A Model-Based Approach to Bridging Plasma and Dried Blood Spot Concentration Data for Phase 3 Verubecestat Trials

In-clinic venous dried blood spot (DBS) pharmacokinetic (PK) sampling was incorporated into two phase 3 studies of verubecestat for Alzheimer's disease (EPOCH [NCT01739348] and APECS [NCT01953601]), as a potential alternative to plasma PK sampling. Initially, plasma and DBS PK samples were collected concurrently to better understand the DBS-plasma verubecestat concentration relationship, with the intention of discontinuing DBS or plasma sampling following interim analysis. Following initial analyses and comparison of results with prespecified selection criteria, plasma PK sampling was discontinued; however, a stability issue resulting in generally lower DBS verubecestat concentrations with longer collection-to-assay times was subsequently discovered (associated with non-compliance in DBS sample handling), prompting reintroduction of plasma sampling. To enable inclusion of DBS data in population PK analyses, a conversion algorithm for calculating plasma-equivalent concentrations (accounting for DBS sample instability) was developed using paired (time-matched) plasma and DBS data from the EPOCH study. Verubecestat population PK models developed from pooled phase 1/1b and EPOCH data using either (1) plasma-only data or (2) plasma and plasma-equivalent concentrations (calculated from non-paired DBS samples) yielded similar results. The algorithm robustness was demonstrated using DBS data from paired samples from the APECS study and comparison between plasma and plasma-equivalent concentrations. The population PK model was updated with APECS data (both plasma and, if no plasma sample available, plasma equivalents). The results demonstrated similar PK in the two phase 3 populations and exposures consistent with expectations from phase 1 data. This case study illustrates challenges with employing new sampling techniques in large, global trials and describes lessons learned.