The window of normalization in murine models is relatively short and occurs soon after administration of bevacizumab compared with that in humans [29], which may explain several paradoxical findings reported recently and suggest that the schedule and dosing of combination therapies warrant considerable attention. models were established with H157, H460 and A549 with primary resistance to erlotinib and treated with erlotinib Roy-Bz plus bevacizumab or each agent alone. Erlotinib concentrations in tumors were determined by high-performance liquid chromatography. Results Roy-Bz Bevacizumab alone did not inhibit NSCLC cell growth in vitro. In primarily erlotinib-resistant NSCLC cells, the levels of VEGF protein were highest in H157 cell followed in order by H460 and A549 cells. In vivo, bevacizumab alone significantly inhibited tumor growth only in xenograft models with high (H157) and/or moderate (H460) levels of VEGF protein. A combination of erlotinib and bevacizumab partially reversed resistance to erlotinib in H157 xenografts (high VEGF level) with increasing intratumoral erlotinib concentrations, but not in KLF5 H460 (moderate) or A549 (low) xenografts. Conclusions These results support that combined with anti-VEGF therapy could enhance antitumor activity of anti-EGFR therapy and/or partially reverse resistance to EGFR TKI, by increasing EGFR TKI concentration in specific tumors that express high levels of VEGF protein. Electronic supplementary material The online version of this article (doi:10.1007/s00280-014-2610-x) contains supplementary material, which is available to authorized users. and are tumor length and width, respectively. Tumor growth inhibition (TGI, ?%) formula is (TuGcontrol?TuGtest)/TuGcontrol??100?%, where TuG?=?final tumor size-pretreatment tumor size. Determination of intratumoral erlotinib concentration by HPLC Erlotinib levels in homogenized tumor tissues were determined by reverse-phase high-performance liquid chromatography (HPLC) with UV detection at 345?nm. Separation was achieved on a Waters Symmetry C18 column (150??4.6?mm, 5.0?m; Waters, Milford, MA) preceded by the use of a Symmetry C18 Guard column (3.9??20?mm). The mobile phase was 50?mM potassium phosphate buffer (pH 4.8) containing 0.2?% triethylamine and acetonitrile (60:40, v/v), with 1.0?mL/min flow rate at 25?C. Sample pretreatment involved mixing 500?L of tumor tissue homogenate with 80?L of internal standard (70?g/mL of midazolam in methanol) and 5?mL of tert-butyl methyl ether for 10?min. After centrifugation (650?g, 10?min, 4?C), the organic top layer was transferred to a clean tube and dried under nitrogen gas at 37?C. The residue was dissolved in 250?L of mobile phase. The solution was centrifuged (4,000?g, 30?min) and the supernatant was passed through a microporous membrane filter (Millex-GV 0.22-m filters, Millipore Corp., Bedford, MA). Insoluble materials were removed by filtration, and the filtrate was analyzed by high-performance liquid chromatography. The calibration curves were linear over a concentration range of 20C4,000?ng/mL (test and/or MannCWhitney test were used for comparison of two groups and one-way analysis of variance (ANOVA) test was for more than three groups. erlotinib, bevacizumab Next, we applied bevacizumab alone or plus erlotinib to the NSCLC cells in vitro. As reported previously [21], bevacizumab alone did not inhibit the growth of the tested NSCLC cells in vitro (Fig.?1c). Growth inhibition with bevacizumab (10?ng/mL) plus erlotinib (1?mol/L) was similar to that with erlotinib alone (1?mol/L; Fig.?1d) in the six NSCLC cells (test was used to compare tumor volume at the last measurement between the groups (T/C): **test was used to compare bevacizumab Roy-Bz and vehicle treatment in each model: *erlotinib, bevacizumab We also examined the levels of human VEGF protein in tumor tissues. Consistent with the previous observations in vitro, the level of VEGF protein in the H157 tumor tissue was highest, followed in order by H460 Roy-Bz and A549 tumor tissues (erlotinib, bevacizumab Concentration of erlotinib in tumor tissues of xenograft models Roy-Bz Previous studies have shown that bevacizumab can enhance drug delivery to tumors [15]; however, this remains controversial [16, 17]. According to a previous study [27], the erlotinib concentration in mouse tumors reaches its peak concentration within 1?h after p.o. administration and declines rapidly for the next 6?h. Therefore, we excised tumor samples in athymic mice 1?h after administrating erlotinib p.o. on the last day of treatment and observed the changes in intratumoral erlotinib concentration. Erlotinib concentrations in the H157, H460 and A549 tumor tissues treated with erlotinib alone or plus bevacizumab reached 3.98??0.65 and 7.61??1.28?g/g (test was used to compare erlotinib and combination.