Efficacy and safety of standard of care with/without bevacizumab for platinum‐resistant ovarian/fallopian tube/peritoneal cancer previously treated with bevacizumab: The Japanese Gynecologic Oncology Group study JGOG3023

Abstract We investigated the efficacy and safety of further bevacizumab therapy in patients with platinum‐resistant ovarian cancer whose disease had progressed after bevacizumab plus chemotherapy. In this multicenter, open‐label, phase II trial (JGOG3023), patients were randomized 1:1 to a single‐agent chemotherapy alone (either pegylated liposomal doxorubicin [40 or 50 mg/m2 administered intravenously], topotecan [1.25 mg/m2 intravenously], paclitaxel [80 mg/m2 intravenously], or gemcitabine [1000 mg/m2 intravenously]) or single‐agent chemotherapy + bevacizumab (15 mg/m2 intravenously). The primary endpoint was investigator‐assessed progression‐free survival (PFS) according to RECIST version 1.1. Secondary endpoints were overall survival (OS), objective response rate (ORR), and response rate according to Gynecological Cancer Intergroup cancer antigen 125 criteria. In total, 103 patients were allocated to chemotherapy (n = 51) or chemotherapy + bevacizumab (n = 52). Median investigator‐assessed PFS was 3.1 and 4.0 mo in each group, respectively (hazard ratio [HR] = 0.54, 95% confidence interval [CI]: 0.32‐0.90, P = .0082). Median OS was 11.3 and 15.3 mo in each group, respectively (HR = 0.67, 95% CI: 0.38‐1.17, P = .1556). Respective ORRs were 13.7% and 25.0% (P = .0599) and response rates were 16.7% and 21.4% (P = .8273). The incidence of grade ≥3 treatment‐related AEs was 42.0% in the chemotherapy group and 54.9% in the chemotherapy + bevacizumab group; AEs were well tolerated, with only 2 and 12 events leading to discontinuation of therapy, respectively. Bevacizumab was effective beyond progressive disease and AEs were manageable. The observed improvement in PFS requires further verification.


| INTRODUC TI ON
Globally, the incidence of ovarian cancer is increasing every year, 1 and because most cases are diagnosed at an advanced stage, these patients tend to have a poor prognosis. The standard of care for ovarian cancer includes surgery plus chemotherapy with platinum and taxane agents. 2 However, 25% of patients at first relapse experience platinum resistance, and almost all patients who experience recurrence or progressive disease subsequently develop platinum-resistant disease. 3 Platinum-resistant recurrent ovarian cancer has a poor prognosis and a short OS of less than 12 mo. 4 There are limited treatment options for these patients, and available treatments do not necessarily prolong survival. The most common treatments for platinum-resistant ovarian cancer are PLD, topotecan, paclitaxel, and gemcitabine (GEM). 5 However, overall response rates are no greater than 15% and the median PFS is only 3-4 mo. 4 Therefore, new treatments are needed for platinumresistant ovarian cancer.
Bevacizumab is a recombinant humanized monoclonal antibody that limits angiogenesis by inhibiting vascular endothelial growth factor. Bevacizumab is currently the standard for both the primary treatment of ovarian cancer 6,7 and the treatment of subsequent relapse. [8][9][10] Previous trials have demonstrated the clinical usefulness of administering bevacizumab beyond progressive disease in locally recurrent or metastatic breast cancer and in advanced or recurrent colorectal cancer. [11][12][13] The efficacy of continuing bevacizumab beyond progressive disease (ie, re-treatment with bevacizumab after progression of disease when previously treated with bevacizumab) was demonstrated in patients with platinum-sensitive recurrent ovarian cancer in a randomized phase III study; the study results showed a PFS of 11.8 mo and 8.8 mo with and without bevacizumab, respectively, with no unexpected toxicity. 14 Currently, however, there are no clinical trials of bevacizumab beyond progressive disease in patients with platinum-resistant recurrent ovarian cancer.
The present study sought to investigate the efficacy and safety of chemotherapy with or without bevacizumab in patients with platinum-resistant recurrent epithelial ovarian, fallopian tube, or primary peritoneal cancer who were previously treated with bevacizumab for front-line or platinum-sensitive ovarian cancer.

| Study design
Details of the study design and rationale have been published previously. 15 In brief, this was an open-label, parallel-arm, randomized, phase II trial conducted at Japanese Gynecologic Oncology Group institutions (JGOG3023).
Patients with platinum-resistant recurrent ovarian cancer and recurrence after bevacizumab therapy in combination with chemotherapy were randomly assigned 1:1 by dynamic randomization (minimization method) to treatment with single-agent chemotherapy (control arm: chemotherapy group) or single-agent chemotherapy plus bevacizumab (experimental arm: chemotherapy + bevacizumab group). Platinum resistance was defined as progression within 6 mo after the completion of at least 3 cycles of platinum-based chemotherapy, including bevacizumab. This definition therefore also includes patients that were characterized with platinum-refractory disease, which was defined as progression within 28 d of completing platinum-based chemotherapy. Patients with a measurable lesion confirmed by diagnostic imaging or with a cancer antigen (CA)125 level more than 2 times higher than normal were considered eligible. overall survival (OS), objective response rate (ORR), and response rate according to Gynecological Cancer Intergroup cancer antigen 125 criteria. In total, 103 patients were allocated to chemotherapy (n = 51) or chemotherapy + bevacizumab (n = 52). Details of dosing schedules, criteria for discontinuation/modification of the chemotherapy regimens, and prohibited/allowed concomitant drugs have been described in detail previously. 15 Briefly, the dosing schedule for each chemotherapy regimen was as follows and each cycle was repeated until disease progression: PLD was administered intravenously at 40 mg/m 2 or 50 mg/m 2 , 1 mg/min on Day 1 with a cycle equal to 28 d; topotecan was administered intravenously at 1.25 mg/m 2 for more than 30 min on Days 1, 2,

| Patients
A full list of the inclusion and exclusion criteria have been described previously. 15 The critical inclusion criteria were as follows: age ≥20 y with histologically confirmed epithelial ovarian, fallopian tube, or primary peritoneal carcinoma; treated with ≥3 cycles of bevacizumab + platinum chemotherapy; progression occurred <6 mo after completion of platinum treatment; ECOG PS of 0-2; and RECIST progression with measurable lesion, or patients with non-measurable disease who could be evaluated based on Gynecological Cancer Intergroup (GCIG) CA125 criteria.
Critical exclusion criteria were as follows: ovarian borderline malignant tumor; history of other clinically active malignancy within 5 y of enrollment; ≥4 previous anticancer regimens; history of bowel obstruction (including sub-occlusive disease, related to the underlying disease, and history of abdominal fistula, gastrointestinal perforation, or intra-abdominal abscess); pelvic examination showing evidence of rectosigmoid involvement, computed tomography showing bowel involvement, or symptoms of bowel obstruction; surgery within 28 d prior to the study start or need for major surgery during study treatment; need of palliative radiotherapy <14 d prior to the study start; current or recent treatment with another investigational drug within 30 d of the first study treatment dosing; known hypersensitivity to any of the study drugs or excipients; pregnancy, lactation, or childbearing potential in women not using highly effective contraception; and judgment by the principal investigator of inappropriateness to participate in the trial.

| Efficacy
The primary endpoint was investigator-assessed PFS (at study

| Safety
The incidence and severity of AEs and treatment-related AEs were evaluated according to the Common Toxicity Criteria for Adverse Events v4.0 JCOG Japanese version (CTCAE v4.0-JCOG). 18

| Statistical analysis
Details of the statistical analysis have been described previously. 15 In summary, the sample size was set to compare PFS, the primary endpoint, and the median PFS of the control group was assumed to be 3 mo. The HR of the treatment group to the control group was expected to be 0.7; the median PFS of the treatment group corresponding to an HR equal to 0.7 was approximately 4.3 mo. The survival time of each group was assumed to follow an exponential distribution. With an allocation ratio to the chemotherapy and chemotherapy + bevacizumab groups set at 1:1, 90 events would be required to ensure a statistical power of ≥80% in the log-rank test with a 1-sided significance level of 20%. As it was anticipated that it | 243 SHOJI et al.
would be difficult to secure a sufficient sample size with a two-sided significance level of .05, we set the 1-sided significance level at .2 to improve detection power. We assumed that the objectives of this randomized phase 2 clinical trial justify this approach. Considering a registration period of 24 mo, which was initially planned, and a minimum PFS observation period of 6 mo, 97 patients would be required to observe 90 events. Therefore, the target number of patients was determined to be 106 (approximately 53 patients per group).
The analysis populations were the ITT set (defined as patients who were registered to this study and randomly allocated to either of the groups), the PPS (defined as patients among the ITT set who received the protocol therapy at least once and had no serious protocol deviation), and the SAS (defined as patients who were registered to this study and received the protocol therapy at least once).
Descriptive statistics were used for baseline demographic and clinical characteristics, with n (%) for categorical variables and mean ± SD for continuous variables. In each group, the median PFS and OS and their respective 95% CI (two-sided) were estimated by the Kaplan-Meier method. Statistical significance between the groups was evaluated by the stratified log-rank test with a significance level of 20% (one-sided) for PFS and 5% (two-sided) for OS. A stratified Cox proportional hazard model was used for estimating the adjusted HR and its 95% CI (two-sided). A multivariable Cox proportional hazard model was constructed using treatment allocation, each factor, and their interactions as explanatory variables to test whether the coefficient of the interaction term was zero. A two-sided significance level of 20% was used. ORR was evaluated in patients with evaluable response according to the RECIST v1.1 criteria. A Cochran-Mantel-Haenszel test was used to assess between-group statistical significance, with a significance level of 5% (two-sided). ORR and its 95% Clopper-Pearson CI (two-sided) were calculated for each group.
A multivariable logistic regression model was used to estimate the adjusted odds ratio between treatment groups. The median number of paracentesis procedures performed and the percentage of patients who underwent paracentesis were determined. Statistical significance between the groups was calculated by the van Elteren test (stratified Wilcoxon rank sum test), with a significance level of 5% (two-sided). The statistical software used for statistical analysis was SAS version 9.3 (SAS Institute Inc, Cary, NC, USA).

| Patient disposition and characteristics
The study was conducted from June 2015 to November 2019. The enrollment period was initially planned for 2 y but was extended to 3.5 y to reach the sample size required; the follow-up period was 6 mo after completion of patient registration. The patient disposition is shown in Figure 1. In total, 103 patients were enrolled and allocated to the chemotherapy group (n = 51) and chemotherapy + bevacizumab group (n = 52); all of these patients were included in the ITT analysis set. Two patients in the chemotherapy + bevacizumab group were discontinued before starting protocol treatment. Two patients were misallocated to the chemotherapy group and received bevacizumab. These patients were included in the chemotherapy group for efficacy analyses and in the chemotherapy + bevacizumab group for safety analyses. One patient allocated to the chemotherapy + bevacizumab group received chemotherapy alone (without bevacizumab).
This patient was included in the chemotherapy + bevacizumab group for efficacy analysis and the chemotherapy group for safety analysis.
The efficacy analyses were done using the ITT set. For the safety analysis, 50 patients (chemotherapy group) and 51 patients (chemotherapy + bevacizumab group) were included in the SAS.
The baseline demographic and clinical characteristics of the study patients are summarized in Table 1. The mean ± SD age of patients was 60.7 ± 12.15 y in the chemotherapy group and 60.3 ± 9.71 y in the chemotherapy + bevacizumab group. Serous carcinoma was the most common histological category in both groups. Twelve patients in each group had a platinum-free interval of zero during the study treatment and use of bevacizumab as a front-line therapy was similar in both groups. Baseline characteristics were well balanced between groups.

| Primary endpoint
The median investigator-assessed PFS (primary endpoint) was

| Safety
A summary of AEs and treatment-related AEs is shown in Table 2.
The overall incidence of AEs was 100% in the chemotherapy group and 98.0% in the chemotherapy + bevacizumab group, and that of treatment-related AEs was 96.0% in the chemotherapy group and 96.1% in the chemotherapy + bevacizumab group. The AEs were generally manageable; only 2 patients in the chemotherapy group and 12 patients in the chemotherapy + bevacizumab group discontinued treatment due to AE.
There were more treatment-related grade ≥3 AEs in the chemotherapy + bevacizumab group compared with the chemotherapy group (54.9% vs 42.0%). One patient in the chemotherapy + bevacizumab group died due to an AE (infection or infestation -other). The patient had a fever of 39°C during Cycle 2 of PLD + bevacizumab therapy and received appropriate inpatient management. Further chemotherapy was cancelled. On Day 28 of the second administration of bevacizumab, the patient experienced a massive melena, which was considered an AE possibly attributable to bevacizumab.
Later, the patient presented with neutrophil count decreased and grade 4 platelet count decreased. Although symptomatic treatment was provided, the patient died of bacteremia. Neutrophil count decreased and platelet count decreased were AEs possibly attributable to PLD; therefore, the attending physician reported that the death was attributable to both bevacizumab and PLD.
The incidence of grade 4 AEs was low and included stroke (n = 1) in the chemotherapy group, and duodenal perforation (n = 1) and F I G U R E 1 Patient disposition. Data are n (%). a The reasons for discontinuation were as follows. In 1 case, the dosing regimen assigned to the patient was not approved. In another case, the patient's protein to creatinine ratio (1.3) was found to be above that established as an eligibility criterion for this study (≤1.0). b Two patients were misallocated to the chemotherapy group and received bevacizumab; these patients were included in the chemotherapy group for efficacy analyses and in the chemotherapy + bevacizumab group for safety analyses. One patient allocated to the chemotherapy + bevacizumab group received chemotherapy alone (without bevacizumab) and was included in the chemotherapy + bevacizumab group for efficacy analysis and the chemotherapy group for safety analysis. BEV, bevacizumab; CT, chemotherapy; ITT, intention to treat

| D ISCUSS I ON
Overall survival is short in patients with recurrent ovarian cancer, and low response rates are observed with current treatments. 4,5 However, the recent QUADRA trial demonstrated the clinical efficacy of niraparib in patients with recurrent ovarian cancer. 19 Although the trial enrolled a broad population that included In comparison, our study differs from the AURELIA trial as it includes patients with progressive platinum-resistant and recurrent cancer who had previously received bevacizumab treatment. As such, we describe the efficacy and safety of chemotherapy + bevacizumab in this population and show that this is a treatment option in which none was previously established. Of note, the safety profile in our study is also consistent with that shown in the AURELIA trial.
PFS was 4 mo in the chemotherapy + bevacizumab group in this study compared with 6.7 mo in the AURELIA trial. 10      included and because of the relatively small size of the study, which was expected, due to the relative rarity of platinum-resistant recurrent ovarian cancer in patients who received prior treatment with bevacizumab. Nonetheless, our data provide information relevant to the management of this disease, and our findings should be confirmed in a larger, more geographically inclusive, phase III study.
The results of this phase II study demonstrated the efficacy and manageable toxicity of continuing bevacizumab beyond progressive disease in patients with platinum-resistant recurrent ovarian cancer previously treated with bevacizumab for front-line or platinum-sensitive recurrent ovarian cancer. However, although an improvement in PFS was observed, further verification is required.