Four RCTs (4,246 participants) were included, sample sizes ranged from 361 to 1,873 participants. Two trials were rated at a low risk of bias for all criteria, apart from allocation concealment. A third trial was rated as being at risk of bias in its blinding. The fourth trial was only available as a conference paper, and was rated unclear for all quality criteria.
Objective response rate: Over the four trials, the pooled odds ratio indicated that combining bevacizumab with conventional chemotherapy significantly increased the objective response rate, compared with chemotherapy alone (OR 2.17, 95% CI 1.51 to 3.10; Ι²=84%). Significant heterogeneity was present. In the subgroup analysis, the addition of bevacizumab was significantly beneficial for both first-line therapy (OR 1.90, 95% CI 1.17 to 3.06; Ι²=90%) and patients with recurrent ovarian cancer (OR 2.77, 95% CI 2.00 to 3.83; Ι²=0).
Progression-free survival: In the four trials, the addition of bevacizumab significantly improved progression-free survival, compared with chemotherapy alone (HR 0.69, 95% CI 0.52 to 0.87; Ι²=92%); significant heterogeneity was present. The subgroup analysis found a significant increase in progression-free survival, with bevacizumab, for both first-line therapy patients (HR 0.83, 95% CI 0.71 to 0.95; Ι²=71%) and patients with recurrent ovarian cancer (HR 0.48, 95% CI 0.41 to 0.56; Ι²=0).
Overall survival: Three trials, with 3,885 participants, reported overall survival. There was no significant difference between participants receiving conventional chemotherapy and those who also received bevacizumab, regardless of patient treatment status.
Adverse events: Across all four trials, compared with controls, patients receiving bevacizumab had a 2.7 times greater risk of a grade two or higher gastrointestinal event (95% CI 1.58 to 4.76; Ι²=0); a 4.6 times greater risk of grade two or higher hypertension (95% CI 3.74 to 5.74; Ι²=84%); a 4.9 times greater risk of grade three or higher proteinuria (95% CI 2.62 to 9.07; Ι²=58%); and a 2.0 times greater risk of arterial thrombosis (95% CI 1.21 to 3.29; Ι²=30%). There was no difference in the risk of venous thromboembolism.
There was no indication of publication bias from the Begg and Egger tests.