Gene expression profiling identifies EPHB4 as a potential predictive biomarker in colorectal cancer patients treated with bevacizumab

Irene Guijarro-Muñoz, Antonio Sánchez, Esther Martínez-Martínez, Jose M. García, Clara Salas, Mariano Provencio, Luis Álvarez-Vallina, Laura Sanz

The anti-VEGF monoclonal antibody bevacizumab was approved in 2004 as a first-line treatment for metastatic colorectal cancer (CRC) in combination with chemotherapy and provided proof of principle for antiangiogenic therapy. However, there is no biomarker that can help to select patients who may benefit from bevacizumab in order to improve cost-effectiveness and therapeutic outcomes. The aim of this study was to compare gene expression profiles in CRC patients treated with bevacizumab who responded to the treatment with those that did not respond, in an effort to identify potential predictive biomarkers. RNA isolated from formalin-fixed paraffin-embedded tumor specimens of patients treated with bevacizumab was subjected to gene expression analysis with quantitative RT-PCR arrays profiling 84 genes implicated in the angiogenic process. Data were validated at the protein level using immunohistochemistry. We identified a gene, EPHB4, whose expression was significantly increased in nonresponders (p = 0.048, Mann-Whitney test). Furthermore, high EPHB4 tumor levels were associated with decreased median overall survival (16 months vs 48, Log-rank p = 0.012). This was not observed in a control group of CRC patients treated only with chemotherapy, suggesting that EPHB4 constitutes a potential predictive biomarker and not a mere prognostic one. These data support the notion of a potential synergy between EPHB4-EFNB2 and VEGF-VEGFR pathways, making patients with high EPHB4 expression more resistant to VEGF blocking. Therefore, determination of EPHB4 levels in CRC samples could be useful for the prediction of response to bevacizumab.