AN INTEGRATED MECHANISTIC AMYLOID PATHWAY MODEL INFORMED BY CLINICAL BIOMARKER DATA FROM VERUBECESTAT AND OTHER BACE INHIBITORS

Previous amyloid models have largely focused on empirically describing soluble CSF biomarkers from clinical programs or have integrated the biological steps of amyloid in theoretical models informed by literature values and/or animal data. In this analysis, the BACE inhibitor program at Merck was leveraged to fully inform an integrated mechanistic amyloid pathway model based on clinical data alone. Serial CSF biomarker data (Aβ40, Aβ42, sAPPβ, Aβ oligomers (AβO), infused 13C leucine) collected in healthy volunteers or Alzheimers (AD) patients via lumbar catherization in 3 Phase I studies for verubecestat (MK-8931) and one Phase I study each for MK-8511 and MK-8277 were integrated into a single dataset. A mechanistic model describing the Aβ pathways in brain from precursor APP pool to monomeric forms and to aggregate forms was developed, including the transit to CSF (Figure 1). BACE inhibition was added as EMAX term acting on the initial cleavage step of APP. Plaque formation was added to the model as a slow build-up process from AβO. The model provided a good simultaneous description of all study results, needing only to account for different baseline levels in the studies as well as separate IC50 parameters for each compound (4.3, 9.7, 11.9 nM for MK-8931, MK-8277, and MK-8511). This suggests that the model captures the underlying system of pathways well. No significant differences in IC50 or EMAX between AD patients and healthy subjects were found in the analysis, and the extent of the inhibition (EMAX) approached 100 % (98 %). Oligomeric Aβ equilibrated rapidly with monomers, while plaque turnover was slow.