High Expression of Tumor Necrosis Factor-related Apoptosis-inducing Ligand Is Associated with Favorable Ovarian Cancer Survival¹

Epithelial ovarian cancer is among the leading causes of cancer deaths for females in the United States and Western Europe. Similar to other types of human cancers, ovarian cancers are thought to arise because of accumulation of mutations in genes that regulate cellular proliferation and apoptosis. In this regard alterations in several genes, including the p53 tumor suppressor gene and HER-2/neu oncogene, have been identified in some ovarian cancers (1). A pattern of molecular alterations characteristic of all epithelial ovarian cancers has not been identified, however. The molecular pathogenesis of these cancers is heterogeneous, and this is reflected in the variability of clinical characteristics such as histological type, differentiation, potential for invasion and metastasis, response to therapy, and outcome.

Most ovarian cancers are found to have metastasized extensively in the peritoneal cavity at diagnosis because effective screening and early detection techniques do not exist. About 70% of women with advanced ovarian cancer have a complete clinical response after surgical cytoreduction and platinum-based chemotherapy, but median survival is only about 3 years (2). There is considerable variation in the extent of response to therapy between patients, however. Some women succumb relatively quickly, whereas others who are treated in an identical manner may live more than 10 years. As noted above, the wide variation in outcome is likely determined by the heterogeneity of the molecular alterations that underlie the development of ovarian cancers.

Prior attempts to identify molecular prognostic markers have focused on single genes, but this approach is time-consuming and relatively inefficient. The recent development of microarray technology has provided the opportunity to simultaneously evaluate the expression of thousands of genes in ovarian cancers. Several groups have used arrays to demonstrate differences in gene expression between normal ovarian epithelium and ovarian cancers and also between various histological types of ovarian cancer (3, 4, 5, 6, 7). We (J. M. L., R. W., J. M., A. B.) recently used microarrays to determine whether distinct patterns of gene expression could be identified that predict long- or short-term survival in women with advanced-stage serous ovarian cancers.³ Several individual genes involved in growth and apoptosis were shown to be differentially expressed. Expression of the TRAIL⁴ gene was 7.4-fold higher in ovarian cancers than in normal ovarian epithelium and 1.5-fold higher in cancers of long-term survivors compared with short-term survivors.

The TRAIL gene, which is located on chromosome 3q26, is a member of the TNF cytokine family. It was initially identified by homology to the COOH-terminal extracellular domain of other TNF family members, such as fas ligand, TNF-α, and lymphotoxin-α (8). Subsequently, TRAIL has been shown to induce apoptosis in a variety of human and mouse transformed or malignant cells, whereas normal cells are not affected (9). Selective induction of apoptosis in cancer cells suggests a potential therapeutic use for TRAIL (10, 11). In ovarian cancers Cuello et al. (12) found that a majority of 12 chemoresistant cell lines were resistant to TRAIL alone but also that some were sensitive to the combination of TRAIL and cytotoxic drugs such as cis-platinum. In view of these in vitro data and our microarray data suggesting that high TRAIL expression correlates with favorable survival in advanced ovarian cancers, we sought to confirm the association between TRAIL expression and outcome in a larger cohort of patients.

One hundred and twenty epithelial ovarian cancers and eight NOSE samples were obtained with institutional review board-approved informed consent from patients treated by the Division of Gynecologic Oncology at Duke University Medical Center. Eleven of the cancers were early stage (I/II), and 109 were advanced stage (III/IV). Histopathological subtypes included 95 serous, 6 endometrioid, 5 mucinous, 4 clear cell, and 10 mixed type. All patients with advanced-stage disease were treated with primary surgical cytoreduction (52 optimal, 57 suboptimal) followed by platinum-based chemotherapy.

Total RNA was extracted using the RNeasy RNA extraction kit (Qiagen Inc., Valencia, CA), and a 1-μg aliquot subject to a reverse transcription reaction was extracted with the Roche First Strand cDNA kit (Roche, Basel, Switzerland). cDNA was quantified by fluorometry, and 5 ng of cDNA were subjected to 40 cycles of QRT-PCR in the Roche Light Cycler (Roche) according to the manufacturer’s instructions in the presence of intron-spanning cDNA primers to the housekeeping gene, β-microglobulin, to confirm adequate cDNA normalization. In the same way, QRT-PCR was used to determine relative expression of the TRAIL gene in 120 epithelial ovarian cancers using the intron-spanning cDNA primers 5′-aga-cct-gcg-tgc-tga-tcg-tg-3′ and 5′-tta-ttt-tgc-ggc-cca-gag-cc-3′, 10 units of AdvanTaq Plus DNA Polymerase with PCR reaction buffer (Clontech, Palo Alto, CA), deoxynucleoside triphosphate, and Syber Green. Serial dilutions of the sample demonstrating highest TRAIL expression were used to create a concentration curve, and relative expression levels were calculated for each sample from this curve. Statistical analysis was performed using Student’s t test and Kaplan-Meier log-rank survival and Spearman correlation coefficient analysis.

QRT-PCR was used to determine relative expression of the TRAIL gene in 120 epithelial ovarian cancers and 8 NOSE samples (Table 1; Fig. 1). For each sample, TRAIL expression was calculated as a relative value from a serial dilution curve of the cancer demonstrating the highest TRAIL expression level. Mean relative TRAIL expression in 120 ovarian cancers was approximately 10-fold higher than that in 8 NOSE samples (1.28 versus 0.13; P < 0.001). Moderate and poorly differentiated (grade II/III) cancers expressed 3.1-fold higher TRAIL levels than well-differentiated (grade I) cancers (P = 0.0001), although no difference was seen between moderate (grade II) and poorly (grade III) differentiated cancers (1.45 versus 1.23; P = 0.61). Although overall serous cancers expressed higher levels of TRAIL relative to other histopathological subtypes (1.09 versus 0.57; P = 0.03), no statistical difference was identified between TRAIL expression in serous cancers compared with other individual histopathological subtypes (endometrioid, mucinous, or clear cell cancer). Expression in 11 early-stage (I/II) and 109 advanced-stage (III/IV) cancers was 6.1-fold (P = 0.12) and 10.1-fold (P < 0.0001) higher, respectively, than that in normal ovarian surface epithelium. In advanced-stage (III/IV) cancers, TRAIL expression was 1.7-fold higher than that in early-stage (I/II) cancers (P = 0.26). There was no relationship between TRAIL expression and extent of cytoreductive surgery (optimal versus suboptimal).

High TRAIL expression measured by QRT-PCR was associated with prolonged survival. For all stages, relative TRAIL expression was 2.3-fold higher (1.52 versus 0.66) in cancers from women who lived >5 years (n = 36) than in cancers from those who died in <1 year (n = 22; Spearman correlation coefficient of 0.26; P = 0.03). Among advanced-stage (III/IV) cases, relative TRAIL expression was 2.2-fold higher (1.6 versus 0.9) in cancers of women who lived >5 years (n = 32) than in cancers from those who lived <1 year (n = 20; P = 0.18). In Fig. 2, Kaplan-Meier survival analysis of the 109 patients with advanced-stage ovarian cancer demonstrates the association between high TRAIL expression and favorable survival (P = 0.14, log-rank test). In the subset of patients with advanced serous cancers, 29 long-term survivors (>5 years) demonstrated 1.8-fold higher (1.52 versus 0.68) relative TRAIL expression than 14 short-term survivors (<1 year; P = 0.08).

We found that ovarian cancer tissue samples express significantly more TRAIL mRNA than cultured normal human ovarian epithelial cells. Lower TRAIL expression in the normal ovarian epithelial cells examined in this study could be attributable to their increased proliferation in monolayer culture, whereas these cells normally are relatively quiescent in vivo. However, it also has been shown previously that TRAIL expression is higher in malignant breast cells compared with normal breast tissue (13). Immunostaining for TRAIL was seen in 52% of human breast cancers but not in normal breast tissue. In contrast, TRAIL mRNA was identified using QRT-PCR in essentially all normal and malignant breast tissues, suggesting that whereas TRAIL mRNA is expressed in all cases, translation to protein occurs predominantly in cancers. Although we identified TRAIL mRNA in both normal and malignant ovarian epithelial samples, significantly higher levels were seen in the cancers. The possibility that translation to protein also may be greater in cancers warrants further study.

TRAIL is a member of a family of “death ligands,” including TNF and fas ligand, that are capable of inducing apoptosis. Under most circumstances, normal cells are relatively resistant to TRAIL-induced cell death, whereas cancer cells often are more sensitive (9, 14). In addition, synergy between cytotoxic agents and TRAIL has been observed, which suggests the potential use of TRAIL as an adjunct to conventional cancer chemotherapy (10, 11). In this regard, recombinant forms of TRAIL have been evaluated in animal models and shown to prolong overall survival in mice inoculated with breast cancer cells (15, 16). Phase I human clinical trials using TRAIL are ongoing.

In ovarian cancers, Cuello et al. (12) demonstrated that the combination of TRAIL and chemotherapeutic agents resulted in a significant increase in apoptosis and overall growth inhibition in ovarian cancer cells. Using expression arrays, we had found previously that high TRAIL expression appeared to predict favorable survival in 18 patients with advanced serous ovarian cancers (7). In the current study, we have demonstrated in a much larger group of women with ovarian cancer that high TRAIL expression correlates with favorable survival. The association between TRAIL expression and survival was evident when comparing long-term survivors (>5 years) with short-term survivors (<1 year). In addition, among the entire group of patients, those whose cancers had high expression of TRAIL mRNA had more favorable median survival.

In view of the favorable effect of TRAIL expression on survival, it is somewhat surprising that TRAIL expression was higher in advanced cancers relative to early-stage cases. Our sample size was not large enough to permit definite conclusions regarding these observed differences. It is possible, however, that low TRAIL expression may contribute to the lack of efficacy of adjuvant chemotherapy in early-stage cancers described in previous studies (17). This must be interpreted with some caution, however, given the small number of early-stage cancers analyzed in our set and more recent data demonstrating a benefit to adjuvant chemotherapy in early-stage disease (18).

The biological actions of TRAIL are regulated by an intricate array of cell membrane receptors and downstream signal transduction pathways (14, 19). TRAIL mediates apoptosis by binding to multiple cell surface receptors, including DR4/TRAIL-R1, Killer/DR5/TRAIL-R2/TRICK2, TRID/DcR1/TRAIL-R3/LIT, and TRUNDD/DcR2/TRAIL-R4. TRAIL binding with DR4 and KILLER/DR5 induces activation of the caspase pathway and subsequent apoptosis. In contrast, trail receptor without an intracellular domain, which lacks a cytoplasmic death domain, and trail receptor with a truncated death domain, which has a truncated cytoplasmic death domain, are reported to act as nonfunctional “decoy receptors” that compete for TRAIL and protect cells from TRAIL-mediated apoptosis. To take full therapeutic advantage of this apoptotic cascade, characterization of each step in the pathway is essential. Our findings to date suggest that TRAIL may have some utility in the treatment of ovarian cancer and that pharmacological manipulation of the TRAIL pathway may improve survival for patients with this disease. The use of TRAIL to enhance chemosensitivity of ovarian cancers represents an appealing molecular therapeutic strategy worthy of further investigation.