Paclitaxel With and Without Pazopanib for Persistent or Recurrent Ovarian Cancer: A Randomized Clinical Trial

Key Points

Question  Does the addition of pazopanib to weekly paclitaxel improve progression-free survival in women with recurrent epithelial ovarian cancer?

Findings  In this randomized phase 2 clinical trial that included 106 women, the median (range) progression-free survival was 7.5 (90% CI, 5.9-9.0) months for paclitaxel with pazopanib vs 6.2 (90% CI, 5.6-8.3) months for paclitaxel plus placebo, which was not statistically significant.

Meaning  The addition of pazopanib to weekly paclitaxel does not improve progression-free survival in women with recurrent epithelial ovarian cancer.

Importance  Ovarian cancer is the leading cause of gynecologic cancer deaths in the United States. Pazopanib is an oral, multitarget kinase inhibitor of vascular endothelial growth factor receptors 1, 2, and 3; platelet-derived growth factor receptors α and β; and proto-oncogene receptor tyrosine kinase (c-KIT).

Objective  To estimate the progression-free survival (PFS) hazard ratio (HR) of weekly paclitaxel and pazopanib compared with weekly paclitaxel and placebo in women with recurrent ovarian cancer. Secondary objectives included frequency and severity of adverse events, proportion responding, and overall survival (OS) in each arm. Translational research objectives included exploring the association between possible biomarkers and single-nucleotide polymorphisms in vascular endothelial growth factor A, interleukin 8, and hypoxia-inducible factor 1α; and PFS, OS, and proportion responding.

Design, Setting, and Participants  A randomized, placebo-controlled, double-blind phase 2 study was conducted at 26 participating institutions. Patients were enrolled between December 12, 2011, and April 22, 2013. Data were frozen on August 11, 2014. Participants were patients with persistent or recurrent epithelial ovarian, fallopian tube, or primary peritoneal carcinoma with 1 to 3 prior regimens and performance status of 0 to 2. One hundred six patients enrolled; 100 were evaluable for toxic effects.

Interventions  All patients received paclitaxel 80 mg/m² intravenously on days 1, 8, and 15 every 28 days and were randomized 1:1 to pazopanib 800 mg orally daily or placebo.

Main Outcomes and Measures  The primary end point was PFS. The study was designed to detect a 37.5% reduction in the hazard with 80% power (α = 10%).

Results  A total of 106 women (median age [range], 61 [35-87] years; 88 [83%] white) were enrolled. Study arms were well balanced for age, performance status, measurable disease, and prior bevacizumab. Proportion responding was 14 of 44 (31.8%) vs 10 of 44 (22.7%) for pazopanib plus paclitaxel vs paclitaxel alone. Median PFS was 7.5 vs 6.2 months for pazopanib plus paclitaxel vs paclitaxel alone, respectively (HR, 0.84; 90% CI, 0.57-1.22; P = .20). Median OS was 20.7 vs 23.3 months for pazopanib plus paclitaxel vs paclitaxel alone (HR, 1.04; 90% CI, 0.60-1.79; P = .90). Severe hypertension was more common on the pazopanib plus paclitaxel arm (relative risk, 12.0; 95% CI, 1.62-88.84). More patients discontinued treatment on the paclitaxel arm for disease progression (34 of 52 [65.4%] vs 17 of 54 [31.5%]), and more on the pazopanib plus paclitaxel arm for adverse events (20 of 54 [37%] vs 5 of 52 [9.6%]). No association was found between single-nucleotide polymorphisms (interleukin 8 and hypoxia-inducible factor 1α) and OS and proportion responding. Patients with VEGFA CC genotype may be more resistant to weekly paclitaxel than those with the AC or AA genotype, with 1 of 14 (7%), 3 of 15 (20%), and 4 of 8 (50%) responding, respectively.

Conclusions and Relevance  The combination of pazopanib plus paclitaxel is not superior to paclitaxel in women with recurrent ovarian cancer.

Trial Registration  clinicaltrials.gov Identifier: NCT01468909

Ovarian cancer is the leading cause of gynecologic cancer deaths and the fifth most common cause of cancer deaths in women in the United States.¹ While there are several active cytotoxic agents for the treatment of recurrent ovarian cancer, median survival after recurrence is less than 3 years, highlighting the need for testing novel agents in this population.^2-4 Of the single agents used in platinum-resistant disease, weekly paclitaxel appears to be the most active and serves as a credible partner for novel combinations.^5-8

Multiple reports have shown that angiogenesis is associated with worse survival for patients with ovarian cancer.^9-13 Bevacizumab, a monoclonal antibody that targets vascular endothelial growth factor (VEGF), gained U.S. Food and Drug Administration approval based on the results of the AURELIA (Avastin Use in Platinum-Resistant Epithelial Ovarian Cancer) trial,⁴ in which bevacizumab in combination with chemotherapy improved progression-free survival (PFS) by 3.3 months compared with chemotherapy alone in women with platinum-resistant ovarian cancer following 1 to 2 prior regimens.¹⁴

Pazopanib is an oral multitarget tyrosine kinase inhibitor of VEGF receptors 1, 2, and 3; platelet-derived growth factor (PDGF) receptors α and β; and proto-oncogene receptor tyrosine kinase (c-KIT). Friedlander et al¹⁵ studied pazopanib in 36 patients with low-volume recurrent ovarian cancer. The cancer antigen (CA) 125 proportion responding was 31%, median (range) time to response was 29 (27-57) days, and median response duration was 113 (95% CI, 58-150) days. Pazopanib has also been studied as a maintenance therapy after first-line chemotherapy in women with ovarian cancer, demonstrating an improvement in median PFS of 5.6 months (17.9 months vs 12.3 months; P = .002).¹⁶

The primary objective of this trial was to estimate the PFS hazard ratio (HR) of the combination of weekly paclitaxel and pazopanib compared with weekly paclitaxel and placebo in women with persistent or recurrent ovarian cancer. Secondary objectives included determining the frequency and severity of adverse events and comparing proportion responding, overall survival (OS), and duration of response in each arm. Translational research objectives included exploring the association between possible biomarkers and single-nucleotide polymorphisms (SNPs) in vascular endothelial growth factor A (VEGFA), interleukin 8 (IL-8), and hypoxia inducible factor 1α (HIF1A); and PFS, OS, and proportion responding.

All patients provided informed consent prior to enrollment, and approval was provided by the institutional review boards at all participating institutions. Patients were 18 years or older and had recurrent or persistent epithelial ovarian, fallopian tube, or primary peritoneal cancer. Patients must have received at least 1 platinum-based chemotherapy regimen for the management of primary disease and up to 2 additional cytotoxic regimens with no more than 1 nonplatinum, nontaxane regimen. Treatment with weekly paclitaxel for recurrent or persistent disease was prohibited. Patients were allowed to receive a biologic or targeted agent for primary disease treatment, but biologic or targeted agents targeting the VEGF and/or PDGF pathways were not permitted for recurrent or persistent disease. Patients were required to have measurable or detectable disease. Patients with measurable disease had at least 1 target lesion to be used to assess response as defined by Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1.¹⁷ Detectable disease included patients with ascites, pleural effusion, or solid and/or cystic abnormalities on radiographic imaging that did not meet RECIST criteria for target lesions. Patients who received 1 prior regimen could have a Gynecologic Oncology Group (GOG) performance status of 0 to 2. Patients who received 2 to 3 prior regimens were required to have a GOG performance status of 0 to 1. Patients needed adequate hematologic, renal, hepatic, and thyroid function.

Exclusion criteria included clinically significant cardiac disease, uncontrolled hypertension, increased risk of gastrointestinal bleeding or perforation, brain metastases, and history of cerebrovascular accident within 6 months prior to initiation of study treatment. Additional eligibility and exclusion criteria are available in the study protocol in the eAppendix (Supplement 1).

This study was a national, randomized, double-blind, placebo-controlled phase 2b trial of weekly paclitaxel with or without pazopanib. Patients were randomized 1:1 using a permuted block design (block size, 1) and stratified by their platinum-free interval (≤182 days vs >182 days), measurable disease status (measurable vs detectable), and prior use of bevacizumab therapy (none vs prior). Randomization was performed by the GOG Statistical and Data Center in Buffalo, New York, using a computer-generated random allocation sequence with an algorithm that required a seed. The seed was kept at the statistical center and not communicated to members outside the office (which helped conceal the assignments). Patients were randomized to paclitaxel 80 mg/m² weekly on days 1, 8, and 15 (1 hour intravenous infusion) with placebo or pazopanib 800 mg orally daily every 28 days. The required absolute neutrophil count on day 1 was at least 1500 cells/μL and the required platelet count was at least 100 000/μL. Day 8 and day 15 paclitaxel treatment was given if absolute neutrophil count was at least 1000 cells/μL and the platelet count was at least 75 000/μL. Patients who required more than 2 dose reductions of paclitaxel were removed from the study. Patients requiring more than 2 dose reductions of pazopanib or placebo discontinued pazopanib or placebo (with continuation of paclitaxel, if appropriate, until unacceptable toxic effects or progression of disease). Prophylactic growth factors were allowed if patients experienced recurrent neutropenic complications after treatment modifications. Patients were removed from the study treatment for any toxic effect requiring a dose interruption more than 14 days. Treatment was continued until disease progression or adverse effects prohibited further therapy.

The primary objective was to estimate the stratified PFS HR of weekly paclitaxel and pazopanib compared with weekly paclitaxel and placebo in patients with persistent or recurrent ovarian, fallopian tube, or primary peritoneal cancer. Progression-free survival was defined as the time from randomization to time of progression or death, whichever occurred first, or date of last contact. Secondary end points included adverse events as assessed by Common Terminology Criteria for Adverse Events version 4.0, the proportion responding (RECIST version 1.1), and OS, defined as the time from study entry to the time of death or the date of last contact. Patients had computed tomography (CT) scans or magnetic resonance imaging at baseline and every other cycle (or every 8 weeks if off treatment prior to progression) for the first 6 months, then every 3 months thereafter until disease progression. Confirmation of complete and partial responses was required at 4 weeks or longer from initial documentation. Cancer antigen 125 was not used to assess response or progression in this study. An intent-to-treat analysis was performed for PFS and OS, and Kaplan-Meier curves were generated. To assess associations between severe toxic effects and the treatment regimen, toxic effects were dichotomized into severe vs not severe. Severe toxic effects were defined as grade 4 to 5 hematological toxic effects or grade 3 to 5 nonhematological toxic effects. An exact χ² test was used to determine significance.¹⁸ Statistical analysis was performed using SAS version 9.4 (SAS Institute).

Because this is a phase 2 study, the probabilities of type I and type II errors were 10% and 20%, respectively. Eighty-four events were targeted at the final analysis for power of 80% to detect a 37.5% reduction in the HR. This would result in an increase of the proportion of patients surviving progression free at 5 months from 50% in the placebo arm to 65% in the pazopanib arm. An interim futility analysis according to Wie and colleagues¹⁹ was performed at the 61st event. The final rejection boundary was adjusted according to Jennison and Turnbull²⁰ owing to small sample size. When the final analysis was conducted at 84 events, the study’s realized power was 81%. To assure data maturation in a timely manner, approximately 110 patients were targeted for accrual to the clinical trial. If patients were lost prior to progression, they were censored at the date last seen. Patients who were removed from the study because of toxic effects prior to progression and who were not started on a new treatment continued to be followed with CT scans to determine date of progression. Patients were censored at the time they started a new treatment regimen if this occurred prior to documented progression.

High levels of interleukin 6 (IL-6), IL-8, and osteopontin were hypothesized to be negatively prognostic for PFS. These analyses were carried out as a priori hypothesis tests without adjustment for α. The rest were exploratory analyses, conducted on various biomarkers with PFS, OS, and proportion responding. If a P value was less than .05, the association was deemed “suggestive.” There were many tests, so any suggestion should be considered hypothesis generating. Given the sample size, power was relatively low, so negative results should not be considered definitive.

Quantification of plasma markers by chemiluminescence enzyme-linked immunosorbent assay (ELISA) was performed as follows. Plasma was collected from 77 available patients at the following points: before cycles 1, 2, and 6 for responders, or at the time the patient went off treatment before cycle 6. Plasma IL-8, IL-6, VEGF, and osteopontin were measured by ELISA based on the manufacturer’s instructions. All plasma samples were run in duplicates.

Ten mL of whole venous blood was used for DNA extraction and SNP analysis. Deoxyribonucleic acid was isolated from peripheral blood mononuclear cells using the Human Whole Blood Genomic DNA Extraction Kit (Qiagen). Single-nucleotide polymorphisms, including VEGFA (rs699947, rs833061), IL-8 (rs4073), and HIF1A (rs11549467), were assessed using TaqMan SNP Genotyping Assays (Applied Biosystems).

There were 106 patients (median age [range], 61 [35-87]; 88 [83%] white) enrolled between December 12, 2011, and April 22, 2013, at 26 institutions (Figure 1). Seven patients remained on treatment when data were frozen on August 11, 2014, 3 in the placebo arm and 4 in the combination arm. Fifty-two women were randomized to the placebo arm and 54 to the pazopanib arm. Baseline characteristics were similar between the 2 groups (eTable 1 in Supplement 2). The majority of women had 1 prior regimen, measurable disease, and were platinum resistant.

At the time of analysis, median (range) follow-up time was 17.7 (0.1-26.5) months, and 84 PFS events had occurred: 40 in the pazopanib arm and 44 in the placebo arm. At data cutoff, 24 patients (44.4%) in the pazopanib arm and 20 patients (38.5%) in the placebo arm had died. Patients received a median (range) of 5.5 (1.0-25.0) cycles on the placebo arm, and 5.0 (1.0-24.0) cycles on the pazopanib arm. About one-third of patients on both arms received more than 6 cycles (16 of 54 [29.6%] on the pazopanib arm vs 17 of 52 [32.7%] on the placebo arm). Progression-free survival was not statistically different between the 2 groups; the medians were 7.5 months (90% CI, 5.5-9.0) on the pazopanib arm compared with 6.2 months (90% CI, 5.6-8.3) on the placebo arm (HR, 0.84; 90% CI, 0.57-1.22; P = .22) (Figure 2). A log-rank test was performed to assess the cumulative number of cycles administered by regimen; there were no significant differences. Assessments of CT scans were performed at similar times in both groups (eFigure 1 in Supplement 2). A subset analysis of PFS between the pazopanib arm and the paclitaxel arm was performed for both platinum-resistant (HR, 0.73; 95% CI, 0.40-1.34) and platinum-sensitive (HR, 1.03; 95% CI, 0.55-1.91) disease. There was no difference in OS between the 2 groups; median OS was 20.7 months (90% CI, 17.0-22.6) in the pazopanib arm vs 23.3 months (17.3-not reached) in the placebo arm (HR, 1.04; 90% CI, 0.6-1.79; P = .91) (Figure 3).

Forty-four patients in each group had measurable disease and were evaluable for response. There were no significant differences in the proportion responding between the 2 groups; the response rate was 14 of 44 (31.8%) in the pazopanib arm and 10 of 44 (22.7%) in the placebo arm (Table). The relative probability of responding to pazopanib vs placebo was 1.4 (95% CI, 0.70-2.81).

Thirty-four of 52 patients (66%) receiving placebo stopped therapy because of disease progression, compared with 17 of 54 patients (31.5%) receiving pazopanib. A similar number of patients stopped therapy in both arms owing to refusal, 8 patients (15.4%) vs 7 patients (13%) in the placebo and pazopanib arms, respectively. Five of 52 patients (10%) on the placebo arm and 20 of 54 (37%) on the pazopanib arm stopped protocol therapy because of adverse events (P = .001). The most common adverse events leading to treatment discontinuation were neutropenia and neuropathy. There was one 68-year-old patient who had a treatment-related death on the pazopanib arm listed as “sudden death not otherwise specified” 1.3 months after study entry.

The most common grade 3 and 4 adverse event was neutropenia, though febrile neutropenia was uncommon (eTable 2 in Supplement 2). Vascular disorders were more common among women treated with pazopanib, notably severe hypertension (risk ratio, 12.0; 95% CI, 1.6-88.8). However, no patients discontinued therapy because of hypertension. There were 2 grade 3 bowel perforations on the pazopanib arm, both managed conservatively.

Assessment of plasma markers was available for 77 patients. Baseline levels of IL-6 and osteopontin levels were not associated with treatment response using Wilcoxon analysis (eFigure 2 in Supplement 2). However, we found that patients with lower baseline levels of IL-8 were randomized to the experimental arm vs the control arm. There were no suggestive differences in biomarker levels within each treatment over time.

Next, we evaluated selected SNPs (VEGFA, IL-8, and HIF1A) for PFS, OS, and proportion responding by the log-rank exact χ² test and the Cochran-Armitage trend test.^18,21,22 There was no association between IL-8 and HIF1A and OS and proportion responding. Interestingly, we found a suggestive association between VEGFA and response in the control group (P = .03) (eFigure 3 in Supplement 1). Patients with the VEGFA CC genotype may be more resistant to weekly paclitaxel than those with the AC or AA genotype, with 7%, 20%, and 50% responding, respectively (eFigure 4 in Supplement 1). However, among those treated with pazopanib, there was no association between VEGF SNPs and OS. We observed point estimates in the CC genotype yielding the highest proportion responding (55%) in comparison with the AC and AA genotypes, with 24% and 33% responding in these groups, respectively (eFigure 4 in Supplement 1). The Hardy-Weinberg equation was evaluated for the SNPs, and there was no deviation noted (data not shown).

This study indicates that the combination of weekly paclitaxel with pazopanib is not superior to weekly paclitaxel alone in women with recurrent ovarian, fallopian tube, and primary peritoneal cancer. There were no significant differences in PFS, OS, or proportion responding between the 2 arms. More patients stopped treatment on the control arm because of progression of disease, whereas more patients on the pazopanib arm stopped therapy because of adverse events. The addition of pazopanib to weekly paclitaxel increased toxic effects, especially severe hypertension. This is a known adverse effect of antiangiogenic agents, and it has been seen in other trials studying pazopanib.^15,16,23 Grade 3 to 4 neutropenia also increased in the combination arm, though febrile neutropenia was rare. Pignata et al²³ recently published the results of a similar randomized, open-label, phase 2 trial, MITO-11 (Weekly Paclitaxel With or Without Pazopanib in Platinum-Resistant or Refractory Ovarian Cancer). In that trial, a statistically significant increase in neutropenia was observed but was attributed to the greater number of cycles administered to patients receiving the combination. In our study, the median number of cycles was similar in both groups. More likely, the increased adverse effects are related to an increase in systemic exposure to paclitaxel when coadministered with pazopanib. Two phase 1 studies^24,25 showed a 36% to 38% increase in systemic exposure of paclitaxel when given with pazopanib vs paclitaxel alone. The combination of pazopanib with paclitaxel appears to be more toxic than other antiangiogenic agents combined with paclitaxel.^4,24 Given that these women have incurable cancer, toxic effects are of concern, as they may have a detrimental effect on quality of life.

The MITO-11 study²³ demonstrated a 2.86-month improvement in PFS for women on the combination arm (6.35 months vs 3.49 months; P = .001). There are several notable differences between that trial and our trial. First, MITO-11 was not placebo controlled, and therefore both physicians and patients knew who was receiving pazopanib. This introduces possible bias into these results, particularly on PFS and proportion responding. Second, all patients in MITO-11 had platinum-resistant or platinum-refractory disease, whereas about half of the patients in the current study had platinum-sensitive disease. Third, no patients in MITO-11 had received prior bevacizumab, and no more than 2 previous lines of therapy were allowed. About 20% of patients in our study had a history of bevacizumab use. This is relevant, given bevacizumab has been shown to improve PFS in the upfront setting and in patients with recurrent ovarian cancer.^4,26-29

Identification of markers predictive to VEGF- and VEGFR-targeted therapy in cancer has yielded inconsistent results. It has been reported³⁰ that high levels of IL-6, IL-8, and osteopontin were negatively correlated with PFS in metastatic renal cell carcinoma and were also predictive of greater benefits from pazopanib treatment. Similarly, a recent abstract³¹ presented at the American Society of Clinical Oncology demonstrated that patients with ovarian cancer with high IL-6 levels treated with bevacizumab in combination with carboplatin and paclitaxel had longer PFS and OS compared with placebo. In addition, both osteopontin and IL-6 were negative prognostic markers for both PFS and OS. Our correlative studies showed that there were no significant changes in the biomarkers over time within the treatment and placebo arms, which may be reflective of lack of superiority of weekly paclitaxel with pazopanib compared with weekly paclitaxel alone in these patients.

Interestingly, we found a trend between different genotypes of VEGFA SNPs and treatment response in the placebo group. Those patients with the CC genotype might be more resistant than those with the AC or AA genotype, suggesting that VEGFArs699947 genotypes might be useful molecular determinants for weekly paclitaxel treatment in ovarian cancer. However, whether the baseline IL-8 and VEGFA SNPs could be used as markers predictive to pazopanib treatment will require larger studies.

The main limitation of this study is the sample size. While this is the largest study to date that we know of in women with recurrent ovarian cancer receiving weekly paclitaxel and pazopanib, it is possible that the results were affected by a type II error. The study has 1 chance in 5 of rejecting a regimen with a clinically significant impact. The point estimates for PFS and the relative probability of responding, although not statistically significant, do favor the experimental regimen with a maximum likelihood reduction in the hazard rate by 16% and a potential increase in the probability of responding by 40%. Unfortunately, there is no evidence to favor the experimental regimen by OS at this time.

The addition of pazopanib to weekly paclitaxel in women with recurrent ovarian cancer increased toxic effects and did not significantly improve patient outcomes. Results from this study do not support further investigation of this combination in this patient population at these doses and schedule.

Corresponding Author: Debra L. Richardson, MD, Section of Gynecologic Oncology, Stephenson Cancer Center, 800 NE 10th St, Suite 5050, Oklahoma City, OK 73104 (debra-richardson@ouhsc.edu).

Accepted for Publication: September 20, 2017.

Published Online: December 14, 2017. doi:10.1001/jamaoncol.2017.4218

Author Contributions: Dr Richardson had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Richardson, Sill, Coleman, Sood, Hanjani, Lankes, Aghajanian.

Acquisition, analysis, or interpretation of data: Richardson, Sill, Coleman, Sood, Pearl, Kehoe, Carney, Van Le, Zhou, Alvarez-Secord, Gray, Landrum, Lankes, Hu, Aghajanian.

Drafting of the manuscript: Richardson, Sill, Coleman, Lankes, Hu.

Critical revision of the manuscript for important intellectual content: Sill, Coleman, Sood, Pearl, Kehoe, Carney, Hanjani, Van Le, Zhou, Alvarez-Secord, Gray, Landrum, Aghajanian.

Statistical analysis: Sill, Coleman.

Administrative, technical, or material support: Richardson, Coleman, Sood, Landrum, Lankes.

Study supervision: Richardson, Coleman, Sood, Carney, Lankes.

Conflict of Interest Disclosures: Dr Richardson has received honoraria for advisory boards from Genentech and AstraZeneca. Dr Coleman has grant support from Roche/Genentech, AstraZeneca, Clovis, AbbVie, Merck, and Janssen. Dr Zhou is on the speaker’s bureau for Clovis. Dr Alvarez Secord has received clinical trial grant funding from AbbVie, Amgen, Astellas Pharma Inc, Astex Pharmaceuticals Inc, AstraZeneca, Boehringer Ingelheim, Bristol Myers Squibb, Eisai, Endocyte, Exelixis, Incyte, Merck, Prima Biomed, Roche/Genentech, and TESARO. She has also received honoraria for advisory boards from Alexion, Astex, AstraZeneca, Boehringer Ingelheim, GSK, Clovis, Janssen/Johnson & Johnson, Precision Therapeutics, Roche/Genentech, and TESARO. Dr Aghajanian has received honoraria for Focus study steering committee from Mateon Therapeutics, and honoraria for advisory boards for Clovis, Cerulean Pharma, Bayer, VentiRX, and AstraZeneca.

Funding/Support: This study was supported by National Cancer Institute grants to the Gynecologic Oncology Group Tissue Bank (grant No. U10 CA27469, U24 CA114793, U10 CA180868), NRG Oncology (grant No. 1U10 CA180822), and NRG Operations (grant No. U10CA180868). The translational research was supported in part by the National Institutes of Health (grant No. P50 CA083639) and the National Cancer Institute’s National Clinical Trials Network (award No. U10 CA180858).

Role of the Funder/Sponsor: The funders/sponsors had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Additional Information: The following Gynecologic Oncology Group member institutions participated in the primary treatment studies: Stony Brook University Medical Center, University of Texas Southwestern Medical Center, University of Hawaii, Abington Memorial Hospital, University of North Carolina at Chapel Hill, The Hospital of Central Connecticut, Duke University Medical Center, Fred Hutchinson Cancer Research Center, University of Oklahoma Health Sciences Center, University of Alabama at Birmingham, Walter Reed National Military Medical Center, Abramson Cancer Center of The University of Pennsylvania, Washington University School of Medicine, Ohio State University Comprehensive Cancer Center, Michigan Cancer Research Consortium Community Clinical Oncology Program (CCOP), Wichita CCOP, Women’s Cancer Center of Nevada, Moffitt Cancer Center and Research Institute, Women and Infants Hospital, Georgia Center for Oncology Research and Education, Cancer Research for the Ozarks National Cancer Institute Community Oncology Research Program (NCORP), Greenville Health System Cancer Institute/Greenville CCOP, Rush University Medical Center, Cleveland Clinic Foundation, Wisconsin National Cancer Institute Community Oncology Research Program, and Iowa-Wide Oncology Research Coalition NCORP.

Meeting Presentation: This study was presented in part at the 15th Biennial International Gynecologic Cancer Society Meeting; November 8-11, 2014; Melbourne, Australia.

Additional Contributions: We thank Yunjie Sun, PhD, for technical support in performing the ELISA and SNP genotyping assays, Sandra Dascomb, MS, and Kim Blaser for their contributions at the NRG Oncology SDMC Buffalo Office, and Mary McNulty, AAS (GOG Tissue Bank), for assistance with specimen distribution. The contributors did not receive any compensation for their assistance.