BACE inhibition causes rapid, regional, and non-progressive volume reduction in Alzheimer's disease brain.

In the phase 3 EPOCH trial (Clinicaltrials.gov; NCT01739348), treatment with the BACE inhibitor verubecestat failed to improve cognition in patients with mild-to-moderate Alzheimer's disease, but was associated with reduced hippocampal volume after 78 weeks as assessed by MRI. The aims of the present exploratory analyses were to: (i) characterize the effect of verubecestat on brain volume by evaluating the time course of volumetric MRI changes for a variety of brain regions; and (ii) understand the mechanism through which verubecestat might cause hippocampal (and other brain region) volume loss by assessing its relationship to measures of amyloid, neurodegeneration, and cognition. Participants were aged 55-85 years with probable Alzheimer's disease dementia and a Mini Mental State Examination score ≥15 and ≤26. MRIs were obtained at baseline and at Weeks 13, 26, 52 and 78 of treatment. MRIs were segmented using Freesurfer and analysed using a tensor-based morphometry method. PET amyloid data were obtained with 18F-flutemetamol (Vizamyl®) at baseline and Week 78. Standardized uptake value ratios were generated with subcortical white matter as a reference region. Neurofilament light chain in the CSF was assessed as a biomarker of neurodegeneration. Compared with placebo, verubecestat showed increased MRI brain volume loss at Week 13 with no evidence of additional loss through Week 78. The verubecestat-related volumetric MRI loss occurred predominantly in amyloid-rich brain regions. Correlations between amyloid burden at baseline and verubecestat-related volumetric MRI reductions were not significant (r = 0.05 to 0.26, P-values > 0.27). There were no significant differences between verubecestat and placebo in changes from baseline in CSF levels of neurofilament light chain at Week 78 (increases of 7.2 and 14.6 pg/ml for verubecestat versus 19.7 pg/ml for placebo, P-values ≥ 0.1). There was a moderate correlation between volumetric MRI changes and cognitive decline in all groups including placebo at Week 78 (e.g. r = -0.45 to -0.55, P < 0.001 for whole brain), but the correlations were smaller at Week 13 and significant only for the verubecestat groups (e.g. r = -0.15 and -0.11, P < 0.04 for whole brain). Our results suggest that the verubecestat-associated MRI brain volume loss is not due to generalized, progressive neurodegeneration, but may be mediated by specific effects on BACE-related amyloid processes.