Advice
Evidence review: efficacy
During the development of this evidence summary, a search was carried out for published studies using bevacizumab 7.5 mg/kg in combination with paclitaxel and carboplatin for treating ovarian cancer. One open-label randomised controlled trial (RCT) that assessed the efficacy of bevacizumab 7.5 mg/kg for treating ovarian cancer (ICON7) was identified, with the quality of life results also reported separately (Stark et al. 2013). They form the evidence base for this summary. Other RCTs such as Burger et al. (2011) and economic analyses were excluded because they assessed the licensed 15 mg/kg dose of bevacizumab.
ICON7 was a multicentre, randomised, open-label controlled phase III trial that recruited 1528 women who had undergone surgery for epithelial ovarian, fallopian tube or primary peritoneal cancer and who had adequate coagulation values and bone marrow, liver and renal function with no plans for further surgery before disease progression. Women were eligible for the trial if they had either high-risk, early-stage disease (International Federation of Gynaecology and Obstetrics [FIGO] stage I or IIA and clear-cell or grade 3 tumours – maximum 10% of trial population) or advanced disease (FIGO stage IIB to IV). In addition, women needed to have a reasonable level of daily function as measured on the Eastern Cooperative Oncology Group (ECOG) performance status (scale of 0–4 with 0 indicating normal activity and 4 indicating bedridden and possibly needing hospitalisation). To be eligible for the trial, women had to have an ECOG performance status of between 0 and 2. Most women (94%) had a good performance status of 0 or 1. The median age of participants was 57 years. A total of 90% of the women had epithelial ovarian cancer; 9% had high-risk, early-stage disease; 30% were at high risk for progression; 21% had FIGO stage III, IIIA or IIIB disease; 70% had FIGO stage IIIC or IV disease; 69% had serious histological type; and 26% had more than 1.0 cm of residual disease after debulking surgery.
Women were randomised centrally on a 1:1 basis using an interactive phone or web-based system, therefore allocation to treatment was concealed. Randomisation was stratified according to Gynaecological Cancer Intergroup (GCIG) group, the intended time of initiation of post-surgical chemotherapy (4 weeks or less, or more than 4 weeks), cancer stage and residual disease to ensure the groups were well matched in terms of these baseline characteristics.
The trial treatment period lasted up to 54 weeks and women were randomised to 1 of 2 treatment arms:
-
The standard chemotherapy group (n=764): paclitaxel 175 mg/m2 of body surface area and carboplatin at a target area under the curve of 5 or 6 mg/ml per minute every 3 weeks for 6 cycles.
-
The bevacizumab group (n=764): the same regimen as the standard chemotherapy group plus bevacizumab 7.5 mg/kg every 3 weeks for 5 or 6 cycles, followed by continued treatment with bevacizumab alone for an additional 12 cycles or until disease progression (cycle 1 of bevacizumab was omitted to avoid delayed wound healing if chemotherapy was started within 4 weeks of surgery).
More than 90% of women in each group received 6 cycles of chemotherapy and 62% of the women in the bevacizumab group had 18 cycles. A total of 5% of women (n=48 in standard chemotherapy group and n=27 in the bevacizumab group) received additional chemotherapy or bevacizumab before disease progression.
The primary outcome was progression-free survival (calculated from the date of randomisation until the date of disease progression or death, whichever occurred first). Analysis of efficacy for progression-free survival was performed on the intention-to-treat population (all women who had been randomly assigned to treatment) using an un-stratified log-rank test.
Women were assessed for disease progression using the Response Evaluation Criteria in Solid Tumours (RECIST) guidelines, using a combination of clinical assessment, computerised tomography (CT) scan, magnetic resonance imaging (MRI) scan and cancer antigen 125 (CA-125) levels.
Secondary outcomes included overall survival, biological progression-free survival, response to therapy, toxicity and quality of life.
Pre-planned analyses included a log-rank test that stratified for factors used for randomisation, and interaction analyses to explore the difference in the relative size of treatment effects in subgroups according to baseline characteristics, high risk for progression, and stratification factors.
After submission of the primary progression-free survival analysis, regulatory authorities requested an overall survival analysis with at least 365 deaths and an update on progression-free survival using the same data set. Results for both the primary and updated analysis are presented below.
Progression-free survival
Primary analysis: intention-to-treat population
In the primary analysis (data cut-off February 2010), after a median follow-up of 19 months, an improvement in median progression-free survival of 1.7 months was seen in the bevacizumab group compared with the standard chemotherapy group (19 months compared with 17.3 months respectively). The estimated hazard ratio (HR) for progression or death in the bevacizumab group was 0.81 (95% confidence interval [CI] 0.70 to 0.94, p=0.004). The Kaplan‑Meier curves for progression-free survival crossed over demonstrating non-proportional hazards. A test for non-proportional hazards confirmed this (p<0.0001). The conventional hazard ratio is not meaningful when calculated using data that demonstrates non-proportional hazards. To better estimate the effect of bevacizumab on progression-free survival, restricted mean values were estimated. Using all data obtained up to 36 months after randomisation, an improvement in restricted mean progression-free survival of 1.5 months was estimated for the bevacizumab group compared with the standard chemotherapy group (21.8 months compared with 20.3 months respectively).
Primary analysis: subgroup at high risk for progression
The trial reported on a pre-specified subgroup of the women with advanced ovarian cancer who had FIGO stage III and more than 1.0 cm of residual disease after debulking surgery or who had FIGO stage IV disease. They termed these women at 'high risk for progression' (n=465). No other subgroup analyses were presented.
In the primary analysis for this subgroup of women, after a median follow-up period of 19 months, median progression-free survival improved by 5.4 months in the bevacizumab group compared with the standard chemotherapy group (15.9 months compared with10.5 months respectively; HR 0.68, 95% CI 0.55 to 0.85, p<0.001). However, the number of women who had disease progression or who died was similar in both groups (bevacizumab: n=173 [74.9%] compared with standard chemotherapy: n=158 [67.5%]).
Restricted mean progression-free survival estimated using data up to 36 months after randomisation showed an improvement of 3.4 months in the bevacizumab group compared with the standard chemotherapy group (16.5 months compared with 13.1 months respectively; no p value reported).
Updated analysis: intention-to-treat population
An updated analysis was performed (data cut-off November 2010) after a median follow-up period of 28 months. This showed an improvement in median progression-free survival of 2.4 months in the bevacizumab group compared with the standard chemotherapy group (19.8 months compared with 17.4 months respectively; HR 0.87, 95% CI 0.77 to 0.99, p=0.04). However, a similar number of women in each group had either disease progression or died (bevacizumab: n=470 [61.5%] compared with standard chemotherapy: n=464 [60.7%]).
In the updated analysis, restricted mean progression-free survival using data up to 36 months after randomisation showed an improvement of 1.9 months using bevacizumab compared with standard chemotherapy (22.5 months compared with 20.6 months respectively). Restricted mean progression-free survival was also estimated using data up to 42 months after randomisation, which showed a similar improvement of 1.7 months on bevacizumab compared with standard chemotherapy (24.1 months compared with 22.4 months respectively). No statistical tests were reported for these comparisons.
Updated analysis: subgroup at high risk for progression
In the updated analysis, after a median follow-up period of 28 months, median progression-free survival improved by 5.5 months on bevacizumab compared with standard chemotherapy for the subgroup of women at 'high risk for progression' (16.0 months compared with 10.5 months respectively; HR 0.73, 95% CI 0.60 to 0.93, p=0.002). However, the number of women who had disease progression or who died was similar in both groups (bevacizumab: n=190 [82.3%] compared with standard chemotherapy: n=196 [83.8%]).
Updated analysis of restricted mean progression-free survival using data up to 36 months after randomisation for this subgroup of women, showed an improvement of 3.5 months in the bevacizumab group compared with the standard chemotherapy group (17.6 months compared with 14.1 months respectively; no p value reported). A similar level of improvement in restricted mean progression-free survival of 3.6 months was observed using data up to 42 months after randomisation (18.1 months compared with 14.5 months respectively; no p value reported).
Further subgroup analysis was provided in a manufacturer's submission reported in the NICE technology appraisal on bevacizumab in combination with paclitaxel and carboplatin for first-line treatment of advanced ovarian cancer (NICE technology appraisal guidance 284). After an average of 28 months, women with stage III suboptimally debulked disease had a median improvement in progression-free survival of 6.8 months using bevacizumab compared with standard chemotherapy (16.9 months [n=140] compared with 10.1 months [n=154] respectively; HR 0.67, 95% CI 0.52 to 0.87, p value not reported). No statistically significant difference between treatment groups was observed for women with stage III optimally debulked cancer: median increase in progression-free survival of 1.6 months on bevacizumab compared with standard chemotherapy (19.3 months [n=383] compared with 17.7 months [n=368] respectively; HR 0.89, 95% CI 0.74 to 1.07). In addition, no statistically significant difference between treatment groups was observed for women with stage IV cancer: median increase in progression-free survival of 3.4 months on bevacizumab compared with standard chemotherapy (13.5 months [n=104] compared with 10.1 months [n=97] respectively; HR 0.74, 95% CI 0.55 to 1.01).
In the supplementary appendix of ICON7, only the hazard ratio was reported for the subgroup of women with inoperable stage III or IV disease, which showed a statistically significant improvement for those receiving bevacizumab (HR 0.66, 95% CI 0.48 to 0.91, p<0.011).
Overall survival
Primary analysis: intention-to-treat population
In the primary analysis, after a median follow-up of 19 months 241 deaths had occurred (111/764 in the bevacizumab group and 130/764 in the standard chemotherapy group): 231 (96%) were disease related, 5 treatment related and 5 due to other causes. There was no statistically significant difference between the 2 treatment groups for overall survival (HR 0.81, 95% CI 0.63 to 1.04, p=0.098). These results are interim results and should be interpreted with caution.
Updated analysis: intention-to-treat population
In the updated analysis, after a median follow-up of 28 months 378 deaths had occurred (178/764 in the bevacizumab group and 200/764 in the standard chemotherapy group). There was no statistically significant difference between the 2 treatment groups for overall survival (HR 0.85, 95% CI 0.69 to 1.04, p=0.11). However, only half of the pre-specified number of deaths (n=715) that was needed for an 80% power to detect a 23% increase in median overall survival had occurred at this time and so the results of these analyses should be interpreted with caution. The complete overall survival results are expected to be published at the end of 2013.
Updated analysis: subgroup at high risk for progression
Results for the subgroup at 'high risk for progression' (which was defined as women with FIGO stage III and >1.0 cm residual disease after debulking or women with FIGO stage IV) showed a statistically significant improvement in overall survival for the bevacizumab group after a median follow-up of 28 months. There were 79 deaths out of 231 women in the bevacizumab group compared with 109 deaths out of 234 women in the standard chemotherapy group. After a median follow up of 28 months, median survival still favoured bevacizumab with an improvement of 7.8 months compared with standard chemotherapy (36.6 months compared with 28.8 months respectively; HR 0.64, 95% CI 0.48 to 0.85, p=0.002). However, in the NICE technology appraisal on bevacizumab in combination with paclitaxel and carboplatin for first-line treatment of advanced ovarian cancer (NICE technology appraisal guidance 284) it was noted that, 'taking into account the shape of the Kaplan–Meier curve from the interim analysis of the high-risk patients, it is likely that the mean overall survival benefit would be much less than the median'. In addition, only half of the pre-specified number of deaths (n=715) that would be needed for an 80% power to detect a 23% increase in median overall survival had occurred at this time and so the results of these analyses should be interpreted with caution.
Quality of life
Health-related quality of life was measured using the European Organisation for the Research and Treatment of Cancer (EORTC) quality-of-life questionnaire ovarian cancer module (QLQ-OV28) and QLQ-C30 questionnaires. The OV28 contains 28 questions that assess symptoms, chemotherapy side effects, peripheral neuropathy, body image, attitude to disease and treatment, and sexual functioning. The QLQ-C30 contains 30 questions grouped into global health status, 5 function scales and 9 symptom scales.
Participants in the trial completed quality of life questionnaires on their own before the administration of treatment or medical consultation at baseline, before each chemotherapy cycle, then every 6 weeks for the rest of year 1 and every 3 months in year 2 unless women had disease progression. Data are due to be collected at 3 years for all those women who are still alive. The primary outcome was comparison of mean global quality of life in the standard chemotherapy and bevacizumab groups at 54 weeks using the EORTC-QLQ-C30 global quality of life scale on a translated scale from 0 (worst) to 100 (best).
The final results from the analysis of these questionnaires are due at the end of 2013. Preliminary results showed statistically significantly higher mean global quality of life score in the standard chemotherapy group compared with the bevacizumab group at 18 weeks (71.1 [n=594] compared with 66.9 [n=678] respectively, p<0.01) and 54 weeks (74.5 [n=454] compared with 69.5 [n=562] respectively, p<0.01). A quality of life difference of more than 10 points was defined as clinically significant by the study authors and these results fell short of this threshold. There was no statistically significant difference at 76 weeks (standard chemotherapy: 73.7 and bevacizumab: 72.6).
The Stark et al. (2013) study changed the definition for clinical significance of difference in quality of life in March 2011 after the publication of a review (Cocks et al. 2011). For the QLQ-C30, a small clinically important difference was defined as 4 to 7 points and a moderate difference was 10 to 15 points. For the QLQ-OV28, no evidence of a clinically relevant effect size was found and so a difference of 0.3 standard deviations was used. There were discrepancies between the number of questionnaire responses reported in the study by Stark et al. (2013) and the ICON7 trial, which may have affected the accuracy of the results.
After 18 weeks, a small clinically important difference in the mean global quality of life score of 5.1 points was seen favouring standard chemotherapy compared with bevacizumab (64.4 compared with 59.2 respectively; 95% CI 2.9 to 7.4, p<0.0001). Mean global quality of life at 54 weeks in women whose disease had not yet progressed also showed a small clinically important increase for women in the standard chemotherapy group by an average of 6.4 points compared with those in the bevacizumab group (76.1 compared with 69.7 respectively; 95% CI 3.7 to 9.0, p<0.0001). The number of women whose global quality of life score improved by at least 10 points between baseline and 54 weeks was also higher in the standard chemotherapy group compared with the bevacizumab group (66% [221/333] compared with 56% [250/444]; odds ratio [OR] 0.58, 95% CI 0.42 to 0.80, p=0.001).
Exploratory analysis of subscales showed clinically small but statistically significant reductions in quality of life for women in the bevacizumab group for role functioning, financial worries, attitudes to disease or treatment, hormonal symptoms and rash (all p<0.01).
There are no data available yet for quality of life after disease progression.
The reason for the small reduction in quality of life in the bevacizumab group remains unclear, and was not shown to be due to resolution timing of ascites, abdominal wall wound healing, or life disruption with continued chemotherapy.