The University of Texas M. D. Anderson Cancer Center
SPORE in Ovarian Cancer
Robert C. Bast, Jr., M.D.
Vice President, Translational Research
Professor, Department of Experimental Therapeutics
Harry Carothers Wiess Distinguished University Chair for Cancer Research
University of Texas, MD Anderson Cancer Center
1515 Holcombe Blvd., Box 355
Houston, TX 77030
Tel: (713) 792-7743
Fax: (713) 792-7864
PROJECT 1
Our overall goal in Project 1 is to develop an effective strategy for early detection of ovarian cancer. To this end, our specific aims are to 1) identify an optimally sensitive panel of known and novel markers for ovarian cancer that, in aggregate, detect >90% of early-stage cancers 2) develop and test statistical techniques which permit utilization of multiple markers over time to detect early ovarian cancer, increasing sensitivity without compromising specificity and 3) conduct a screening trial in women at average risk to determine whether the CA 125 algorithm achieves a positive predictive value of 10% and to build a bank of serum and plasma specimens from the same volunteers over a period of years.
We are evaluating the CA125 algorithm for early detection of ovarian cancer and assembling a serum/plasma bank to evaluate specificity of multiple markers over time. All five of our screening sites (Providence, Des Moines, M. D. Anderson, UTHSC-Family Practice and Baylor-Dallas) have expanded their screening activities, including our new site at Baylor Research Institute in Dallas, TX. In addition, we are collaborating with investigators at the University of Puerto Rico who hold a U54 in collaboration with investigators at UTMDACC. As of June 12, 2008, we have now accrued 2415 healthy postmenopausal women and obtained 7145 serum/plasma specimens. Seventy-one transvaginal ultrasounds have been performed prompted by the ROC algorithm. Currently, less than 2% of screened women have been referred for TVS, consistent with our goals for specificity. Five asymptomatic women have had operations based on the CA125 algorithm and subsequent ultrasound removing two benign cystadenomas, one stage I borderline ovarian cancer and two invasive cancers in stage IIA and stage IIC respectively as our goal was to prompt no more than 10 operations to diagnose 1 ovarian cancer, it appears that the positive predictive value is indeed greater than 10%.
During the last year, we have continued to collaborate with Dr. Anna Lokshin at the University of Pittsburgh Cancer Institute to evaluate multiple biomarkers on the Luminex multiplex platform. The multimarker panel that provided the highest diagnostic power, 92% sensitivity at 98% specificity, was comprised of 4 biomarkers: CA 125, EGFR, HE4, and VCAM-1. For comparison, the sensitivity of CA 125 alone (>35U/ml) for early stage disease was 58% at 98% specificity. This panel was selective for ovarian cancer, showing 36% sensitivity for cases with benign pelvic disease, 8% sensitivity for breast cancer, 4% sensitivity for colorectal cancer, and 45% sensitivity for lung cancer. Thus, a panel of CA 125, EGFR, HE4, and sVCAM-1, could serve as an initial stage in a screening strategy for epithelial ovarian cancer prompting ultrasound in a small fraction of women. With a specificity of 98%, only 2% of women would require referral for transvaginal sonography (TVS) in a two-stage screening strategy.
Collaborative studies have also been conducted with investigators from Vermillion using SELDI proteomic techniques in our laboratory at M. D. Anderson and with Dr. Zhen Zhang and Dr. Dan Chan at Johns Hopkins. Again, our goal was to identify a panel of proteomic markers that might compliment CA125 to achieve 90% sensitivity for early stage disease at 98% specificity. Samples in both a training and test set were analyzed using the CA125II immunoassay and SELDI-TOF-MS protocols to measure 7 proteins [transthyretin, Apo-A1, transferrin, hepcidin, ß2M, CTAP3, and ITIH4. In the training set analysis, a combination of CA125 and the panel of 7 biomarkers achieved a sensitivity of 80% at 98% specificity (CA125 alone = 68 % sensitivity). In the validation set, CA125 exhibited a sensitivity of 79% for stage I, whereas the marker panel plus CA125 produced a sensitivity of 87% at 98% specificity (P= 0.015, McNemar's test). In addition, we have assayed the seven peptides against the panel of more than 900 specimens assembled by the SPORE collaborative Prostate, Lung, Colon, and Ovary (PLCO)/Early Detection Research Network (EDRN) project with the other SPOREs to evaluate multiple markers and to gain access to preclinical samples from the PLCO trial.
Although serum assays have been evaluated most extensively, detection of reliable biomarkers in urine could provide a more convenient test for mass screening in the community. In urine, we have found that levels of soluble mesothelin related protein were elevated in 40% of patients with stage I-II ovarian cancer when values were corrected for glomerular filtration rate. Likewise, in collaboration with Dr. Patricia Kruk at the University of South Florida, we have found that BCL2 is elevated in 19% of urines from stage I-II ovarian cancer patients at the 95th percentile. When used in combination at these cutoffs, 18 of 32 stage I-II cancers (56%) could be identified with one or the other biomarker using these cutoffs. Using SELDI, Thirteen peaks detected in urine distinguished healthy individuals from patients with stage I-II ovarian cancer (P<0.05) and an overlapping set with 13 peaks distinguished patients with benign ovarian disease from stage I–II ovarian cancer. The best of these markers could also improve the performance of serum CA125. An algorithm is currently being developed to utilize multiple proteomic markers in urine with SMRP and BCL2. Our goal is to achieve 90% sensitivity at 95% specificity.
PROJECT 2
The overall goal of this project is to develop new strategies for targeting blood vessels in ovarian cancers.
Our aim to characterize the toxicity and biologically-active dose of VEGF-Trap (high-affinity VEGF decoy receptor) in combination with taxane chemotherapy in ovarian cancer patients is being explored as a phase I/II clinical study of VEGF Trap, now called aflibercept. This is in combination with docetaxel in patients with recurrent and measurable ovarian cancer in whom no more than 2 previous regimens have been administered. The phase I portion of the trial has been completed identifying no DLT at the maximum dose (Level III) of 6 mg/kg. Although not a specific goal of the phase I study, responses have been observed in several patients, including 2 confirmed partial responses. Of the 3 other phase I patients still on therapy, stable disease has been observed, however, all have had declining CA125 values. Therefore, results with the phase I clinical trial with VEGF-Trap in combination with docetaxel appear encouraging. The Phase II study has opened and has enrolled 3 patients to date. Enrollment in this aspect of the trial has been brisk and we plan to open the two-stage designed phase II in two other partnering institutions. Evaluation for clinical efficacy will commence when approximately 24 patients have been enrolled. The maximum accrual in this trial will be 58 patients.
With regard to pre-clinical findings, we have discovered novel mechanisms by which perivascular cells (pericytes) provide a survival advantage for endothelial cells. In a series of laboratory studies using a well-characterized murine model of ovarian carcinoma, we evaluated the efficacy of strategies that target both endothelial cells (VEGF blockers) and pericytes (PDGF-BB signaling blockers). To determine the impact of vessel maturation on anti-angiogenic therapy, we developed a fully orthotopic model of ovarian cancer by injecting tumor cells directly into the ovary. Relative to normal vessels, tumor vasculature was characterized by loosely attached pericytes in reduced density. PDGF-BB was expressed predominantly by the endothelial and cancer cells whereas PDGFRβ was present in pericyte-like cells. PDGF-BB significantly increased migration of and VEGF production by pericyte-like cells while PDGFRβ blockade abrogated these effects. Moreover, we demonstrated that pericytes provide a survival advantage for tumor-associated endothelial cells. Consistent with this finding, dual-targeting of pericytes and endothelial cells (via VEGF (VEGF-Trap) and PDGF-BB (PDGF-Trap)) targeted therapy was more effective in inhibiting in vivo tumor growth and regression of formed tumors than either agent alone. Additional experiments are underway to identify the interactions between endothelial cells and pericytes.
We further identified that focal adhesion kinase (FAK) is rapidly activated in pericytes in response to PDGF-BB stimulation. Moreover, FAK plays a functional significant role in regulating VEGF production by pericytes in response to PDGF-BB stimulation. Targeting FAK with a small molecule inhibitor (TAE226) was highly effective in reducing tumor growth, both alone and in combination with taxane-based chemotherapy. This work has been published in Cancer Research over the last year. Additionally, we have recently discovered that neuropilins are expressed on pericytes and will carry out experiment during the upcoming year to understand their functional role in different cell populations in the tumor microenvironment.
In addition, we have identified novel genes that are overexpressed in tumor vasculature and may serve as new targets for anti-angiogenesis approaches. We have characterized the role of a polycomb group protein enhancer of Zeste homologue 2 (EZH2) in tumor vasculature. In collaboration with Dr. Michael Birrer at the NCI, we performed pathway analyses and predicted that EZH2 expression in the tumor vasculature would be increased in response to VEGF stimulation. Over the last year, we have demonstrated that indeed EZH2 promoter activity is increased in response to VEGF treatment. We have uncovered novel roles of EZH2 in regulating tumor angiogenesis and are close to submitting a manuscript regarding these findings.
Furthermore, we have found that metronomic chemotherapy (frequent administration of chemotherapeutic agents at substantially lower doses with no prolonged drug-free breaks) is highly effective in combination with anti-angiogenesis agents. We studied the effect of metronomic chemotherapy in combination with VEGF & PDGF blockers in ovarian carcinoma using in vivo experimental mouse models of advanced human ovarian cancer. Over the last year, we have examined the role of recently described Dll4 (delta-like ligand 4: a novel angiogenic ligand of the Notch receptor), inhibition in ovarian cancer. Dll4 is highly expressed in tumor vessels. Our preliminary data indicate that when Dll4 inhibition (using Dll4-Fc) is administered alone in our orthotopic mouse model of human ovarian cancer, it resulted in slightly increased ascites, modest decreases in tumor burden, and abnormal vasculature. In striking contrast, when Dll4 FC is coupled with VEGF inhibition (using a soluble VEGF decoy receptor, the VEGF Trap), tumor burden is strikingly inhibited and little ascites is seen. This combination anti-angiogenesis regimen holds promise as a novel therapeutic modality; however, the mechanisms of its potency need to be determined and will be examined further during this year.
Finally, over the last year, we have demonstrated that topotecan is a potent inhibitor of HIF1-alpha. Moreover, topotecan demonstrates metronomic activity in that it inhibits endothelial cells at substantially lower doses than those required for tumor cell inhibition. We have also carried out in vivo experiments and demonstrated that topotecan is efficacious even at metronomic doses compared to MTD based dosing. We anticipate completing the mechanism-oriented studies for this project and hope to submit a manuscript.
PROJECT 3
One aim of this project was to complete the phase I/II trial of E1A gene therapy combined with chemotherapy. Between October 2005 and May 2007, we entered 18 patients onto the Phase I/II trial (ID02-231) of IV Paclitaxel +/- IP Liposomal-E1A, with nine randomized to each treatment arm. Responses to treatment were observed in both arms of the trial. Toxicity was tolerable and dose-limiting toxicity was intraperitoneal pain. As biopsies of cancer deposits before and after treatment for correlative studies constituted a secondary endpoint of the trial, a separate consent was required by our IRB and few biopsies were actually obtained. In May 2007, we were notified by the company from whom we had purchased the liposomal E1A that the lipid in the liposomes had deteriorated during storage and no longer met the quality assurance standards agreed upon with the FDA when the drug was first formulated. Consequently, the only drug supply had aged and no longer met FDA standards. Therefore, we suspended accrual to the study, and the trial was officially closed by the Data Safety Monitoring Board at M.D. Anderson in October 2007.
Although it is disappointing that we had to stop the ongoing trial, we have encouraging pre-clinical data that allowed us to propose a new trial for platinum-resistant ovarian cancer patients using a liposome that will permit weekly i.v. injection avoiding the inconvenience to the patient of i.p. administration. We expect the RAC submission will be in late 2008 or early 2009.
Another aim of our project is to develop an ovarian cancer- specific gene delivery system and expression vector. Through inter-SPORE collaborations with Breast and Pancreatic Cancers in MDACC, we successfully developed an expression system that targets cancer tissues but not normal organs, namely, T-VISA (hTERT-VP16-GAL4-WPRE integrated systemic amplifier). The T-VISA expression vector exhibited ovarian cancer-specificity in cell culture and in an animal model. With robust expression of T-VISA in ovarian cancer cells in both cell culture and animal systems, we have constructed T-VISA-E1A to test its specificity and cell killing activity in ovarian cancer cells. As expected, T-VISA-E1A expression suppressed growth in several different ovarian cancer cell lines, but not in normal cells. To test the toxicity of T-VISA-E1A in vivo, T-VISA-E1A was systemically administered via i.v. injection, we have observed low or virtually no toxicity and still retained strong tumor suppression in the treated animals. This new i.v. liposome formulation with ovarian cancer-specific expression vector, T-VISA-E1A provides an ideal candidate to move into a Phase I clinical trial.
Finally, we are also working to develop an effective combination of E1A gene therapy with other agents in a preclinical ovarian cancer model.
PROJECT 4
In our first aim, we described the development and antitumor activity of a novel RGDS-conjugated LY294002 pro-drug termed SF1126. The PI3K inhibitor LY294002 is not a viable drug candidate for clinical development due to poor solubility and a short half-life. In contrast, SF1126 is water soluble, has favorable pharmokinetics, and is well tolerated in murine systems. The capacity of SF1126 to inhibit U87MG and PC3 tumor growth was enhanced by the RGDS integrin (alpha v beta 3/alpha 5 beta 1) binding component. Antitumor activity of SF1126 was associated with the accumulation of SF1126 in tumor tissue and with pharmacodynamic knockdown of phosphorylated AKT in vivo. SF1126 exhibits both antitumor and antiangiogenic activity and is a viable pan PI3K inhibitor for assessment in human cancer clinical trials.
To facilitate better identification of baseline biomarkers for efficacy of PI3K pathway and other inhibitors in vitro, we have almost fully completed a comprehensive analysis of 49 ovarian cancer cell lines using 500K single nucleotide polymorphism (SNP) profiling, exon array profiling, selected gene sequencing and RPPA. These ovarian cancer cell lines are also being treated with a panel of targeted inhibitors including PDK1, PI3K, AKT, mTOR, membrane receptor, PLK1 and AURKB inhibitors to determine differential sensitivity. For example, preliminary data indicate that PI3K inhibitors are highly active against ovarian cancer cells with genomic and proteomic aberrations in the PI3K pathway. A publicly accessible website that will store the comprehensive data derived from this ovarian cancer cell line project is under construction.
Our clinical trial ‘A Pilot Pharmacodynamic Trial Using Molecular Marker and Imaging Studies As Primary Endpoints To Determine The Optimal Biological Dosage Of Perifosine, Orally Available AKT PH domain Inhibitor, Combined With Docetaxel In Patients With Relapsed Epithelial Ovarian Cancer’ is currently ongoing. Patient enrollment (ten in total) to the first two of three perifosine dose levels is completed. Tumor tissue, hair follicles, blood, DC-MRI and CT-PET imaging are being collected in each patient before and soon after starting clinical trial therapy. We have not reached dose limiting toxicities or the optimal biologic dose of perifosine and patient enrollment (ten patients) at dose level three (the maximum dose level) has begun. We have also initiated negotiations with a number of companies for follow on trials using PI3K inhibitors.
We will complete the comprehensive ovarian cancer cell line project described above and make these data publicly available. We will also complete the perifosine clinical trial along with the associated correlative studies. If we can demonstrate that PI3K biomarker modulation occurs at the highest perifosine dose level, a phase II efficacy study of perifosine and docetaxel is planned. In case we cannot demonstrate PI3K biomarker modulation by the AKT inhibitor perifosine, a phase I/II optimal biologic dosing and efficacy study of the PI3K inhibitor SF1126 in combination with docetaxel is currently being planned in ovarian cancer. To facilitate this, we are in discussion with Semafore Pharmaceuticals. Funding for our expanded clinical trial efforts will be obtained from multiple sources in addition to SPORE funding. For example, Dr. Hennessy, a co-PI on this SPORE project, has obtained funding for expansion of our clinical trial efforts through a CDA from ASCO.
PROJECT 5
The overall goals of this project are to develop strategies to identify ovarian cancer patients least likely to respond to modern platinum/taxane-based treatments and to develop therapeutic strategies to attack cancer cells in these patients. We have identified genes in regions of amplification at 8q24, 11q13, and/or 20q11-q13 that are associated with poor patient outcome and chemoresistant disease. We are currently developing siRNA and/or small molecule inhibitors directed against two of the most promising genes as therapeutic agents for the treatment of chemoresistant ovarian cancer.
PVT1: In the past year, it has become clear that the region at chromosome 8q24 encoding PVT1 is variably transcribed into RNA and that some transcripts extend into intronic regions and encode putative miRNAs. We have explored PVT1 as a possible target since chemically modified antisense oligonucleotides complementary to specific miRNAs, so-called ‘Antagomirs’, are attractive therapeutic compositions that can be introduced clinically. We found that the number of viable cells in hsa-miR-1204 AntagmiR-transfected HEY cells decreased 40% relative to that of cells treated with AntagmiR control at 48h. We also investigated the effects of PVT1-encoded miRNAs on cell apoptosis in HEY cells in which PVT1 is amplified and over expressed. We found increased apoptosis in hsa-miR-1204 AntagmiR transfected cells. Importantly, the hsa-miR-1204 AntagmiR had no effect on OV90 cells in which the PVT1 locus is not amplified. Studies of the effects of siRNAs against PVT1 developed last year also suppressed hsa-miR-1204 expression suggesting that this may be one mechanism through which these siRNAs act.
We also initiated assessment of the effects of over expressing miR-1204 at the PVT1 locus. We showed that PVT1 over expression dramatically increased soft agar clonogenicity of Rat1a-Myc cells. The growth rates of Rat1a-Myc-GFP and those expressing another gene locus reported by Huppi (Rat1a-Myc-PVT1 huppi) were, however, virtually the same in colony numbers when they were grown adherent to plastic dishes. These results indicate that PVT1 miR-1204 cooperates with c-MYC to potentiate Rat1a cells colony growth.
To explore the hypotheses that over expression of miRs encoded in the PVT1 transcript contribute to ovarian cancer tumorigenesis by suppressing apoptotic responses normally elicited by amplification of the proximal oncogene, MYC and the miRs contribute to resistance of drugs that normally operate though apoptotic mechanisms, we have assessed responses to liposomal siRNA targeted therapies in orthotopic models of ovarian cancer in nude mice using cell lines with and without amplification at the PVT1 locus. The PVT1 amplified tumor, HeyA8, responded to the PVT1 siRNA therapy as well as docetaxel for the treatment of ovarian cancer, and siPVT1 plus docetaxel was synergistic. However, the PVT1-nonamplified line responded only to docetaxel. These results support our hypothesis that PVT1 amplification acts by suppressing an apoptotic response and encourages further development of therapeutic approaches to attack ovarian tumors that have amplified the PVT1 locus. We are currently investigating the use of liposomal PVT1 AntagmiRs in vivo as alternative agents for the treatment of chemoresistant ovarian cancer.
PFDN4: We have continued assessment of the PFDN4 gene, a chaperone subunit, which maps a region of recurrent amplification on chromosome 20q13. In support of a role for PFDN4 in folding of functional tubulin (one of the best-known client proteins of prefoldin), we have demonstrated that siRNA mediated silencing of PFDN4 gene expression results in microtubule cytoskeleton malformations in vitro. Of potential clinical significance, we have confirmed that patients whose tumors stained positively for PFDN4 protein had median survival times of 1.84 years as compared to patients whose tumors weakly expressed PFDN4 8.94 years (P = 0.001). Furthermore, ovarian cancer cell lines that are chemoresistant to taxanes or platinum-based agents, express PFDN4 at higher levels compared to their chemosensitive counterparts. Taken together, these data suggest that PFDN4 overexpression contributes to chemoresistance and poor clinical outcome for ovarian cancer patients.
Based on the known role for prefoldins in tubulin synthesis, we have focused on the effect of PFDN4-mediated gene silencing as both an alternative and compliment to conventional taxane therapy for the treatment of ovarian cancer. For these studies we have used an orthotopic model of ovarian cancer in nude mice using both taxane sensitive and resistant cell lines. Mice bearing intraperitoneal ovarian tumors were treated with liposomal PFDN4 siRNA, both alone and in combination with docetaxel. In taxane-sensitive tumors (HeyA8), PFDN4 siRNA therapy worked as well as docetaxel for the treatment of ovarian cancer, where a 64% reduction in tumor size was observed. When used in combination with docetaxel, PFDN4 siRNA therapy resulted in a >90% reduction in tumor size as compared to control. This data was confirmed using a second ovarian cancer cell line, SKOV3-ip1. In a taxane-resistant model of ovarian cancer (HeyA8-MDR), PFDN4 siRNA therapy resulted in a modest 30% reduction in tumor size, whereas combination therapy with docetaxel resulted in an ~70% reduction in tumor size, as compared to either control or single-agent taxane treated tumors. These data suggest that PFDN4 therapy may, in fact, resensitize taxane-resistant tumors to the effects of docetaxel.
There is growing evidence that prefoldin family members, including PFDN4, play a role in additional protein complexes, including those mediating DNA repair. This suggests that PFDN4 over expression may further contribute to chemoresistance to agents such as cisplatin and topotecan, which trigger cancer cell apoptosis by inducing DNA damage. We are currently characterizing whether PFDN4-based therapy can enhance the effects of cisplatin and topotecan on SKOV3-ip1 tumor growth.
Other targets: Finally, we are continuing in vitro characterization of additional genes whose amplifications are associated with poor patient survival, including ZNF217 (a putative zinc-finger transcription factor) and EVI1 (ecotropic viral integration site 1). We anticipate that these genes also represent therapeutic targets for the more effective treatment of chemoresistant ovarian cancer.
PATHOLOGY CORE
In the previous 12 months, the Pathology Core has collected tissue, blood, urine, and/or ascites from 445 patients undergoing surgery by a gynecologic surgeon at MDACC. The core has collected from 96 women with primary ovarian or peritoneal serous carcinoma. Additionally, the Core has collected tissues and/or fluids from 80 women with either pathologically verified normal ovaries or benign ovarian processes, such as ovarian cysts or endometriosis. The Core has provided to SPORE researchers 3,500 H&E stained slides and corresponding 8,000 unstained slides for use in immunohistochemistry and in situ hybridization experiments. This includes distribution of ovarian cancer tissue microarray slides.
The Core has also maintained interactions with the various SPORE projects. The Core continues to provide samples of serum, urine, frozen ovarian cancer, and frozen normal ovary to support the proteomic experiments in Project 1. These experiments are designed to identify early molecular markers of ovarian cancer. To support the experiments examining angiogenic factors in ovarian cancer in Project 2, the Core has provided unstained sections of ovarian cancers from paraffin-embedded material and OCT-embedded frozen samples. The Core provided unstained sections of ovarian cancer tissues for immunohistochemistry experiments in Project 3. For Projects 4 and 5, the Core has provided frozen tissues of ovarian cancers and normal ovaries and ovarian cancer tissue microarrays to support the FISH, CGH, protein lysate array, and immunohistochemistry experiments.
THE ADMINISTRATIVE CORE
Robert C. Bast, Jr., M.D. (SPORE Principal Investigator) and David M. Gershenson, M.D. (SPORE Co-Principal Investigator) direct this Core and co-chair the Executive Committee of the SPORE, which comprises all Co-Principal Investigators and all Core Directors. The Executive Committee meets monthly to review the SPORE’s scientific progress and fiscal status. The Committee identifies problems and barriers and assures that all goals are met within a realistic time and within the approved budget limitations.
The Administrative Core performed the following specific responsibilities: (1) oversight of all activities of the SPORE, including Projects and Core Resources; (2) compliance with all general governmental and NCI regulations and requirements; (3) communication and consultation with the NCI SPORE Project Officer and other staff and preparation of all required reports (4) coordination of data quality control and quality assurance issues; (5) maintenance of fiscal and budgetary functions; (6) convening of monthly meetings of the SPORE Executive Committee; (7) administration of the Developmental Research Program; (8) administration of the Career Development Program; (9) coordination of SPORE Internal/External Advisory Board Committee Retreat; (10) convening of monthly meetings of the entire Executive Committee ; (11) convening of monthly scientific meetings with presentations by Project Leaders, Developmental Project investigators, and Career Development investigators; (12) assists with the preparation of all reports, abstracts, and manuscripts resulting from the SPORE; (13) coordination with other organ-site SPORES and ovarian cancer SPORES to promote and maintain communication and integration, including the distribution of materials, electronic communications, and conference calls; (14) coordination for all attendees of the annual SPORE Workshop, which includes the distribution of material from NCI and travel, hotel and meeting registration; and (15) coordination of NCI Ovarian SPORE PI monthly conference calls. On April 25, 2008, an External/Internal Advisory Meeting was held at MD Anderson.
http://www.mdanderson.org/departments/ovarianspore/
|
