarian cancer samples, revealing the majority of IRF-immunoreactivity was detectable in the cytoplasm of ovarian cancer cells. Only occasionally was a faint nuclear staining for IRF-1 detected in tumor-infiltrating immune cells. Furthermore, our in vitro investigations showed that IRF-1 and IRF-2 are constitutively expressed in all ovarian cancer cell lines tested. When compared with HPMC and OSE, constitutive expression of IRF-1 was found to be either up- or down-regulated in the various ovarian cancer cell lines; surprisingly, expression of IRF-2 was not significantly up-regulated in any of the cell lines studied. In vitro treatment of ovarian cancer cells with IFN-γ, IL-1 and TNF-α clearly demonstrated responsiveness with regard to the expression of both IRFs and suggests a critical role of immune cell-derived pro-inflammatory cytokines in regulating the expression of IRF-1 and IRF-2 in the tumor. This was indirectly corroborated by the observation that high IRF-1 and IRF-2 expressing tumors exhibited a considerable higher accumulation of not otherwise specified inflammatory cells (data not shown). However, the considerable differences in constitutive expression of both IRFs (e.g. HOC-7 and HTB-77 in comparison to the other ovarian cancer cell lines investigated) and the variety in responsiveness to various cytokines between the cell lines tested, suggest that to a certain extent other phenotypic mechanisms such as chromosomal aberrations or epigenetic phenomena (i.e. methylation of CpG island in the promoter region) contribute to the regulation of IRF expression.

IRF-1 and IRF-2 were originally identified as a transcriptional activator and a repressor, respectively, of IFN-inducible genes. Gene transfection studies have revealed that IRF-1 exhibits tumor-suppressive properties, whereas IRF-2 antagonizes the function of IRF-1, either by competing for binding to the same DNA sequences in the promoter regions of target genes or by repressing involved co-activators. IRF-2 is therefore considered to act as an oncoprotein. Moreover, the molecular background of the growth-promoting and anti-apoptotic effects of IRF-2 has been largely substantiated by its direct stimulatory effect on the expression of histone H4, cyclin-D1 and Bcl-2 as well as by its down-regulating effect on activated caspase-9.16, 17

Here we report, however, that univariate survival analyses show not only high intratumoral IRF-1 expression but also high-level IRF-2 expression to be associated with improved disease-free and overall survival in a large cohort of 138 ovarian cancer patients. This meaningful, albeit unexpected, finding regarding IRF-2 stands in contrast to the data on solid tumors published so far. Wang and colleagues recently provided evidence that IRF-1 was significantly down-regulated and IRF-2 was over-expressed in human esophageal cancers when compared with matched normal esophageal tissues. Furthermore, statistical data indicated that IRF-2 expression was tightly correlated with disease progression in esophageal cancer.16

Moreover, besides the association between high-level IRF-2 expression and improved survival in univariate analyses, other of our findings argue against an oncogenic role of IRF-2 in ovarian cancer: (i) in contrast to other solid tumors such as diffusely infiltrating astrocytomas, breast and as already mentioned esophageal cancers,16, 18, 19 the expression of IRF-2 was not found to be up-regulated in either ovarian cancer tissue or established ovarian cancer cell lines, (ii) significantly higher levels of IRF-2 mRNA were found in early-stage when compared with advanced-stage ovarian cancers, and (iii) IRF-2 expression was inversely related to residual disease after primary debulking surgery. Concerning the latter point, there is excellent evidence to show that ovarian cancers, whose complete surgical clearance is impossible at primary surgery, exhibit a more aggressive phenotype with a completely different molecular signature than do cancers allowing optimal cytoreduction.20

As our findings regarding IRF-2 conflicted with its generally accepted oncogenic properties, we performed an additional sequencing of the PCR product demonstrated at 76 bp to evaluate the accuracy of the detected templates. This sequencing confirmed that the product we demonstrated with quantitative RT-PCR was indeed the IRF-2-specific mRNA (data not shown), and further supported the hypothesis that, when compared with other solid tumors, IRF-2 appears to play a different role in the biology of ovarian cancer.

Nevertheless, the reason why IRF-2 does not behave like an oncoprotein in this cancer entity remains speculative. As far as we understand, on the one hand, this could be because of a distinct malignant phenotype immanent to ovarian cancer, where IRF-2 is not constitutively over-expressed and plays no decisive role in oncogenesis. On the other hand, the basic oncogenic properties of IRF-2, which may also exist in ovarian cancer, could be masked by the overwhelming effects of local anti-tumor immune defense, including a disproportionate up-regulation of IRF-1 in cancer cells. This could be especially true in ovarian cancer, which in comparison to other solid tumors, has been shown to be particularly exposed to and influenced by the local host's immune system, because of its unique dissemination throughout the peritoneal cavity.2–4, 21

In addition to both IRFs, elevated intratumoral content of IFN-γ mRNA was also predictive of improved patient overall survival, but was without prognostic value with regard to disease-free survival in univariate analyses. In the stepwise Cox regression hazard model, IRF-1 retained independent prognostic significance for disease-free as well as for overall survival, while IFN-γ was revealed to have significant independent prognostic value only for overall survival. However, it is worth mentioning that for overall survival, the predictive power of the expression of IRF-1 mRNA was greater (hazard ratio: 0.34, than was that estimated for IFN-γ mRNA (hazard ratio: 0.46). Interestingly, the favorable clinical outcome in patients bearing cancers with strong IRF-1 expression was abrogated when transcripts of FoxP3 were found to be concomitantly elevated. This phenomenon, which is obviously caused by a number of tumor-infiltrating Tregs sufficient to reverse clinical outcome, has also been observed in patients with high-level IFN-γ-expressing tumors22 and corroborates the findings of Curiel et al.3 who demonstrated that high recruitment of Tregs into the microenvironment of ovarian carcinomas fosters immune privilege and independently predicts reduced survival.

The lack of a significant association between intratumoral IRF-2 and IFN-γ in the examined ovarian cancer samples as well as the fact that IRF-1 and IFN-γ were found to be independent of each other in predicting overall survival may substantiate our in vitro-findings and those of others23, 24 showing that the expression of both IRFs is not exclusively regulated by IFN-γ, but may be significantly influenced by other locally acting cytokines such as IL-1 and TNF-α, but also by growth factors like EGF. In this context, it should be emphasized that especially IRF-1, as a relevant prognosticator, should not be regarded as a simple second-messenger of IFN-γ, but as a variable that reflects local anti-tumor immune response and other events influencing tumor growth, namely more precisely than does IFN-γ.

In summary, quantitation of IRF-1 and IRF-2 using real-time PCR in ovarian cancer samples and cell lines revealed that IRF-2 cannot be regarded as a classic oncoprotein in ovarian cancer. Among the investigated parameters reflecting activated local anti-tumor defense, intratumoral IRF-1 mRNA was shown to be the most powerful variable predicting favorable clinical outcome in terms of disease-free and overall survival. Most notably, the presented data give new insights into the complexity of the crosstalk between the host immune system and cancer cells and underline the clinical importance of local immune-modulatory influences in the biology of ovarian cancer.

References

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. References
  • 1
    IARC: GLOBOCAN. Cancer incidence, mortality and prevalence worldwide, 2002. Available at http://www-dep.iarc.fr/.
  • 2
    Zhang L,Conejo-Garcia JR,Katsaros D,Gimotty PA,Massobrio M,Regnani G,Makrigiannakis A,Gray H,Schlienger K,Liebman MN,Rubin SC,Coukos G. Intratumoral T cells, recurrence, and survival in epithelial ovarian cancer. N Engl J Med 2003; 348: 20313.
  • 3
    Curiel TJ,Coukos G,Zou L,Avarez X,Cheng P,Mottram P,Evdemon-Hogan M,Conejo-Garcia JR,Zhang L,Burrow M,Zhu Y,Wei S, et al. Specific recruitment of regulatory T cells in ovarian carcinoma fosters immune privilege and predicts reduced survival. Nat Med 2004; 10: 9429.
  • 4
    Marth C,Fiegl H,Zeimet AG,Müller-Holzner E,Deibl M,Doppler W,Daxenbichler G. Interferon-gamma expression is an independent prognostic factor in ovarian cancer. Am J Obstet Gynecol 2004; 191: 1598605.
  • 5
    Pujade-Lauraine E,Guastalla JP,Colombo N,Devillier P,François E,Fumoleau P,Monnier A,Nooy M,Mignot L,Bugat R,Marques C,Mousseau M, et al. Intraperitoneal recombinant interferon gamma in ovarian cancer patients with residual disease at second-look laparotomy. J Clin Oncol 1996; 14: 34350.
  • 6
    Windbichler GH,Hausmaninger H,Stummvoll W,Graf AH,Kainz C,Lahodny J,Denison U,Müller-Holzner E,Marth C. Interferon-gamma in the first-line therapy of ovarian cancer: a randomized phase III trial. Br J Cancer 2000; 82: 113844.
  • 7
    Alberts DS,Marth C,Alvarez RD,Johnson G,Bidzinski M,Kardatzke DR,Bradford WZ,Loutit J,Kirn DH,Clouser MC,Markman M. Randomized phase 3 trial of interferon gamma-1b plus standard carboplatin/paclitaxel versus carboplatin/paclitaxel alone for first-line treatment of advanced ovarian and primary peritoneal carcinomas: results from a prospectively designed analysis of progression-free survival. Gynecol Oncol 2008; 109: 17481.
  • 8
    Tanaka N,Taniguchi T. The interferon regulatory factors and oncogenesis. Semin Cancer Biol 2000; 10: 7381.
  • 9
    Taniguchi T,Ogasawara K,Takaoka A,Tanaka N. IRF family of transcription factors as regulators of host defense. Annu Rev Immunol 2001; 19: 62355.
  • 10
    Harada H,Kitagawa M,Tanaka N,Yamamoto H,Harada K,Ishihara M,Taniguchi T. Anti-oncogenic and oncogenic potentials of interferon regulatory factors-1 and -2. Science 1993; 259: 9714.
  • 11
    Lowney JK,Boucher LD,Swanson PE,Doherty GM. Interferon regulatory factor-1 and -2 expression in human melanoma specimens. Ann Surg Oncol 1999; 6: 6048.
  • 12
    Connett JM,Badri L,Giordano TJ,Connett WC,Doherty GM. Interferon regulatory factor 1 (IRF-1) and IRF-2 expression in breast cancer tissue microarrays. J Interferon Cytokine Res 2005; 25: 58794.
  • 13
    Kim EJ,Lee JM,Namkoong SE,Um SJ,Park JS. Interferon regulatory factor-1 mediates interferon-gamma-induced apoptosis in ovarian carcinoma cells. J Cell Biochem 2002; 85: 36980.
  • 14
    Zeimet AG,Marth C,Offner FA,Obrist P,Uhl-Steidl M,Feichtinger H,Stadlmann S,Daxenbichler G,Dapunt O. Human peritoneal mesothelial cells are more potent than ovarian cancer cells in producing tumor marker CA-125. Gynecol Oncol 1996; 62: 3849.
  • 15
    Siemens CH,Auersperg N. Serial propagation of human ovarian surface epithelium in tissue culture. J Cell Physiol 1988; 134: 34756.
  • 16
    Wang Y,Liu DP,Chen PP,Koeffler HP,Tong XJ,Xie D. Involvement of IFN regulatory factor (IRF)-1 and IRF-2 in the formation and progression of human esophageal cancers. Cancer Res 2007; 67: 253549.
  • 17
    Vaughan PS,van der Meijden CMV,Aziz F,Harada H,Taniguchi T,van Wijnen AJ,Stein JL,Stein GS. Cell cycle regulation of histone H4 gene transcription requires the oncogenic factor IRF-2. J Biol Chem 1998; 273: 1949.
  • 18
    Yoshino A,Katayama Y,Yokoyama T,Watanabe T,Ogino A,Ota T,Komine C,Fukushima T,Kusama K. Therapeutic implications of interferon regulatory factor (IRF)-1 and IRF-2 in diffusely infiltrating astrocytomas (DIA): response to interferon (IFN)-beta in glioblastoma cells and prognostic value for DIA. J Neurooncol 2005; 74: 24960.
  • 19
    Doherty GM,Boucher L,Sorenson K,Lowney J. Interferon regulatory factor expression in human breast cancer. Ann Surg 2001; 233: 6239.
  • 20
    Berchuck A,Iversen ES,Lancaster JM,Dressman HK,West M,Nevins JR,Marks JR. Prediction of optimal versus suboptimal cytoreduction of advanced-stage serous ovarian cancer with the use of microarrays. Am J Obstet Gynecol 2004; 190: 91025.
  • 21
    Zeimet AG,Widschwendter M,Knabbe C,Fuchs D,Herold M,Müller-Holzner E,Daxenbichler G,Offner FA,Dapunt O,Marth C. Ascitic interleukin-12 is an independent prognostic factor in ovarian cancer. J Clin Oncol 1998; 16: 18618.
  • 22
    Wolf D,Wolf AM,Rumpold H,Fiegl H,Zeimet AG,Muller-Holzner E,Deibl M,Gastl G,Gunsilius E,Marth C. The expression of the regulatory T cell-specific forkhead box transcription factor FoxP3 is associated with poor prognosis in ovarian cancer. Clin Cancer Res 2005; 11: 832631.
  • 23
    Mori K,Stone S,Khaodhiar L,Braverman LE,DeVito WJ. Induction of transcription factor interferon regulatory factor-1 by interferon-gamma (IFN gamma) and tumor necrosis factor-alpha (TNF alpha) in FRTL-5 cells. J Cell Biochem 1999; 74: 2119.
  • 24
    Geller DA,Nguyen D,Shapiro RA,Nussler A,Di Silvio M,Freeswick P,Wang SC,Tweardy DJ,Simmons RL,Billiar TR. Cytokine induction of interferon regulatory factor-1 in hepatocytes. Surgery 1993; 114: 23542.