P2‐005: Evidence from Model‐Based Assessments of Compounds Tested in Completed Clinical Trials Indicates That The Amyloid Hypothesis Has Not Been Adequately Tested

It remains an open question whether the failed efficacy trials for the gamma secretase inhibitors (semagacestat and avagacestat) and, at most, modest findings for antibodies targeting soluble Aβ (solanezumab) demonstrate that the amyloid hypothesis has been disproven. A model-based analysis was conducted to characterize the degree of modulation of Aβ tested in these trials to inform this question. A semi-mechanistic mathematical pharmacokinetic/pharmacodynamic model describing inhibition of brain Aβ production and distribution to CSF previously developed to describe CSF Aβ biomarker data from the BACE inhibitor verubecestat (MK-8931) Phase 1 program was extended to semagacestat, avagacestat, and solanezumab. Clinical data on CSF Aβ levels and drug exposure over time were digitized from the public domain. The model-estimated median steady-state % reduction (averaged over the dosing interval) in brain Aβ40 was 29.5%, 14.8%, 70.8%, and 86.8% for daily administered 50 mg avagacestat, 140 mg semagacestat, 12 mg and 40 mg verubecestat. Thus the two gamma secretase inhibitor trials likely tested a degree of Aβ reduction that was less than that associated with human genomic evidence (∼50% Aβ increase with early onset due to trisomy 21 and 40% Aβ reduction with protective BACE1 mutant). Given the genomic alterations are present at birth, it is likely that a therapeutic intervention would require a greater level of perturbation to be effective. Alterations in free and total Aβ in plasma and CSF with solanezumab were well described by a model with slowed clearance of the Aβ-mAb complex and suggested limited CNS penetration and modulation of CSF Aβ. Overall, these results indicate that the completed trials did not robustly test the amyloid hypothesis. Ongoing verubecestat trials of 12 and 40 mg are testing a substantially greater reduction in brain production of Aβ and therefore will constitute a more robust test of the amyloid hypothesis.