control group was 58 days, whereas the mice receiving the RTK inhibitors gefitinib and the RTK combination demonstrated a significantly improved survival (p = 0.0001) with median survival of 94 and 90 days respectively (Figure 5A). Mice treated with sunitnib alone had a median survival of 63 days and was not significantly different from the control group (p = 0.13). The results also demonstrate that addition of sunitinib to gefitinib had no significant impact on median survival of mice treated with gefitinib alone (p = 0.18). The in vivo study was also repeated in a different genetic background using a rat syngeneic gliosarcoma model. 9L gliosarcoma 1 mm3 tumor pieces were implanted intracranially in 32 Fisher 344 rats. The rats were divided into four groups of 8 each and were gavaged with gefitinib alone, sunitinib alone, gefitinib and sunitinib combination and PBS. The animals were treated every Monday, Wednesday and Friday as described earlier and the doses were adjusted for the rats as described earlier. Surprisingly, in this study none of the drugs including gefitinib and the gefitinib and sunitib combination showed any efficacy (Figure 5B). These results demonstrate that the outcome of drug efficacy testing in animal models is very much influenced by the underlying genetic backbone of the tumor cell line, and very likely as well the ability for drugs to reach intracranial tumors.
Figure 5. A: Mice implanted intracranially with 020913 GBM oncosphere cells were treated with gefitinib (75 mg/kg), sunitinib (15 mg/kg) and a combination of gefitinib (75 mg/ kg) and sunitinib (15 mg/kg). Gefitinib alone could significantly (p = 0.0001) improve survival in the animals compared to control animals. Sunitinib did not show any efficacy either when used alone (p = 0.13, compared to control) or when combined with gefitinib (p = 0.18, compared to gefitinib alone). Figure 5B: Rats were implanted intracranially with 9L tumor 1 mm3 tumor pieces. Rats were then treated with gefitinib (50 mg/kg), sunitinib (8 mg/kg) and a combination of gefitinib (50 mg/kg) and sunitinib (8 mg/kg). None of the drugs including the combination of gefitinib and sunitinib showed any efficacy (p = 0.9). doi:10.1371/journal.pone.0044372.g005
to single agent therapy, and test this combination in preclinical animal models of glioblastoma. Monotherapy of RTK targeting agents have been largely ineffective and there is enough in vitro experimental evidence to support the use of combination therapy targeting multiple tyrosine kinases [7,8,9,10,11]. We first identified possible effective RTK combination using in vitro cell proliferation studies. Next we planned to test efficacy in improved preclinical animal models at FDA approved doses to try and mimic what might be achievable in a clinical trial. In this study, gefitinib and sunitinib was the best in vitro combination, based on its ability to reduce proliferation and kill GBM oncospheres. The pattern of effective inhibitor combinations suggests that successful simultaneous inhibition of EGFR and PDGFR and other tyrosine kinases was necessary. In spite of the in vitro prediction, our results in vivo differed quite significantly. We did achieve a survival benefit in animals, but evidence indicated this was only for the gefitinib, and only in the cell line with EGFR amplification, where there is existing data to suggest that a single EGFR inhibitor might have a modest survival benefit in those tumors most dependent on EGFR signaling [12].
The results of the in vivo efficacy studies demonstrate that gefitinib alone could improve survival in 020913 GBM xenograft models by 62% compared to untreated controls, whereas the same drug was a completely ineffective when tested at similar concentrations in a syngeneic 9L rat gliosarcoma model. The differences in the results could be attributed to the genetic makeup of the cells. 020913 cells are human GBM derived neurosphere line that has always been propagated in a serum free media supplemented with EGF and FGF [13]. It is possible that the cell culture conditions would select the cells that are more dependent on EGF and FGF for their growth. Moreover, 020913 cells have EGFR amplification and therefore these cells would be more responsive towards EGFR inhibitors such as gefitinib. On the contrary, 9L cells are grown in serum containing medium and have no specific dependence on EGF for growth and may not be inhibited by mere EGFR inhibition. Sunitinib was the only RTK inhibitor that induced apoptosis in GBM oncosphere cells (Figure 3), whereas all the other RTK inhibitors were cytostatic. However, sunitinib failed to demonstrate any efficacy in our preclinical GBM animal models affirms the recently published phase II clinical trial data demonstrating limited efficacy of sunitinib in GBM patients [14]. The most likely explanation for this is that sunitinib cannot reach effective intracranial tumor concentrations at these doses. Sunitinib failed to work in vivo in 9L cells that are significantly more sensitive to sunitinib in vitro (9L cells have at least 5?0 times lower IC50 for sunitinib compared to GBM oncosphere lines, Table S1). Our observations are consistent with a previous report demonstrating responsiveness of GBM patients co-expressing EGFRvIII and PTEN to EGFR inhibitors [12]. Guillamo et al. demonstrated that loss of PTEN makes GBM xenografts resistant to gefitinib in an ex vivo brain slice model [15]. Similarly, EGFR copy number has been shown to be a predictor of gefitinib related survival benefit in advanced non small cell lung cancer patients [16]. It is also likely that other EGFR mutations, such as point mutations within EGFR or PIK3CA or PTEN and the activation of PI3K/AKT pathway need to be considered when considering response to EGFR inhibitors [17]. Clinical trial evaluating erlotinib, another EGFR inhibitor had insufficient activity in GBM patients and no clear biomarker could be identified that was associated with a response to erlotinib [18]. However, a study by Sarkaria et al. [19] identified two GBM xenografts that were sensitive to erlotinib. Sarkaria et al. concluded that amplification or mutation of EGFR and presence of wild type PTEN was required but not enough for ensuring sensitivity to erlotinib. One other key difference between our study and the one by Sarkaria et al. and Van den Bent et al. is the choice of inhibitors. Although, both gefitinib and erlotinib target EGFR, they are different small molecules and have a differing pharmacokinetic profile. Gefitinib can cross blood brain barrier effectively and dephosphorylate EGFR [20], whereas erlotinib is a substrate for p-glycoprotein and breast cancer resistance protein and therefore has a limited brain penetration [21]. Immunoblotting analysis of 020913 cells when treated with different RTK inhibitors in vitro suggests that to achieve complete growth inhibition of GBM oncosphere lines all of the three pathways involving AKT, STAT3 and MAPK that are downstream of RTKs should be inhibited. Activation of any one of these pathways may enable the oncosphere cells to develop a resistance mechanism and regrow- or simply continue growth. Gefitinib or sunitinib or the other RTK inhibitors when used as single agents could only inhibit p-AKT and not p-STAT3 (Figure 4) and the cells could regain their growth potential immediately after the drug was withdrawn (Figure 3). These observations suggests that while devising a rational
