Comparison of Various Nonclinical In Vivo/In Vitro Models to Assess CNS Penetration of TAK-285, a Novel HER2/EGFR Inhibitor

TAK-285 is an orally available novel HER2/EGFR inhibitor that is currently in clinical development for HER2 over-expressing (HER2+) breast cancer.  Since approximately one third of patients with HER2+ metastatic breast cancer develop brain metastasis, treatment options for patients whose brain metastases have progressed after radiotherapy are urgently needed.   The currently approved HER2-targeted agents, trastuzumab and lapatinib, have generally shown limited activity against brain metastases, likely due to the limited brain penetration of these drugs. In this work, we used and compared various nonclinical tools to assess the CNS penetration potential of TAK-285.  In a Caco-2 cell model, TAK-285 showed high permeability and was not a substrate for P-gp and BCRP.  In comparison, and consistent with literature, lapatinib showed low permeability and were found to be substrates for P-gp/BCRP.   The brain penetration of TAK-285 and lapatinib was assessed in vivo in Sprague-Dawley rats.  The brain-to-plasma AUC ratio (B/P) for total (free plus bound) TAK-285 and lapatinib was 0.202 and 0.0243, respectively, indicating distribution of TAK-285 into brain tissue in contrast to negligible CNS penetration of lapatinib across an intact BBB (B/P being close to the blood volume fraction of brain).  Additional nonclinical studies were performed to assess the unbound TAK-285 concentration in rat brain. A tissue binding study was conducted using equilibrium dialysis with rat brain homogenates and the results, taken together with the total B/P ratio, suggested that there was virtually no unbound TAK-285 available in the brain tissue.  To assess the potential relevance of this finding, a definitive in vivo rat microdialysis study was performed with dual probes to measure unbound TAK-285 in the brain interstitial fluid (ISF) and in blood simultaneously.  The results showed that the exposure of unbound TAK-285 in brain ISF was about 24% of that in plasma indicating that the brain tissue binding method significantly underestimated unbound TAK-285 brain concentrations.  To enable subsequent evaluations of CNS distribution of TAK-285 in humans, recognizing that brain microdialysis is not feasible for use in the clinical setting, a rat CSF pharmacokinetic study was conducted to collect and measure the unbound CSF concentration of TAK-285.  Consistent with the microdialysis findings and contrary to the tissue binding results, a substantial amount of unbound TAK-285 (approximately 13% of that in plasma) was found in the CSF.  The generally similar CSF and brain ISF distributional characteristics of TAK-285 support use of unbound CSF exposure in the clinical setting as a surrogate of brain ISF exposure. A clinical study was recently conducted with CSF collection from patients at steady-state conditions after multiple dose oral administration of TAK-285.  The human CSF results were in good accordance with the expectations from the rat CSF study and indicated the lack of substantial barriers to CNS penetration of TAK-285 in humans.