OBJECTIVETo establish subcutaneous xenograft models of gastric cancer in nude mice and to screen the predictive biomarkers of bevacizumab effectiveness.

METHODSSubcutaneous xenograft models were established using BGC823 gastric cancer cell line in 20 male 4-week old BALB/C-nu/nu nude mice and were randomly divided into four groups, bevacizumab group(15 mg/kg), 5-FU group(15 mg/kg), combined group and control group, with 5 mice in each group. Bevacizumab and 5-FU were administered intraperitoneally every other day for three weeks. After treatment, tumor size and inhibition rate were calculated. Expression of CD31 was examined by immunohistochemistry for evaluation of microvascular density(MVD). Levels of human vascular endothelial growth factor(VEGF), basic fibroblast growth factor (bFGF), placental growth factor (PIGF) and interleukin 8(IL-8) were tested by enzyme linked immunosorbent assay(ELISA).

RESULTSCompared to the control group, bevacizumab group and combined group had a significantly lower MVD(5.2±1.0 and 4.3±1.2 vs. 13.8±1.6, P<0.05), a smaller tumor volume [(305.6±184.1) mm(3) and (242.2±71.4) mm(3) vs.(1535.2±625.1) mm(3), P<0.05], and lower levels of VEGF and IL-8 in tumor tissues [VEGF:(351.6±84.1) ng/L and (242.2±71.4) ng/L vs. (1256.7±702.1) ng/L, P<0.05); IL-8:(20 903±1485) ng/L and (27 489±6772) ng/L vs. (57 032±2437) ng/L, P<0.05]. The above parameters were not significantly different between 5-FU group and control group(all P>0.05). Levels of bFGF and IGF were not significantly different among four groups as well(all P>0.05).

CONCLUSIONVEGF and IL-8 may be used to be biomarkers candidates to predict bevacizumab effectiveness on human gastric cancer.