Clinical gamma-secretase inhibitors (GSIs) alter beta-amyloid 40/42 ratio in brain and show CSF beta-amyloid rebound in vivo

To provide a detailed PK/PD in vivo assessment of gamma secretase inhibition in the search for clues to cognitive worsening in clinical trials. Wild type mice (129/sve) and tg2576 transgenic mice were assessed in acute multi-dose time course studies with LY-450139, BMS-708163, and GSI-953 examining A b lowering in brain, plasma, and CSF compartments in order to provide a PK/PD understanding of GSI inhibition. In brain A b levels were assessed in guanidine extracts with immunoassays configured for x-40 and x-42, while in plasma and CSF only the x-40 assay was used. In wt mouse, the efficacy at lowering brain A b x-40 always exceeded that of A b x-42, which resulted in a maximum drop in the ratio of A b 40/42 of 1.7, 2.5, and 2.95 for BMS-708163, LY-450139, and GSI-953, respectively. In tg2576 mouse, a similar phenomenon was observed in a head to head comparison of LY-450139 and BMS-708163 with a maximum ratio drop of A b 40/42 of 2.1 and 2.2, respectively. In addition to observing the plasma re-bound phenomenon, which is well documented, we also observed a similar phenomenon with the CSF compartment. In wild type mouse, LY-450139 showed a significant increase in A b x-40 at 130% of control. GSI-953 showed a non-significant trend increase of 128% of control, BMS-708163 did not display CSF re-bound. In our analysis of GS inhibition using an in vitro membrane assay we do not see an alteration in A b 40/42 ratio. Our data demonstrates that treatment of clinically relevant GSI's, alter the ratio of A b 40/42 in vivo, in both wild type and tg2576 mice, a phenomenon which is also observed with AD causing presenilin mutations. Further work is needed to evaluate this apparent disconnect between in vitro and in vivo A b alterations and how this may relate to the apparent cognitive worsening seen in the clinic with these compounds.