Abstract
BACKGROUND: Vascular endothelial growth factor inhibitors (VEGFi) are compromised by a lack of validated biomarkers. Previously we showed that changes in the concentration of plasma Tie2 (pTie2) was a response biomarker for bevacizumab. Here, we investigated whether pTie2 can predict response and progression cross-tumour for generic VEGFi treatment. PATIENTS AND METHODS: Patients (n = 124) with advanced biliary tract cancer (ABC) received cisplatin/gemcitabine with cediranib or placebo (ABC-03 trial). Concentrations of pTie2 were measured longitudinally from before treatment until disease progression. Data from patients with ovarian cancer (n = 92, ICON7 trial) and patients with colorectal cancer (CRC) (n = 70, Travastin trial) were also included. RESULTS: Cediranib-treated ABC patients were deconvoluted into distinct groups where in one group pTie2 trajectories resembled those seen in placebo-treated patients and in another pTie2 significantly reduced (t-test P = 2.7 × 10(-14)). Using the 95% confidence interval for these two groups, we defined a vascular complete response (vCR) as a 24% reduction in pTie2 within 9 weeks; vascular no response (vNR) as a 7% increase in pTie2, and a vascular partial response (between these limits). vCR cediranib-treated patients had significantly improved progression-free survival (8.8 versus 7.5 months, restricted mean ratio 0.73, P = 0.012) and overall survival (18.8 versus 12.1 months, hazard ratio 0.49, P = 0.02). By integrating data across ovarian cancer, CRC and ABC, we show that (i) patients with vNR do not benefit from VEGFi and (ii) Tie2-defined vascular progression occurs sufficiently in advance of radiological progressive disease that changes in treatment could be offered to prevent clinical deterioration. CONCLUSION: pTie2 is the first cross-tumour, generic VEGFi, vascular response biomarker to guide optimum use of VEGFi in clinical practice.