Thy-1 predicts poor prognosis and is associated with self-renewal in ovarian cancer

Ovarian cancer is the leading cause of gynecologic cancer-related death in the United States [1]. Currently there are no effective screening modalities available and over 80% of patients present with advanced stage disease [2]. Even with aggressive cytoreductive surgery and adjuvant chemotherapy, over 80% of patients with advanced stage disease will recur [3]. Furthermore, recurrent disease often demonstrates increasing resistance to conventional chemotherapy and contributes to the high mortality of this disease [4].

Cancer stem cells (CSCs) are cancer cells that retain the ability to self-renew and exhibit increased proliferation and chemoresistance [5]. In ovarian cancer, CSCs have been suggested as a means of chemoresistance and aggressive malignant behavior and thus are attractive therapeutic targets [6]. Successful identification of CSCs in ovarian cancer may be helpful in determining and subsequently targeting mechanisms of chemoresistance and recurrence in the future. Several markers have been implicated in ovarian cancer, including CD133, CD44, CD24, CD117, EpCAM and ALDH [7]. We previously transduced ovarian cancer cells (A2780) with a NANOG-GFP reporter system to identify ovarian CSCs based on GFP intensity [8, 9]. Using this platform, we performed a high-throughput flow cytometry screen to compare expression of 242 cell surface markers in ovarian CSCs (GFP-positive) and non-CSCs (GFP-negative) and identified CD55 as a CSC marker and a driver of self-renewal and chemoresistance pathways [10‐12]. Our high-throughput screen also identified a second protein that was more highly expressed in CSCs, Thy-1.

Thy-1 (also known as CD90) is a glycosylphosphatidylinositol (GPI) anchored protein that localizes to lipid rafts at the cell surface [13, 14]. Investigation of the role of Thy-1 in ovarian cancer is limited. Abeysingh et al. investigated the effect of Thy-1 overexpression on tumorigenicity of the SKOV3 established cell line and suggested that Thy-1 regulates differentiation and acts as a putative tumor suppressor [15, 16]. This is in stark contrast to more recent discovery that Thy-1 is a CSC marker in glioblastoma as well as hepatocellular, pancreatic, and gallbladder cancer and promotes tumorigenicity and self-renewal [17‐21]. Our high-throughput screen suggested Thy-1 as a putative CSC marker in ovarian cancer. Our primary objective was to validate this finding with in vitro studies and to correlate with clinical outcomes in women with ovarian cancer. Our secondary objective was to evaluate whether we could target Thy-1 to negatively impact cancer cell growth.

While a majority of ovarian cancers demonstrate sensitivity to standard chemotherapy, recurrent disease is pervasive, which is why women who present with advanced disease (stage III or IV) currently face survival rates between 19 to 47% at 5 years [24]. The recurrence pattern and eventual development of chemoresistance in ovarian cancer make it an apt target for study of cancer stem cells (CSCs). While several CSC markers have been identified in ovarian cancer cells, none are specific or independently sufficient to consistently identify ovarian CSCs and more investigation is warranted to better characterize ovarian CSCs [25]. We identified Thy-1 as a cell surface marker more highly expressed in a GFP-labeled CSC population and were able to subsequently demonstrate increased proliferation and self-renewal capacity that was reversed by Thy-1 knockdown. Thy-1 may contribute to better identification of the ovarian CSC signature and may help guide efforts in the identification and eventual targeting of CSCs.

Ovarian CSC identification may not only have prognostic implications, but has the potential to guide management decisions. Meng et al. previously found that women with serous ovarian cancer and ascites specimens with higher expression of the CSC marker CD44 had significantly poorer progression-free survival (6 vs. 18 mos, P = 0.01) [26]. This is a clinically significant finding, and if validated on a larger scale could help to guide decision-making regarding treatment goals and the role of maintenance therapy. In our bioinformatics analysis, we found that high Thy-1 expression was associated with poorer progression-free survival in both serous and endometrioid ovarian cancers. We identified a particularly compelling difference in progression-free survival for women with endometrioid ovarian tumors with high expression of Thy-1 (Fig. 3c). In women with endometrioid ovarian tumors, women with high expression of Thy-1 were more than 3 times more likely to experience recurrent disease than women with low expression of Thy-1. This trend was consistent in both early stage and late state endometrioid ovarian cancers, although due to limited numbers of endometrioid ovarian tumors, power was limited (Additional file 1: Figure S1). This association is especially relevant in the case of endometrioid tumors because these are frequently well-differentiated (low grade), often confined to the ovary at diagnosis (stage 1a or 1b), and when early stage are associated with a 5-year overall survival of over 90% [27]. Given this, low grade and early stage tumors may be spared adjuvant treatment with chemotherapy and younger patients are sometimes offered fertility-sparing surgery to balance increased surgical morbidity against recurrence risk. In this setting, knowledge of increased CSC expression and corresponding increased recurrence risk may be an indication for closer surveillance or more aggressive clinical management.

To evaluate levels of RNA expression for Thy-1 in human tissue, we selected a small cohort of 7 women with stage IIIc high grade serous ovarian cancer that had undergone surgery and received adjuvant therapy with surveillance at our institution. This was intended as a pilot study and we deliberately selected patients with variable progression-free intervals following platinum-based chemotherapy. We noted incidentally that women with weak Thy-1 staining experienced longer than expected progression-free survival (Table 1). However, the limited number of patients precludes meaningful clinical outcome comparisons. Also, our clinical database included patients undergoing surgery in 2006–2007 and care patterns were different during this time period. For example, of these 7 patients, Patient 6 was known to be BRCA2+ and Patient 1 tested negative for BRCA, however all others had unknown BRCA status. BRCA status is known to be associated with platinum sensitivity and longer progression-free intervals. Additionally, Patients 1 and 5 received adjuvant IP therapy, and both Patients 2 and 6 were enrolled in GOG 218, a blinded study that randomized patients to receive adjuvant carboplatin and paclitaxel with bevacizumab or placebo [28]. Further appropriately powered studies are needed to demonstrate whether RNA in situ hybridization carries prognostic value for ovarian cancer; however, our data do support that expression is differential.

Beyond prognostic value, CSC markers are potential targets for preventing recurrence and neutralizing chemoresistance. After treatment with neoadjuvant chemotherapy, patient tissue samples of residual ovarian cancer have been shown to display enriched stem cell populations [29]. Interestingly, both patients in our small cohort that were treated with neoadjuvant chemotherapy had strong RNA expression of Thy-1, which raises the question of whether neoadjuvant chemotherapy enriches Thy-1 expressing cell populations. CSCs are known to have multiple features that are thought to contribute to chemoresistance including more efficient DNA protection and repair mechanisms, activation of survival pathways, and inactivation of apoptotic pathways [25]. Chau et al. previously identified a population of CD117+ (c-kit) ovarian cancer cells and demonstrated reduced resistance to cisplatin and paclitaxel in a murine model with c-kit knockdown and treatment with imatinib, a tyrosine kinase inhibitor known to target c-kit [30]. Zhang et al. previously used miRNA to target Thy-1+ CSCs in a murine model of hepatocellular carcinoma [31]. We were able to demonstrate decreased proliferation and self-renewal after Thy-1 knockdown in ovarian cancer cells (Fig. 4b, c), suggesting a potential for targeting. The ability to specifically target CSCs in ovarian cancer would hopefully allow oncologists to reduce recurrence risk by targeting those cells that are able to evade conventional systemic chemotherapy and retain the ability to grow and metastasize.

Despite promising results in vitro and in vivo models, early translational studies in ovarian cancer have been disappointing. However, the relapsing nature of ovarian cancer begs the question whether targeted therapy is being appropriately investigated. For example, early clinical trials of imatinib (which targets c-kit, CD117) in women with ovarian cancer have failed to show a survival benefit, even when evaluated in women with increased expression of associated biomarkers [32‐35]. However, each of those studies evaluated efficacy of an oral drug in women with platinum-resistant, active recurrent metastatic disease. If CSCs are promoters of cell population survival and recurrence, perhaps these targeted agents are best studied as maintenance therapies for women with minimal to no visible residual disease. Or, perhaps CSC-targeted therapy should be studied in the same spaces in which microscopic residual disease exists, such as the peritoneal cavity. Both intraperitoneal delivery of chemotherapy and intraoperative delivery of heated intraperitoneal chemotherapy have been shown in prospective randomized trials to prolong the disease-free interval in ovarian cancer, but the impact that these treatment modalities have on CSCs is unknown [36, 37]. It would be valuable in future prospective studies to collect and analyze tissue with the goal of understanding how treatment impacts CSC populations and whether this correlates with survival outcomes.

In summary, we identified Thy-1 as a surface marker of CSCs in ovarian cancer and further demonstrated increased proliferative and self-renewal capability in Thy-1 expressing cells that was reversible with Thy-1 knockdown (Figs. 1,4). We were additionally able to validate these findings by identifying poorer clinical outcomes for women with ovarian cancer and high expression of Thy-1 (Fig. 3). The paradox of ovarian cancer is that although we have generally efficacious chemotherapy agents in the primary setting, we fail to target microscopic cells that escape primary therapy and progress into recurrent, almost universally fatal, disease. It is imperative that we further investigate CSC markers in ovarian cancer and understand the role of CSCs as predictors and mediators of recurrence so that we can target these cells with the goal of keeping patients disease-free.