Protocol. This trial protocol has been provided by the authors to give readers additional information about their work.

Transcription

1 Protocol This trial protocol has been provided by the authors to give readers additional information about their work. Protocol for: Burger RA, Brady MF, Bookman MA, et al. Incorporation of bevacizumab in the primary treatment of ovarian cancer. N Engl J Med 2011;365:

2 GYNECOLOGIC ONCOLOGY GROUP ADMINISTRATIVE OFFICE FOUR PENN CENTER 1600 JOHN F. KENNEDY BLVD SUITE 1020 PHILADELPHIA, PA FAX Philip J. DiSaia, M.D. Chairman John R. Kellner Executive Director of Operations PROTOCOL GOG-0218 A PHASE III TRIAL OF CARBOPLATIN AND PACLITAXEL PLUS PLACEBO VERSUS CARBOPLATIN AND PACLITAXEL PLUS CONCURRENT BEVACIZUMAB (RHUMAB VEGF, NSC #704865, IND #7921) FOLLOWED BY PLACEBO, VERSUS CARBOPLATIN AND PACLITAXEL PLUS CONCURRENT AND EXTENDED BEVACIZUMAB, IN WOMEN WITH NEWLY DIAGNOSED, PREVIOUSLY UNTREATED, SUBOPTIMAL ADVANCED STAGE EPITHELIAL OVARIAN AND PERITONEAL PRIMARY CANCER NCI-SUPPLIED AGENT(S): BEVACIZUMAB/ PLACEBO (NSC #704865, IND #7921) NCI Version POINTS: PER CAPITA TRANSLATIONAL RESEARCH PER CAPITA Award based on specimen submissions. Distribution: Frozen tumor-3 points, tumor block-2 points (2 nd choice tumor sections-1 point), frozen serum-1 point.. MEMBERSHIP- Bonus membership point will be awarded for submission of satisfactory frozen tumor, tumor block and frozen serum STUDY CHAIR STUDY CO-CHAIR NURSE CONTACT ROBERT A. BURGER, M.D. GINI FLEMING, M.D. CAROL COWAN, R.N. UCI MED CTR, IRVINE UNIVERSITY OF CHICAGO UCI MEDICAL CTR CHAO FAM COMP CA CTR SECT OF MED/ONC (MC 2115) GYN/ONC DEPT DIVISION OF GYN ONC 5841 S. MARYLAND AVE (RM I-211) 101 THE CITY DRIVE BLDG. 56, RM. 264 CHICAGO IL BUILDING 56 (714) (773) ORANGE CA FAX: (888) FAX: (773) (714) raburger@uci.edu gfleming@medicine.bsd.uchicago.edu FAX: (714) ccowan@uci.edu DEVELOPMENTAL THERAPEUTICS CO-CHAIR MICHAEL A. BOOKMAN, M.D. SEE GOG WEBSITE DIRECTORY STATISTICIAN MARK F. BRADY, PH.D SEE GOG WEBSITE DIRECTORY TRANSLATIONAL SCIENTIFIC COLLABORATORS MICHAEL BIRRER, MD, PHD NATIONAL CANCER INSTITUTE ROBERT A. BURGER, MD UNIVERSITY OF CALIFORNIA, IRVINE JOHN P. FRUEHAUF, MD, PHD UNIVERSITY OF CALIFORNIA, IRVINE TRANSLATIONAL RESEARCH CO-CHAIR MICHAEL BIRRER, M.D., PH.D. SEE GOG WEBSITE DIRECTORY QUALITY OF LIFE CO-CHAIR: BRADLEY J. MONK, M.D. SEE GOG WEBSITE DIRECTORY TRANSLATIONAL RESEARCH SCIENTIST KATHLEEN M DARCY, PHD TRANSLATIONAL RESEARCH SCIENTIST SEE GOG WEBSITE DIRECTORY STUDY PATHOLOGIST SHARON LIANG, M.D.,PHD SEE GOG WEBSITE DIRECTORY This protocol was designed and developed by the Gynecologic Oncology Group (GOG). It is intended to be used only in conjunction with institution-specific IRB approval for study entry. No other use or reproduction is authorized by GOG nor does GOG assume any responsibility for unauthorized use of this protocol. JUNE 14, 2005 PAGE 1 OF 2

3 Patient enrollments from institutions that are not aligned with GOG will be conducted via the NCI Cancer Trials Support Unit (CTSU) and all data should be sent to CTSU Data Operations unless otherwise specified in the CTSU logistical appendix. CTSU will use the GOG-0218 number as required for reporting to GOG and NCI and when registering patients through the GOG Registrar. CTSU participants and institutions will be instructed to use the GOG-0218 study number on all data forms. CANCER TRIALS SUPPORT UNIT (CTSU) ADDRESS AND CONTACT INFORMATION To submit site registration documents: For patient enrollments: To mail study forms or data: CTSU Regulatory Office Coalition of National Cancer Cooperative Groups 1818 Market Street, Suite 1100 Philadelphia, PA Phone: Fax CTSU Data Operations Center Voice Mail Fax [For CTSU patient enrollments that must be completed within approximately one hour or other extenuating circumstances, call Please use the number for ALL other CTSU patient enrollments.] Westat CTSU Data Operations Center 1441 W. Montgomery Avenue Rockville, MD For treatment-related questions contact the Study Chair, for eligibility-related questions contact the Statistical and Data Center of the GOG at The option remains to contact CTSU Help Desk for assistance in obtaining a response from the Group. All other questions (including forms-specific questions) should be communicated by phone or e- mail to: CTSU General Information Line , or ctsucontact@westat.com. All calls and correspondence will be triaged to the appropriate CTSU representative. The CTSU Public Web site is located at: The CTSU Registered Member Web site is located at See appendix VIII for CTSU logistical information. PAGE 2 OF 2 JUNE 14, 2005

4 GOG-0218 ELIGIBILITY Epithelial ovarian or peritoneal primary cancer FIGO Stages III-IV suboptimal Randomization (cycle = 21 days): Arm I (standard chemotherapy) SCHEMA Phase A Re-registration Phase B Chemotherapy * day 1 every 21 days x 6 cycles Placebo (for Bevacizumab) ** day 1 every 21 days beginning with cycle 2 x 5 cycles Placebo (for Bevacizumab) ** day 1 every 21 days beginning with cycle 7 and continuing for 15 months (22 cycles) max total treatment time Arm II (concurrent bevacizumab) Phase A Re-registration Phase B Chemotherapy * day 1 every 21 days x 6 cycles Bevacizumab ** day 1 every 21 days beginning with cycle 2 x 5 cycles Placebo (for Bevacizumab) ** day 1 every 21 days beginning with cycle 7 and continuing for 15 months (22 cycles) max total treatment time Arm III (extended bevacizumab) Phase A Re-registration Phase B Chemotherapy * day 1 every 21 days x 6 cycles Bevacizumab ** day 1 every 21 days beginning with cycle 2 x 5 cycles Bevacizumab ** day 1 every 21 days beginning with cycle 7 and continuing for 15 months (22 cycles) max total treatment time *Paclitaxel 175mg/m 2 IV over 3 hours followed by Carboplatin AUC 6 IV over 30 minutes day 1 of cycles 1 through 6 only (Note: docetaxel 75mg/m 2 IV over 1 hour may be substituted for paclitaxel [see sections 2.65, 5.322, and 6.51].) **Bevacizumab / Placebo 15mg/kg IV day 1 of each cycle beginning with cycle 2

5 GOG-0218 OUTCOME MEASURES Primary Endpoint: -Overall Survival (OS) Secondary Endpoints: -Progression-Free Survival (PFS) -Response Rate (RR) -Toxicity -Quality of Life -Translational Research - Please see Section 7.2 as well as Appendix VI (Specimen Procedures) and Appendix VII (Laboratory Procedures) for details regarding the specimen requirements and laboratory testing for this protocol. Patients treated on this trial will not be eligible for therapy on clinical trials evaluating consolidation or maintenance therapy while on or off study.

6 GOG-0218 TABLE OF CONTENTS PAGE 1.0 OBJECTIVES BACKGROUND AND RATIONALE PATIENT ELIGIBILITY AND EXCLUSIONS STUDY MODALITIES TREATMENT PLAN AND ENTRY/RANDOMIZATION PROCEDURE TREATMENT MODIFICATIONS STUDY PARAMETERS EVALUATION CRITERIA DURATION OF STUDY STUDY MONITORING AND REPORTING PROCEDURES STATISTICAL CONSIDERATIONS BIBLIOGRAPHY 87 SUGGESTED PATIENT INFORMED CONSENT APPENDIX I -FIGO Stage Grouping for Primary Carcinoma of the Ovary (1985) APPENDIX II -Glossary of Terms APPENDIX III -NCI Standard Language Involving Agent(s) Covered by a Clinical Trials Agreement or a Cooperative Research and Development Agreement APPENDIX IV -GOG Web-Based Registration and Randomization APPENDIX V -Anaphylaxis Precautions APPENDIX VI -Specimens Procedures for Translational Research APPENDIX VII -Laboratory Procedures for Translational Research APPENDIX VIII -CTSU Logistical Guidelines

7 -1- GOG OBJECTIVES This is a phase III randomized study to evaluate new treatment programs for patients with advanced stage (International Federation of Gynecologic Oncology, FIGO, III-IV, Appendix I) epithelial ovarian or peritoneal primary cancer. 1.1 Primary Objectives 1.11 To determine if the addition of 5 concurrent cycles of Bevacizumab to 6 cycles of standard therapy (carboplatin and paclitaxel) [Arm II] reduces the death rate when compared to 6 cycles of standard therapy alone [Arm I] in women with newly diagnosed suboptimal advanced epithelial ovarian and peritoneal primary cancer To determine if the addition of 5 concurrent cycles plus extended Bevacizumab for 15 months total treatment time to 6 cycles of standard therapy (carboplatin and paclitaxel) [Arm III] reduces the death rate when compared to 6 cycles of standard therapy [Arm I] in women with newly diagnosed suboptimal advanced epithelial ovarian and peritoneal primary cancer. 1.2 Secondary Objectives 1.21 In the event that both Arm II and Arm III regimens are superior to the Arm I regimen with respect to overall survival, to determine whether the Arm III regimen reduces the death rate when compared to the Arm II regimen 1.22 To determine whether the Arm II or Arm III regimen increases the duration of progression-free survival (PFS) when compared with the Arm I regimen To compare each of the experimental regimens to the Arm I regimen with respect to the incidence of severe toxicities or serious adverse events To determine the impact on Quality of Life (QOL, as measured by the FACT-O TOI) following treatment with the above regimens. 1.3 Translational Research Objectives 1.31 To assess the relationship between angiogenic markers and clinical outcome including tumor response, progression-free survival and overall survival in patients randomized to standard cytotoxic chemotherapy (paclitaxel and carboplatin) without Bevacizumab, with concurrent Bevacizumab or with extended Bevacizumab To assess the predictive value of a set of genes whose expression correlates with survival of patients with advanced stage suboptimal epithelial ovarian or peritoneal primary cancer.

8 2.0 BACKGROUND AND RATIONALE -2- GOG Standard Management of Advanced Ovarian and Peritoneal primary Carcinoma After initial surgical diagnosis, staging and cytoreduction, the standard primary systemic chemotherapy for women with advanced epithelial ovarian and peritoneal primary cancer consists of chemotherapy with a platinum and taxane combination, 1,2 usually carboplatin 3-6 and paclitaxel. While significant advances have been made in patient management, this disease still carries the highest fatality to case ratio for all gynecologic malignancies diagnosed in the United States. It is estimated that in 2004, 25,580 new cases will have been diagnosed and 16,090 women will have died of the disease. 7 Over the past two decades, there have been only modest improvements in overall 5-year survival, and while 5-year survival has increased steadily from 30% to 50% overall, it has improved by only 5%, from 20% to only 25% for women with advanced-stage tumors. Clearly improvements are needed in primary therapeutic strategies. 2.2 New Therapeutic Strategies to Improve Outcomes GOG-0182-ICON5 was a 5-arm randomized clinical trial comparing standard therapy (carboplatin and paclitaxel) with four investigational arms incorporating gemcitabine, topotecan and liposomal doxorubicin, either in combination or in sequence with paclitaxel and carboplatin. Major ovarian cancer clinical trials groups throughout the world participated in this study, including the MRC ICON investigators in the United Kingdom, European Institute of Oncology in Italy, and the Australia-New Zealand GOG Consortium. This international collaboration provided a unique opportunity to accrue large numbers of patients in a timely manner which facilitated the simultaneous evaluation of multiple agents in a prospective randomized trial. With international participation, accrual exceeded 1,200 patients per year, and the trial reached its targeted accrual goal within four years of activation. While the results of GOG-0182-ICON5 will help establish optimum chemotherapy for previously untreated patients with advanced ovarian and peritoneal primary cancer, the next generation of clinical trials will explore the impact of molecular targeted therapies in conjunction with chemotherapy. In particular, growth factor signal transduction inhibitors and anti-angiogenic agents as single agents and in combination with cytotoxic drugs are currently undergoing phase I and II trials in women with these tumors. Many of these agents have been shown to have cytostatic effects and have shown synergy with chemotherapy in experimental models of human cancer. In addition, since it is postulated that such biologic agents may also have a role in maintenance therapy, the general approach in phase III trials will be the evaluation of the impact on outcome of active biologic agents in combination with standard cytotoxic therapy plus or minus extended single agent administration, compared with standard cytotoxic therapy alone, in patients with advanced disease.

9 -3- GOG Rationale for Angiogenesis -Targeted Therapeutics Angiogenesis is one of the cardinal processes leading to invasion and metastasis of solid tumors. The angiogenic-signaling pathway may be triggered by the release of angiogenic promoters such as vascular endothelial growth factor (VEGF) from tumor cells into the local microenvironment. There is accumulating evidence that angiogenesis plays a central role in ovarian cancer disease progression and prognosis Given that a direct relationship has been demonstrated between the expression of biomarkers of angiogenesis and the behavior of epithelial ovarian cancer, it would seem implicit that pharmacological inhibitors of angiogenesis could arrest tumor progression Neutralizing anti-vegf monoclonal antibodies have demonstrated therapeutic activity in a variety of pre-clinical solid tumor models. 18, Role of Bevacizumab, an Anti-VEGF Monoclonal Antibody, in Epithelial Ovarian and Peritoneal primary Cancer Therapy Bevacizumab is a recombinant humanized version of a murine anti-human VEGF monoclonal antibody, named rhumab VEGF. Bevacizumab has been advanced into clinical development for use as a single agent to induce tumor growth inhibition in patients with solid tumors and for use in combination with cytotoxic chemotherapy to delay the time to disease progression in patients with metastatic solid tumors. 20 A current phase II trial of single agent Bevacizumab for patients with recurrent epithelial ovarian and peritoneal primary cancer having received one to two prior chemotherapy regimens is ongoing in the GOG (GOG-0170-D). Accrual to this twostage designed trial was completed early in September The trial advanced into its second accrual stage based on objective responses reported in the first cohort of 28 patients. The interim preliminary analysis revealed that approximately 25% demonstrated objective responses by RECIST criteria, and approximately 1/3 of patients were on treatment without evidence of progression at 6 months. This appeared to be an extremely unusual observation for a compound presumed to be at best cytostatic when administered as a single agent. The final analysis of this trial is planned for the first quarter of Experience with Combination Bevacizumab - Cytotoxic Therapy Evidence from pre-clinical studies and recent phase II and III clinical trials in other solid tumors has demonstrated enhanced anti-tumor activity of traditional cytotoxic regimens, when combined with Bevacizumab. For example, Devore and colleagues reported on a three-arm phase II randomized trial of carboplatin/paclitaxel at with or without Bevacizumab (7.5 mg/kg or 15 mg/kg dose levels) every 21 days until disease progression, in 99 patients with stages IIIB and IV non-small cell lung cancer. 21 Response rates were 21.9 percent (7/32 patients) in the low dose and 42.9 percent (14/35 patients) in the high dose Bevacizumab combination arms, compared to a response rate of 31.3 percent (10/32 patients) in the chemotherapy alone arm. A phase II/III trial in this patient population has been conducted by ECOG; the final analysis of this study is pending.

10 -4- GOG-0218 More importantly, a recently reported phase III trial, AVF2107, of over 800 previously untreated patients with previously untreated metastatic colorectal cancer randomized to receive either Bevacizumab for one year plus the Saltz chemotherapy regimen (5-FU/Leucovorin/CPT-11, IFL) or the Saltz regimen plus placebo for one year met its primary endpoint of improving overall survival. 22 The magnitude of benefit observed far exceeded what the study was designed to demonstrate. The trial also met the secondary endpoints of progression-free survival, response rate, and duration of response (see table below). Hazard IFL/ Bevacizumab IFL/ placebo Ratio (n = 403) (n = 412) (p-value) Response Rate 44.9% 34.7% (0.0029) Median TTP 10.6 months 6.2 months ( ) Median Survival 20.3 months 15.6 months 0.65 ( ) Bleeding, thrombosis, asymptomatic proteinuria and hypertension were identified in phase II studies as possible safety events, but only Grade 3 hypertension and arterial thrombosis events were clearly increased in this phase III study. Preliminary results from a more recent, large, randomized phase III trial for patients with advanced colorectal cancer who had previously received treatment show that those who received Bevacizumab in combination with an oxaliplatin regimen known as FOLFOX4 (oxaliplatin, 5-fluorouracil and leucovorin) had a significantly prolonged survival over patients who received FOLFOX4 alone. The Data Monitoring Committee overseeing the trial, known as E3200, recommended that the results of a recent interim analysis be made public because the study had met its primary endpoint of demonstrating improved overall survival, which was 17% longer in the Bevacizumab arm. Specifically, the median overall survival in the Bevacizumab plus FOLFOX4 arm was 12.5 months compared to 10.7 months for patients treated with FOLFOX4 alone. There was a 26 percent reduction in the risk of death (hazard ratio of 0.74) for patients in this study who received Bevacizumab plus FOLFOX4 compared to those who received FOLFOX4 alone. Treatment toxicities observed in this study were consistent with those adverse effects observed in other clinical trials in which Bevacizumab was combined with chemotherapy. These included hypertension and bleeding as more predominant in the Bevacizumab arm. 23

11 -5- GOG-0218 Multiple phase I-III trials, such as those cited above, have demonstrated the safety and tolerability of Bevacizumab with traditional schedules and dosing of carboplatin and paclitaxel. 2.6 Rationale for Clinical Trial Design Bevacizumab was selected for evaluation in combination with standard chemotherapy based on preliminary phase II single agent data obtained in patients with recurrent epithelial ovarian and peritoneal primary cancers (see Section 2.4) and results from a phase III clinical trial in patients with metastatic colorectal cancer demonstrating a survival benefit to patients receiving Bevacizumab with standard cytotoxic chemotherapy compared with patients receiving standard chemotherapy alone (see Section 2.5). Based on the mechanism of action of Bevacizumab, there may be benefit of extended therapy with this agent until disease progression, in extending PFS or OS in this patient population. However, it is unclear whether additional benefit of Bevacizumab beyond the general duration of standard primary chemotherapy exists. Therefore, two experimental arms were selected to compare with standard cytotoxic chemotherapy with paclitaxel and carboplatin: one incorporating 5 cycles of Bevacizumab (concurrent Bevacizumab) and the other with Bevacizumab for 15 months total treatment time (extended Bevacizumab) Study Population A study population limited to those patients with suboptimal residual tumors was selected because this group carries the poorest prognosis of those with advanced disease and would allow conservation of sample size needed to demonstrate a survival benefit associated with Bevacizumab therapy Sample Size Considerations The sample size of the proposed study was estimated with overall survival (OS) as the primary endpoint, while taking steps to protect the integrity of progression-free survival (PFS) as a secondary endpoint. At the recently held 3 rd International Ovarian Cancer Consensus Conference, September 2004, in Black Forest, Germany, an agreement was reached that PFS is a reproducible surrogate of overall survival in this population. 24 In keeping with this agreement, there will be a planned final analysis of PFS incorporated into the statistical analysis plan that is being submitted to the FDA under the Special Protocol Assessment process Optimum Number of Cytotoxic Chemotherapy Cycles It has become common practice to administer six to eight cycles of initial chemotherapy in phase III clinical trials for patients with advanced-stage ovarian cancer. The optimal number of cycles has not been defined, but there is no evidence that more than 4 cycles is associated with an improvement in long-term outcomes. Length of therapy has not been prospectively evaluated with a combination of platinum and paclitaxel. From available data, it is reasonable to conclude that the absolute number of cycles within a clinically relevant range of between 6 and 8 is unlikely to have a measurable impact on long-term disease

12 -6- GOG-0218 control. At present, there are no prospective data to indicate that dose intensity, cumulative dose delivery, or number of cycles has an impact on long-term outcomes following primary therapy with platinum and paclitaxel. There is, however, evidence of increased risk of severe adverse effects of treatment with the combination of paclitaxel and carboplatin beyond the traditional 6 cycles. These effects include cumulative platelet toxicity and increased risks of severe hypersensitivity, particularly related to carboplatin, as well as increased risk of high Grade neuropathy related to paclitaxel. The above counterbalancing factors serve as the rationale for 6 cycles of induction chemotherapy in the current trial Delay of Initial Treatment with Bevacizumab Because of the concern of potential wound complications related to Bevacizumab, in this trial Bevacizumab/placebo therapy will begin at the start of cycle number 2 of carboplatin and paclitaxel combination chemotherapy Selected Substitution of Docetaxel for Paclitaxel Publication of results from GOG Protocol and a confirmatory European trial 26 led to adoption of paclitaxel and carboplatin as the standard primary therapy for patients with advanced epithelial and peritoneal primary cancer. However, it is estimated that on the order of 5% of patients in the population eligible for participation in the current trial will develop peripheral neuropathy or refractory acute hypersensitivity infusion reactions which would necessitate discontinuation of paclitaxel. Docetaxel is a novel taxane with reduced potential for neurotoxicity compared with paclitaxel. In addition, docetaxel has been safely substituted for paclitaxel in patients experiencing sever acute hypersensitivity to paclitaxel refractory where re-challenge is either unsuccessful or deemed unsafe. With regard to efficacy, there is evidence that docetaxel is an alternative treatment option to paclitaxel for patients with epithelial ovarian and peritoneal primary cancer. Docetaxel has been combined with cisplatin or carboplatin extensively in phase II and III clinical trials. These studies have demonstrated activity in a variety of tumor types (non-small cell lung, breast, head and neck, bladder, gastric, and gynecologic malignancies) and show that combinations of these drugs are safe and feasible Docetaxel has substantial activity against platinum-refractory ovarian carcinoma 54 and is also active as primary therapy in ovarian cancer. 30,43,49,51 A phase III randomized trial (SCOTROC) of docetaxel and carboplatin versus paclitaxel and carboplatin in patients with advanced epithelial ovarian cancer has recently been published. 50 In this trial, patients received carboplatin at an AUC of 5 with either docetaxel at 75 mg/m 2 1-hour IV infusion or paclitaxel at 175 mg/m 2 3-hour IV infusion. Results of this trial demonstrated no significant difference in median progression-free survival (15.0 months versus 14.8 months), two year overall survival (64.2% versus 68.9%) or objective tumor response (58.7% versus 59.5%) for the combination of docetaxel and carboplatin versus the combination of paclitaxel and carboplatin, respectively. While docetaxel and carboplatin produced more neutropenia

13 -7- GOG-0218 (Grade 3-4 neutropenia 94% for docetaxel and carboplatin versus 84% for paclitaxel and carboplatin, p <.001) and neutropenic complications than treatment with paclitaxel-carboplatin, the docetaxel and carboplatin regimen was significantly less neurotoxic (Grade 2 neurosensory toxicity in 11% for docetaxel and carboplatin versus 30% for paclitaxel and carboplatin, p.001). The results of the SCOTROC trial have led many oncologists to select substitution of docetaxel for paclitaxel in first line therapy for patients with advanced epithelial and peritoneal primary cancer. Thus, in order to optimize cytotoxic therapy in all arms of the current trial, reduce the likelihood of protocol violations and avoid imbalances in the type of taxane utilized in each treatment arm, in the current trial docetaxel will be selectively substituted for paclitaxel in circumstances in which peripheral neuropathy or hypersensitivity warrants discontinuation of paclitaxel (Section 6.51 and Section 6.62, respectively) Post-Remission Therapy It is expected that all of the chemotherapy regimens employed in this trial will achieve an overall response rate of greater than 75%. However, as many as 90% of patients with suboptimal stages III and IV epithelial ovarian and peritoneal primary cancer in clinical complete remission will ultimately recur and die of disease. Therefore, a number of strategies are under active consideration to delay or prevent recurrence. Among these strategies include consolidation treatment with cytotoxic, hormonal, or biologic targeted agents. For example, recent data have revealed that continuation of single-agent paclitaxel on a monthly schedule for 12 cycles significantly extended progression-free survival. 55 Certainly consolidation therapy has been implemented variably in clinical practice outside clinical trials with the decision based on physician and patient preference, with no evidence that overall survival is influenced by either treatment of patients in complete clinical remission or for that matter, at the time of clinical disease progression.

14 -8- GOG-0218 Due to the lack of evidence that any current consolidation approach is associated with an improvement in overall survival, and our desire to preserve the integrity of the progression-free interval as a secondary endpoint of clinical importance, the current trial design will control for the potential use of consolidation therapies, including agents unique to the experimental regimen, as well as the potential for investigator assessment bias. In order to accomplish this goal, the arms will be placebo-controlled. Based on the mechanism of action of Bevacizumab, there may be benefit of extended therapy with this agent, justifying the inclusion of an extended therapy arm in the current trial. Specifically, a total duration of 15 months of treatment for the extended Bevacizumab/placebo therapy arm was chosen since it approximates the median PFS in this population of women with stages III and IV epithelial and peritoneal primary cancer with suboptimal residual disease, based on data from recent GOG phase III trials (see Section 11.0). While it was considered important for the placebo control to be maintained for this entire duration in all three arms of the trial, at the same time extending a placebo beyond 15 months was felt to be impractical, unethical and cost-prohibitive. Finally, patients treated on this trial will not be eligible for therapy on clinical trials evaluating consolidation or maintenance therapy while on or off study Role of Secondary Surgery Continued uncertainty exists as to whether second-look surgical procedures contribute to the overall management of patients with ovarian cancer. Some investigators and institutions do not recommend a second-look operation whereas other investigators feel that, in some patients, a second-look procedure may be useful to identify patients with small-volume residual disease who are candidates for additional treatment. The uncertain benefit of second-look surgery has been reflected in current treatment guidelines, where it has been designated as an optional procedure (NCCN). In community practice outside of clinical trials, the frequency of second-look surgery has declined. The nonuniform application of second-look surgery and the ability to document subclinical residual disease has the potential for confounding primary endpoints on this clinical trial, such as determination of the PFS. In the absence of clear evidence that this procedure provides benefit, second-look surgery for patients in clinical complete remission will not be permitted on the current study. Based on results from GOG-0152 demonstrating that interval secondary cytoreductive surgery did not improve progression-free or overall survival in patients with advanced disease who had previously undergone maximal primary cytoreduction, 56 and the potential for increased surgical morbidity from delayed wound healing in patients who undergo major surgical procedures while on treatment with Bevacizumab, interval cytoreductive surgery will not be permitted for patients enrolled on the current trial.

15 -9- GOG CA-125 as a Biologic Marker of Progressive Disease Serum levels of CA-125, a tumor-associated glycoprotein antigen, are elevated in 80% of patients with epithelial ovarian cancer. 57 CA-125 has been monitored, often on a frequent basis, to verify response to therapy, presence of residual disease, and as early evidence of recurrence. However, CA-125 is not entirely tumor specific, and can be elevated in a variety of benign conditions, such as endometriosis, uterine fibroids, and pelvic inflammation; this is particularly true in pre-menopausal women. In addition, levels of CA-125 can be discordant with tumor response, both as false-positive and false-negative trends; the influence of biologic agents on these inaccuracies is unclear. Nonetheless, because imaging modalities such as contrast computed tomography appear to be relatively insensitive in detecting disease progression, it has been standard practice for patients and physicians interpret a progressive rise in CA-125 post-therapy as evidence of recurrent or progressive disease, and will make therapeutic decisions based solely on the CA-125. This has complicated the assessment of PFS in prior randomized trials, as patients will receive new therapy prior to clinical documentation of progressive disease on the basis of physical examination or radiographic findings. The current randomized trial will employ a conservative formula to define progressive disease based on serial elevation of CA (in addition to other standard definitions in the management of solid tumors), but only following completion of initial chemotherapy. Although imperfect, it is preferable to apply uniform criteria that include CA-125 rather than absorb uncharacterized events that would compromise the secondary endpoint of PFS. Progression during the period of cytotoxic chemotherapy will require radiographic or physical confirmation. 2.7 Quality of Life (QoL) This trial will help determine if anti-vegf therapy, when combined with standard chemotherapy, prolongs OS and PFS after suboptimal cytoreductive surgery of epithelial ovarian and peritoneal primary cancer. In addition, this study will determine the optimal schedule of anti-vegf therapy. Patient-reported outcomes may differ when anti-vegf therapy is added to standard paclitaxel and carboplatin chemotherapy. Specifically, the primary objective of measuring QoL in this trial is to determine if the addition of anti-vegf therapy reduces disease related symptoms (improves QoL) more quickly and for more prolonged periods of time than chemotherapy alone. In addition, other objectives of measuring QoL include determining if anti-vegf therapy alters QoL as a result of treatment related toxicity not captured though traditional physicianreported measures. Preliminary data from GOG Protocol 0170-D suggest that Bevacizumab may, among responders, not only reduce tumor volume as measured though traditional disease response monitoring, but may also clear ascites and pleural effusions leading to reduced abdominal bloating and pain thus improving QoL. Indeed, VEGF appears to be

16 -10- GOG-0218 obligatory for ascites formation by increasing vascular permeability. 63 Thus, neutralization of VEGF activity could perhaps dramatically improve QoL after just one or two doses of Bevacizumab by reducing malignant ascites formation. Moreover, since taxanes have anti-angiogenesis activity, it is hypothesized that a combination of Bevacizumab and paclitaxel could be synergistic. 64 Unfortunately, QoL was not closely monitored in GOG Protocol 0170-D and future trials such as the current clinical trial studying Bevacizumab in ovarian cancer require QoL measures to evaluate these important endpoints. 65 Since most women with suboptimal, advanced stage ovarian cancer will succumb to their malignancy and since many regimens have similar efficacy, differences in QoL may help determine the optimal treatment regimen in this setting. In addition, systematic documentation of QoL among those enrolled onto this trial may assist in providing information to future non-trial patients regarding the expected effects of therapy as they make their treatment choices. To date, four completed phase III studies in the upfront treatment of ovarian cancer have implemented QoL outcome measures in their study design, and in every instance QoL was helpful in determining the best regimen. For example, OV.10 established the benefit of paclitaxel in treating ovarian carcinoma, 26,66 the AGO trial established the benefit of carboplatin, 67 and the SCOTROC trial established the role of docetaxel 50 and in all of these studies QoL was an important endpoint. More recently, GOG-0152 was the first prospective trial to study QoL in ovarian cancer performed in the GOG. This study included suboptimal patients similar to the current study and demonstrated the feasibility of obtaining high quality QoL data from this population within this cooperative group. 68 GOG-0152 again illustrated the critical importance of measuring QoL. This study showed that endpoints useful in evaluating optimal therapies in the upfront management of ovarian cancer may be missed if only physician reported endpoints are measured. For example, this study found important difference in neurotoxicity between regimens by measuring QoL and demonstrated that baseline QoL (as measured by the FACT-O) was prognostic of overall survival. Importantly, the recently completed study, GOG-0182, which was the predecessor to the currently proposed study, did not contain a QoL component. If this trial shows no difference in the anti-tumor activity between the six regimens studied in this clinical trial, the opportunity to pick the best regimen will be missed because QoL was not measured. In the current trial, QoL will be assessed using the Trial Outcome Index of the Functional Assessment of Cancer Therapy-Ovary (FACT-O TOI). 69,70 This 26-item summary score captures the FACT-G QOL dimensions of Physical Well-Being (7 items), Functional Well-Being (7 items), and the Ovarian Cancer Subscale (12 item). By combining these three subscales, one is assured of capturing the full range of physical aspects of QOL in advanced ovarian cancer, including pain, fatigue, abdominal symptoms and functional status. By combining questions GP4, O1, and O3, which assess abdominal pain, swelling and cramps respectively, a comprehensive patient reported assessment of disease related abdominal symptoms including ascites can be evaluated. Also, the abdominal pain module piloted in GOG Protocol 0172 will be included.

17 -11- GOG-0218 The timing of the QoL assessments is critical to capture data useful in discriminating subtle differences between regimens. This is complicated by the fact that the acute affects of cytotoxic therapy may cause a decrease in QoL. In order to capture early difference in QoL as a result of anti-angiogenesis therapy with Bevacizumab, assessment time points during this trial will be weighted toward the early part of this study. In addition, since some subjects may only complete a few cycles of therapy, it is important to have early assessment points. Finally, in order to avoid the confounding affects of acute chemotherapy related toxicity, questionnaires will be completed just before (21 days after the last dose) the next cycle of chemotherapy and focus on QoL within the last seven days. Thus, assessments will be made: 1. Prior to cycle 1 (t = 0 weeks) 2. Prior to cycle 4 (after 3 doses of chemotherapy and 2 doses of Bevacizumab/placebo, t= 9 weeks), to assess immediate changes in QoL 3. Prior to cycle 7 (after 6 doses of chemotherapy and 5 doses of Bevacizumab/placebo, t= 18 weeks) to assess intermediate changes in QoL 4. Prior to cycle 13 (t= 36 weeks), 6 months after chemotherapy 5. Prior to cycle 21 (t= 60 weeks), completion of study therapy 6. 6 months after completion of study therapy (t= 84 weeks) 2.8 Translational Research Related to Anti-VEGF Therapy 2.81 Markers of Angiogenesis Angiogenesis is one of the cardinal processes leading to invasion and metastasis of solid tumors. There are more than 19 known angiogenic growth factors and at least 30 known angiogenesis inhibitors in the body, and more than 300 exogenous angiogenesis inhibitors have been discovered to date. Vascular endothelial growth factor (VEGF) and basic fibroblast growth factor (bfgf) are among the most well studied angiogenic growth factors. In tumors, angiogenesis has been studied by quantifying the tumor blood micro-vessel density (MVD) determined immunohistochemically using antibodies to CD31, a protein expressed on the surface of vascular endothelial cells. MVD has been shown to predict the response of gastric adenocarcinomas to taxane-based therapy. 71 In addition to MVD, most angiogenesis studies also evaluate VEGF, which has been shown to promote neovascularization and stimulate endothelial cell survival. 72 VEGF levels were also found to correlate with MVD in endometrial and cervical, but not ovarian cancers. 73,75 In ovarian cancer, higher VEGF levels, but not MVD, were found to significantly correlate with decreased patient survival. 74 Multivariate analysis demonstrated that VEGF was an independent prognostic indicator of overall survival, 74 while the prognostic significance of MVD alone for ovarian cancer was less strong. 74,76

18 -12- GOG-0218 Immunohistochemistry will be utilized to evaluate the expression of CD-31 and VEGF in previously untreated primary or metastatic tumor tissue. Expression of these angiogenic markers will be examined in conventional unstained tumor sections compared with tissue microarrays (TMAs). An analysis will be undertaken to assess the relationship between tumor tissue expression of angiogenic markers and clinical outcome including tumor response, progressionfree survival and overall survival in this patient population in the TMAs created for GOG-0218 if appropriate. If not, conventional unstained tissue sections will be used to examine the relationship between the angiogenic markers and clinical outcome in patients participating in this randomized treatment protocol. The exact choice of angiogenic markers/assays to be evaluated/performed in pretreatment tumor and/or serum specimens will be reevaluated based on evolving data in the field. The GOG has embarked on a programmatic examination of angiogenic markers as predictors of clinical outcome including tumor response and survival in ovarian and cervical cancer. The first part of this plan incorporated exploratory translational research objectives into Phase II (GOG-0170D, GOG-0229B and GOG-0231-B) and Phase III (GOG-0157, GOG-0175 and GOG-0191) treatment protocols, a pilot protocol (GOG-9911) and a GOG Tissue Bank project. The laboratory data for these studies will continue to be analyzed, reported and published during the next few years. The second part of this plan will incorporate research hypotheses generated from these exploratory studies into definitive translational research objectives that can be tested and validated in randomized phase III treatment protocols activated or under development in ovarian cancer and cervical cancer including GOG-0198, GOG-0212, GOG-0213, GOG-0218 and GOG At the appropriate time, the current translational research objectives in GOG-0218 will be amended to incorporate a definitive translational research objective regarding angiogenic markers that can be tested and validated using the specimens submitted for this protocol. A summary of the relevant laboratory data will also be provided to establish the background and rationale for that amendment Genomic Analysis The genomic research component of this protocol will focus on the refinement and validation of genes whose expression predict for survival in patients with advanced stage ovarian cancer. Despite the fact that 80% of advanced ovarian cancers (stages III/IV) respond to primary treatment with surgery and chemotherapy, the disease usually recurs and is ultimately fatal. Though most patients die within two years of diagnosis, a subset of patients, even with clinically and morphologically indistinguishable diseases, develop a more chronic form of ovarian cancer, and may survive five years or more with treatment. It is possible that patients with indolent cancer should be monitored and treated differently from patients with rapidly progressing ovarian cancer. At this point, clinicians do not have the tools to predict the clinical course of the disease at the time of initial diagnosis.

19 -13- GOG-0218 Transcription profiling is a large-scale gene-expression analysis-technology, which has been widely used to identify differentially expressed genes and molecular signatures in many biological processes In the past five years, over 600 manuscripts on expression profiling of cancers using microarray technology have been published, illustrating the recognized utility of this approach in exploring questions of tumor biology and clinical correlates. The principles of class prediction and class discovery as they apply to the molecular classification of human cancers were exemplified by Golub et al., who used oligonucleotide microarrays to monitor gene expression in acute leukemias as a test case. 78 Class prediction identified and validated a subset of informative genes whose expression was highly correlated with previously defined classes. Further, subsequent studies have utilized these approaches to provide proof of the molecular profiling principle as well as to gain novel insights into clinical cancer problems. Using a specialized, lymphoid cell-specific cdna microarray, Alizadeh et al. performed expression profiling of diffuse large B-cell lymphomas and identified two molecularly distinct forms of this malignancy that correlated with overall survival. 77 We hypothesize that transcription profiling will provide us with gene signatures for advanced stage ovarian cancer patients who are likely to develop aggressive recurrent disease and shortened survival, and thus may need to be treated with unconventional regimes. Thus, these new discoveries and technologies provide an ideal opportunity to determine the molecular and biochemical basis for the survival of patients with advanced stage suboptimal epithelial ovarian and peritoneal primary cancers. 2.9 Inclusion of Women and Minorities The Gynecologic Oncology Group and GOG participating institutions will not exclude potential subjects from participating in this or any study solely on the basis of ethnic origin or socioeconomic status. Every attempt will be made to enter all eligible patients into this protocol and therefore address the study objectives in a patient population representative of the entire epithelial ovarian and peritoneal primary cancer population treated by participating institutions.

20 3.0 PATIENT ELIGIBILITY AND EXCLUSIONS 3.1 Eligible Patients -14- GOG Patients with a histologic diagnosis of epithelial ovarian cancer or peritoneal primary carcinoma, FIGO stages III or IV (see Appendix I), with suboptimal ( 1 cm maximal diameter any remaining lesion) residual disease following initial surgery and with appropriate tissue available for histologic evaluation. The minimum surgery required was an abdominal surgery providing tissue for histologic evaluation and establishing and documenting the primary site and stage, as well as a maximal effort at tumor debulking. If additional surgery was performed, it should have been in accordance with appropriate surgery for ovarian or peritoneal carcinoma described in the GOG Surgical Procedures Manual ( However, the surgeon is not required to have performed all of the items contained in this section of the GOG Surgical Procedures Manual Patients with the following histologic epithelial cell types are eligible: Serous adenocarcinoma, endometrioid adenocarcinoma, mucinous adenocarcinoma, undifferentiated carcinoma, clear cell adenocarcinoma, mixed epithelial carcinoma, transitional cell carcinoma, malignant Brenner's Tumor, or adenocarcinoma not otherwise specified (N.O.S.) Patients must have adequate: Bone marrow function: Absolute neutrophil count (ANC) greater than or equal to 1,500/µl, equivalent to Common Toxicity Criteria for Adverse Events v3.0 (CTCAE) Grade Platelets greater than or equal to 100,000/µl. (CTCAE Grade 0-1) Renal function: Creatinine 1.5 x institutional upper limit normal (ULN), CTCAE Grade Hepatic function: Bilirubin less than or equal to 1.5 x ULN (CTCAE Grade 1) SGOT and alkaline phosphatase less than or equal to 2.5 x ULN (CTCAE Grade 1) Neurologic function: Neuropathy (sensory and motor) less than or equal to CTCAE Grade Blood coagulation parameters: PT such that international normalized ratio (INR) is 1.5 (or an in-range INR, usually between 2 and 3, if a patient is on a stable dose of therapeutic warfarin) and a PTT 1.2 times the upper limit of normal.

21 -15- GOG Patients with a GOG Performance Status of 0, 1, or Patients must be entered between 1 and 12 weeks after initial surgery performed for the combined purpose of diagnosis, staging and cytoreduction Patients with measurable (see Section 8.11) and non-measurable (see Section 8.12) disease are eligible. Patients may or may not have cancer-related symptoms Patients who have met the pre-entry requirements specified in Section An approved informed consent and authorization permitting release of personal health information must be signed by the patient or guardian. 3.2 Ineligible Patients 3.21 Patients with a current diagnosis of borderline epithelial ovarian tumor (formerly tumors of low malignant potential ) or recurrent invasive epithelial ovarian cancer treated with surgery only (such as those with stage Ia or Ib low Grade lesions) are not eligible. Patients with a prior diagnosis of a borderline tumor that was surgically resected and who subsequently develop an unrelated, new invasive epithelial ovarian or peritoneal primary cancer are eligible, provided that they have not received prior chemotherapy for any ovarian tumor Patients who have received prior radiotherapy to any portion of the abdominal cavity or pelvis are excluded. Prior radiation for localized cancer of the breast, head and neck, or skin is permitted, provided that it was completed more than three years prior to registration, and the patient remains free of recurrent or metastatic disease Patients who have received prior chemotherapy for any abdominal or pelvic tumor are excluded. Patients may have received prior adjuvant chemotherapy for localized breast cancer, provided that it was completed more than three years prior to registration, and that the patient remains free of recurrent or metastatic disease Patients who have received any targeted therapy (including but not limited to vaccines, antibodies, tyrosine kinase inhibitors) or hormonal therapy for management of their ovarian, peritoneal primary or fallopian tube carcinoma Patients with synchronous primary endometrial cancer, or a past history of primary endometrial cancer, are excluded, unless all of the following conditions are met: Stage not greater than I-B; no more than superficial myometrial invasion, without vascular or lymphatic invasion; no poorly differentiated subtypes, including papillary serous, clear cell or other FIGO Grade 3 lesions.

22 -16- GOG With the exception of non-melanoma skin cancer and other specific malignancies as noted above, patients with other invasive malignancies who had (or have) any evidence of the other cancer present within the last five years or whose previous cancer treatment contraindicates this protocol therapy are excluded Patients with acute hepatitis or active infection that requires parenteral antibiotics Patients with serious, non-healing wound, ulcer, or bone fracture. This includes history of abdominal fistula, gastrointestinal perforation or intra-abdominal abscess within 28 days. Patients with granulating incisions healing by secondary intention with no evidence of fascial dehiscence or infection are eligible but require weekly wound examinations (see Section 7.1) Patients with active bleeding or pathologic conditions that carry high risk of bleeding, such as known bleeding disorder, coagulopathy, or tumor involving major vessels Patients with history or evidence upon physical examination of CNS disease, including primary brain tumor, seizures not controlled with standard medical therapy, any brain metastases, or history of cerebrovascular accident (CVA, stroke), transient ischemic attack (TIA) or subarrachnoid hemorrhage within six months of the first date of treatment on this study Patients with clinically significant cardiovascular disease. This includes: Uncontrolled hypertension, defined as systolic 150 mm Hg or diastolic 90 mm Hg Myocardial infarction or unstable angina 6 months prior to registration New York Heart Association (NYHA) Grade II or greater congestive heart failure (Appendix II) Serious cardiac arrhythmia requiring medication CTCAE Grade 2 or greater peripheral vascular disease (at least brief ( 24 hrs) episodes of ischemia managed non-surgically and without permanent deficit) History of CVA within six months Patients with known hypersensitivity to Chinese hamster ovary cell products or other recombinant human or humanized antibodies.

23 -17- GOG Patients with clinically significant proteinuria. Urine protein should be screened by urine protein-creatinine ratio (UPCR). The UPCR has been found to correlate directly with the amount of protein excreted in a 24 hour urine collection Specifically, a UPCR of 1.0 is equivalent to 1.0 gram of protein in a 24 hour urine collection. Obtain at least 4 ml of a random urine sample in a sterile container (does not have to be a 24 hour urine). Send sample to lab with request for urine protein and creatinine levels [separate requests]. The lab will measure protein concentration (mg/dl) and creatinine concentration (mg/dl) The UPCR is derived as follows: protein concentration (mg/dl)/creatinine (mg/dl). Patients must have a UPCR < 1.0 to allow participation in the study Patients with or with anticipation of invasive procedures as defined below: Major surgical procedure, open biopsy or significant traumatic injury within 28 days prior to the first date of Bevacizumab/placebo therapy (cycle 2) Major surgical procedure anticipated during the course of the study Core biopsy within 7 days prior to the first date of Bevacizumab/placebo therapy (cycle 2) Patients with GOG Performance Grade of 3 or Patients who are pregnant or nursing. To date, no fetal studies in animals or humans have been performed. The possibility of harm to a fetus is likely. Bevacizumab specifically inhibits VEGF, which is responsible for formation of new blood vessels during development, and antibodies can cross the placenta. Therefore, Bevacizumab should not be administered to pregnant women. Subjects will be apprised of the large potential risk to a developing fetus. It is not known whether Bevacizumab is excreted in human milk. Because many drugs are excreted in human milk, Bevacizumab should not be administered to nursing women. Patients of childbearing potential must agree to use contraceptive measures during study therapy and for at least six months after completion of Bevacizumab therapy Patients under the age of Patients who have received prior therapy with any anti-vegf drug, including Bevacizumab.

24 -18- GOG STUDY MODALITIES 4.1 Paclitaxel (NSC #673089) 4.11 Formulation: Paclitaxel is supplied as a 6mg/mL non-aqueous solution in multidose vials containing 30mg/5mL, 100mg/16.7mL, or 300mg/50mL of paclitaxel. In addition to 6mg of paclitaxel, each ml of sterile non-pyrogenic solution contains 527mg of purified Cremophor EL (polyoxyethylated castor oil) and 49.7% (v/v) dehydrated alcohol, USP Storage: Unopened vials of paclitaxel are stable to the date indicated on the package when stored between 20 to 25 C (68 to 77 F). Protect from light Preparation: Paclitaxel must be diluted prior to infusion. Paclitaxel should be diluted in 0.9% Sodium Chloride for Injection, USP; 5% Dextrose Injection, USP; 5% Dextrose and 0.9% Sodium Chloride Injection, USP; or 5% Dextrose in Ringer s Injection to a final concentration of 0.3 to 1.2mg/mL. The solutions are physically and chemically stable for up to 27 hours at ambient temperature (approximately 25 C / 77 F) and room lighting conditions. NOTE: In order to minimize patient exposure to the plasticizer DEHP, which may be leached from PVC infusion bags or sets, diluted paclitaxel solutions should be stored in bottles (glass, polypropylene) or plastic (polypropylene, polyolefin) bags and administered through polyethylene-lined administration sets. Paclitaxel should be administered through an inline filter with a microporous membrane not greater than 0.22 microns. Use of filter devices such as IVEX- 2 or IVEX-HP, which incorporate short inlet and outlet PVC-coated tubing, has not resulted in significant leaching of DEHP. All patients should be premedicated with corticosteroids, diphenhydramine, and H 2 antagonists prior to paclitaxel administration in order to prevent severe hypersensitivity reactions. Patients who experience severe hypersensitivity reactions to paclitaxel should not be rechallenged with the drug Adverse Effects: Consult the package insert for the most current and complete information Supplier: Commercially available both from Bristol-Myers Squibb Oncology as well as generic manufacturers. Consult the American Hospital Formulary Service Drug Information guide, Facts and Comparisons, or the package insert for additional information. 4.2 Carboplatin (Paraplatin - NSC #241240) 4.21 Formulation: Carboplatin is supplied as a sterile, pyrogen-free, 10mg/mL aqueous solution in multi-dose vials containing 50mg/5mL, 150mg/15mL, 450mg/45mL, or 600g/60mL of carboplatin.

25 -19- GOG Storage: Unopened vials of carboplatin are stable to the date indicated on the package when stored at 25 C (77 F). Excursions from 15 to 30 C (59 to 86 F) are permitted. Protect from light. Carboplatin multidose vials maintain microbial, chemical, and physical stability for up to 14 days at 25 C following multiple needle entries Preparation: Carboplatin aqueous solution can be further diluted to concentrations as low as 0.5mg/mL with 5% Dextrose in Water or 0.9% Sodium Chloride for Injection, USP. When prepared as directed, carboplatin aqueous solutions are stable for 8 hours at room temperature (25 C / 77 F). Since no antibacterial preservative is contained in the formulation, it is recommended that carboplatin solutions be discarded 8 hours after dilution. Calvert Formula for Carboplatin (AUC) Dosing total dose (mg) = target AUC (in mg/ml/minute) * [GFR (in ml/minute) + 25] NOTE: Aluminum reacts with carboplatin causing precipitate formation and loss of potency; therefore, needles or intravenous sets containing aluminum parts that may come in contact with the drug must NOT be used for the preparation or administration of carboplatin Adverse Effects: Consult the package insert for the most current and complete information Supplier: Commercially available both from Bristol-Myers Squibb Oncology as well as generic manufacturers. Consult the American Hospital Formulary Service Drug Information guide, Facts and Comparisons, or the package insert for additional information. 4.3 Bevacizumab (NSC #704865, IND #7921) or Placebo All investigators who receive a copy of the protocol should also obtain a copy of the Investigator s Brochure (IB). IB s are available from the Pharmaceutical Management Branch, CTEP, DCTD, NCI and may be obtained by ing the IB Coordinator (ibcoordinator@mail.nih.gov) or by calling the IB Coordinator at Description: Bevacizumab is a recombinant humanized anti-vegf monoclonal antibody, consisting of 93% human and 7% murine amino acid sequences. The agent is composed of human IgG framework and murine antigen-binding complementarity-determining regions. Bevacizumab blocks the binding of vascular endothelial growth factor (VEGF) to its receptors resulting in inhibition of angiogenesis How Supplied: Bevacizumab and Placebo are supplied as a clear to slightly opalescent, sterile liquid ready for parenteral administration. For Bevacizumab, each 100mg (25mg/mL 4mL fill) glass vial contains bevacizumab with phosphate, trehalose, polysorbate 20, and Sterile Water for Injection, USP. For Placebo, each 0mg (0mg/mL 4mL fill) glass vial contains phosphate, trehalose, polysorbate 20, and Sterile Water for Injection, USP.

26 -20- GOG-0218 Bevacizumab being provided in this study is for investigational use only and is not the commercially marketed product. The commercially available bevacizumab is marketed under the name Avastin. Investigational bevacizumab and commercially available Avastin are produced at separate facilities. Although investigational bevacizumab and Avastin are expected to be very similar it is possible that some differences may exist. For further details and molecule characterization, see the updated bevacizumab Investigator Brochure Storage and Stability: Bevacizumab is shipped on blue ice for next day delivery. On receipt, bevacizumab should be stored in the refrigerator (2 o to 8 o C) and should remain refrigerated until just prior to use. Do not freeze. Do not shake. Shelf-life studies of bevacizumab are continuing. Investigators will be notified when lots have expired. The sterile single use vials contain no antibacterial preservatives; therefore, vials should be discarded eight hours after initial entry Preparation: Vials contain no preservative and are intended for single use only. The calculated dose should be placed in a sterile, empty IV bag and diluted with a sufficient amount of 0.9% Sodium Chloride for Injection to obtain a final volume of 100 ml. Once diluted in 0.9% Sodium Chloride for Injection, the bevacizumab solution must be administered within 8 hours Administration: Bevacizumab is administered intravenously as a continuous infusion. The initial dose should be administered over a minimum of 90 minutes. If no adverse reactions occur after the initial dose, the second dose should be administered over a minimum of 60 minutes. If no adverse reactions occur after the second dose, all subsequent doses should be administered over a minimum of 30 minutes. If infusion-related adverse reactions occur, all subsequent infusions should be administered over the shortest period that was well tolerated Clinical Supplies: Bevacizumab (NSC ) and matching placebo will be provided free of charge by Genentech and distributed by the Pharmaceutical Management Branch (PMB), Cancer Therapy Evaluation Program (CTEP), Division of Cancer Treatment and Diagnosis (DCTD), National Cancer Institute (NCI). Bevacizumab and matching placebo will be supplied in 4 ml fill glass vials each containing 100 mg (Bevacizumab) or 0 mg (Placebo for Bevacizumab) of bevacizumab. The blinded, patient-specific vials will be sealed in a cardboard box with a tamper-evident seal. For Phase A (cycles 2 through 6), each box will be labeled with the protocol number (i.e., GOG-0218 ) the box number (i.e., Box 1 of 2 and Box 2 of 2 ) the number of vials (e.g., 48 vials )

27 -21- GOG-0218 the patient ID number (e.g., "; where 999 indicates the GOG assigned institution code for the registering site, 0218 indicates the protocol number, and 001 indicates the patient sequence number for the registering site) the patient initials (i.e., first initial, middle initial, last initial [e.g., FML ]) the agent identification (i.e., Phase A - Bevacizumab 100 mg or Placebo ) a blank line for the pharmacist to enter the patient s name storage instructions (i.e., Store in refrigerator [2 8 C]. Do not freeze. Do not shake. ) emergency contact instructions a Julian date At the completion of chemotherapy cycle 6, ALL remaining PHASE A clinical supplies of bevacizumab / placebo should be returned to PMB (see Drug Returns below). For Phase B (cycles 7 through 22), each box will be labeled with the protocol number (i.e., GOG-0218 ) the box number (i.e., Box 1 of 2 and Box 2 of 2 ) the number of vials (e.g., 48 vials ) the patient ID number (e.g., "; where 999 indicates the GOG assigned institution code for the registering site, 0218 indicates the protocol number, and 001 indicates the patient sequence number for the registering site) the patient initials (i.e., first initial, middle initial, last initial [e.g., FML ]) the agent identification (i.e., Phase B - Bevacizumab 100 mg or Placebo ) a blank line for the pharmacist to enter the patient s name storage instructions (i.e., Store in refrigerator [2 8 C]. Do not freeze. Do not shake. ) emergency contact instructions a Julian date The Julian date indicates the day the box was labeled and shipped and is composed of the last two digits of the calendar year (e.g., 2005 = 05, 2006 = 06) and a day count (e.g., January 1 = 001, December 31 = 365). For example, a box labeled and shipped on January 1, 2005 would have a Julian date of and a box labeled and shipped on December 31, 2006 would have a Julian date of The Julian date will be used by PMB for recalls. When a lot expires, PMB will determine the last date the expired lot was shipped and will recall all vials (i.e., both Bevacizumab and Placebo) shipped on or before that date thus eliminating any chance of breaking the blind. Questions about drug orders, transfers, returns, or accountability should be addressed to the PMB by calling (301) Monday through Friday between 8:30am and 4:30pm Eastern Time.

28 -22- GOG Drug Ordering: Phase A (beginning with cycle 2 and continuing through cycle 6 [during chemotherapy]) No blinded starter supplies will be available for this phase. Blinded, patientspecific supplies for Phase A will be sent to the registering investigator at the time of randomization. This randomization will be performed by the GOG Statistical and Data Center in Buffalo, NY. The assigned patient ID number must be recorded by the registering institution for proper vial dispersion. Once a patient has been registered with the GOG Statistical and Data Center, the GOG Statistical and Data Center will electronically transmit a clinical drug request for that patient to the PMB. This request will be entered and transmitted by the GOG Statistical and Data Center the day the patient is registered and will be processed by the PMB the next business day and shipped the following business day. All shipments will be sent on blue ice by FedEx (generally one to two day delivery). Thus, if a patient is registered on Monday, GOG would enter a clinical drug request for that patient on Monday and PMB would process the request on Tuesday and ship the drug on Wednesday. Both United States and Canadian sites could expect to receive their order either Thursday or Friday. Note that PMB will only send blue ice shipments on Monday through Thursday for delivery on Tuesday through Friday. Thus, if a patient is registered on Wednesday, the order will be processed on Thursday and shipped the following Monday for delivery on Tuesday or Wednesday. The initial request will be for 48 vials (minimum of 2 courses / 2 cycles / 6 week supply). Three weeks after the initial electronic request (i.e., three weeks before needed), sites may reorder an additional 3 courses / 3 cycles / 9 week supply to complete Phase A by completing an NCI Clinical Drug Request form and faxing it to the PMB at The NCI Clinical Drug Request form is available on the CTEP home page ( or by calling the PMB at The assigned patient ID number (e.g., " "), the patient initials (e.g., "FML"), the number of vials remaining from the initial shipment, and the patient s weight (in KG) should be entered on each order. All drug orders will be shipped directly to the physician registering the patient. Phase B (beginning with cycle 7 [first cycle post chemotherapy] and continuing through cycle 22 [15 months total treatment duration]) In order to obtain Phase B clinical supplies, patients must be re-registered with the GOG Statistical and Data Center (see section 5.1) at the completion of chemotherapy (cycle 6). No blinded starter supplies will be available for this phase. Blinded, patient-specific supplies for Phase B will be sent to the registering investigator at the time of re-registration. This re-registration will be performed by the GOG Statistical and Data Center in Buffalo, NY. The patient ID number will NOT change. Once a patient has been re-registered with the GOG Statistical and Data Center, the GOG Statistical and Data Center will electronically transmit a clinical drug request for that patient to the PMB. This request will be entered and transmitted by the GOG Statistical and Data Center the day the patient is re-registered and will be processed by the PMB the next business day and shipped the following business day. All shipments will be sent

29 Patient Randomized with GOG -23- GOG-0218 on blue ice by FedEx (generally one to two day delivery). Thus, if a patient is re-registered on Monday, GOG would enter a clinical drug request for that patient on Monday and PMB would process the request on Tuesday and ship the drug on Wednesday. Both United States and Canadian sites could expect to receive their order either Thursday or Friday. Note that PMB will only send blue ice shipments on Monday through Thursday for delivery on Tuesday through Friday. Thus, if a patient is registered on Wednesday, the order will be processed on Thursday and shipped the following Monday for delivery on Tuesday or Wednesday. The initial request will be for 48 vials (minimum of 2 courses / 2 cycles / 6 week supply). Three weeks after the initial electronic request (i.e., three weeks before needed), sites may reorder an additional 2 courses / 2 cycles / 6 week supply by completing an NCI Clinical Drug Request form and faxing it to the PMB at The NCI Clinical Drug Request form is available on the CTEP home page ( or by calling the PMB at The assigned patient ID number (e.g., " "), the patient initials (e.g., "FML"), the number of vials remaining from the prior shipment, and the patient s weight (in KG) should be entered on each order. All drug orders will be shipped directly to the physician reregistering the patient. GOG-0218 Shipment Schedule Initial e-order Transmitted by GOG Initial e-order Received and Approved by PMB Initial Order Shipped By PMB Initial Order Received at Site * Monday Monday Tuesday Wednesday Thursday Tuesday Tuesday Wednesday Thursday Friday Wednesday Wednesday Thursday Monday Tuesday Thursday Thursday Friday Monday Tuesday Friday Friday Monday Tuesday Wednesday * arrival time approximate / shipments sent by Federal Express 4.38 Drug Transfers: Vials MAY NOT be transferred from one patient to another patient or from one protocol to another protocol. All other transfers (e.g., a patient moves from one participating clinical site to another participating clinical site, the principal investigator at a given clinical site changes) must be approved in advance by the PMB. To obtain an approval for transfer, investigators should complete and submit to the PMB (fax number ) a Transfer Investigational Agent Form available on the CTEP home page ( or by calling the PMB at The patient ID number (e.g., " ) and the patient initials (e.g., "FML") should be entered in the "Received on NCI Protocol No." and the "Transferred to NCI Protocol No." fields in addition to the protocol number (i.e., GOG-0218") Drug Returns: Only unreconstituted drug supplies should be returned to the PMB. When it is necessary to return study drug (e.g., sealed Phase A vials remaining when a patient completes cycle 6, sealed Phase A or Phase B vials remaining when a patient permanently discontinues protocol treatment, expired vials recalled by the PMB), investigators should return the study drug to the

30 -24- GOG-0218 PMB using the NCI Return Drug List available on the CTEP home page ( or by calling the PMB at The patient ID number (e.g., " ") and the patient initials (e.g., "FML") should be entered in the "Lot Number" field. A separate line item is required for EACH patient ID number returned Drug Accountability: The investigator, or a responsible party designated by the investigator, must maintain a careful record of the receipt, disposition, and return of all drugs received from the PMB using the NCI Investigational Agent Accountability Record available on the CTEP home page ( or by calling the PMB at A separate NCI Investigational Agent Accountability Record must be maintained for each patient ID number (e.g., " ") on this protocol Emergency Unblinding: In the event of an emergency during normal business hours (Monday through Friday 9:00 am to 5:00 pm Eastern Time), contact the GOG Statistical and Data Center by phone at At all other times, dial pager: Following the tone indicate a telephone number for a return phone call. A staff member from the GOG Statistical and Data Center will return your call. Remember, this is only in the event of an emergency! This procedure is to be used by the physician when the physician needs to know whether the patient is taking Bevacizumab or a placebo to manage the acute illness. Patients should be instructed that if they have any questions or symptoms they should contact the treating physician s office. The GOG Statistical and Data Center will require the protocol number (i.e., GOG-0218 ), the patient ID number (e.g., ), and the patient initials (e.g., FML ) to unblind the patient Adverse Effects: General Information on Adverse Effects of Bevacizumab The most serious adverse effects associated with Bevacizumab have been gastrointestinal perforations, wound healing complications, hemorrhage, hypertensive crises, arterial thrombotic complications, nephrotic syndrome and congestive heart failure (see specifics below). The most common severe (NCI-CTC Grade 3-4) adverse events among 1032 patients receiving Bevacizumab in Genentech-sponsored studies were asthenia, pain, hypertension, diarrhea, and leukopenia. The most common adverse events of any severity among the 742 patients receiving Bevacizumab in Genentech-sponsored studies were asthenia, pain, abdominal pain, headache, hypertension, diarrhea, nausea, vomiting, anorexia, stomatitis, constipation, upper respiratory infection,

31 -25- GOG-0218 epistaxis, dyspnea, exfoliative dermatitis, and proteinuria. Because clinical trials are conducted under widely varying conditions, adverse reaction rates observed in the clinical trials of a drug cannot be directly compared to rates in the clinical trials of another drug and may not reflect the rates observed in practice. The adverse reaction information from clinical trials does, however, provide a basis for identifying the adverse events that appear to be related to drug use and for approximating rates. A total of 1032 patients with metastatic colorectal cancer (n = 568) and with other cancers (n = 464) received Bevacizumab either as a single agent (n = 157) or in combination with chemotherapy (n = 875) in Genentech-sponsored clinical trials. All adverse events were collected in 742 of the 1032 patients; for the remaining 290, all NCI-CTC Grade 3 and 4 adverse events and only selected Grade 1 and 2 adverse events (hypertension, proteinuria, thromboembolic events) were collected. Adverse events across all Genentech-sponsored studies were used to further characterize specific adverse events. Comparative data on adverse experiences, except where indicated, are limited to Study AVF2107, a randomized, active-controlled study in 897 patients receiving initial treatment for metastatic colorectal cancer. 22 All NCI-CTC Grade 3 and 4 adverse events and selected Grade 1 and 2 adverse events (hypertension, proteinuria, thromboembolic events) were reported for the overall study population. In Study AVF2107, the median age was 60, 60% were male, 78% had colon primary lesion, and 29% had prior adjuvant or neoadjuvant chemotherapy. The median duration of exposure to Bevacizumab in Study AVF2107 was 8 months in Arm 2 and 7 months in Arm 3. All adverse events, including all NCI-CTC Grade 1 and 2 events were reported in a subset of 309 patients. The baseline entry characteristics in the 309 patient safety subset were similar to the overall study population and well-balanced across the three study arms. Severe and life-threatening (NCI-CTC Grade 3 and 4) adverse events, which occurred at a higher incidence ( 2%) in patients receiving bolus- IFL plus Bevacizumab as compared to bolus-ifl plus placebo, are presented in the Table 4A.

32 -26- GOG-0218 Table 4A NCI-CTC Grade 3 and 4 Adverse Events in Study1 (Occurring at Higher Incidence (> 2%) in Bevacizumab vs. Control) Arm 1 IFL+Placebo (n = 396) Grade 3 4 Events 295 (74%) 340 (87%) Body as a Whole Asthenia 28 (7%) 38 (10%) Abdominal Pain 20 (5%) 32 (8%) Pain 21 (5%) 30 (8%) Cardiovascular Deep Vein Thrombosis 19 (5%) 34 (9%) Hypertension 10 (2%) 46 (12%) Intra-Arterial Thrombosis 5 (1%) 13 (3%) Syncope 4 (1%) 11 (3%) Digestive Diarrhea 99 (25%) 133 (34%) Constipation 9 (2%) 14 (4%) Hemic/Lymphatic Leukopenia 122 (31%) 145 (37%) Neutropenia a 41 (14%) 58 (21%) Arm 2 IFL+Bevacizumab (n = 392) a Central laboratories were collected on Days 1 and 21 of each cycle. Neutrophil counts are available in 303 patients in Arm 1 and 276 in Arm 2. Adverse events of any severity, which occurred at a higher incidence ( 5%) in the initial phase of the study in patients receiving Bevacizumab (bolus-ifl plus Bevacizumab or 5-FU/LV plus Bevacizumab) as compared to the bolus-ifl plus placebo arm, are presented in Table 4B.

33 -27- GOG-0218 Body as a Whole Table 4B NCI-CTC Grade 1 4 Adverse Events in Study AVF2107 Subset (Occurring at Higher Incidence (> 5%) in Bevacizumab vs. Control) Arm 1 IFL + Placebo (n = 98) Arm 2 IFL + Bevacizumab (n = 102) Asthenia 68 (70%) 75 (74%) 80 (73%) Pain 54 (55%) 62 (61%) 67 (62%) Abdominal Pain 54 (55%) 62 (61%) 55 (50%) Headache 19 (19%) 27 (26%) 30 (26%) Cardiovascular Hypertension 14 (14%) 23 (23%) 37 (34%) Hypotension 7 (7%) 15 (15%) 8 (7%) Deep Vein Thrombosis 3 (3%) 9 (9%) 6 (6%) Digestive Vomiting 46 (47%) 53 (52%) 51 (47%) Anorexia 29 (30%) 44 (43%) 38 (35%) Constipation 28 (29%) 41 (40%) 32 (29%) Stomatitis 18 (18%) 33 (32%) 33 (30%) Dyspepsia 15 (15%) 25 (24%) 19 (17%) Weight Loss 10 (10%) 15 (15%) 18 (16%) Flatulence 10 (10%) 11 (11%) 21 (19%) GI Hemorrhage 6 (6%) 25 (24%) 21 (19%) Dry Mouth 2 (2%) 7 (7%) 4 (4%) Colitis 1 (1%) 6 (6%) 1 (1%) Hemic/Lymphatic Thrombocytopenia 0 5 (5%) 5 (5%) Metabolic/Nutrition Hypokalemia 11 (11%) 12 (12%) 18 (16%) Bilirubinemia 0 1 (1%) 7 (6%) Musculoskeletal Myalgia 7 (7%) 8 (8%) 16 (15%) Arm 3 5-FU/LV + Bevacizumab (n = 109)

34 -28- GOG-0218 Nervous Dizziness 20 (20%) 27 (26%) 21 (19%) Confusion 1 (1%) 1 (1%) 6 (6%) Abnormal Gait 0 1 (1%) 5 (5%) Respiratory Upper Respiratory Infection 38 (39%) 48 (47%) 44 (40%) Dyspnea 15 (15%) 26 (26%) 27 (25%) Epistaxis 10 (10%) 36 (35%) 35 (32%) Voice Alteration 2 (2%) 9 (9%) 6 (6%) Skin/Appendages Alopecia 25 (26%) 33 (32%) 6 (6%) Dry Skin 7 (7%) 7 (7%) 22 (20%) Exfoliative Dermatitis 3 (3%) 3 (3%) 21 (19%) Nail Disorder 3 (3%) 2 (2%) 9 (8%) Skin Discoloration 3 (3%) 2 (2%) 17 (16%) Skin Ulcer 1 (1%) 6 (6%) 7 (6%) Special Senses Taste Disorder 9 (9%) 14 (14%) 23 (21%) Excess Lacrimation 2 (2%) 6 (6%) 20 (18%) Urogenital Proteinuria 24 (24%) 37 (36%) 39 (36%) Urinary Frequency/Urgency 1 (1%) 3 (3%) 6 (6%) Specific Severe Adverse Effects of Bevacizumab Mucocutaneous Hemorrhage Two distinct patterns of bleeding have occurred in patients receiving Bevacizumab. The first is minor hemorrhage, most commonly Grade 1 epistaxis. The second is serious, and in some cases fatal, hemorrhagic events. Serious hemorrhagic events occurred primarily in patients with non-small cell lung cancer, an indication for which Bevacizumab is not approved. In a randomized study in patients with non-small cell lung cancer

35 -29- GOG-0218 receiving chemotherapy with or without Bevacizumab, four of 13 (31%) Bevacizumab -treated patients with squamous cell histology and two of 53 (4%) Bevacizumab -treated patients with non-squamous histology experienced life-threatening or fatal pulmonary hemorrhage as compared to none of the 32 (0%) patients receiving chemotherapy alone. Of the patients experiencing events of life-threatening pulmonary hemorrhage, many had cavitation and/or necrosis of the tumor, either preexisting or developing during Bevacizumab therapy. These serious hemorrhagic events occurred suddenly and presented as major or massive hemoptysis. In Study AVF2107, both serious and non-serious hemorrhagic events occurred at a higher incidence in patients receiving Bevacizumab. In the 309 patients in which Grade 1-4 events were collected, epistaxis was common and reported in 35% of patients receiving bolus-ifl plus Bevacizumab compared with 10% of patients receiving bolus-ifl plus placebo. These events were generally mild in severity (NCI-CTC Grade 1) and resolved without medical intervention. Other mild to moderate hemorrhagic events reported more frequently in patients receiving bolus-ifl plus Bevacizumab when compared to those receiving bolus-ifl plus placebo included gastrointestinal hemorrhage (24% vs. 6%), minor gum bleeding (2% vs. 0), and vaginal hemorrhage (4% vs. 2%). The risk of central nervous system (CNS) bleeding in patients with CNS metastases receiving Bevacizumab has not been evaluated because these patients were excluded from Genentechsponsored studies following development of CNS hemorrhage in a patient with a CNS metastasis in Phase 1 studies. However, CNS bleeding events reported in patients receiving Bevacizumab without evidence of CNS metastases were uncommon and included subarachnoid hemorrhage and hemorrhagic stroke. Patients with serious hemorrhage i.e., requiring medical intervention, should have Bevacizumab treatment discontinued and receive aggressive medical management. Patients with recent hemoptysis should not receive Bevacizumab Thromboembolism In Study AVF2107, 18% of patients receiving bolus-ifl plus Bevacizumab and 15% of patients receiving bolus-ifl plus placebo experienced a Grade 3-4 thromboembolic event. The

36 -30- GOG-0218 incidence of the following Grade 3 and 4 thromboembolic events were higher in patients receiving bolus-ifl plus Bevacizumab as compared to patients receiving bolus-ifl plus placebo: cerebrovascular events (4 vs. 0 patients), myocardial infarction (6 vs. 3), deep venous thrombosis (34 vs. 19), and intra-arterial thrombosis (13 vs. 5). In contrast, the incidence of pulmonary embolism was higher in patients receiving bolus-ifl plus placebo (16 vs. 20). In Study AVF2107, 53 of 392 (14%) patients who received bolus-ifl plus Bevacizumab and 30 of 396 (8%) patients who received bolus-ifl plus placebo had a thromboembolic event and received full-dose warfarin. Two patients in each treatment arm (four total) developed bleeding complications. In the two patients treated with full-dose warfarin and Bevacizumab, these events were associated with marked elevations in their INR. Eleven of 53 (21%) patients receiving bolus-ifl plus Bevacizumab and one of 30 (3%) patients receiving bolus-ifl developed an additional thromboembolic event. In the AVF2107 study, there was a 1% incidence of arterial thromboembolic events (which include myocardial infarction, transient ischemia attack, cerebrovascular accident/stroke and angina/unstable angina) in the IFL + placebo arm versus 3% in the IFL + Bevacizumab arm. A pooled analysis of the rate of arterial TE events from 5 randomized studies showed that treatment with Bevacizumab increased the risk of these events two- to three-fold (up to 5%). Furthermore, certain baseline characteristics, specifically age 65 years and prior arterial TE event, conferred additional risk Gastrointestinal Perforations/Wound Healing Complications Gastrointestinal perforation and wound dehiscence, complicated by intra-abdominal abscesses, occurred at an increased incidence in patients receiving Bevacizumab as compared to controls. Bevacizumab has also been shown to impair wound healing in pre-clinical animal models. In Study AVF2107, one of 396 (0.3%) patients receiving bolus- IFL plus placebo, six of 392 (2%) patients receiving bolus-ifl plus Bevacizumab, and four of 109 (4%) patients receiving 5- FU/LV plus Bevacizumab developed gastrointestinal perforation, in some instances with fatal outcome. These episodes occurred with or without intra-abdominal abscesses and at various time points during treatment. The typical presentation was reported as abdominal pain associated with symptoms such as constipation and vomiting.

37 -31- GOG-0218 In addition, two of 396 (0.5%) patients receiving bolus-ifl plus placebo, four of 392 (1%) patients receiving bolus-ifl plus Bevacizumab, and one of 109 (1%) patients receiving 5-FU/LV plus Bevacizumab developed a wound dehiscence during study treatment. The appropriate interval between surgery and subsequent initiation of Bevacizumab required to avoid the risks of impaired wound healing has not been determined. In Study AVF2107, the clinical protocol did not permit initiation of Bevacizumab for at least 28 days following surgery. There was one patient (among 501 patients receiving Bevacizumab on Study AVF2107) in whom an anastomotic dehiscence occurred when Bevacizumab was initiated per protocol. In this patient, the interval between surgery and initiation of Bevacizumab was greater than two months. Similarly, the appropriate interval between termination of Bevacizumab and subsequent elective surgery required to avoid the risks of impaired wound healing has not been determined. In Study AVF2107, 39 patients who were receiving bolus-ifl plus Bevacizumab underwent surgery following Bevacizumab therapy and, of these patients, six (15%) had wound healing/bleeding complications. In the same study, 25 patients in the bolus-ifl arm underwent surgery and, of these patients, one of 25 (4%) had wound healing/bleeding complications. The longest interval between last dose of study drug and dehiscence was 56 days; this occurred in a patient on the bolus-ifl plus Bevacizumab arm. The interval between termination of Bevacizumab and subsequent elective surgery should take into consideration the calculated half-life of Bevacizumab (approximately 20 days). Bevacizumab therapy should not be initiated for at least 28 days following major surgery. The surgical incision should be fully healed prior to initiation of Bevacizumab. Because of the potential for impaired wound healing, Bevacizumab should be suspended prior to elective surgery. The appropriate interval between the last dose of Bevacizumab and elective surgery is unknown; however, the half-life of Bevacizumab is estimated to be 20 days and the interval chosen should take into consideration the half-life of the drug. Bevacizumab therapy should be discontinued in patients with gastrointestinal perforation or wound dehiscence requiring medical intervention.

38 Hypertension a (>150/100 mmhg) Severe Hypertension a (>200/110 mmhg) Hypertension -32- GOG-0218 The incidence of hypertension and severe hypertension was increased in patients receiving Bevacizumab in Study AVF2107 (see Table 4C). Table 4C Incidence of Hypertension and Severe Hypertension in Study AVF2107 Arm 1 IFL + Placebo (n = 394) Arm 2 IFL + Bevacizumab (n = 392) Arm 3 5-FU/LV + Bevacizumab (n = 109) 43% 60% 67% 2% 7% 10% a This includes patients with either a systolic or diastolic reading greater than the cutoff value on one or more occasions. Among patients with severe hypertension in the Bevacizumab arms, slightly over half the patients (51%) had a diastolic reading greater than 110 associated with a systolic reading less than 200. Hypertensive crisis with end organ damage has rarely been reported. Medication classes used for management of patients with Grade 3 hypertension receiving Bevacizumab included angiotensinconverting enzyme inhibitors, beta-blockers, diuretics, and calcium channel blockers. Four months after discontinuation of therapy, persistent hypertension was present in 18 of 26 patients that received bolus-ifl plus Bevacizumab and 8 of 10 patients that received bolus-ifl plus placebo. Across all clinical studies (n = 1032), development or worsening of hypertension resulted in hospitalization or discontinuation of Bevacizumab in 17 patients. Four of these 17 patients developed hypertensive encephalopathy. Severe hypertension was complicated by subarachnoid hemorrhage in one patient.

39 -33- GOG-0218 Bevacizumab should be permanently discontinued in patients with hypertensive crisis. Temporary suspension is recommended in patients with severe hypertension that is not controlled with medical management Proteinuria In Study AVF2107, both the incidence and severity of proteinuria (defined as a urine dipstick reading of 1+ or greater) was increased in patients receiving Bevacizumab as compared to those receiving bolus-ifl plus placebo. Urinary dipstick readings of 2+ or greater occurred in 14% of patients receiving bolus-ifl plus placebo, 17% receiving bolus-ifl plus Bevacizumab, and in 28% of patients receiving 5-FU/LV plus Bevacizumab. 24-hour urine collections were obtained in patients with new onset or worsening proteinuria. None of the 118 patients receiving bolus- IFL plus placebo, three of 158 patients (2%) receiving bolus-ifl plus Bevacizumab, and two of 50 (4%) patients receiving 5- FU/LV plus Bevacizumab who had a 24-hour collection experienced NCI-CTC Grade 3 proteinuria ( 3.5 gm protein/24 hours). In a dose-ranging, placebo-controlled, randomized study of Bevacizumab in patients with metastatic renal cell carcinoma, an indication for which Bevacizumab is not approved, 24-hour urine collections were obtained in approximately half the patients enrolled. Among patients in whom 24-hour urine collections were obtained, four of 19 (21%) patients receiving Bevacizumab at 10 mg/kg every two weeks, two of 14 (14%) receiving Bevacizumab at 3 mg/kg every two weeks, and none of the 15 placebo patients experienced NCI-CTC Grade 3 proteinuria ( 3.5 gm protein/24 hours). Nephrotic syndrome occurred in five of 1032 (0.5%) patients receiving Bevacizumab in Genentech-sponsored studies. One patient died and one required dialysis. In three patients, proteinuria decreased in severity several months after discontinuation of Bevacizumab. No patient had normalization of urinary protein levels (by 24-hour urine) following discontinuation of Bevacizumab.

40 -34- GOG Congestive Heart Failure Congestive heart failure (CHF), defined as NCI-CTC Grade 2-4 left ventricular dysfunction, was reported in 22 of 1032 (2%) patients receiving Bevacizumab in Genentech-sponsored studies. Congestive heart failure occurred in six of 44 (14%) patients receiving Bevacizumab and concurrent anthracyclines. Congestive heart failure occurred in 13 of 299 (4%) patients who received prior anthracyclines and/or left chest wall irradiation. In a controlled study, the incidence was higher in patients receiving Bevacizumab plus chemotherapy as compared to patients receiving chemotherapy alone. The safety of continuation or resumption of Bevacizumab in patients with cardiac dysfunction has not been studied Infusion Hypersensivity Reactions Infusion reactions with the first dose of Bevacizumab were uncommon ( 3%). Severe reactions during the infusion of Bevacizumab occurred in two patients. One patient developed stridor and wheezing during their first dose. A second patient, receiving paclitaxel followed by Bevacizumab, developed a Grade 3 hypersensitivity reaction requiring hospitalization during their third infusion of Bevacizumab. Both patients responded to medical management. Information on re-challenge is not available. Bevacizumab infusion should be interrupted in all patients with severe infusion reactions and appropriate medical therapy administered. There are no data regarding the most appropriate method of identification of patients who may safely be retreated with Bevacizumab after experiencing a severe infusion reaction.

41 -35- GOG Other Serious Adverse Events The following other serious adverse events are considered unusual in cancer patients receiving cytotoxic chemotherapy and occurred in at least one subject treated with Bevacizumab in clinical studies. Body as a Whole: polyserositis Digestive: intestinal obstruction, intestinal necrosis, mesenteric venous occlusion, anastomotic ulceration Hemic and lymphatic: pancytopenia Metabolic and nutritional disorders: hyponatremia. Urogenital: ureteral stricture 4.5 Docetaxel (Taxotere RP-56976, NSC #628503) 4.51 Formulation: Docetaxel is supplied as a sterile, non-pyrogenic, non-aqueous viscous solution in single dose vials containing 20mg/0.5mL or 80mg/2mL of docetaxel. Each ml contains 40mg docetaxel (anhydrous) and 1040mg polysorbate Docetaxel requires dilution prior to use. A sterile, non-pyrogenic, single dose diluent is supplied for this purpose. The diluent for docetaxel contains 13% (w/w) ethanol in water for injection and is supplied in vials Storage: Unopened vials of docetaxel are stable to the date indicated on the package when stored between 2 and 25 C (36 and 77 F). Protect from light Preparation: Docetaxel must be combined with its supplied diluent (final concentration = 10mg/mL) and then further diluted prior to infusion. Docetaxel should be diluted in 0.9% Sodium Chloride for Injection, USP or 5% Dextrose Injection, USP to produce a final concentration of 0.3 to 0.74mg/mL. The fully prepared docetaxel infusion solution should be used within 4 hours (including the infusion duration). NOTE: In order to minimize patient exposure to the plasticizer DEHP, which may be leached from PVC infusion bags or sets, the final docetaxel dilution for infusion should be stored in bottles (glass, polypropylene) or plastic (polypropylene, polyolefin) bags and administered through polyethylene-lined administration sets. All patients should be premedicated with oral corticosteroids for 3 days starting 1 day prior to docetaxel administration in order to reduce the incidence and

42 -36- GOG-0218 severity of fluid retention as well as the severity of hypersensitivity reactions Adverse Effects: Consult the package insert for the most current and complete information Supplier: Commercially available from Aventis. Consult the American Hospital Formulary Service Drug Information guide, Facts and Comparisons, or the package insert for additional information. 4.6 Quality of Life Measures The FACT-O TOI has been selected as the multidimensional, combined generic and disease-specific QOL questionnaire for use with ovarian cancer patients. The questionnaire is a 26-item self-report measure developed specifically for cancer patients and designed to be used in a variety of settings, including clinical trials. FACT-O TOI displays the QOL measures in the following order, recognizing the need for ease of administration and scoring: (1) FACT-G, (2) Additional Concerns: Ovarian component, (3) Additional items: stomach pain (from Protocol GOG-0172).

43 -37- GOG TREATMENT PLAN AND ENTRY/RANDOMIZATION PROCEDURE Before patient entries will be accepted, an official signed CTSU IRB Certification Form and a CTSU IRB/Regulatory Approval Transmittal Sheet (forms can be downloaded at must be received by the CTSU Regulatory Office. These forms can be faxed or mailed to: CTSU Regulatory Office Coalition of National Cancer Cooperative Groups 1818 Market Street, Suite 1100 Philadelphia, PA FAX Two Phase Registration / Randomization 5.11 Phase A - Initial Registration / Randomization When a suitable candidate has been obtained for protocol entry, the following steps should be taken: An approved informed consent and authorization permitting release of personal health information must be signed by the patient or guardian Make certain all eligibility requirements according to Section 3.0 have been met Phase A - Initial Registration / Randomization Fast Fact Sheet data should be gathered. With this data in hand, the GOG Statistical and Data Center should be called at: , Monday through Friday, 9 am to 5 pm EST/EDT Entry/Randomization will take place on the telephone after consideration of Fast Fact Sheet data The institution will enter the patient's initials (e.g. FML), GOG number, and, under assigned regimen, the term Phase A, in the appropriate place in their Log Book to verify the patient's entry into Phase A GOG Web-Based Registration and Randomization See Appendix IV for specific instructions regarding web-based registration and randomization. This is a randomized, double-blind trial. Blinded, phase and patient-specific clinical supplies of bevacizumab / placebo will be requested from the Pharmaceutical Management Branch, CTEP, NCI by the GOG Statistical and Data Center at the time of the initial registration / randomization (Phase A) and should arrive at the clinical site within approximately seven to ten days (see Section 4.3).

44 -38- GOG Phase B - Completion of Chemotherapy / Re-registration Phase B - Completion of Chemotherapy / Re-registration Fast Fact Sheet data should be gathered. With this data in hand, the GOG Statistical and Data Center should be called at: , Monday through Friday, 9 am to 5 pm EST/EDT Entry/Randomization will take place on the telephone after consideration of Fast Fact Sheet data The institution will enter the patient's initials (e.g. FML), GOG number, and, under assigned regimen, the term Phase B, in the appropriate place in their Log Book to verify the patient's entry into Phase B GOG Web-Based Registration and Randomization See Appendix IV for specific instructions regarding web-based registration and randomization. This is a randomized, double-blind trial. Blinded, phase and patient-specific clinical supplies of bevacizumab / placebo will be requested from the Pharmaceutical Management Branch, CTEP, NCI by the GOG Statistical and Data Center at the time of the completion of chemotherapy / re-registration (Phase B) and should arrive at the clinical site within approximately seven to ten days (see Section 4.3). 5.2 Procedures for International Collaborators If this study is activated for international collaboration, an appropriate Appendix will be included to specify procedures that will be employed for patient registration and randomization.

45 -39- GOG Treatment Plan 5.31 Therapeutic Program Patients will be randomized to one of three treatment regimens in equal proportions: Arm Phase A Regimen* Schedule Phase B Regimen* Schedule I Chemotherapy** q 21 days x 6 cycles q 21 days until Placebo*** q 21 days x 5 15 months total Then Placebo*** cycles beginning treatment time with cycle 2 of has elapsed Chemotherapy II Chemotherapy** q 21 days x 6 cycles Bevacizumab*** q 21 days x 5 cycles beginning with cycle 2 of Chemotherapy Then Placebo*** q 21 days until 15 months total treatment time has elapsed III Chemotherapy** q 21 days x 6 cycles Bevacizumab*** q 21 days x 5 cycles beginning with cycle 2 of Chemotherapy Then Bevacizumab*** q 21 days until 15 months total treatment time has elapsed NOTE: Cycle = 21 days NOTE: Patient re-registration required to move from Phase A to Phase B (see Section 5.1). * All treatment doses on day 1 ** Paclitaxel 175 mg/m 2 IV over 3 hours followed by Carboplatin AUC 6 IV over 30 minutes (Note : docetaxel 75mg/m 2 IV over 1 hour may be substituted for paclitaxel [see sections 2.65, 5.322, and 6.51].) *** Bevacizumab / Placebo 15mg/kg IV This is a randomized, double-blind trial. Blinded, phase and patient-specific clinical supplies of bevacizumab / placebo will be requested from the Pharmaceutical Management Branch, CTEP, NCI by the GOG Statistical and Data Center at the time of the initial registration / randomization (Phase A) AND at the time of the completion of chemotherapy / re-registration (Phase B) and should arrive at the clinical site within approximately seven to ten days (see Section 4.3) Methods of Chemotherapy and Study Drug (Bevacizumab / Placebo) Administration Biometric considerations in dose calculation Maximum body surface area used for Carboplatin, Paclitaxel and Docetaxel dose calculations will be 2.0 m 2 as per GOG

46 -40- GOG-0218 Chemotherapy Procedure Manual Bevacizumab / Placebo will be dosed at 15 mg/kg, with no maximum to total mg Sequence and timing of drug administration: Paclitaxel will be infused over 3 hours. Due to the risk of immediate hypersensitivity reaction, paclitaxel should always be the first drug to be infused during any combination. (Note, for circumstances in which docetaxel should be substituted for paclitaxel: Docetaxel will be administered as a 1 hour IV infusion at a starting dose of 75 mg/m 2 see Section and Section 6.62) Carboplatin will be administered as a 30 minute infusion, following paclitaxel (or docetaxel) administration. Bevacizumab/placebo administration will be as a continuous intravenous infusion following carboplatin infusion. Anaphylaxis precautions should be observed during Bevacizumab/placebo administration (Appendix V). DO NOT ADMINISTER AS AN IV PUSH OR BOLUS. The initial dose should be delivered over 90 minutes as an IV infusion. If the first infusion is well tolerated, the second infusion may be administered over 60 minutes. If the 60- minute infusion is well tolerated, all subsequent infusions may be administered over 30 minutes Pre-Medication: Paclitaxel For all courses where paclitaxel is to be administered, it is recommended that a preparative regimen be employed one hour prior to the treatment regimen on that day, to reduce the risk associated with hypersensitivity reactions to these drugs. This regimen should include a standard dose of dexamethasone (either IV or PO), an anti-histamine H1 (diphenhydramine mg IVP or orally, or an equivalent dose of an alternate H 1 blocker such as loratadine or fexofenadine), and a standard dose of antihistamine H2 IVP (such as cimetidine, ranitidine, or famotidine). In the event of a prior Bevacizumab/placebo hypersensitivity reaction, the prophylactic regimen should be modified as suggested in below Suggested Prophylaxis in Event of Prior Bevacizumab/ Placebo Infusion Reaction In the event of a prior Bevacizumab / placebo hypersensitivity

47 -41- GOG-0218 reaction, the following recommended prophylactic regimen is recommended upon re-exposure: H 1 blocker (diphenhydramine mg IVP or orally one hour prior to injection; or an equivalent dose of an alternate H 1 blocker such as loratadine 10 mg or fexofenadine 60 mg). H 2 blocker (famotidine 20 mg IVP or orally one hour prior to injection; or an equivalent dose of an alternate H 2 blocker). Dexamethasone (10 mg administered PO 12 and 6 hours prior to Bevacizumab injection) Docetaxel For all courses where docetaxel is to be administered, (see Section 6.51) it is recommended that patients be premedicated with dexamethasone 8 mg orally taken the night before, morning of, and evening after each treatment (total dose, 24 mg/wk), and an anti-histamine H1 (diphenhydramine mg IVP or orally, or an equivalent dose of an alternate H 1 blocker such as loratadine or fexofenadine) one hour prior to docetaxel. In the event of a prior Bevacizumab/placebo hypersensitivity reaction, the prophylactic regimen should be modified as suggested above in section Antiemetic Regimens It is anticipated that nausea and vomiting may be a significant side effect of each regimen. The following representative antiemetic regimens are suggested: Ondansetron 8-32 mg IV 30 minutes prior to administration of chemotherapy and dexamethasone mg IV 30 minutes prior to drug administration or, Granisetron 10 mcg/kg IV (or 2 mg PO) 30 minutes prior to chemotherapy, with or without lorazepam mg IV 30 minutes prior to chemotherapy Dosing of Carboplatin The carboplatin dose will be calculated to reach a target area under the curve (AUC) of concentration x time according to the Calvert formula using an estimated glomerular filtration rate (GFR) from the Jelliffe formula.

48 -42- GOG-0218 The initial dose of carboplatin must be calculated using GFR. In the absence of new renal obstruction or other renal toxicity greater than or equal to CTC Grade 2 (serum creatinine 1.5 x ULN), the dose of carboplatin will not be recalculated for subsequent cycles, but will be subject to dose modification as noted. In patients with an abnormally low serum creatinine (less than or equal to 0.6 mg/dl), due to reduced protein intake and/or low muscle mass, the creatinine clearance should be estimated using a minimum value of 0.6 mg/dl. If a more appropriate baseline creatinine value is available within 4 weeks of treatment, that may also be used for the initial estimation of GFR. CALVERT FORMULA: Carboplatin dose (mg) = target AUC x (GFR + 25) For the purposes of this protocol, the GFR is considered to be equivalent to the creatinine clearance. The creatinine clearance is calculated by the method of Jelliffe using the following formula: Ccr = 0.9 x {98 - [0.8 (age - 20)]} Scr Where: Ccr = creatinine clearance in ml/min Age = patient's age in years (from 20-80) Scr = serum creatinine in mg/dl Dosing of Bevacizumab/placebo Bevacizumab will be administered at 15 mg/kg IV. Patient weight at screening will be used to determine the Bevacizumab dose to be used for the duration of the study. If a patient s weight changes by 10% during the course of the study, then the Bevacizumab dose will be recalculated Supportive Care Guidelines for Bevacizumab/placebo: If an infusion-related adverse reaction occurs, the patient should be premedicated (if not already scheduled as described in Section ) for the next course; however, the infusion time for Bevacizumab (or Placebo) may not be decreased for the next infusion. If the next infusion is well tolerated with pre-medication, the infusion time for the next dose may then be decreased by minutes as long as the patient continues to be pre-medicated. If a patient experiences an infusionassociated adverse event with the 60-minute infusion, all subsequent doses should be given over minutes. Similarly, if a patient experiences an infusion-associated adverse event with the 30-minute

49 -43- GOG-0218 infusion, all subsequent doses should be given over minutes Prohibited Concomitant Therapeutic Modalities Prior to documented disease progression, the following therapeutic modalities are prohibited: Reassessment or cytoreductive surgery Anti-neoplastic therapy not otherwise specified in the current protocol, including cytotoxic, biologic, hormonal, or radiation therapy, regardless of indication (treatment of measurable disease or consolidation therapy). 5.4 Quality of Life Assessment Intervals When determining the specific assessment times, the investigator must balance treatment toxicities, the natural history of the disease, and time since initiating therapy along with an acute awareness of the study objectives. The investigators for the proposed study recommend seven assessment points to include: 5.41 Baseline, defined as prior to cycle 1 (t = 0 weeks). This assessment allows a pretreatment baseline against which to compare later results Prior to cycle 4 (t = 9 weeks, after three doses of chemotherapy and two doses Bevacizumab/placebo), defined as second comparison point in which to compare short-term effects. Ideally, this assessment should be obtained shortly before the administration of cycle 4, but could occur up to one week prior to cycle 4. This third cycle comparison is justifiable given the anticipated continued early impact of anti-angiogenesis therapy on QoL, especially abdominal symptoms. Since this is also the first assessment of response, this will allow a correlation of QoL with response in this setting Prior to cycle 7 (t = 18 weeks, after six doses of chemotherapy and five doses Bevacizumab/placebo), defined as final comparison point in which to compare short-term effects. Ideally, this assessment should be obtained shortly before the administration of cycle 7, but could occur up to one week prior to cycle 7. This comparison is justifiable given the anticipated continued early impact of antiangiogenesis therapy on QoL especially abdominal symptoms Prior to cycle 13 (t = 36 weeks, six months after completing cytotoxic chemotherapy), defined as the first assessment six months after the completion of carboplatin and paclitaxel. It is important to note that patients should continue to receive QOL assessments even if they stop the clinical trial early or are on other treatment regimens.

50 -44- GOG Prior to cycle 21 (t = 60 weeks), at the cessation of Bevacizumab/placebo, 15 months from initiation of all systemic therapy. It is important to note that patients should continue to receive QOL assessments even if they stop the clinical trial early or are on other treatment regimens Six months after cessation of Bevacizumab/placebo (t = 84 weeks). This final QoL assessment will allow evaluation of the long term affects of therapy. It is important to note that patients should continue to receive QOL assessments even if they stop the clinical trial early or are on other treatment regimens.

51 -45- GOG TREATMENT MODIFICATIONS In order to maintain dose-intensity and cumulative dose-delivery on this study, reasonable efforts will be made to minimize dose reduction and treatment delays as specified. Any patient whose treatment is delayed must be evaluated on a weekly basis until adequate hematologic and non-hematologic parameters have been met. No dose escalation is planned for this study. 6.1 Individual Dose Modification Levels All modifications are relative to the actual starting doses for the specific Regimen. For application of individual dose modifications, see specific guidelines below. Allowable drug dose levels and instructions are summarized in Tables A, B, C, and D. General Guidelines for Hematologic Toxicity (Section 6.2) Hematologic Nadirs, Table A (Section 6.3) Dose Levels for Docetaxel, Table B (Section 6.3) Delayed Hematologic Recovery, Table C (Section 6.4) Non-Hematologic Toxicity Table D (Section 6.5) 6.2 General Guidelines for Hematologic Toxicity 6.21 Initial treatment modifications will consist of cycle delay and/or dose reduction as directed Treatment decisions will be based on the absolute neutrophil count (ANC) rather than the total white cell count (WBC) Subsequent cycles of therapy will not begin until the ANC is 1500 cells/mm 3 (CTCAE Grade 1) and the platelet count is 100,000/mm 3. Therapy (including Bevacizumab / placebo) will be delayed for a maximum of three weeks until these values are achieved. Patients who fail to recover adequate counts within a three-week delay will no longer receive protocoldirected cytotoxic chemotherapy, but will resume Bevacizumab / placebo alone as otherwise specified, through completion of cycle 6 (Phase A) and after reregistration (Section 5.12), through completion of Phase B. Patients who received G-CSF prior to the current cycle may begin with ANC 1000 cells/mm 3, if clinically appropriate, to allow for transient reductions in ANC after discontinuation of G-CSF. Patients who are delayed more than 7 days may begin with ANC 1000 cells/mm 3, if clinically appropriate; as they will receive G-CSF with subsequent therapy (see Section 6.241) Use of Hematopoietic Cytokines and Protective Agents The use of hematopoietic cytokines and protective reagents are restricted as noted:

52 -46- GOG In general, patients will NOT receive prophylactic filgrastim (G-CSF), PEG-filgrastim (Neulasta), or sargramostim (GM-CSF) unless they experience treatment delays or recurrent neutropenic complications after treatment modifications as specified. In particular, hematopoietic growth factors should not be used to avoid initial chemotherapy dose modifications as stipulated in the protocol. However, patients may also receive growth factors for management of neutropenic complications in accordance with clinical treatment guidelines. If required, it is recommended that growth factors be initiated the day after the last dose of chemotherapy and typically continuing for a minimum of 10 days or until the ANC is sustained above 1000/mm 3. Growth factors should be discontinued if the ANC exceeds 10,000/ mm 3 and should not be used within 72 hours of a subsequent dose of chemotherapy Patients will NOT receive prophylactic thrombopoietic agents unless they experience recurrent Grade 4 thrombocytopenia after treatment modifications as specified below Patients may receive erythropoietin (EPO), iron supplements, and/or transfusions as clinically indicated for management of anemia Patients may NOT receive amifostine or other protective reagents, unless indicated in the study design Dose Modifications for Bevacizumab There will be no dose reduction for Bevacizumab. Treatment should be interrupted or discontinued for certain adverse events, as described below. 6.3 Modifications for Hematologic Toxicity (Nadirs) 6.31 Initial occurrence of dose-limiting neutropenia (defined in 6.32) and dose limiting thrombocytopenia (defined in 6.33) will be performed according to Table A, using the regimen modifications in Table B Dose-Limiting Neutropenia (DLT-ANC) is defined by the occurrence of febrile neutropenia or prolonged Grade 4 neutropenia persisting 7 days. There will be no modifications for uncomplicated Grade 4 neutropenia lasting less than 7 days. Febrile neutropenia is defined within the CTCAE as fever with or without clinically or microbiologically documented infection with ANC less than 1,000 /mm 3 and fever greater than or equal to 38.5 C Dose-limiting thrombocytopenia (DLT-PLT) is defined by any occurrence of Grade 4 thrombocytopenia ( 25,000/mm 3 ) or bleeding associated with Grade 3 thrombocytopenia (25,000 to 50,000/mm 3 ). There will be no modifications for uncomplicated Grade 3 thrombocytopenia.

53 -47- GOG-0218 Table A: Modification Instructions for Dose-Limiting Hematologic Toxicity (In conjunction with Table B when docetaxel substituted for paclitaxel) First Occurrence Second Occurrence Third Occurrence Add G-GSF and maintain all Off Study current drug doses DLT ANC DLT PLT Yes No Reduce carboplatin one AUC unit (and docetaxel one dose level*) Yes Yes Reduce carboplatin one AUC unit (and docetaxel one dose level*) No Yes Reduce carboplatin one AUC unit (and docetaxel one dose level*) Add G-GSF and decrease carboplatin one AUC unit (and docetaxel one dose level*) Decrease carboplatin one AUC unit (and docetaxel one dose Off Study Off Study level*) * See Table B below, for patients in whom docetaxel has been substituted for paclitaxel according to guidelines in Section Table B: Dose Levels for Docetaxel* Starting Dose Level Dose Level -1 Dose Level mg/m 2 65 mg/m 2 55 mg/m 2 * For patients in whom docetaxel has been substituted for paclitaxel according to guidelines in Section Modifications for Delayed Hematologic Recovery: 6.41 Delay on the basis of neutropenia (Delay-ANC) is defined if the ANC is less than 1,500 cells/mm 3 (CTCAE Grade 2 or worse) within 24 hours prior to scheduled therapy, or less than 1,000 cells/ mm 3, if the patient received G-CSF during the previous cycle Delay on the basis of thrombocytopenia (Delay-PLT) is defined if the platelet count is less than 100,000/ mm 3 within 24 hours prior to scheduled therapy Modifications noted below are only required for management of delays in the absence of dose reductions stipulated by nadir DLT-ANC and/or DLT-PLT (as noted above). In other words, if the patient experiences DLT-ANC and Delay- ANC, make the modifications as indicated for the nadir counts without additional modifications based on delayed recovery.

54 -48- GOG-0218 Table C. Modifications for Delayed Hematologic Recovery Category Delay Modification (days) Delay-ANC 1-7 No Change 8-21 Add G-CSF with Next Cycle >21 Off Study Treatment (see Section 6.7) Delay-PLT 1-7 No Change 8-21 Decrease carboplatin one AUC unit (and docetaxel one dose level*) >21 Off Study * For patients in whom docetaxel has been substituted for paclitaxel according to guidelines in Section Adjustments for Non-Hematologic Toxicity Table D: Modifications for Non-Hematologic Toxicity Drug Regimen -2 Level Regimen -1 Level Regimen Starting Dose Paclitaxel 110 mg/m mg/m mg/m 2 Carboplatin Bevacizumab/placebo Not Applicable Not Applicable Not Applicable Docetaxel 55 mg/m 2 65 mg/m 2 75 mg/m Grade 2 (or greater) peripheral neuropathy requires reduction of one dose level in paclitaxel and delay in subsequent therapy for a maximum of three weeks until recovered to Grade 1. If peripheral neuropathy fails to recover to Grade 1 by a maximum delay of three weeks from time therapy is due, then paclitaxel should be withheld from all subsequent chemotherapy cycles and docetaxel substituted for paclitaxel unless medically contraindicated, according to Section In such cases where docetaxel has been substituted for paclitaxel, if CTCAE Grade 3 or 4 peripheral neuropathy occurs during or after the first cycle of docetaxel substitution then subsequent doses of docetaxel will be delayed for a maximum of three weeks until recovered to CTCAE Grade 2. If peripheral neuropathy fails to recover to Grade 2 by a maximum delay of three weeks from time therapy is due, then all docetaxel should be withheld from all subsequent chemotherapy cycles 6.52 Hypertension. Patients receiving Bevacizumab / Placebo should be monitored prior to each dose with measurement of blood pressure (see Section 7.0 Study Parameters). Medication classes used for management of patients with Grade 3 hypertension receiving Bevacizumab/placebo included angiotensin-converting enzyme inhibitors, beta blockers, diuretics, and calcium channel blockers. The goal for blood pressure control should be consistent with general medical practice guidelines (i.e. 140/90 mmhg in general and 130/80 mmhg for patients with diabetes).

55 -49- GOG-0218 For controlled hypertension, defined as systolic 150 mm Hg and diastolic 90 mm Hg, continue Bevacizumab/placebo therapy For uncontrolled hypertension (systolic 150 mm Hg or diastolic 90) or symptomatic hypertension less than CTCAE Grade 4, hold Bevacizumab/placebo treatment, with anti-hypertensive therapy initiated or continued, as in During the period of combination chemotherapy with Bevacizumab/placebo, if hypertension is controlled and symptomatic hypertension has resolved by one week after holding treatment, continue all therapy. During the period of combination chemotherapy with Bevacizumab/placebo, if hypertension remains uncontrolled or symptomatic hypertension, less than CTCAE Grade 4, persists one week after holding treatment, the next treatment cycle should contain paclitaxel and Carboplatin only, if applicable, as otherwise indicated in the protocol, with Bevacizumab/placebo omitted. During the period after completion of chemotherapy, if uncontrolled or symptomatic hypertension has not resolved by three weeks after holding treatment with Bevacizumab/placebo, treatment with Bevacizumab/placebo should be discontinued for the remainder of the study. Bevacizumab/placebo should be discontinued for the remainder of the study in any patient developing CTCAE Grade 4 hypertension Proteinuria. Patients receiving Bevacizumab/placebo should be monitored by urine analysis for urine protein:creatinine (UPC) ratio prior to every other dose of Bevacizumab / placebo: UPC ratio < 3.5 UPC ratio > 3.5 Grade 4 or nephrotic syndrome Continue Bevacizumab / placebo. Hold Bevacizumab / placebo until UPC ratio recovers to < 3.5. If therapy is held for > 2 months due to proteinuria, discontinue Bevacizumab / placebo. Discontinue Bevacizumab / placebo.

56 -50- GOG Hemorrhage. Bevacizumab/placebo will be discontinued in patients with CTCAE Grade 3 hemorrhage and receiving full-dose anticoagulation. For all other patients with CTCAE Grade 3 hemorrhage, Bevacizumab/placebo should be held until ALL of the following criteria are met: bleeding has resolved blood hemoglobin level is stable there is no bleeding diathesis that would increase the risk of therapy there is no anatomical or pathologic condition that can increase the risk of hemorrhage recurrence. Patients who experience delay of resolution according to the above criteria for >3 weeks, recurrence of Grade 3 hemorrhage, or any CTCAE Grade 4 hemorrhage will be taken off Bevacizumab/placebo therapy Thrombosis Arterial Thrombosis Bevacizumab/placebo will be discontinued for CTCAE Grade 3 arterial thrombotic events (including cerebrovascular ischemia, transient ischemic attack, cardiac ischemia/infarction, peripheral or visceral arterial ischemia) or CTCAE Grade 2 arterial thrombotic events new or worsened since beginning Bevacizumab/placebo therapy Venous Thrombosis Treatment with Bevacizumab/placebo will be held for CTCAE Grade 3 or asymptomatic CTCAE Grade 4 venous thrombosis. For patients on therapeutic anticoagulation, PT INR or PTT (whichever appropriate) should be monitored closely during Bevacizumab/placebo therapy. If the planned duration of full-dose anticoagulation is 3 weeks, Bevacizumab/placebo should be held until the full-dose anticoagulation period is over. If the planned duration of full-dose anticoagulation is 3 weeks, Bevacizumab/placebo may be resumed during the period of fulldose anticoagulation if ALL of the following criteria are met (otherwise such patients will be taken off Bevacizumab/placebo therapy): The patient must have an in-range INR (usually between 2 and 3) on a stable dose of warfarin (or other anticoagulant) or on stable dose of heparin prior to restarting treatment The subject must not have pathological conditions that carry high risk of bleeding (e.g. tumor involving major vessels).

57 -51- GOG The subject must not have had hemorrhagic events while on study The patient is benefiting from treatment (no evidence of disease progression). Patients with symptomatic CTCAE Grade 4, or recurrent/worsening venous thromboembolic events after resumption of Bevacizumab/placebo treatment, will be taken off Bevacizumab/placebo therapy Coagulopathy. Bevacizumab/placebo should be held if the coagulation parameters are higher then the intended therapeutic range or for coagulopathy as follows. For CTCAE Grade 3 or 4 coagulopathy: hold treatment, until PT/PTT resolve to Grade 1. For patients with PT/INR therapeutic range while on therapeutic warfarin, treatment with Bevacizumab/placebo will be held until PT/INR is within the therapeutic range. Patients experiencing treatment delay three weeks because of failure to meet the above criteria will be taken off Bevacizumab/placebo therapy Wound Disruption/Intestinal Perforation. Bevacizumab/placebo will be discontinued in the event of wound disruption requiring medical or surgical intervention. This includes intestinal perforation related or unrelated to anastomoses Renal toxicity (associated with reduction in GFR) is not expected as a direct complication of chemotherapy in this untreated patient population using the prescribed dose and schedule of each regimen. As such, there are no specific dose modifications for renal toxicity. However, the target AUC dose of carboplatin must be recalculated each cycle in any patient who develops renal toxicity, defined by serum creatinine greater than 1.5 x institutional upper limit normal (ULN), CTCAE Grade Intestinal obstruction. Bevacizumab/placebo will be held for occurrence of CTCAE Grade 3 toxicity, until resolution to CTCAE Grade 1 and will be permanently discontinued for occurrence of CTCAE Grade 4 toxicity Hepatic toxicity is not expected as a direct complication of chemotherapy in this untreated patient population using the prescribed dose and schedule for each regimen. However, the development of Grade 3 (or greater) elevations in SGOT (AST), SGPT (ALT), alkaline phosphatase or bilirubin requires reduction of one dose level in paclitaxel and withholding of Bevacizumab/placebo, and delay in subsequent therapy for a maximum of three weeks until recovered to Grade There will be no dose modifications for alopecia, nausea, constipation, or diarrhea. It is recommended that routine medical measures be employed to manage nausea, constipation, and diarrhea.

58 -52- GOG In general, the occurrence of a hypersensitivity reaction to paclitaxel, carboplatin, Bevacizumab/placebo or docetaxel is not considered a dose-limiting toxicity. Patients may be retreated at full doses after administration of medication to prevent hypersensitivity reactions, and adjustments in infusion rates should be made. However, if despite these safety measures repeat attempt at infusion of the inciting drug results in a recurrent hypersensitivity reaction, the inciting drug should be discontinued for the remainder of the study. In the event of any CTCAE Grade 3 or 4 allergic or infusional reaction to Bevacizumab / placebo, Bevacizumab / placebo will be permanently discontinued. In the event of recurrent hypersensitivity reaction to paclitaxel, docetaxel should be substituted for paclitaxel, according to guidelines in Sections and Also, please see Appendix V for management of suspected hypersensitivity reactions to Bevacizumab/placebo Potential modifications for other non-hematologic toxicities with an impact on organ function of Grade 2 (or greater) require discussion with one of the study co-chairs except where noted below in Section Special Modifications Study Treatment For any CTCAE Grade 3 non-hematologic adverse event (except controllable nausea/emesis), Bevacizumab/placebo should be held until symptoms resolve to CTCAE Grade 1. If a CTCAE Grade 3 adverse event persists for three weeks or recurs after resumption of Bevacizumab/placebo therapy, the patient will be taken off all protocol therapy For any CTCAE Grade 4 non-hematologic adverse event (except controllable nausea/emesis), the patient will be taken off all protocol therapy.

59 -53- GOG STUDY PARAMETERS 7.1 Observations and Tests The following observations and tests are to be performed and recorded on the appropriate form(s). The specimen requirements are provided in Section 7.2 Observations and Tests History & Physical Blood pressure Toxicity Assessment CBC/Differential/ Platelets Urine Protein- Creatinine Ratio (UPCR) Serum Creatinine Bilirubin, SGOT, Alkaline Phosphatase Ca/PO4/Mg PT/PT INR/PTT Audiogram EKG Radiographic Tumor Measurement Chest X-Ray Serum CA-125 Level QoL Survey Incision Check Pre- Treatment Prior to Initial Study Treatment During Cytotoxic Chemotherapy and Bevacizumab/placebo Treatment Weekly Prior to Each Course Prior to Every Other Course Following Cytotoxic Chemotherapy, During Bevacizumab/place bo Only Treatment Prior to Every Course Prior to Every Other Course Post-Treatment Q 3 Months x 8 (24 Months) then q 6 Months 1 X X X 1 2 X X X 3 X X X 3 X , X X X , , 13 1, X X, X 17

60 -54- GOG Must be obtained within 28 days prior to initiating protocol therapy. 2. Blood pressure should be assessed at least weekly during the first cycle of Bevacizumab/placebo therapy. During the time between treatments, blood pressure assessment may be done at home by the patient at the investigator s discretion, and the investigator or study nurse will be responsible for obtaining these results from the patient. 3. See section Must be obtained within 14 days prior to initiating protocol therapy. 4. Must be obtained within 4 days of re-treatment with protocol therapy. 5. Weekly until counts recover from nadir 6. Urine protein should be screened by UPCR (see Section 3.3 for details). Patients must have a UPCR < 1.0 to allow participation in the study. 7. Patients receiving Bevacizumab/placebo should be monitored by urine analysis for urine protein:creatinine (UPC) ratio prior to every other dose of Bevacizumab / placebo: UPC ratio < 3.5 Continue Bevacizumab / placebo. UPC ratio > 3.5 Grade 4 or nephrotic syndrome Hold Bevacizumab / placebo until UPC ratio recovers to < 3.5. If therapy is held for > 2 months due to proteinuria, discontinue Bevacizumab / placebo. Discontinue Bevacizumab / placebo. 8. When clinically indicated 9. For patients on prophylactic or therapeutic anticoagulation, PT INR should be monitored before each treatment. Treatment should be held for PT INR of > 1.5 on prophylactic warfarin or > therapeutic range if on full-dose wafarin. 10. For patients with a history of hearing loss; repeat as clinically indicated 11. An initial CT scan or MRI of at least the abdomen and pelvis is required to establish post-surgical baseline for the extent of residual disease within 4 weeks of registration and beginning treatment. 12. Follow-Up Radiographic Assessment of Disease. In the absence of disease progression by criteria in Section 8.3, imaging using the same modality and encompassing the same field as in the initial pre-treatment evaluation should be repeated with the following schedule: a) After cycle 3 (before cycle 4) of paclitaxel-carboplatin b) After cycle 6 of paclitaxel-carboplatin (before cycle 7, Bevacizumab/placebo) c) Every three months after completion of chemotherapy during treatment with Bevacizumab/placebo d) During or after completion of all protocol therapy, as clinically indicated at any time for clinical suspicion of progressive disease, including rising serum CA-125 levels not meeting criteria for disease progression in and of themselves according to section 8.3 If based on any of these evaluations a response (CR or PR) is documented, a same modality imaging study should be performed after more than 4 weeks but within 3 months in order to confirm persistence of response by RECIST criteria. Regardless of the level of response confirmed, imaging will be repeated according to the schedule above. Imaging assessments as part of this protocol should be discontinued if disease progression is confirmed according to guidelines in section 8.3, regardless of means of confirmation, except that when disease progression is defined by CA-125 criteria alone, imaging using the same modality and encompassing the same field as in the initial pre-treatment evaluation should be obtained within two weeks that such progression is documented. 13. Not required if CT or MRI of chest already performed at pre-treatment baseline. 14. Baseline pre-chemotherapy value is required. When available, also include pre-surgical value. 15. Progression can be based upon serum CA-125, only during the period following completion of cytotoxic chemotherapy, if one of the three conditions are met: 1. Patients with elevated CA-125 pretreatment and normalization of CA-125 must show evidence of CA-125 greater than or equal to two times the upper normal limit on two occasions at least one week apart or 2. Patients with elevated CA-125 pretreatment, which never normalizes must show evidence of CA-125 greater than or equal to two times the nadir value on two occasions at least one week apart or 3. Patients with CA-125 in the normal range pretreatment must show evidence of CA-125 greater than or equal to two times the upper normal limit on two occasions at least one week apart. When disease progression is defined by CA-125 criteria alone, imaging using the same modality and encompassing the same field as in the initial pre-treatment evaluation should be obtained within 2 weeks that such progression is documented (see Section 7.1).If the patient does not meet criteria for disease progression on the basis of CA-125 elevations, then CA-125 monitoring should be continued according to schedule. 16. See Sections 4.5 and 5.4. QoL surveys are to be obtained at 6 time points: baseline, defined as prior to cycle 1; prior to cycle 4 of chemotherapy; prior to cycle 7; six months after completing cytotoxic chemotherapy; at the cessation of Bevacizumab/placebo (15 months from initiation of all systemic therapy); and six months after cessation of Bevacizumab/placebo. 17. See Section Patients with granulating incisions healing by secondary intention with no evidence of fascial dehiscence or infection are eligible but require weekly wound examinations until complete closure. Any occurrence of fascial dehiscence or deterioration related to the incision should be addressed according to guidelines for treatment modification in Section 6.57 and Adverse Events reporting in Section 10.3.

61 7.2 Specimen Requirements and Laboratory Testing -55- GOG Archival formalin-fixed and paraffin-embedded primary or metastatic tumor tissue collected prior to initiating front-line chemotherapy will be a mandatory requirement for patients enrolled at GOG and CTSU institutions within the United States and an optional requirement for patients enrolled at GOG institutions outside the United States as described in Appendix VI unless the patient does not consent to release her tumor tissue for this research study. In this situation, the reason must be documented on the fast-fact sheet to insure that the case is assigned to the appropriate form group and to avoid a protocol violation for this study. Every attempt should be made to provide a tumor block (see bolded paragraph on the top of the next page for suggestions if the block can not be provided on a permanent basis). If it is not possible to provide a block on a permanent or temporary basis, the back-up option will be to provide fifteen unstained sections, 10 m in thickness, on clean autoclaved glass slides suitable for laser capture microdissection, RNA extractions and genomic analysis, and a 50 m thick scroll of tumor in an autoclaved microfuge tube or sterile cryovial suitable for RNA extraction and genomic analysis. The type of tumor tissue (primary or metastatic) and specimen (block or sections) will need to be specified on the specimen transmittal form (Form SP) submitted for FT01 for GOG-0218 to insure that the appropriate per capita points are distributed for each patient. If sections are submitted instead of a tumor block, the reason must be stated in field 15 on the SP Form for this specimen (i.e., the Pathology Department at your institution is prohibited by local or state law from releasing blocks on a permanent or temporary basis for any reason). In addition, be aware that sections is not one of the choices provided for Items shipped (field 9 on Form SP), please select Other and specify sections on slides and in tube/vial. When a paraffin block is submitted, the GOG Tissue Bank will prepare appropriate standard unstained sections and thick sections on slides as well as a 50 m thick scroll in an autoclaved microfuge tube or sterile cryovial. The GOG Tissue Bank will also take cores from appropriate tumor blocks to create a series of tissue microarrays (TMAs) for this study. The plan will be to create two types of clinical outcome TMAs for this protocol. The first will contain tumor cores from patients who experienced a clinical response (complete or partial response), stable disease or increasing disease. The second will contain tumor cores from patients who experienced short survival, intermediate survival or long survival. The specific types of the TMAs that can be created will depend on the tumor blocks that are submitted for this protocol and the clinical outcomes observed for these cases. Since three to four cores from the same paraffin block are needed to reflect staining in a conventional tissue section, both of the clinical outcome TMAs will be generated in two sets of quadruplicate core. Each TMA will contain 250 cores representing 200 individual cases and 50 controls. Each set will represent 200 independent (unique) cases with the same 50 controls. This will allow one set to be used for screening or exploratory analyses and the other for validation. Additional sets will be prepared if sufficient blocks

62 -56- GOG-0218 associated with these clinical outcomes are available. Incorporation of the same controls on each of these TMAs will allow investigators to evaluate the performance of the individual arrays and allow inferences to be drawn across arrays when certain criteria are satisfied. If your institution can not permanently provide a tumor block for this research study, please urge the Pathology Department to allow a tumor block to be submitted to the GOG Tissue Bank on a temporary basis. In this case, please state in field 15 on the SP Form for FT01 that the tumor block must be returned after the unstained sections and cores are obtained. Creation of these clinical outcome TMAs for suboptimally-debulked advanced stage epithelial ovarian or peritoneal primary carcinoma will leverage the value of the tumor blocks submitted for GOG-0218 and establish an enduring resource for ovarian cancer research to study biomarkers of tumor response and survival following front-line treatment with standard cytotoxic chemotherapy with paclitaxel and carboplatin only, standard cytotoxic chemotherapy with paclitaxel and carboplatin plus 6 cycles of Bevacizumab (concurrent Bevacizumab) or standard cytotoxic chemotherapy with paclitaxel and carboplatin plus Bevacizumab for 15 months total treatment time (extended Bevacizumab). The formalin-fixed and paraffin-embedded tumor tissue will be used for immunohistochemistry assays and genomic analysis. Specifically, unstained sections on charged slides will be distributed to Dr. Robert Burger at the University of California Irvine Medical Center to compare the immunohistochemical expression of CD-31 and VEGF in conventional tumor tissue sections compared with TMAs. In addition, unstained sections on clean slides and 50 m thick tumor scroll will be distributed in batches to Dr. Michael Birrer at the National Cancer Institute for genomic analysis Frozen primary or metastatic tumor tissue collected prior to initiating frontline chemotherapy will be an optional requirement from each patient who is enrolled on this protocol as described in Appendix VI. This tumor specimen must be snap-frozen or OCT-embedded and frozen. The type of tumor tissue (primary or metastatic) and specimen (piece of snapfrozen tissue or OCT-mold with embedded frozen tissue) will need to be specified on the specimen transmittal form (Form SP) submitted for RT01 for GOG-0218 to insure that the appropriate per capita points are distributed for each patient. Frozen primary or metastatic tumor tissue will be distributed in batches to Dr. Michael Birrer at the National Cancer Institute for genomic analysis. To insure that the appropriate per capita points are distributed for each patient, (1) specify whether or not you plan to submit the optional frozen tumor specimen on the fast fact sheet for each patient you enroll on GOG-0218 to make sure the case is assigned to the proper form group, and (2) submit the SP Form for RT01 for GG When frozen tumor is submitted, indicate whether the item being shipped is an OCT-mold or a piece of snap frozen tumor in field

63 Required Specimens (Specimen Codes) 1 Archival Formalin-Fixed and Paraffin-Embedded Primary or Metastatic Tumor: 1 st choice: Tumor Block 2 nd choice: Tumor Sections - 15 unstained thick sections on slides and 50 m scroll in tube/vial (FT01) Frozen Tumor Tissue: snapfrozen or OCT-embedded and frozen (RT01) Frozen Pre-Treatment Serum (SB01) -57- GOG on Form SP. If both are being submitted, select Other and enter OCT-mold and piece in the specify field. In addition, indicate if the tumor is primary or metastatic in field 22 on Form SP Pre-treatment serum collected prior to initiating front-line chemotherapy will be an optional requirement from each patient who is enrolled on this protocol as described in Appendix VI. One aliquot of pre-treatment serum will be distributed in batches to Dr. Robert Burger at the University of California Irvine Medical Center to quantify VEGF level using an enzyme linked immunosorbent assay (ELISA). The exact choice of the other angiogenic markers / assays to be evaluated / performed in the other aliquots of pre-treatment serum will be reevaluated based on evolving data in the field. To insure that the appropriate per capita points are distributed for each patient, (1) specify whether or not you plan to submit the optional frozen serum specimen on the fast fact sheet for each patient you enroll on GOG-0218 to make sure the case is assigned to the proper form group, and (2) submit the SP Form for SB01 for this protocol. Quick Scan Summary of the Specimen Requirements for GOG Form SP Label in Forms Tracking System 2 SP-FT SP-RT SP-SB Collection Time Points and Requirements Collected prior to initiating front-line chemotherapy. Mandatory for patients enrolled at GOG and CTSU institutions within the United States and optional for patients enrolled at GOG institutions outside the United States. Collected prior to initiating front-line chemotherapy. Optional requirement for all patients. Collected prior to initiating front-line chemotherapy. Deadlines and Recommendations 2 Ship FT01 to the GOG Tissue Bank using your own shipping container within 8 weeks of study entry. Submit a copy of Form SP for FT01 to the Statistical and Data Center within 8 weeks of study entry. Ship RT01 to the GOG Tissue Bank using a Single-Chamber Specimen Kit within 8 weeks of study entry. Submit a copy of Form SP for RT01 to the Statistical and Data Center within 8 weeks of study entry. Ship SB01 to the GOG Tissue Bank using a Single-Chamber Specimen Kit within 8 weeks of study entry. Optional requirement for all Submit a copy of Form SP for SB01 patients. to the Statistical and Data Center within 8 weeks of study entry. 1 Label each specimen with the protocol number (GOG-0218), a GOG Bank ID (# # # # - # # G # # #), a specimen code (see above) and the collection date (mm/dd/yyyy). 2 Please complete Form SP for EACH of these three specimens and include a copy when the specimen is submitted to the GOG Tissue Bank as described in Appendix VI.

64 -58- GOG EVALUATION CRITERIA Response and progression will be evaluated in this study using the new international criteria proposed by the Response Evaluation Criteria in Solid Tumors (RECIST) Committee. 80 Changes in only the largest diameter (unidimensional measurement) of the tumor lesions are used in the RECIST criteria. Note: Lesions are either measurable or non-measurable using the criteria provided below. The term evaluable in reference to measurability will not be used because it does not provide additional meaning or accuracy. 8.1 Definitions 8.11 Measurable Disease Measurable lesions are defined as those that can be accurately measured in at least one dimension (longest diameter to be recorded) as 20 mm with conventional techniques (CT, MRI, x-ray) or as 10 mm with spiral CT scan. All tumor measurements must be recorded in millimeters (or decimal fractions of centimeters) Non-Measurable Disease All other lesions (or sites of disease), including small lesions (longest diameter <20 mm with conventional techniques or <10 mm using spiral CT scan), are considered non-measurable disease. Bone lesions, leptomeningeal disease, ascites, pleural/pericardial effusions, lymphangitis cutis/pulmonis, inflammatory breast disease, abdominal masses (not followed by CT or MRI), and cystic lesions are all non-measurable Target Lesions All measurable lesions up to a maximum of five lesions per organ and 10 lesions in total, representative of all involved organs, should be identified as target lesions and recorded and measured at baseline. Target lesions should be selected on the basis of their size (lesions with the longest diameter) and their suitability for accurate repeated measurements (either by imaging techniques or clinically). A sum of the longest diameter (LD) for all target lesions will be calculated and reported as the baseline sum LD. The baseline sum LD will be used as reference by which to characterize the objective tumor response Non-target lesions All other lesions (or sites of disease) should be identified as non-target lesions and should also be recorded at baseline. Non-target lesions include measurable lesions that exceed the maximum numbers per organ or total of all involved organs as well as non-measurable lesions. Measurements of these lesions are not required but the presence or absence of each should be noted throughout follow-up.

65 8.2 Guidelines for Evaluation of Measurable Disease 8.21 Method of Measurement -59- GOG General Aspects of Tumor Measurement All measurements should be taken and recorded in metric notation using a ruler or calipers. All baseline evaluations should be performed as closely as possible to the beginning of treatment and never more than 4 weeks before the beginning of the treatment. Note: Tumor lesions that are situated in a previously irradiated area might or might not be considered measurable. The same method of assessment and the same technique should be used to characterize each identified and reported lesion at baseline and during follow-up. Imaging-based evaluation is preferred to evaluation by clinical examination when both methods have been used to assess the anti-tumor effect of a treatment Specific Methods of Tumor Measurement Clinical lesions. Clinical lesions will only be considered measurable when they are superficial (e.g., skin nodules and palpable lymph nodes). In the case of skin lesions, documentation by color photography, including a ruler to estimate the size of the lesion, is recommended Chest x-ray. Lesions on chest x-ray are acceptable as measurable lesions when they are clearly defined and surrounded by aerated lung. However, CT is preferable Conventional CT and MRI. These techniques should be performed with cuts of 10 mm or less in slice thickness contiguously. Spiral CT should be performed using a 5 mm contiguous reconstruction algorithm. This applies to tumors of the chest, abdomen, and pelvis. Head and neck tumors and those of extremities usually require specific protocols. PET scanning information will not be evidence of disease progression or measurable disease. PET CT Fusion studies may not meet technical requirements. Any CT used must used criteria for assessing according to RECIST.

66 8.3 Response Criteria -60- GOG Ultrasound (US). When the primary endpoint of the study is objective response evaluation, US should not be used to measure tumor lesions. It is, however, a possible alternative to clinical measurements of superficial palpable lymph nodes, subcutaneous lesions, and thyroid nodules. US might also be useful to confirm the complete disappearance of superficial lesions usually assessed by clinical examination Endoscopy, Laparoscopy. The utilization of these techniques for objective tumor evaluation has not yet been fully and widely validated. Their uses in this specific context require sophisticated equipment and a high level of expertise that may only be available in some centers. Therefore, the utilization of such techniques for objective tumor response should be restricted to validation purposes in reference centers. However, such techniques can be useful to confirm complete pathological response when biopsies are obtained Tumor markers. Tumor markers alone cannot be used to assess response. If markers are initially above the upper normal limit, they must normalize for a patient to be considered in complete clinical response Cytology, Histology. These techniques can be used to differentiate between partial responses (PR) and complete responses (CR) in rare cases (e.g., residual lesions in tumor types, such as germ cell tumors, where known residual benign tumors can remain) Evaluation of Target Lesions The cytological confirmation of the neoplastic origin of any effusion that appears or worsens during treatment when the measurable tumor has met criteria for response or stable disease is mandatory to differentiate between response or stable disease (an effusion may be a side effect of the treatment) and progressive disease Complete Response (CR): Disappearance of all target lesions Partial Response (PR): At least a 30% decrease in the sum of the longest diameter (LD) of target lesions, taking as reference the baseline sum LD

67 -61- GOG Progressive Disease (PD): At least a 20% increase in the sum of the LD of target lesions, taking as reference the smallest sum LD recorded since the treatment started or the appearance of one or more new lesions Stable Disease (SD): Neither sufficient shrinkage to qualify for PR nor sufficient increase to qualify for PD, taking as reference the smallest sum LD since the treatment started 8.32 Evaluation of non-target lesions Complete Response (CR): Disappearance of all non-target lesions and normalization of tumor marker level Note: If serum CA-125 levels are initially above the upper normal limit, they must normalize for a patient to be considered in complete clinical response Incomplete Response/ Stable Disease (SD): Persistence of one or more non-target lesion(s) and/or maintenance of tumor marker level above the normal limits Progressive Disease (PD):Appearance of one or more new lesions and/or unequivocal progression of existing non-target lesions Although a clear progression of non-target lesions only is exceptional, in such circumstances the opinion of the treating physician should prevail, and the progression status should be confirmed at a later time by the review panel (or study chair) Progression Based On Serum CA-125 Progression can be based upon serum CA-125, only during the period following completion of cytotoxic chemotherapy, if one of the three conditions are met: 1. Patients with elevated CA-125 pretreatment and normalization of CA-125 must show evidence of CA-125 greater than or equal to two times the upper normal limit on two occasions at least one week apart or 2. Patients with elevated CA-125 pretreatment, which never normalizes must show evidence of CA-125 greater than or equal to two times the nadir value on two occasions at least one week apart

68 -62- GOG-0218 or 3. Patients with CA-125 in the normal range pretreatment must show evidence of CA-125 greater than or equal to two times the upper normal limit on two occasions at least one week apart When disease progression is defined by CA-125 criteria alone, imaging using the same modality and encompassing the same field as in the initial pretreatment evaluation should be obtained within 2 weeks that such progression is documented (see Section 7.1) Evaluation of Best Overall Response Progression Based on Development or Worsening of Ascites or Pleural Effusions Suspected progression based solely on developing or worsening ascites or pleural effusions must be verified cytologically The best overall response is the best response recorded from the start of the treatment until disease progression/recurrence (taking as reference for progressive disease the smallest measurements recorded since the treatment started). The patient's best response assignment will depend on the achievement of both measurement and confirmation criteria (see Sections 8.31 and 8.41). Target Lesions Non-Target Lesions New Lesions Overall Response CR CR No CR CR Incomplete response/sd No PR PR Non-PD No PR SD Non-PD No SD PD Any Yes or No PD Any PD Yes or No PD Any Any Yes PD

69 -63- GOG-0218 Note: Patients with a global deterioration of health status requiring discontinuation of treatment without objective evidence of disease progression at that time should be classified as having symptomatic deterioration. Every effort should be made to document the objective progression, even after discontinuation of treatment. In some circumstances, it may be difficult to distinguish residual disease from normal tissue. When the evaluation of complete response depends on this determination, it is recommended that the residual lesion be investigated (e.g., fine needle aspirate/biopsy) before confirming the complete response status. 8.4 Confirmatory Measurement/Duration of Response 8.41 Confirmation In order for a patient to be assigned a status of PR or CR, changes in tumor measurements must be confirmed by repeat assessments that should be performed no less than 4 weeks after the criteria for response are first met. In the case of SD, follow-up measurements must have met the SD criteria at least once after study entry at a minimum interval of 8 weeks (see Section 8.33) Duration of Overall Response The duration of overall response is measured from the time measurement criteria are met for CR or PR (whichever is first recorded) until the first date that recurrent or progressive disease is objectively documented (taking as reference for progressive disease the smallest measurements recorded since the treatment started). The duration of overall CR is measured from the time measurement criteria are first met for CR until the first date that recurrent disease is objectively documented. 8.5 Definitions Related to Evaluation Unrelated to Objective Response 8.51 Overall Survival is the observed length of life from entry into the study to death or the date of last contact Progression-Free Survival (measurable disease studies) is the period from study entry until disease progression, death or date of last contact. The defined date of disease progression will depend on the method of determination as follows: For disease progression defined by imaging or palpation of at least a 20% increase in the sum of the LD of target lesions, the appearance of one or more new lesions, or unequivocal progression of existing non-

70 -64- GOG-0218 target lesions, the date of progression will be defined as the date such lesions were first found to be progressed by imaging or palpation For disease progression defined by development or worsening of ascites or pleural effusions, the date of progression will be defined as the date of cytologic verification For disease progression defined by CA125 criteria alone, the date of progression will be defined as the first date of the initial CA125 of greater than or equal to two times the nadir value or upper limit of normal, whichever of these is applicable. Given that imaging using the same modality and encompassing the same field as in the initial pretreatment evaluation is required within 2 weeks of the confirmatory (second) CA125 value, if imaging criteria are met for progression, then the date of progression would be defined as the date of the imaging study (as in 8.521) Recurrence-Free Survival (non-measurable disease studies) is the period from study entry until disease recurrence, death or date of last contact Subjective Parameters including performance status, specific symptoms, and side effects are graded according to the CTCAE v3.0.

71 -65- GOG DURATION OF STUDY 9.1 Patients will receive treatment until disease progression, intolerable toxicity intervenes, or completion of Bevacizumab/placebo therapy, whichever comes first. The patient may voluntarily withdraw from the study at any time. No form of therapy targeted against a patient s cancer other than that specified in this protocol will be administered until disease progression. 9.2 All patients will be followed for disease status and toxicity (with completion of all required case report forms) until death or voluntary withdrawal from study. In addition, following study therapy, patients will be monitored for delayed toxicity for a period of one year with Q forms submitted to the GOG Statistical and Data Center, unless consent is withdrawn. 9.3 Adequate Duration of Study to Evaluate Toxicity. The minimal length of trial to evaluate toxicity is defined as receiving one course of therapy and receiving any followup information for evaluation of toxicity.

72 -66- GOG STUDY MONITORING AND REPORTING PROCEDURES 10.1 A consent form must be signed by the patient or guardian prior to entry into study. Current FDA, NCI and institutional regulations concerning informed consent will be followed Before patient entries will be accepted, an official signed CTSU IRB Certification Form and a CTSU IRB/Regulatory Approval Transmittal Sheet (forms can be downloaded at must be received by the CTSU Regulatory Office. These forms can be faxed or mailed to: CTSU Regulatory Office Coalition of National Cancer Cooperative Groups 1818 Market Street, Suite 1100 Philadelphia, PA FAX ADVERSE EVENT REPORTING FOR AN INVESTIGATIONAL AGENT Definition of Adverse Events (AE) An adverse event (AE) is any unfavorable and unintended sign (including an abnormal laboratory finding), symptom, or disease that occurs in a patient administered a medical treatment, whether the event is considered related or unrelated to the medical treatment. This study will utilize the Common Terminology Criteria for Adverse Events version 3.0 (CTCAE v3.0) for defining and grading specific adverse events. A copy of the CTCAE v3.0 can be downloaded from the CTEP home page at A GOG CTC Manual is also available on the GOG member web site ( under MANUALS) and can be mailed to the institution registering a patient to this study if requested Reporting Adverse Events Depending on the phase of the study, use of investigational agents, and role of the pharmaceutical sponsor, an AE report may need to reach multiple destinations. For patients participating on a GOG trial, the primary report should either be submitted by using the CTEP automated system for expedited reporting (AdEERS). All initial AdEERS submissions are reviewed by GOG prior to final submission. As such, submitting a report through AdEERS also serves as notification to GOG. Specific situations that require reporting through AdEERS are summarized in the table that follows Phase 2 and 3 Trials Utilizing an Agent under a CTEP IND: AdEERS Expedited Reporting Requirements for Adverse Events That Occur Within 30 Days 1 of the Last Dose of the Investigational Agent

73 -67- GOG-0218 Unrelated Unlikely Possible Probable Definite Grade 1 Grade 2 Grade 2 Grade 3 Grade 3 Unexpected Expected Unexpected and Expected Unexpected Expected Not Required Not Required Not Required 7 Calendar Days Not Required Not Required With Hospitalization 7 Calendar Days 7 Calendar Days Without Hospitalization Not Required 7 Calendar Days With Hospitalization 7 Calendar Days 7 Calendar Days Without Hospitalization Not Required Not Required Grades 4 & 5 2 Grades 4 & 5 2 Unexpected Expected 7 Calendar Days 24-Hrs; 3 Calendar Days 7 Calendar Days 7 Calendar Days 1 Adverse events with attribution of possible, probable, or definite that occur greater than 30 days after the last dose of treatment with an agent under a CTEP IND require reporting as follows: AdEERS 24-hour notification followed by complete report within 3 calendar days for: Grade 4 and Grade 5 unexpected events AdEERS 7 calendar day report: Grade 3 unexpected events with hospitalization or prolongation of hospitalization Grade 5 expected events 2 Although an AdEERS 24-hour notification is not required for death clearly related to progressive disease, a full report is required as outlined in the table. March 2005 Note: All deaths on study require both routine and expedited reporting regardless of causality. Attribution to treatment or other cause must be provided. Expedited AE reporting timelines defined: 24 hours; 3 calendar days The investigator must initially report the AE via AdEERS within 24 hours of learning of the event followed by a complete AdEERS report within 3 calendar days of the initial 24-hour report. 7 calendar days - A complete AdEERS report on the AE must be submitted within 7 calendar days of the investigator learning of the event. Any medical event equivalent to CTCAE grade 3, 4, or 5 that precipitates hospitalization (or prolongation of existing hospitalization) must be reported regardless of attribution and designation as expected or unexpected with the exception of any events identified as protocolspecific expedited adverse event reporting exclusions. Any event that results in persistent or significant disabilities/incapacities, congenital anomalies, or birth defects must be reported via AdEERS if the event occurs following treatment with an agent under a CTEP IND. Use the NCI protocol number and the protocol-specific patient ID provided during trial registration on all reports.

74 -68- GOG-0218 Additional Instructions or Exceptions to AdEERS Expedited Reporting Requirements for Phase 2 and 3 Trials Utilizing an Agent under a CTEP-IND: All Grade 2, 3, or 4 myelosuppression (including neutropenia, anemia, and thrombocytopenia) that does not require hospitalization is exempt from expedited reporting Procedures for Expedited Adverse Event Reporting: AdEERS Expedited Reports: Expedited reports are to be submitted using AdEERS available at The NCI guidelines for expedited adverse event reporting requirements are also available at this site. Please consult these guidelines for secondary malignancy (including AML, MDS) reporting requirements. For the purposes of expedited reporting of adverse events to CTEP, unexpected events are those not listed in the Agent Specific Adverse Event List (ASAEL). The ASAEL is a subset of AEs within the Comprehensive Adverse Event and Potential Risks List (CAEPR). This list of events is based on CTEP s clinical experience with this agent and defines expected Grade 2 and 3 AEs not requiring hospitalization as exempt from expedited reporting. The CAEPR is a complete list of reported and/or potential AEs associated with an agent under a CTEP IND. For questions or comments regarding the ASAEL or CAEPR, please contact the AdEERS MD Help Desk at adeersmd@tech-res.com GOG Expedited Reports: Submitting an AdEERS report to CTEP satisfies the GOG requirements for reporting adverse events. All adverse reactions will be immediately directed to the Study Chair for further action Automated CDUS reporting For studies using investigational agents, the GOG Statistical and Data Center (SDC) routinely reports adverse events electronically to the CTEP Clinical Data Update System (CDUS Version 3.0). The SDC submits this data quarterly. The AEs reported through AdEERS will also be included with the quarterly CDUS data submissions IDB: The requirement for timely reporting of adverse events to the study sponsor is specified in the Statement of Investigator, Form FDA In signing the FDA-1572, the investigator assumes the responsibility for reporting AE s to the NCI. In compliance with FDA regulations as contained in 21CFR , AE s should be reported by the investigator. Sufficient information should be provided so that the IDB physicians can make an independent assessment of the AE and determine if an expedited report to the FDA is warranted and if so, submit the report within the timeframe stipulated in the CFR.

75 -69- GOG-0218 Comprehensive Adverse Events and Potential Risks List (CAEPR) for Bevacizumab (NSC #704865) The Comprehensive Adverse Event and Potential Risks list (CAEPR) provides a single, complete list of reported and/or potential adverse events (AE) associated with an agent using a uniform presentation of events by body system. In addition to the comprehensive list, a subset, the Agent Specific Adverse Event List (ASAEL), appears in a separate column and is identified with bold and italicized text. This subset of AEs (ASAEL) contains events that are considered expected for expedited reporting purposes only. Refer to the CTEP, NCI Guidelines: Adverse Event Reporting Requirements for further clarification. The CAEPR does not provide frequency data; refer to the Investigator s Brochure for this information. Below is the CAEPR for Bevacizumab. Version 1.0, January 21, Category (Body System) Adverse Events with Possible Relationship to Bevacizumab (CTCAE v3.0 Term) Agent Specific Adverse Event List (ASAEL) ALLERGY/IMMUNOLOGY Allergic reaction/hypersensitivity (including drug fever) BLOOD/BONE MARROW Leukocytes (total WBC) Neutrophils/granulocytes (ANC/AGC) CARDIAC ARRHYTHMIA Supraventricular arrhythmia NOS CARDIAC GENERAL Cardiac ischemia/infarction Hypertension Hypotension Left ventricular diastolic dysfunction Left ventricular systolic dysfunction CONSTITUTIONAL SYMPTOMS Fatigue (asthenia, lethargy, malaise) Fever (in the absence of neutropenia, where neutropenia is defined as ANC <1.0 x 10e9/L) Rigors/chills Weight loss DERMATOLOGY/SKIN Rash/desquamation Ulceration Urticaria (hives, welts, wheals) Wound complication, non-infectious Allergic reaction/hypersensitivity (including drug fever) Cardiac ischemia/infarction Hypertension Fever (in the absence of neutropenia, where neutropenia is defined as ANC <1.0 x 10e9/L) Rigors/chills Rash/desquamation Urticaria (hives, welts, wheals)

76 -70- GOG-0218 GASTROINTESTINAL Anorexia Colitis Constipation Diarrhea Fistula, GI - Select Heartburn/dyspepsia Leak, GI: large bowel Mucositis/stomatitis (functional/symptomatic) - Select Nausea Perforation, GI - Select Vomiting HEMORRHAGE/BLEEDING Hemorrhage GI - Select Hemorrhage, CNS Hemorrhage, GU: vagina Hemorrhage, pulmonary/upper respiratory: lung Hemorrhage, pulmonary/upper respiratory: nose Hemorrhage/Bleeding - Other (varicesgastric/esophagus) INFECTION Infection with normal ANC or Grade 1 or 2 neutrophils - Select METABOLIC/LABORATORY Alkaline phosphatase ALT, SGPT (serum glutamic pyruvic transaminase) AST, SGOT (serum glutamic oxaloacetic transaminase) Bilirubin (hyperbilirubinemia) Creatinine Proteinuria NEUROLOGY CNS cerebrovascular ischemia PAIN Pain - abdomen NOS Pain - chest/thorax NOS Pain - head/headache Pain - joint Pain - muscle Pain NOS PULMONARY/UPPER RESPIRATORY Bronchospasm, wheezing Cough Mucositis/stomatitis (functional/symptomatic) - Select Hemorrhage GI - Select Hemorrhage, CNS Hemorrhage, pulmonary/upper respiratory: lung Hemorrhage, pulmonary/upper respiratory: nose Infection with normal ANC or Grade 1 or 2 neutrophils - Select Proteinuria CNS cerebrovascular ischemia Pain - chest/thorax NOS Pain - head/headache Pain - joint Cough

77 -71- GOG-0218 Dyspnea (shortness of breath) Nasal cavity/paranasal sinus reactions Voice changes RENAL/GENITOURINARY Renal/Genitourinary - Other (nephrotic syndrome) SYNDROMES Cytokine release syndrome/acute infusion reaction VASCULAR Thrombosis/thrombus/embolism Visceral arterial ischemia (nonmyocardial) Voice changes Cytokine release syndrome/acute infusion reaction Thrombosis/thrombus/embolism 1 This table will be updated as the toxicity profile of the agent is revised. Updates will be distributed to all Principal Investigators at the time of revision. The current version can be obtained by contacting ADEERSMD@tech-res.com. Your name, the name of the investigator, the protocol and the agent should be included in the . Additional selected AEs reported on Bevacizumab trials but with the relationship to Bevacizumab still undetermined: Blood/Bone Marrow hemoglobin; idiopathic thrombocytopenia purpura; platelet Cardiac arrhythmia ventricular arrhythmia Cardiac general cardiac arrest; pericardial effusion Coagulation DIC Death sudden death (cause unknown) Dermatology/Skin hypopigmentation Gastrointestinal bowel obstruction; rectal abscess/necrosis; taste alteration Metabolism hyperglycemia; hypoglycemia; hypomagnesemia; hyponatremia Musculoskeletal/Soft Tissue aseptic necrotic bone; gait/walking; myasthenia gravis Neurological aseptic meningitis; confusion; encephalopathy; peripheral neuropathy; seizure; syncope Ocular/Visual alexia; cataract; homonymous hemianopsia; watery eye Pulmonary/Upper Respiratory ARDS; pneumonitis/pulmonary infiltrates; pneumothorax Renal/Genitourinary urinary frequency (for a more complete listing of reported AEs, please refer to the Investigator s Brochure) Notes: Bevacizumab in combination with other agents could cause an exacerbation of any adverse event currently known to be caused by the other agent, or the combination may result in events never previously associated with either agent.

78 10.4 GOG DATA MANAGEMENT FORMS Form -72- GOG-0218 The following forms must be completed and submitted to the GOG Statistical and Data Center (SDC) in accordance with the schedule below. All forms except: F-form, Pathology report, OP report and QOL forms should be submitted via the SDC Electronic Data Entry System (SEDES) which is available through the GOG website ( Quality of life questionnaires are to be completed on Scantron forms and submitted by mail. Pathology material (F-form, path report and slides) should be submitted together via mail. Weeks Due within Event Copies * Comments Form R and OSO 4 Registration 2 Submit via SEDES Form C 4 Registration 2 Submit via SEDES Operative Report 4 Registration 2 Form DR 4 Registration 2 Submit via SEDES Form DRB 4 Registration 2 Submit via SEDES Form D2M*** 4 Registration 2 Submit via SEDES Primary disease: Form F Pathology Report Slides Registration Registration Registration Completion of each cycle of therapy Form D2R cycle 1 Subsequent cycles Form D2M*** 2 Clinical response assessment Form T 2 Beginning of each subsequent cycle Form Q0 2 Completion of study Rx and change in Rx Form Q 2 Disease progression; death; normal followup 3 3 ** 2 1 Submit together to SDC via postal mail Submit via SEDES 2 Submit via SEDES 1 Submit via SEDES 1 Submit via SEDES 1 quarterly for 2 years, semiannually for 3 more years, Submit via SEDES 1 Submit Coversheet and Scantron form via postal mail Quality of Life Form and Coversheet 2 Date Completed **** Specimen Consent Application 1 Registration N/A Required if biologic specimens are submitted. Form SP-FT for 8 Registration 1 Mandatory for GOG and CTSU archival formalin-fixed and Institutions within the United paraffin-embedded primary or States and Optional for GOG metastatic tumor tissue (FT01) Institutions outside the United Form SP-RT for frozen primary or metastatic tumor tissue (RT01) Form SP-SB for pretreatment serum (SB01) Surgical CRF 8 30 days after any surgical procedure performed on patients while on study States 8 Registration 1 Optional Requirement for all patients 8 Registration 1 Optional Requirement for all patients 2 Submit via SEDES

79 -73- GOG-0218 * The number of required copies including the original form which must be sent to the Statistical and Data Center, if not completed on-line through SEDES. ** At least one representative stained slide (or slides) documenting the primary. *** The original form is retained in the clinic and data is appended to this form each time the disease is reassessed. Two copies of the form are submitted to the Statistical and Data Center at baseline and one copy each time clinical response is evaluated. **** QOL is assessed at the following time points: prior to randomization, prior to cycle 4 (9 weeks after starting treatment), prior to cycle 7 (18 weeks after starting treatment), prior to cycle 13 (36 weeks after starting treatment), prior to cycle 21(60 weeks after starting treatment), and 6 months after completing study treatment (84 weeks after starting treatment). The time in parenthesis refer to patients removed from study treatment prior to completing the entire regimen. Use only Scantron forms with the header GOG Protocol Additional QoL forms are be provided by the SDC upon request. Cover sheet must be submitted together with the Scantron form. If assessment is not performed, a cover sheet is still required. This study will be monitored by the Abbreviated Clinical Data System (CDUS) Version 1.x. CDUS data will be submitted quarterly to CTEP by electronic means.

80 11.0 STATISTICAL CONSIDERATIONS -74- GOG Randomization: This study is a three-arm, double-blind, placebo-controlled randomized clinical trial. All individuals enrolled onto the study will receive standard treatment consisting of 6 cycles of carboplatin and paclitaxel (CT). The two experimental regimens will consist of standard therapy combined with either Bevacizumab for 5 cycles (CTB5) or Bevacizumab for 15 months (CTB+). A dynamic allocation procedure will be used that tends to allocate CT, CTB5 and CTB+ in the ratio of 1:1:1 within the following stratification factors: Stage of disease (stage III vs stage IV) Initial performance status (0 vs 1 or 2). The assigned study regimen is not revealed until after the patient has been successfully registered onto the study. Interim and final reports include an accounting of all patients registered onto the study, regardless of their eligibility status or compliance to the assigned treatment Efficacy and toxicity measures: The principle observations for evaluating the therapeutic effects of treatment are: Primary efficacy endpoint: Overall survival Secondary efficacy endpoint: Progression-free survival (PFS) Safety endpoints: frequency and severity of adverse effects (Common Terminology Criteria for Adverse Events -version 3.0) Accrual goal, accrual rate and study duration: The targeted accrual is 2000 patients. It is anticipated that at least 650 patients per year can be enrolled from GOG treatment centers. Therefore, the anticipated time to accrue the targeted sample size is slightly more than three years. The first objective of this study is to compare each of the experimental regimens to the standard regimen. Assuming a constant death rate (median survival 30 months) and a uniform accrual rate, the anticipated time from initiating the study until survival matures sufficiently for these two comparisons is slightly more than four years. If both of the experimental regimens are superior to the standard regimen, then a second study objective is to compare the two experimental regimens to each other. The anticipated time from initiating the study until the survival matures sufficiently for this comparison is about six years. Expected median duration of survival and PFS on standard treatment The expected median duration of overall survival for women with newly diagnosed, advanced stage, suboptimally debulked, epithelial ovarian cancer treated with a standard platinum-taxane regimen is 30 months. The expected median time to first progression or death (PFS) is approximately 14 months. Treatment efficacy: Primary hypotheses, overall type I error and power

81 -75- GOG-0218 The primary analyses of overall survival will include all patients enrolled onto the study regardless of eligibility or compliance to their assigned study regimen. Patients will be grouped by their randomized treatment for intention-to-treat analyses (ITT). The first objective of this study is to determine whether Bevacizumab (CTB5 or CTB+) reduces the overall death rate when compared to the standard treatment (CT). Each of the null hypotheses: H 01 : 01 = CT / CTB5 1 and H 02 : 02 = CT / CTB+ 1, will be assessed separately, where. is the death rate for the indicated treatment. The treatment regimens will be compared with two distinct logrank tests each of which includes all of the patients enrolled into the study stratified by stage of disease (III vs IV) and initial performance status (0 vs 1 or 2). Since the CTB5 and CTB+ treatment regimens are equivalent over the first six cycles of therapy (initial 4.5 months) these two treatment groups can be combined during this interval and share information in order to increase the statistical power of each experimental-to-standard treatment comparison. Specifically, all of the patients will be included in the analysis of H 01 however, the times at risk for those patients, who are randomized to receive CTB+ and survive longer than 4.5 months, will be censored at 4.5 months. Similarly, while assessing H 02 the times at risk for those patients who are randomized to receive CTB5 and survive longer than 4.5 months, will be censored at 4.5 months. The study design will limit the overall type I error to 2.5% (one-tail) for these two comparisons accounting for the planned interim analyses and the correlation between these comparisons induced by using a common reference group and sharing patients on the experimental regimens. Assuming that the correlation between estimated hazard ratios is 0.5, the one-tail type I error allocated to each of these comparisons is 1.35% including the error spent due to interim analyses 87. If one of these experimental regimens truly decreases the death rate 23% (ie, increases median survival from 30 to 39 months), this study design provides approximately an 83% chance of correctly classifying that regimen superior to the standard regimen (CT). In the event that both experimental regimens are deemed superior to the standard regimen, the two experimental regimens (CTB5 and CTB+) will be compared to each other. In order to ensure adequate statistical power, additional patient follow-up following the evaluation of H 01 and H 02 will be required. The null hypothesis (H 03: 03 = CTB5 / CTB+ 1) will be assessed with a logrank test stratified by stage of disease (III vs IV) and initial performance status (0 vs 1 and 2). Only those patients randomized to receive CTB5 or CTB+ and survive longer than 4.5 months will be included in this comparison. The type I error for testing this hypothesis will be limited to 0.05 (one-tail test). If the prolonged regimen of Bevacizumab (CTB+) reduces the death rate 17% relative to the shorter-duration Bevacizumab regimen (CTB5) (ie, increases median survival from 39 to 47 months) then the study design provides an 80% chance of declaring the CTB+ regimen more effective than the CTB5 regimen. Interim analyses Two interim analyses of overall survival are planned. The first will occur when there are at least 196 deaths reported among those randomized to receive the standard regimen and the second when there are at least 262 deaths reported among those randomized to the standard regimen. These time points are expected to be approximately 60% (523 deaths on all three arms) and 80% (698 deaths on all three

82 -76- GOG-0218 arms) of the full information time (873 deaths on all three arms) for the survival analyses when the alternative hypothesis, that both experimental regimens reduce the death rate 23%, is true. Assuming a constant failure rate and uniform accrual the first and second interim analyses are expected to occur approximately 35 and 41 months after initiating the study depending on the accrual rate. The O Brien-Fleming-like group sequential boundary function will be used to adjust for sequential testing in the event that the interim analysis can not be scheduled at the time when there are precisely the required number of events reported. An -spending function as described by Lan and Demets (1983) will be used with the information fraction calculated as the number deaths at the interim analysis among those patients randomized to the standard regimen to the full information of 327 deaths. The interim analyses will include an assessment of treatment efficacy. For example, provided the interim analysis occurs at precisely 60% and 80% of the information time, each experimental regimen will be compared to the standard regimen with the previously described stratified logrank test and alpha set to (1 st interim analysis), (2 nd interim analysis and (final analysis) If the study is in the accrual phase and either of these null hypotheses are rejected, then consideration will be given to terminating accrual to the standard regimen. The interim analyses will also include a futility assessment. If the stratified logrank procedure indicates that the death rate on an experimental arm exceeds the death rate on the standard arm then consideration will be given to terminating accrual to that experimental arm. This futility analysis increases the overall type II error slightly. The results of interim analyses are scheduled to be reviewed by the GOG Data Monitoring Committee (DMC) at its semi-annual meetings. This committee meets in January and July each year The precise dates for these meetings are set more than one year in advance by individuals who have no knowledge of efficiacy results. Eight weeks prior to each of these meetings, the database is locked in order to prepare a progress report. If the prerequisite number of events has been attained, an interim analysis is also prepared and presented to the DMC at their next scheduled meeting. The decision to terminate accrual to any particular regimen includes consideration of toxicities, treatment compliance, progression-free survival and results from external studies. Additionally, the GOG Data Safety and Monitoring Board (DSMB) reviews accumulating summaries of toxicities and all serious adverse event (SAE) reports on an ongoing basis (not efficacy results). This committee also reviews those deaths in which study treatment may have been a contributing cause. The DSMB reports to the DMC and it may recommend study amendments pertaining to patient safety. The boundaries for assessing statistical significance will not be altered by the DMC s decision to terminate the accrual onto a particular study regimen, Final analysis (experimental regimens vs Standard regimen) The first component of the final analysis will compare the standard arm to each of the experimental arms with respect to overall survival. This analysis will occur when there are at least 327 deaths observed among those randomized to receive the standard regimen. If Bevacizumab truly reduces the death rate 23% then the expected number of deaths for each of these pair-wise comparisons is 600 (273 on each experimental arm and 327 deaths on the standard arm).

83 -77- GOG-0218 First, the death rate for each of the experimental regimens will be compared to the standard regimen with the previously described logrank test and the type I error limited to 1.35% including the type I error spent for the interim analyses. The final analysis will also include exploratory analyses to assess the consistency of the treatment effect across subgroups of patients determined by presence of clinically measurable of disease (clinically measurable vs non-measurable), site of primary disease (ovarian vs extra-ovarian), stage of disease (III vs IV), histologic cell type (papillary serous vs mucinous vs clear cell vs other cell types), Grade (1 and 2 vs 3) and age ( 60 vs 60 years). The exploratory analysis will include an estimate of the treatment hazard ratios among only those patients deemed eligible for the study. Final analysis (short vs prolong treatment) In the event that both of the experimental regimens are deemed superior to the standard regimen then the two experimental regimens will be compared to each other. This analysis will be performed once there are at least 710 deaths among those patients who were randomized to receive either CTB5 or CTB+ and survive at least 4.5 months. These treatment groups will be compared with the previously described stratified logrank test and the critical value set to 1.64 (p=0.05 for a one-tail test). Analysis of PFS An analysis of progression-free survival (PFS) will be performed at the first DMC meeting after at least 274 PFS events have been reported among those allocated to the standard regimen. Additionally, this final analysis of PFS will take place no earlier than the first interim analysis of overall survival. The primary purpose of this analysis is to seek accelerated FDA approval for Bevacizumab. The null hypotheses: H 01 : CT / CTB5 1 and H 02 : CT / CTB+ 1, where is the PFS event rate, will be assessed with two stratified logrank tests. Each of these logrank tests will include all of the patients enrolled to the study. When assessing H 01 those patients randomized to receive CTB+ and surviving progression-free longer than 4.5 months will have their PFS time censored at 4.5 months. Similarly, when H 02 is assessed, those patients randomized to CTB5 with PFS times longer than 4.5 months will have their PFS censored at 4.5 months. Each logrank test will be stratified by stage of disease (III vs IV) and initial performance status (0 vs 1 and 2). Assuming uniform accrual, a constant PFS event rate, and Bevacizumab increases the median time to progression 30% in both experimental arms, then it is estimated that there will be 736 PFS events at the time of the scheduled interim analysis of survival. That is, for each pair-wise comparison, the number of PFS events is 274 in the standard arm, the expected number of events is 231 in each experimental arm and 10-15% of the patients randomized to the experimental regimens will have uncensored PFS times less than 4.5 months and therefore contribute to both comparisons. The critical z-value will be set to 2.18 in order to limit the overall type I error to 2.5% (based on 10,000 simulated trials) for one-sided testing of both H 01 and H 02, accounting for the common reference arm and the shared patients in the experimental arms. It is estimated that the statistical power to detect a hazard ratio ( CT / CTB5 or CT / CTB+ ) of 1.3 at the interim analysis is 82% (based on 10,000 simulated trials).

84 -78- GOG Safety analyses The National Cancer Institute (NCI) Common Terminology Criteria for Adverse Events (CTCAE) criteria version 3.0 will be used to classify toxicities observed during treatment. The severity of each toxicity will be assessed according to the NCI CTCAE 3.0 grading system. Patients will be tabulated according to their maximum severity for each organ system or preferred term. Safety endpoints will be summarized with descriptive statistics for the patients in the safety analysis dataset. The safety analysis dataset will include all patients enrolled to the study who receive any of their assigned study treatment and the patients will be grouped by their assigned treatment. Patients who do not receive any of their assigned study treatment will not be included in these analyses Quality of Life Analyses The principal measure used in this study to assess the quality of life (QoL) is the selfadministered FACT-O TOI for ovarian cancer patients. Each patient will be asked to complete the FACT-O TOI at the following time points during their participation in the study: Prior to cycle Prior to cycle 4 (9 weeks after starting treatment), Prior to cycle 7 (18 weeks after staring treatment), Prior to cycle 13 (36 weeks after starting treatment), Prior to cycle 21 (last treatment cycle, 60 weeks after starting treatment), Six months after study treatment (84 weeks after starting treatment). The times in parentheses indicate the assessment points for those patients who do not complete the entire study regimen. Construct and content The Functional Assessment of Cancer Therapy scale developed for ovarian cancer (FACT-O TOI) is a tool that provides a general QoL score. It consists of 3 subscale: physical well being (7 items), functional well being (7 items) and the Ovarian Cancer subscale (12 items) Hypotheses and analyses The principal QoL question is: Are the FACT-O TOI scores reported by patients at specific time points during treatment independent of the randomized treatment? This question will be broken into three separate hypotheses involving one pair of treatment groups at a time: H 01 : T CT = T CTB5, H 02 : T CT = T CTB+ and H 03 : T CTB5 = T CTB+, where T (.) is a vector of mean TOI scores evaluated at specific time points for the patients treated according to the indicated treatment regimen. For the primary analysis these hypotheses will be assessed with mixed models, adjusting for pretreatment TOI score and age and patients will be included in these analyses regardless of the amount of study treatment they received. For the primary analyses patients will be categorized by their randomized treatment group rather than the treatment received. Analyses which classify patients by the actual treatment they received will be considered exploratory.

85 -79- GOG-0218 H 01 : T CT = T CTB5 : The primary analysis comparing the self-reported TOI scores for patients receiving CT to those receiving CTB5 will focus on the scores assessed prior to cycle 4 and prior to cycle 7. These time points are considered appropriate since the immediate cumulative impact of Bevacizumab treatment on TOI scores, if there is any, should be apparent by this time. Including subsequent assessments points could washout early differences between these treatment regimens, if the impact of treatment wanes after Bevacizumab is stopped. It is anticipated that relatively few patients will withdraw from the study treatment prior to the 7 th cycle of treatment. Since the cumulative dose of drug in the two Bevacizumab containing regimens are identical up to these points in time for the purpose of assessing this hypothesis, the mean scores for all of the patients randomized to either CTB5 or CTB+ will be combined and compared to the standard regimen (CT). H 02 : T CT = T CTB+ : The primary analysis comparing the self-reported TOI scores for patients receiving CT to those receiving CTB+ will focus on the scores assessed prior to cycle 13 and prior to cycle 21 (the last cycle of treatment). These time points are considered appropriate since the cumulative impact of prolonged Bevacizumab treatment on TOI scores, if there is any, should be apparent by these times. This comparison will include only those patients randomized to either CT or CTB+ regardless of the amount of study treatment received. H 03 : T CTB5 = T CTB+ : The primary analysis comparing the self-reported TOI scores for patients randomized to CTB5 to those randomized to CTB+ will focus on the scores assessed prior to cycle 13 and prior to cycle 21 (the last cycle of treatment). These time points are considered appropriate since the cumulative impact of prolonged Bevacizumab treatment on TOI scores, if there is any, will be apparent by this time. This comparison will include those patients randomized to either CTB5 or CTB+ regardless of the amount of study treatment received. Multiplicity of Outcomes The overall type I error for these three QoL hypotheses (H 01, H 02, and H 03 ) will be limited to 5% (two-tail). In order to account for multiple hypotheses, the type I error will be allocated equally to each hypothesis. Specifically, the significance level will be set to (0.05/3) for each treatment comparison with two-tail tests. Missing information Patient death, noncompliance, missed appointments, and patient illiteracy, can cause missing information. One or more of the QoL assessments may be missing for an individual on any occasion. Missing information is troublesome; particularly in studies involving repeated patient assessments. Data management procedures will be used to reduce missing data. To this end, a calendar of events which lists the dates for the required QoL assessments for each patient will be made available to the patient's health care provider as soon as the patient has been registered onto this study. Also, the clinic staff will use the GOG web-based forms tracking system to obtain reminders of the upcoming QoL assessments.

86 -80- GOG-0218 At semi-annual group meetings the data managers and nurses will be given presentations, which describe the goals of this study and stress the importance of obtaining complete assessments. A study contact person will be designated to answer questions that arise throughout the study. Spanish and English versions of the FACT-O are available. Women who are unable to read or have difficulty reading will not be required to participate in the QoL part of this study. Also, any woman, who does not wish to participate in the QoL portion of this study, can refuse and remain eligible for the therapeutic portion of the study. TOI Scoring Within an individual assessment one or more items may not be answered. A subscale score will be computed as long as more than 50% of subscale items have a valid response. A subscale score S i with N i items will be calculated as: S i N * i j 1 ( * s ij j 1 ij ij ) Where δ ij is equal to 1 when the jth item has a valid response, otherwise it is equal to 0 and s ij is the response score of the jth item. The total FACT-O TOI score is the sum of the subscale scores. Statistical Power Considerations The GOG has completed a trial in which 415 patients with advanced ovarian cancer were treated with platinum and paclitaxel for 6 cycles every 21 days. These women reported their self-assessed FACT-O prior to initiating treatment, prior to the 4 th cycle of treatment, following the 6 th cycle of treatment and then 6 months later (GOG-172). Prior to initiating the study treatment, the mean and standard deviation of the FACT-O TOI scores were 67.2 and 15.9, respectively. The mean and standard deviation of the TOI scores prior to cycle 4, after cycle 6 and 12 months after cycle 6 were: (66.6, 15.3), (71.7, 15.6) and (82.7, 14.4), respectively. The correlation between pretreatment assessments and the assessments prior to the 4 th cycles and the 6 th cycle of treatment was about 0.4. The correlation between the pretreatment assessments and the assessments performed 12 months after completing treatment was 0.2. Using these data and assuming there will be a 10% attrition of patients at each of the assessment times: prior to treatment cycle 4 and prior to cycle 7, this study is expected to have approximately 91.4% power for detecting a 2.5 unit true difference in mean TOI scores between treatments when assessing H 01. Similarly, assuming 20% attrition of patients prior to each treatment cycles 13 and 21, this study is expected to have approximately 83% power to detect a 3 unit true difference in mean TOI scores between the treatment groups when assessing either H 02 or H 03. These power calculations are based on 1000 simulated trials. SAS source code for simulations is available upon request.

87 -81- GOG Genomic data analyses Overview The overall objective of the genomic analyses is first to identify genes that are associated with longer survival and then secondly to develop a prognostic index based on the genomic data. One additional objective is to determine whether there are genomic markers that predict which patients respond favorably to Bevacizumab. In general, the primary challenges related to this objective are: a) the need to identify a relatively small number of prognostic genes from among thousands of candidate genes, b) the practicality of having a relatively small number of tissue samples relative to the number of candidate genes. In order to address these challenges this study will utilize a training dataset to develop a prognostic index and a separate and distinct validation dataset. Also, this study will focus on the expression of a relative small number of genes (approximately 200) in which evidence from a previous study indicates that there is an association with overall survival. 91 The association between prognosis and other genes will be evaluated in a similar fashion, but these analyses will be considered part of a secondary analysis. Training and validation sets It is desirable to obtain tissue from all of the patients entered into the randomized portion of this trial. However, patients will not be required to submit tissue in order to participate in the randomized component of this study. In order to establish a training set this study will target a sample of sequentially enrolled eligible and evaluable patients with at least 100 deaths reported. That is, suppose 650 patients are enrolled annually onto the randomized portion of this study and 50% of these patients provide analyzable tissue for the genomic component of this study. Following a cohort of 325 patients (650*0.50) enrolled over the first year of the study for at least one additional year is expected to provide at least 100 deaths to establish the training set. The actual size of the training cohort may be adjusted depending on the proportion of patients providing analyzable tissue, but the minimum number of events will be fixed. A validation cohort will be derived in a similar fashion as the training cohort. That is, the training and validation cohorts will consist of sequentially enrolled eligible patients with analyzable tissue and individuals will not be permitted to be members of both the training and validation cohorts. Genomic expression: Scanning, imaging, measurement of background and spot intensities will be performed according to the Affymetrix protocol. Preliminary analyses: Preliminary analyses will precede the primary analyses. The goal of the preliminary analyses will be to identify procedures for detecting outliers, normalizing measurements and eliminating systematic errors

88 -82- GOG-0218 Analysis of the training dataset: A previous study (Park, DC, unpublished manuscript) has identified approximately 200 genes in which there is some evidence of an association with overall survival in patients with high-grade, late stage ovarian cancer. The primary purpose of this analysis is to refine this set of genes and then propose a prognostic index based on the refined set of genes. Toward this end, a proportional hazards model relating expression levels of each gene to overall survival and will be fitted to the study data. In order to accommodate multiple testing with potentially correlated markers, multivariate permutation methods will be used to identify those genes that are most likely to be prognostic, while limiting the false discovery rate. 95 Specifically, each of the observed times at risk and the corresponding censoring indicator for an individual in the training set will be randomly assigned to one and only one of the expression profiles in the training set in order to define a permuted dataset. Multiple permuted datasets will be defined in a similar fashion. Then a proportional hazards model stratified on randomized treatment will be fitted to each gene using these permuted data sets. For a specified critical p-value, say =0.01, the average number of false positives,, can be calculated from these permuted datasets. The average false discovery proportion is then / where is the number of genes with p-values less than in the original dataset. The value of can be varied in order to limit the average false discovery proportion to less than a prespecified value,. Conditioning the final analysis on 100 events, setting =0.01 and assuming proportional hazards, this study design provides approximately a 92% chance of detecting a normally distributed marker that truly increases the failure rate 1.5 times when comparing patients whose marker values differ by one standard deviation. This design has an 85-90% chance of detecting a marker that is distributed as a gamma (shape parameter between 1 and 4 and unit variance) and truly increases the failure rate 1.4 times when comparing patients whose marker values differ by one standard deviation. In the event that several potentially prognostic genes are identified, an unsupervised analysis like principal components analysis (PCA) (or cluster analysis) will be used to reduce the dimensionality of the gene expression data. In the case of PCA, a smaller number of variables (principal components) will be identified that captures most of the total variation in expression of the putative prognostic genes. A prediction model will then be built using these principal components and a multivariate proportional hazards model stratified by randomized treatment group. Validation of the prognostic index: Using the coefficients from the principal components analysis and the stratified multivariate proportional hazards model developed using the training dataset, a prognostic index will be computed for each individual in the validation dataset. The independence between the prognostic index and overall survival will be assessed with a proportional hazards model fitted to the validation dataset. Ideally, the proportional hazards coefficient of the prognostic index would be 1.0. However, the degree to which this coefficient is less than 1.0 reflects the degree to which the training data was over-fitted. If the prognostic index is deemed prognostic then the relationship between the index and the time to recurrence or death can be displayed as: martingale residuals plots, Schoenfeld residual plots, ROC curves or Kaplan-Meier plots. 96

89 -83- GOG-0218 Predictive index A true prognostic index can be used to distinguish subgroups of patients who are likely to experience different courses in their disease. A predictive index functions similar to a prognostic index except that its function is treatment specific. For example, Her2/neu expression in breast cancers is considered a predictive marker since it predicts a favorable response to certain agents like Herceptin or tamoxifen. Biomarkers that identify biologic pathways which are necessary to respond to a particular treatment are potentially good predictive markers. The procedure for identifying predictive markers is similar to the procedure for identifying prognostics markers. However, rather than identifying markers that are associated with a particular outcome, we attempt to identify markers that are associated with the outcome only when a specific treatment is applied. In other words, in order to identify predictive genomic profiles, the procedure will focus on expression levels that significantly interact with treatment. Specifically, for each gene the proportional hazards model will include expression level, an indicator for Bevacizumab and a cross product term to express an interaction between these variates. Those genes with a significant treatment interaction term are potential predictive biomarkers to be considered for defining an overall predictive index. The procedures for training, index development and validation are similar to those procedures outlined above for prognostic factors. However, since the statistical power for detecting a predictive marker is considerable less than it is for a prognostic marker the minimum required number of events in the training and validation data will be increased Analyses of biomarkers for Angiogenesis Overview of the study design The translational research objectives of this study are to determine whether one or more biologic markers of angiogenesis are associated with either progression-free or overall survival and to develop a potentially prognostic index and possibly a predictive index. There are several putative biomarkers for angiogenesis including: CD-31, TSP-1, CD105, vascular endothelial growth factor (VEGF), basic fibroblastic growth factor (bfgf), angiogenin, TGF- 1, TIMP-1, TIMP-2, thrombopoietin, and VEGF-D. Some of the markers can be measured in paraffin embedded tissue, serum or both. The complete panel of angiogenic biomarkers to be included in this component of this study has not been identified yet. Some of these biomarker expression levels are assessed quantitatively while others are semi-quantitatively. One objective of this study is to assess whether individual angiogenic biomarkers are prognostic for progression-free or overall survival. While there is no specific hypothesis proposed for this component of the study, a general design for this objective can be described. For a specific biomarker a subsample of those patients who are enrolled, eligible and evaluable will identified. Individuals will be selected independently of their PFS or survival outcome (ie, randomly or sequentially by date of enrollment). Generally, the size of the subsample will be large enough to include at least 100 deaths (or PFS events) reported. However, the actual number of events may be adjusted depending on the particular hypothesis being considered or other practical considerations like laboratory costs. Assuming proportional hazards, and conditioning the analysis on 100 events provides approximately a 92% chance of detecting a normally distributed marker that truly increases the failure rate 1.4 times when comparing individuals whose marker values differ by one standard deviation and the

90 -84- GOG-0218 type I error is limited to 5% for a two tail test. The hypothesis that a patient s biomarker value is independent of their overall survival (or PFS) will be assessed with a proportional hazards model. Potential confounders include: age, presence of clinically measurable of disease, site of primary disease, stage of disease, histologic cell type and grade. If a laboratory investigator proposes to evaluate 2 or 3 biomarkers simultaneously, then the design will depend on the specific study hypothesis. That is, if the biomarkers can be considered independent, then the overall type I error can be controlled for their specific study. For example, using a Bonferroni adjustment to limit the experiment-wide type I error to less than 5% for a study of 2 or 3 biomarkers (with 100 events) would reduce the statistical power for each biomarker to 86% and 78%, respectively. If the biomarkers are not considered independent, then the model and the statistical test will depend on the specific study hypothesis. When several biomarkers are studied simultaneously and there is no specific hypothesis, then a training-validation study design will be considered. This approach is described in more detail below. A second objective of this study is to develop a prognostic index utilizing several biomarker measurements from samples collected prior to study treatment. An additional objective is to develop a predictive index. It is hoped that the predictive index will identify those patients most likely to benefit from the addition of Bevacizumab therapy. The overall approach for this study is to develop a prognostic (and predictive) index by modeling the data from a training data set and then validate the index in an independent data set. The modeling procedures for developing a prognostic index are described in the following paragraphs. The procedures for developing a predictive index are similar. Description of the training and validation data sets The patients registered to this study will be allocated to either a training data set or a validation data set. The training and validation datasets will consist of cohorts of sequentially enrolled eligible and evaluable patients. For practical reasons individuals entering the study early will be assigned to the training data set, while those entering the study later will be allocated to the validation data set. The training data set will be considered sufficiently mature to permit developing a prognostic index from the proposed list of markers when there are at least 100 deaths among the first cohort of sequentially enrolled eligible and evaluable patients. Screening markers of angiogenesis The first step toward evaluating these biomarkers is to assess the distribution of each biomarker and the correlation between pairs of biomarkers. Biomarker values that appear to be extreme outliners will be investigated to determine whether there were any anomalies in the handling or processing of the specimen, which may explain the extreme values. Also, highly correlated biomarkers will be noted since these can introduce anomalies into the modeling procedures. In order to visually assess the univariate relationship between each biomarker value and relative death rates, the marker values will be plotted against martingale residuals from a proportional hazards model that does not include the biomarker as a covariate. The martingale residuals may be smoothed over biomarker values with either piece-wise cubic polynomials, penalized curve fitting or kernel smoothers. These plots will be

91 -85- GOG-0218 used to detect departures from linearity and to assess when a more complex model may be necessary to describe the relationship between biomarker values and log hazard ratio. Cross-validation, bootstrapping or a penalized likelihood function will be used to judge the maximum degree of complexity to be considered. If restricted cubic regression splines are used, then functions with not more than four degrees of freedom should be sufficiently flexible to model most relationships. For those pairs of covariates that are highly correlated, consideration will be given to using functional transformations of these markers. For example, a simple approach would use the sum and the difference highly correlated biomarkers for each individual in a proportional hazards model, since these correspond to the first and second principle components for the pair of biomarkers. A plot of beta residuals can be used to assess the influence of each individual on the estimated log hazard ratio. These plots can be used to identify individuals with an unusually large influence. These individuals will be investigated to determine whether there were any anomalies in the handling or processing of the specimen that may explain the unusual values. Multivariate model Using the functional relationships between the biomarker values and the log relative hazards developed during the biomarker screening step, a multivariate model will be constructed. Biomarkers will be eliminated from the model in order to identify a parsimonious model that appears to have some predicative value but is not overly complex. Cross-validation, bootstrapping or a penalized likelihood function will be used to judge parsimony. Covariate Interactions For the purposes of building a prognostic index, only first-order covariate interactions that have a biologic rationale will be considered. That is, if a particular laboratory assay measures a receptor then modeling interactions with potential ligands will be considered. In this case, it is reasonable to expect that the prognostic value of a ligand may depend on the presence of receptors in the tumor. The evaluation of second-order or higher interactions will not be evaluated at this point but considered in exploratory analyses (see below). Biologically, a second-order interaction could exist when two different ligands compete for the same receptor and one switches on and the other switches off cell growth. A covariate may also interact with time. That is, the effect size may depend on the follow-up time. Schoenfeld residual plots vs time are useful for identifying these types of interactions. Fitting the Schoenfeld residuals over time with either piece-wise cubic polynomials, penalized likelihood regression or kernel smoothers may be used to visualize departures from proportional hazards. Missing values It is anticipated that there will not be a significant number of missing biomarker values. Nevertheless, eliminating individuals due to partially missing biomarker values is not desirable, since this may introduce bias or artificially reduce the variance of the biomarker. Therefore, procedures for handling missing values may be necessary. Provided no more than 10% of the values for a particular biomarker are missing, values imputed from the available values can be used. If missing values account for more than

92 -86- GOG %, but not more than 20% of the measures for a particular biomarker, then conditional imputation, which considers the correlation with other biomarkers, will be considered. Finally, if 20% or more of the values for a particular biomarker are missing, multiple imputations can be used to characterize the additional uncertainty in the parameter estimates due to incorporating imputed values in place of unknown values. External validation A prognostic score for each individual in the validation dataset will be computed using the parameter estimates obtained from modeling of the training data set. This score will be modeled with a proportional hazards model in the validation data set. The coefficient estimated from this later model provides an unbiased measure of the value of the prognostic score. If the validation data set includes 100 deaths (or PFS events), then this sample size provides a 92% chance of correctly classifying an index as predicitive if those individuals whose index values are separated by one standard deviation have an asscociated 1.4-fold increase in their death rate. Exploratory analyses While the procedure for building a prognostic score outlined above has flexibility, it is somewhat constrained in order to avoid over-fitting the data in the training set. Overfitting leads to poor prediction and reduces the external validity of the prognostic score. It is not possible, however, to devise a modeling strategy that consistently produces the best prognostic score. Therefore, exploratory analyses will be performed using alternative model building strategies to identify better prognostic scores. One alternative modeling approach is to first reduce the dimensionality of the data with cluster analyses or principal component analysis. In order to avoid bias, it is important that subsequent data modeling procedures do not incorporate any information from the validation dataset. The prognostic score from subsequent models will be assessed relative to the prognostic score developed from the proposed strategy. The c-index computed in the validation data set will be used to compare alternative prognostic scoring procedures. The c-index is the probability that the survival times from two individuals randomly selected from the validation data set can be correctly ranked based on their prognostic scores. A c-index value of 0.5 indicates that the prognostic index is useless, while a value equal to 1.0 indicates a perfect prognostic index. Occasionally, the c-index is transformed so that D xy = 2*(c-index 0.5). This index ranges from 0 to 1 and it is analogous to the Somers rank correlation index for censored data. Predictive index A prognostic index is used to identify those patients who at greater (or lesser) risk of experiencing a specific clinical event. A predictive index is similar to a prognostic index, but depends it also depends on the prescribed treatment. For example, stage is a prognostic factor for patients with breast caner, and estrogen receptor status is a predictive factor since it identifies those patients who are more or less likely to respond to hormonal treatments. One additional goal for this study is to develop a predictive index from the markers of angiogenesis. That is, the focus of the analysis is to determine those markers that are associated with prognosis provided that the patient received Bevacizumab. Therefore, rather than assess the main association between a biomarker and log relative hazard described previously, the proportional hazards model will assess the interaction between biomarker and Bevacizumab treatment when a predictive index is being developed.

93 -87- GOG BIBLIOGRAPHY 1. McGuire WP, Hoskins WJ, Brady MF, Kucera PR, Partridge EE, Look KY et al. Cyclophosphamide and cisplatin compared with paclitaxel and cisplatin in patients with stage III and stage IV ovarian cancer. N Engl J Med 334:1-6, Piccart MJ, Bertelsen K, James K, Cassidy J, Mangioni C, Simonsen E et al. Randomized intergroup trial of cisplatin-paclitaxel versus cisplatin-cyclophosphamide in women with advanced epithelial ovarian cancer: three-year results. J Natl Cancer Inst 92: , Alberts DS, Green S, Hannigan EV, O'Toole R, Stock-Novack D, Anderson P et al. Improved therapeutic index of carboplatin plus cyclophosphamide versus cisplatin plus cyclophosphamide: final report by the Southwest Oncology Group of a phase III randomized trial in stages III and IV ovarian cancer. J Clin Oncol 10:706-17, du Bois A, Luck HJ, Meier W, Adams HP, Mobus V, Costa S et al. A randomized clinical trial of cisplatin/paclitaxel versus carboplatin/paclitaxel as first-line treatment of ovarian cancer. J Natl Cancer Inst Sep.3;95.(17): :1320, Ozols RF, Bundy BN, Greer BE, Fowler JM, Clarke-Pearson D, Burger RA et al. Phase III trial of carboplatin and paclitaxel compared with cisplatin and paclitaxel in patients with optimally resected stage III ovarian cancer: a Gynecologic Oncology Group study. J Clin Oncol 21: , Swenerton K, Jeffrey J, Stuart G, Roy M, Krepart G, Carmichael J et al. Cisplatincyclophosphamide versus carboplatin-cyclophosphamide in advanced ovarian cancer: a randomized phase III study of the National Cancer Institute of Canada Clinical Trials Group. J Clin Oncol 10:718-26, Jemal A, Tiwari RC, Murray T, Ghafoor A, Samuels A, Ward E et al. Cancer statistics, CA Cancer J Clin 54:8-29, Yoneda J, Kuniyasu H, Crispens MA, Price JE, Bucana CD, Fidler IJ. Expression of angiogenesis-related genes and progression of human ovarian carcinomas in nude mice. J Natl Cancer Inst 90:447-54, Nakanishi Y, Kodama J, Yoshinouchi M, Tokumo K, Kamimura S, Okuda H et al. The expression of vascular endothelial growth factor and transforming growth factor-beta associates with angiogenesis in epithelial ovarian cancer. Int J Gynecol Pathol 16:256-62, Gasparini G, Bonoldi E, Viale G, Verderio P, Boracchi P, Panizzoni GA et al. Prognostic and predictive value of tumour angiogenesis in ovarian carcinomas. Int J Cancer 69:205-11, Hollingsworth HC, Kohn EC, Steinberg SM, Rothenberg ML, Merino MJ. Tumor angiogenesis in advanced stage ovarian carcinoma. Am J Pathol 147:33-41, Paley PJ, Staskus KA, Gebhard K, Mohanraj D, Twiggs LB, Carson LF et al. Vascular endothelial growth factor expression in early stage ovarian carcinoma. Cancer 80:98-106, 1997.

94 -88- GOG Alvarez AA, Krigman HR, Whitaker RS, Dodge RK, Rodriguez GC. The prognostic significance of angiogenesis in epithelial ovarian carcinoma. Clin Cancer Res 5:587-91, Gasparini G. The rationale and future potential of angiogenesis inhibitors in neoplasia. Drugs 58:17-38, van Hinsbergh VW, Collen A, Koolwijk P. Angiogenesis and anti-angiogenesis: perspectives for the treatment of solid tumors. Ann Oncol 10 Suppl 4:60-3, Malonne H, Langer I, Kiss R, Atassi G. Mechanisms of tumor angiogenesis and therapeutic implications: angiogenesis inhibitors. Clin Exp Metastasis 17:1-14, Folkman J. Tumor angiogenesis: therapeutic implications. N Engl J Med 285:1182-6, Kim KJ, Li B, Winer J, Armanini M, Gillett N, Phillips HS et al. Inhibition of vascular endothelial growth factor-induced angiogenesis suppresses tumour growth in vivo. Nature 362:841-4, Luo JC, Toyoda M, Shibuya M. Differential inhibition of fluid accumulation and tumor growth in two mouse ascites tumors by an antivascular endothelial growth factor/permeability factor neutralizing antibody. Cancer Res 58: , Presta LG, Chen H, O'Connor SJ, Chisholm V, Meng YG, Krummen L et al. Humanization of an anti-vascular endothelial growth factor monoclonal antibody for the therapy of solid tumors and other disorders. Cancer Res 57:4593-9, Devore et al. Randomized trial of paclitaxel/carboplatin with/without anti-vegf in non-small cell lung cancer. Proc of the Amer Assoc of Clin Oncol Ref Type: Abstract, Hurwitz H, Fehrenbacher L, Novotny W, Cartwright T, Hainsworth J, Heim W et al. Bevacizumab plus irinotecan, fluorouracil, and leucovorin for metastatic colorectal cancer. N Engl J Med Jun.3;350.(23): ;350: , National Cancer Institute Press Office. NIH Press Release: Bevacizumab Combined With Oxaliplatin-Based Chemotherapy Prolongs Survival for Previously Treated Patients With Advanced Colorectal Cancer Ref Type: Personal Communication rd International Ovarian Cancer Consensus Conference. Ann.Oncol Ref Type: In Press 25. McGuire WP, Hoskins WJ, Brady MF, Kucera PR, Partridge EE, Look KY et al. Cyclophosphamide and cisplatin compared with paclitaxel and cisplatin in patients with stage III and stage IV ovarian cancer. N Engl J Med 334:1-6, Piccart MJ, Bertelsen K, James K, Cassidy J, Mangioni C, Simonsen E et al. Randomized intergroup trial of cisplatin-paclitaxel versus cisplatin-cyclophosphamide in women with advanced epithelial ovarian cancer: three-year results. J Natl Cancer Inst 92: , 2000.

95 -89- GOG Bamias A, Deliveliotis C, Karayiannis A, Varkarakis I, Zervas I, Pantazopoulos D et al. Neoadjuvant chemotherapy with docetaxel and cisplatin in patients with high-risk resectable bladder carcinoma: long term results. Eur Urol 46:344-50, Berkenblit A, Tung N, Kim Y, Feyler H, Niloff J, Berghe KV et al. Phase I trial of docetaxel, carboplatin, and gemcitabine as first-line therapy for patients with high-risk epithelial tumors of mullerian origin. Gynecol Oncol 89:486-93, Caponigro F, Massa E, Manzione L, Rosati G, Biglietto M, De Lucia L et al. Docetaxel and cisplatin in locally advanced or metastatic squamous-cell carcinoma of the head and neck: a phase II study of the Southern Italy Cooperative Oncology Group (SICOG). Ann Oncol 12: , Dieras V, Guastalla JP, Ferrero JM, Cure H, Weber B, Winckel P et al. A multicenter phase II study of cisplatin and docetaxel (Taxotere) in the first-line treatment of advanced ovarian cancer: a GINECO study. Cancer Chemother Pharmacol 53:489-95, Firvida JL, Amenedo M, Rodriguez R, Gonzalez A, Salgado M, Ramos M et al. Docetaxel plus fractionated cisplatin is a safe and active schedule as first-line treatment of patients with advanced non-small cell lung cancer: results of a phase II study. Invest New Drugs 22:481-7, Fossella F, Pereira JR, von Pawel J, Pluzanska A, Gorbounova V, Kaukel E et al. Randomized, multinational, phase III study of docetaxel plus platinum combinations versus vinorelbine plus cisplatin for advanced non-small-cell lung cancer: the TAX 326 study group. J Clin Oncol 21: , Gedlicka C, Formanek M, Selzer E, Burian M, Kornfehl J, Fiebiger W et al. Phase II study with docetaxel and cisplatin in the treatment of recurrent and/or metastatic squamous cell carcinoma of the head and neck. Oncology 63:145-50, Georgoulias V, Ardavanis A, Agelidou A, Agelidou M, Chandrinos V, Tsaroucha E et al. Docetaxel versus docetaxel plus cisplatin as front-line treatment of patients with advanced nonsmall-cell lung cancer: a randomized, multicenter phase III trial. J Clin Oncol 22:2602-9, Jahanzeb M, Sarna G, Hirsch R, Radice P, Koletsky A, Martinez M et al. Docetaxel and carboplatin as first-line therapy in advanced non-small cell lung carcinoma: a phase II study. Anticancer Res 24: , Katayama H, Ueoka H, Kiura K, Tabata M, Kozuki T, Tanimoto M et al. Preoperative concurrent chemoradiotherapy with cisplatin and docetaxel in patients with locally advanced non-small-cell lung cancer. Br J Cancer 90:979-84, Kubota K, Watanabe K, Kunitoh H, Noda K, Ichinose Y, Katakami N et al. Phase III randomized trial of docetaxel plus cisplatin versus vindesine plus cisplatin in patients with stage IV non-small-cell lung cancer: the Japanese Taxotere Lung Cancer Study Group. J Clin Oncol 22:254-61, 2004.

96 -90- GOG Lee SH, Kang WK, Park J, Kim HY, Kim JH, Lee SI et al. Combination chemotherapy with epirubicin, docetaxel and cisplatin (EDP) in metastatic or recurrent, unresectable gastric cancer. Br J Cancer 91:18-22, Markman M, Kennedy A, Webster K, Peterson G, Kulp B, Belinson J. Combination chemotherapy with carboplatin and docetaxel in the treatment of cancers of the ovary and fallopian tube and primary carcinoma of the peritoneum. J Clin Oncol 19:1901-5, Mavroudis D, Alexopoulos A, Malamos N, Ardavanis A, Kandylis C, Stavrinidis E et al. Salvage treatment of metastatic breast cancer with docetaxel and carboplatin. A multicenter phase II trial. Oncology 64:207-12, Mitachi Y, Sakata Y, Ohtsu A, Hyodo I, Katsu K, Sairenji M et al. Docetaxel and cisplatin in patients with advanced or recurrent gastric cancer: a multicenter phase I/II study. Gastric Cancer 5:160-7, Obata H, Aoki Y, Watanabe M, Matsushita H, Yahata T, Fujita K et al. Docetaxel and carboplatin combination chemotherapy for recurrent endometrial cancer. Int J Clin Oncol 8:53-5, Pfisterer J, du BA, Wagner U, Quaas J, Blohmer JU, Wallwiener D et al. Docetaxel and carboplatin as first-line chemotherapy in patients with advanced gynecological tumors. A phase I/II trial of the Arbeitsgemeinschaft Gynakologische Onkologie (AGO-OVAR) Ovarian Cancer Study Group. Gynecol.Oncol 92:949-56, Ramalingam S, Dobbs TW, Einzig AI, Wojtowicz-Praga S, Cascino M, Bonomi P et al. Carboplatin and docetaxel in advanced non-small-cell lung cancer: results of a multicenter phase II study. Cancer Chemother Pharmacol 53:439-44, Roth AD. European experience of docetaxel and cisplatin in advanced gastric cancer. Gastric Cancer 5 Suppl 1:27-9, Saitoh S, Sakata Y. Docetaxel and cisplatin in patients with advanced gastric cancer: results of Japanese phase I/II study. Gastric Cancer 5 Suppl 1:23-6, Sengelov L, Kamby C, Lund B, Engelholm SA. Docetaxel and cisplatin in metastatic urothelial cancer: a phase II study. J Clin Oncol 16:3392-7, Specht L, Larsen SK, Hansen HS. Phase II study of docetaxel and cisplatin in patients with recurrent or disseminated squamous-cell carcinoma of the head and neck. Ann Oncol 11:845-9, Vasey PA, Atkinson R, Coleman R, Crawford M, Cruickshank M, Eggleton P et al. Docetaxelcarboplatin as first line chemotherapy for epithelial ovarian cancer. Br J Cancer 84:170-8, Vasey PA, Jayson GC, Gordon A, Gabra H, Coleman R, Atkinson R et al. Phase III randomized trial of docetaxel-carboplatin versus paclitaxel-carboplatin as first-line chemotherapy for ovarian carcinoma. J Natl Cancer Inst 96: , 2004.

97 -91- GOG Vorobiof DA, Rapoport BL, Chasen MR, Cohen GL, Mahomed R, Karime M. Phase II clinical trial of carboplatin and docetaxel in patients with metastatic ovarian cancer: active combination with low incidence of peripheral neuropathy. Int J Gynecol Cancer 13:287-91, Wachters FM, van Putten JW, Boezen HM, Groen HJ. Phase II study of docetaxel and carboplatin as second-line treatment in NSCLC. Lung Cancer 45:255-62, Zalcberg J, Millward M, Bishop J, McKeage M, Zimet A, Toner G et al. Phase II study of docetaxel and cisplatin in advanced non-small-cell lung cancer. J Clin Oncol 16: , Rose PG, Blessing JA, Ball HG, Hoffman J, Warshal D, DeGeest K et al. A phase II study of docetaxel in paclitaxel-resistant ovarian and peritoneal carcinoma: a Gynecologic Oncology Group study. Gynecol Oncol 88:130-5, Markman M, Liu PY, Wilczynski S, Monk B, Copeland LJ, Alvarez RD et al. Phase III randomized trial of 12 versus 3 months of maintenance paclitaxel in patients with advanced ovarian cancer after complete response to platinum and paclitaxel-based chemotherapy: a Southwest Oncology Group and Gynecologic Oncology Group trial. J Clin Oncol 21:2460-5, Rose PG, Nerenstone SR., Monk BJ, Wenzel LB, and Small J. A Phase III Randomized Study of Cisplatin (NSC #119875) and Taxol (Paclitaxel) (NSC #125973) With Interval Secondary Cytoreduction Versus Cisplatin and Paclitaxel in Patients With Suboptimal Stage III Epithelial Ovarian Carcinoma. Proc of the Amer Assoc of Clinl Oncol 21(1), 201a Ref Type: Abstract 57. Bast RC, Jr., Klug TL, St John E, Jenison E, Niloff JM, Lazarus H et al. A radioimmunoassay using a monoclonal antibody to monitor the course of epithelial ovarian cancer. N Engl J Med 309:883-7, Guppy AE, Rustin GJ. CA125 response: can it replace the traditional response criteria in ovarian cancer? Oncologist 7:437-43, Rustin GJ, Marples M, Nelstrop AE, Mahmoudi M, Meyer T. Use of CA-125 to define progression of ovarian cancer in patients with persistently elevated levels. J Clin Oncol 19:4054-7, Rustin GJ. Use of CA-125 to Assess Response to New Agents in Ovarian Cancer Trials. J Clin Oncol 21:187-93, Rustin GJ, Bast RC, Jr., Kelloff GJ, Barrett JC, Carter SK, Nisen PD et al. Use of CA-125 in clinical trial evaluation of new therapeutic drugs for ovarian cancer. Clin Cancer Res 10: , Rustin GJ, Quinn M, Thigpen T, du BA, Pujade-Lauraine E, Jakobsen A et al. Re: New guidelines to evaluate the response to treatment in solid tumors (ovarian cancer). J Natl Cancer Inst 96:487-8, 2004.

98 -92- GOG Bamberger ES, Perrett CW. Angiogenesis in epithelial ovarian cancer. Mol Pathol 55:348-59, Hata K, Osaki M, Dhar DK, Nakayama K, Fujiwaki R, Ito H et al. Evaluation of the antiangiogenic effect of Taxol in a human epithelial ovarian carcinoma cell line. Cancer Chemother Pharmacol 53:68-74, Le T, Leis A, Pahwa P, Wright K, Ali K, Reeder B et al. Quality of life evaluations in patients with ovarian cancer during chemotherapy treatment. Gynecol Oncol 92:839-44, Bezjak A, Tu D, Bacon M, Osoba D, Zee B, Stuart G et al. Quality of life in ovarian cancer patients: comparison of paclitaxel plus cisplatin, with cyclophosphamide plus cisplatin in a randomized study. J Clin Oncol 22: du BA, Luck HJ, Meier W, Adams HP, Mobus V, Costa S et al. A randomized clinical trial of cisplatin/paclitaxel versus carboplatin/paclitaxel as first-line treatment of ovarian cancer. J Natl Cancer Inst 95:1320-9, Wenzel L, Hurwitz H, Morin MJ, Ross R, and Chan JK. Quality of life comparisons in a phase III randomized study of cisplatin and paclitaxel with or without interval secondary cytoreduction in patients with suboptimal stage III epithelial ovarian carcinoma: a Gynecologic Oncology Group study Ref Type: Personal Communication 69. Basen-Engquist K, Bodurka-Bevers D, Fitzgerald MA, Webster K, Cella D, Hu S et al. Reliability and validity of the functional assessment of cancer therapy-ovarian. J Clin Oncol 19: , Cain JM, Wenzel LB, Monk BJ, Cella D. Palliative care and quality of life considerations in the management of ovarian cancer. In: Gershenson DM, McGuire WP. Ovarian Cancer: Controversies in Management. New York: Churchill Livingstone , Giatromanolaki A, Stathopoulos GP, Koukourakis MI, Rigatos S, Vrettou E, Kittas C et al. Angiogenesis and apoptosis-related protein (p53, bcl-2, and bax) expression versus response of gastric adenocarcinomas to paclitaxel and carboplatin chemotherapy. Am J Clin Oncol 24:222-6, Liu W, Reinmuth N, Stoeltzing O, Parikh AA, Fan F, Ahmad SA et al. Antiangiogenic therapy targeting factors that enhance endothelial cell survival. Semin Oncol 29:96-103, Chen CA, Cheng WF, Lee CN, Wei LH, Chu JS, Hsieh FJ et al. Cytosol vascular endothelial growth factor in endometrial carcinoma: correlation with disease-free survival. Gynecol Oncol 80:207-12, Shen GH, Ghazizadeh M, Kawanami O, Shimizu H, Jin E, Araki T et al. Prognostic significance of vascular endothelial growth factor expression in human ovarian carcinoma. Br J Cancer 83: , 2000.

99 -93- GOG Tokumo K, Kodama J, Seki N, Nakanishi Y, Miyagi Y, Kamimura S et al. Different angiogenic pathways in human cervical cancers. Gynecol.Oncol 68:38-44, Obermair A, Wasicky R, Kaider A, Preyer O, Losch A, Leodolter S et al. Prognostic significance of tumor angiogenesis in epithelial ovarian cancer. Cancer Lett 138:175-82, Alizadeh AA, Eisen MB, Davis RE, Ma C, Lossos IS, Rosenwald A et al. Distinct types of diffuse large B-cell lymphoma identified by gene expression profiling. Nature 403:503-11, Golub TR, Slonim DK, Tamayo P, Huard C, Gaasenbeek M, Mesirov JP et al. Molecular classification of cancer: class discovery and class prediction by gene expression monitoring. Science 286:531-7, Lewohl JM, Wang L, Miles MF, Zhang L, Dodd PR, Harris RA. Gene expression in human alcoholism: microarray analysis of frontal cortex. Alcohol Clin Exp Res 24: , Ginsberg JM, Chang BS, Matarese RA, Garella S. Use of single voided urine samples to estimate quantitative proteinuria. N Engl J Med 309:1543-6, Rodby RA, Rohde RD, Sharon Z, Pohl MA, Bain RP, Lewis EJ. The urine protein to creatinine ratio as a predictor of 24-hour urine protein excretion in type 1 diabetic patients with nephropathy. The Collaborative Study Group. Am J Kidney Dis 26:904-9, Schwab SJ, Christensen RL, Dougherty K, Klahr S. Quantitation of proteinuria by the use of protein-to-creatinine ratios in single urine samples. Arch Intern Med 147:943-4, Steinhauslin F, Wauters JP. Quantitation of proteinuria in kidney transplant patients: accuracy of the urinary protein/creatinine ratio. Clin Nephrol 43:110-5, Wilson DM, Anderson RL. Protein-osmolality ratio for the quantitative assessment of proteinuria from a random urinalysis sample. Am J Clin Pathol 100:419-24, Zelmanovitz T, Gross JL, Oliveira J, de Azevedo MJ. Proteinuria is still useful for the screening and diagnosis of overt diabetic nephropathy. Diabetes Care 21:1076-9, Therasse P, Arbuck SG, Eisenhauer EA, Wanders J, Kaplan RS, Rubinstein L et al. New guidelines to evaluate the response to treatment in solid tumors. European Organization for Research and Treatment of Cancer, National Cancer Institute of the United States, National Cancer Institute of Canada. J Natl Cancer Inst 92:205-16, Dunnett CW, A multiple comparison procedure for comparing several treatments with a control, JASA 50(272): , Basen-Engquist K, Bodurka-Bevers D, Fitzgerald MA, Webster K, Cella D, Hu S, Gershenson DM: Reliability and validity of the Functional Assessment of Cancer Therapy-Ovarian. J Clin Oncol 19 (6), , 2001.

100 -94- GOG Cain JM, Wenzel LB, Monk BJ, Cella D. Palliative care and quality of life considerations in the management of ovarian cancer. In Gershenson DM, McGuire WP. Ovarian Cancer: Controversies in Management New York: Churchill Livingstone, , Cella DF, Tulsky DS, Gray G, Sarafian B, Lloyd S, Linn E, Bonomi A, Silberman M, Yellen SB, Winicour P, Brannon J, Eckberg K, Purl S, Blendowski C, Goodman M, Barnicle M, Stewart I, McHale M, Bonomi P, Kaplan E, Taylor S, Thomas C, Harris J. The Functional Assessment of Cancer Therapy (FACT) Scale: Development and validation of the general measure. J Clin Oncol 11 (3), , Park, DC, Hao, K, Bonome, T, Donninger, H, Brady, J, Randonovish, M, Barrett, JC, Koon, CE, Wong, WH, Lee, JY, Welch, WR, Berkowitz, RS, Mok, SC and Birrer, MJ, Identification of a gene signature that can predict long-term survival in patients with high-grade late stage serous ovarian cancer (unpublished manuscript). 92. Schadt, EE, Li, C, Su, C, and Wong, WH, Analyzing high-density oligonucleotide gene expression array data. J of Cell Biochem 80: , Li, C, and Wong WH, Model-based analysis of oligonucleotide arrays: Expression index computation and outlier detection. PNAS 98:31-36, Durbin, BP, Hardin, JS and Hawkins, DM and Rocke DM, A variance-stabilzation transformation for gene-expression microarray data. Bioinformatics 18(1):S105-S110, Benjamini,Y; Hochberg,Y, Controlling the False Discovery Rate: a Practical and Powerful Approach to Multiple Testing. JR Statist Soc B , JHaegerty, PJ, Lumley, T and Pepe, MS, Time-dependent ROC curves for censored survival data and a diagnostic marker. Biometrics 56, , 2000.

101 PROTOCOL GOG-0218 A PHASE III TRIAL OF CARBOPLATIN AND PACLITAXEL PLUS PLACEBO VERSUS CARBOPLATIN AND PACLITAXEL PLUS CONCURRENT BEVACIZUMAB (NSC #704865, IND #7921) FOLLOWED BY PLACEBO, VERSUS CARBOPLATIN AND PACLITAXEL PLUS CONCURRENT AND EXTENDED BEVACIZUMAB, IN WOMEN WITH NEWLY DIAGNOSED, PREVIOUSLY UNTREATED, STAGE III OR IV, EPITHELIAL OVARIAN, PRIMARY PERITONEAL OR FALLOPIAN TUBE CANCER NCI-SUPPLIED AGENT(S): BEVACIZUMAB/ PLACEBO (NSC #704865, IND #7921) (06/26/06) (08/06/07)(10/14/08) NCI Version Date: 05/24/2010 Includes: Revisions 1-8 POINTS: PER CAPITA 30 MEMBERSHIP - 6 TRANSLATIONAL RESEARCH PER CAPITA Award based on specimen submissions. Distribution: Frozen tumor-3 points, tumor block-2 points (2 nd choice tumor sections and scroll-1 point), frozen serum-0.5 point, frozen plasma-0.5 point (06/26/06), and whole blood-0.5 point. (03/16/09) TRANSLATIONAL RESEARCH MEMBERSHIP - Bonus membership point will be awarded for submission of satisfactory frozen tumor, tumor block, frozen serum and frozen plasma. (06/26/06) STUDY CHAIR STUDY CO-CHAIR NURSE CONTACT (10/14//08) ROBERT A. BURGER, M.D. GINI FLEMING, M.D. HEESUN KIM-SUH. FOX CHASE CANCER CENTER UNIVERSITY OF CHICAGO UNIVERSITY OF OKLAHOMA 333 COTTMAN AVE SECT OF MED/ONC (MC 2115) HEALTH SCIENCE CENTER, OB/GYN PHILADELPHIA, PA S. MARYLAND AVE (RM I-211) P.O. BOX 26901, WILLIAMS CHICAGO IL PAVILION - ROOM WP-2470 (215) (773) OKLAHOMA CITY OK FAX: (( FAX: (773) (405) Robert.a.burger@fccc.edu gfleming@medicine.bsd.uchicago.edu FAX: (405) (03/16/09) heesun-kim@ouhsc.edu DEVELOPMENTAL THERAPEUTICS CO-CHAIR TRANSLATIONAL RESEARCH CO-CHAIR MICHAEL A. BOOKMAN, M.D. MICHAEL BIRRER, M.D., PH.D. SEE GOG WEBSITE DIRECTORY SEE GOG WEBSITE DIRECTORY STATISTICIAN MARK F. BRADY, PH.D SEE GOG WEBSITE DIRECTORY TRANSLATIONAL SCIENTIFIC COLLABORATORS MICHAEL BIRRER, MD, PHD NATIONAL CANCER INSTITUTE ROBERT A. BURGER, MD SEE GOG WEBSITE DIRECTORY JOHN P. FRUEHAUF, MD, PHD SEE GOG WEBSITE DIRECTORY DOUGLAS A. LEVINE, M.D. SEE GOG WEBSITE DIRECTORY QUALITY OF LIFE CO-CHAIR: BRADLEY J. MONK, M.D. SEE GOG WEBSITE DIRECTORY TRANSLATIONAL RESEARCH SCIENTIST KATHLEEN M DARCY, PHD TRANSLATIONAL RESEARCH SCIENTIST SEE GOG WEBSITE DIRECTORY STUDY PATHOLOGIST SHARON LIANG, M.D.,PHD SEE GOG WEBSITE DIRECTORY This protocol was designed and developed by the Gynecologic Oncology Group (GOG). It is intended to be used only in conjunction with institution-specific IRB approval for study entry. No other use or reproduction is authorized by GOG nor does GOG assume any responsibility for unauthorized use of this protocol. OPEN TO PATIENT ENTRY SEPTEMBER 26, 2005 REVISED JANUARY 16, 2006 REVISED JUNE 26, 2006 REVISED AUGUST 6, 2007 REVISED OCTOBER 14, 2008 REVISED MARCH 16, REVISED JUNE 1, 2009 SUSPENDED JUNE 29, CLOSED TO PATIENT ENTRY AUGUST 10, 2009 REVISED APRIL 13, 2010 REVISED JUNE 14, 2010 PAGE 1 OF 2