PURPOSE: Bevacizumab improves survival for patients with metastatic colorectal cancer with chemotherapy, but no proven predictive markers exist. The VEGF-A splice form, VEGF(165)b, anti-angiogenic in animal models, binds bevacizumab. We tested the hypothesis that prolonged progression-free survival (PFS) would occur only in patients with low relative VEGF(165)b levels treated with bevacizumab.EXPERIMENTAL DESIGN: Blinded tumor samples from the phase III trial of FOLFOX4 ± bevacizumab were assessed for VEGF(165)b and VEGF(total) by immunohistochemistry and scored relative to normal tissue. A predictive index (PI) was derived from the ratio of VEGF(165)b:VEGF(total) for 44 samples from patients treated with FOLFOX + bevacizumab (arm A) and 53 samples from patients treated with FOLFOX4 (arm B), and PFS, and overall survival (OS) analyzed on the basis of PI relative to median ratio.RESULTS: Unadjusted analysis of PFS showed significantly better outcome for individuals with VEGF(165)b:VEGF(total) ratio scores below median treated with FOLFOX4 + bevacizumab compared with FOLFOX4 alone (median, 8.0 vs. 5.2 months; P <0.02), but no effect of bevacizumab on PFS in patients with VEGF(165)b:VEGF(total) ratio >median (5.9 vs. 6.3 months). These findings held after adjustment for other clinical and demographic features. OS was increased in arm A (median, 13.6 months) compared with arm B (10.6 months) in the low VEGF(165)b group, but this did not reach statistical significance. There was no difference in the high VEGF(165)b:VEGF(total) group between FOLFOX + bevacizumab (10.8 months) and FOLFOX alone (11.3months).CONCLUSION: Low VEGF(165)b:VEGF(total) ratio may be a predictive marker for bevacizumab in metastatic colorectal cancer, and individuals with high relative levels may not benefit. Clin Cancer Res; 1-8. ©2012 AACR.