Potential Bevacizumab Biomarker Identified in Ovarian Cancer

A retrospective analysis of the phase III GOG-0218 trial has identified a potential biomarker of response for bevacizumab in patients with advanced ovarian cancer.

Michael J. Birrer, MD, PhD

A retrospective analysis of the phase III GOG-0218 trial has identified a potential biomarker associated with improvements in progression-free survival (PFS) and overall survival (OS) for patients with advanced ovarian cancer treated with bevacizumab (Avastin) plus chemotherapy, according to findings presented by Michael J. Birrer, MD, PhD, at the 2015 ASCO Annual Meeting.1

In the analysis, the median OS with bevacizumab was 45.6 months compared with 35.9 months with placebo in patients with CD31 high expression (HR = 0.68; 95% CI, 0.52-0.89; P = .0155). In the CD31 low group, the hazard ratio was 1.13. Additionally, high tumor VEGF-A levels suggested a benefit with bevacizumab, although this was not statistically significant (P = .0647).

"This retrospective analysis suggest a positive correlation between high levels of molecular and cellular targets of anti-VEGF and the magnitude of clinical benefit from frontline bevacizumab in epithelial ovarian cancer," Birrer, director of Medical Gynecologic Oncology at the Massachusetts General Hospital, said during his presentation. "The effect of bevacizumab on PFS and OS appeared most pronounced in patients with high CD31."

In the phase III GOG-0218 trial, 1873 patients with untreated stage III/IV ovarian cancer were treated with carboplatin (AUC 6) and paclitaxel (175 mg/m2) with placebo (n = 625), bevacizumab-initiation from cycle 2 through 7 (n = 625), or bevacizumab-continuation starting at cycle 2 and continuing throughout the study (n = 623). Bevacizumab was administered at 15 mg/kg every 3 weeks.

An initial analysis of the GOG-0218 study was published in The New England Journal of Medicine in 2011.2 At this analysis, the median PFS with bevacizumab-continuation was 14.1 months compared with 10.3 months with chemotherapy alone (HR = 0.717; 95% CI, 0.625-0.824; P <.001). The bevacizumab-initiation arm did not demonstrate a statistical difference compared with placebo (HR = 0.908; P = .16). A significant difference in OS was not observed between the arms.

"GOG-0218 did not demonstrated an overall survival difference in any of the arms, but we wanted to see if analyzing the individual biomarkers could tease out a patient population for which a particular treatment improved overall survival," Birrer noted.

Tumor samples were available for 1455 patients treated in the study: 483 from the placebo arm, 475 in the bevacizumab-initiation arm, and 497 in the bevacizumab-continuation group. Outcomes in these patients were similar to those in the full intent-to-treat population, Birrer noted.

The analysis looked specifically at the placebo and the bevacizumab-continuation arms for biomarkers of response, specifically CD31, tumor VEGF-A, VEGFR-2, neuropilin-1, and MET. All testing was conducted by a central laboratory in Belgium, using research-based immunohistochemistry assays on randomly selected sections to account for tumor heterogeneity.

For patients with high CD31 expression, the median PFS was 19.9 months with bevacizumab versus 9.8 months with placebo (HR = 0.40; 95% CI, 0.29-0.54; P = .0025). In the CD31 low group, the median PFS with bevacizumab was 18.0 versus 13.4 months with placebo (HR = 0.80; 95% CI, 0.59-1.07). Other biomarkers were not associated with a significant impact on PFS.

Higher levels of CD31 expression were associated with greater improvements in OS and PFS in patients treated with bevacizumab. In patients with the highest CD31 expression, the median OS was 45.6 months with bevacizumab compared with 34.1 months with placebo (HR = 0.57; 95% CI, 0.39-0.83; P = .0084).

Patients with the highest level of VEGF-A expression experienced a median OS that was not yet reached in the bevacizumab arm compared with 40.6 months with placebo (HR = 0.62; 95% CI, 0.43-0.91; P = .0241).

"There is a direct linear relationship between the microvascular density and the impact of bevacizumab," noted Birrer. "VEGF-A is more complex. We should be analyzing third quartile data for this particular biomarker."

A biomarker for response to bevacizumab has been a long sought after goal. Prior studies had explored plasma-based biomarkers for the use of bevacizumab in the GOG-0218 study; however, they were met with negative findings. In ovarian cancer, the findings suggest CD31 could be predictive of outcomes, although these results require further validation, Birrer cautioned.

"CD31 findings are consistent with smaller samples from two trials in colorectal cancer but contrast with observations in glioblastoma. So, perhaps the predictive effect of CD31 should not be extrapolated across indications," explained Birrer. "The potential predictive value of CD31 microvascular density for PFS and OS benefit and certainly for tumor-associated VEGF-A for overall survival requires further validation in other relevant datasets."

Bevacizumab was first approved in 2004, and is indicated as a treatment for patients with metastatic colorectal cancer, metastatic non-squamous non-small cell lung cancer, persistent, recurrent, or metastatic cervical cancer, recurrent glioblastoma, and metastatic renal cell carcinoma.

In November 2014, the FDA approved bevacizumab in combination with chemotherapy for patients with platinum-resistant recurrent ovarian cancer. This decision was based on a 62% improvement in PFS experienced by patients treated with the combination in the phase III AURELIA trial.3

In the study, the median PFS with bevacizumab was 6.8 versus 3.4 months with chemotherapy alone (HR = 0.38; 95% CI, 0.30-0.49; P <.0001). The objective response rate was 28% versus 13%, with bevacizumab and chemotherapy, respectively. In the bevacizumab arm the median OS was 16.6 versus 13.3 months with chemotherapy (HR=0.89; CI 95%, 0.69-1.14); however, this advantage was not statistically significant.


  1. Birrer MJ, Choi YJ, Brady MF, et al. Retrospective analysis of candidate predictive tumor biomarkers (BMs) for efficacy in the GOG-0218 trial evaluating front-line carboplatin—paclitaxel (CP) with or without bevacizumab (BEV) for epithelial ovarian cancer (EOC). J Clin Oncol. 2015;33 (suppl; abstr 5505).
  2. Burger RA, Brady MF, Bookman MA, et al. Incorporation of Bevacizumab in the Primary Treatment of Ovarian Cancer. N Engl J Med. 2011; 365:2473-2483.
  3. Pujade-Lauraine E, Hilpert F, Weber B, et al. Bevacizumab combined with chemotherapy for platinum-resistant recurrent ovarian cancer: The AURELIA open-label randomized phase III trial. J Clin Oncol. 2014; 32(35):4025