This randomized, open-label study used DCE-MRI to investigate the effect of once-daily oral dosing with vandetanib (100 mg or 300 mg) on tumor perfusion and vascular permeability in 22 patients with advanced colorectal cancer and liver metastases. The primary DCE-MRI variables of iAUC60 and Ktrans did not show any statistically significant changes from baseline for either treatment group. Therefore, the study did not support the hypothesis that vandetanib has effects on tumor vasculature, as defined by changes in gadolinium uptake measured by iAUC60 and Ktrans. The safety and pharmacokinetic profiles of vandetanib were similar to those observed in previous phase I studies [15, 16]. Both vandetanib doses were generally well tolerated with no new toxicities reported. A preliminary assessment of efficacy showed no RECIST objective responses in either treatment group, with five patients in the 300 mg group experiencing a best response of stable disease.
There are several possible explanations for the absence of detectable changes in gadolinium uptake and tumor shrinkage with vandetanib in this setting. Although variations in institutional DCE-MRI protocols and different patient populations do not permit direct comparison, studies of other VEGFR-2 tyrosine kinase inhibitors have demonstrated reductions in iAUC/Ktrans in patients with advanced cancer [7–10]. Therefore, one explanation could be that vandetanib is not sufficiently active versus VEGFR-2 at the two doses investigated. However, this seems unlikely given that vandetanib has previously demonstrated single-agent antitumor activity at 100 mg and 300 mg in NSCLC  and in medullary thyroid cancer [18, 19]; the present study also showed some evidence of antitumor effects (though not tumor shrinkage), with five patients in the 300 mg cohort experiencing stable disease. Inhibition of EGFR (NSCLC) and RET (medullary thyroid cancer) tyrosine kinases is also likely to be contributing to the activity of vandetanib in these tumor types; nevertheless, its relatively greater potency versus VEGFR-2 in vitro  suggests that vandetanib should achieve at least comparable inhibition of VEGFR-2 versus EGFR/RET in vivo. Moreover, in the present study, both vandetanib doses achieved steady-state plasma drug levels that were several-fold greater than the IC50 for inhibition of VEGF-dependent proliferation of human umbilical vein endothelial cells (29 ng/mL) . An anti-VEGFR-2 effect of vandetanib at 100 mg and 300 mg is also supported by an exploratory pharmacodynamic study in patients with breast cancer, which showed inhibition of VEGFR-2 phosphorylation in skin biopsy tissue after 28 days of vandetanib treatment .
A second explanation may be that vandetanib is not active against the tumor vasculature in this particular disease setting. Indeed, the antitumor effects of vandetanib in this group of patients with colorectal cancer were modest compared with its single-agent activity in NSCLC  or medullary thyroid cancer [18, 19]. Furthermore, the canonical changes in plasma VEGF and VEGFR-2 that have been observed with vandetanib in NSCLC  and with other VEGFR tyrosine kinase inhibitors across different tumor types [7, 30] were not seen in the present study. In patients with colorectal cancer, objective tumor responses and effects on gadolinium uptake in tumor vasculature have been observed in single-agent studies of cediranib  and vatalanib . Both of these VEGFR tyrosine kinase inhibitors, as well as bevacizumab, have activity versus VEGFR-1 and VEGFR-2 signaling [32, 33]. In contrast, vandetanib is selective for VEGFR-2 versus VEGFR-1 . It is known that colorectal tumor cells express VEGFR-1 and that autocrine signaling may play a role in tumor cell survival/migration . Activity versus VEGFR-1 may therefore be an important contribution to any effects of antiangiogenic agents on both RECIST assessments and gadolinium uptake in colorectal cancer. In this respect, it is interesting that a recent pan-tumor study with CDP791, a high affinity PEGylated di-Fab conjugate that specifically binds VEGFR-2, showed limited efficacy and no effect on Ktrans .
As discussed above, vandetanib has additional activity versus EGFR and the adverse event profile of vandetanib in this and previous studies [17, 36, 37] is consistent with pharmacodynamic inhibition of both VEGFR (hypertension) and EGFR signaling (rash, diarrhea). Combining inhibition of VEGF (bevacizumab) and EGFR (cetuximab) signaling on a background of chemotherapy has been investigated in two recent colorectal cancer studies, which produced different outcomes. The exploratory efficacy results from the BOND-2 study in irinotecan-refractory, bevacizumab- and cetuximab-naïve patients suggested that adding bevacizumab to cetuximab ± irinotecan may be more effective compared with historical controls . However, the first-line CAIRO-2 study found that adding cetuximab to bevacizumab, capecitabine and oxaliplatin resulted in a significantly shorter PFS . The CAIRO-2 authors speculated that these results may be due to a negative interaction between cetuximab and bevacizumab, and noted that the incidence of hypertension, a relatively common side effect of treatment with bevacizumab and other VEGF signaling inhibitors, was significantly reduced in patients receiving cetuximab. These data suggest, at least in some settings, that the vascular effects associated with VEGF inhibition may be diminished with concomitant EGFR inhibition. Other than vandetanib, AEE788 is the only dual VEGFR and EGFR tyrosine kinase inhibitor in clinical development and it is worth noting that AEE788 also showed no effect on gadolinium uptake in patients with advanced colorectal cancer and liver metastases . An additional factor in the present study is that most patients had received previous treatment with bevacizumab and/or cetuximab, which may have affected responsiveness to subsequent VEGFR-2/EGFR tyrosine kinase inhibition. The mechanism of tumor resistance to the monoclonal antibodies bevacizumab and cetuximab is not well understood and warrants further investigation.
It is also possible that vandetanib treatment may induce hemodynamic changes, such as normalization/remodeling of the tumor vasculature as hypothesized by Jain , that would not necessarily be detected by estimating changes in Ktrans and iAUC60. More complex DCE-MRI approaches such as the St Lawrence and Lee model , which is able to derive independent measurement of blood flow, blood volume and permeability surface area, may be more appropriate for detecting complex changes in tumor vascularity and hemodynamics. Normalization of the tumor vasculature might also be expected to improve tumor oxygenation and blood flow. In this regard, the results from the exploratory assessment of T2* using intrinsic susceptibility MRI merit discussion. Changes in T2* can be used to monitor changes in deoxyhemoglobin and an increase in T2* could result from improved tumor oxygenation and blood flow (i.e., normalization) . However, T2* is influenced by other factors and is therefore a difficult parameter to interpret on its own [25, 43]. In the absence of detectable effects on tumor hemodynamics as measured by DCE-MRI, an increase in T2* could be attributed to an increase in tumor cell death [43, 44]. As such, the significant increase in T2* at vandetanib 300 mg compared with 100 mg in the present study may reflect increased tumor necrosis at the higher dose. Further correlative work is needed to understand the biological basis of changes in T2* in the clinical setting.
Population pharmacokinetic-pharmacodynamic analyses showed no correlation between vandetanib exposure and any of the pharmacodynamic parameters analyzed. Given the long half-life of vandetanib, it may take up to 4 weeks for vandetanib to reach steady state ; in the present study, steady state was attained from day 15 at the earliest, but was mostly from day 22 onwards. It is not fully understood how tumor growth/adaptation during this prolonged period of drug accumulation may affect pharmacodynamic variables.