Longitudinal head-to-head comparison of 11 C-PiB and 18 F-florbetapir PET in a Phase 2/3 clinical trial of anti-amyloid-β monoclonal antibodies in dominantly inherited Alzheimer’s disease

Purpose Pittsburgh Compound-B ( 11 C-PiB) and 18 F-florbetapir are amyloid-β (Aβ) positron emission tomography (PET) radiotracers that have been used as endpoints in Alzheimer’s disease (AD) clinical trials to evaluate the efficacy of anti-Aβ monoclonal antibodies. However, comparing drug effects between and within trials may become complicated if different Aβ radiotracers were used. To study the consequences of using different Aβ radiotracers to measure Aβ clearance, we performed a head-to-head comparison of 11 C-PiB and 18 F-florbetapir in a Phase 2/3 clinical trial of anti-Aβ monoclonal antibodies. Methods Sixty-six mutation-positive participants enrolled in the gantenerumab and placebo arms of the first Dominantly Inherited Alzheimer Network Trials Unit clinical trial (DIAN-TU-001) underwent both 11 C-PiB and 18 F-florbetapir PET imaging at baseline and during at least one follow-up visit. For each PET scan, regional standardized uptake value ratios (SUVRs), regional Centiloids, a global cortical SUVR, and a global cortical Centiloid value were calculated. Longitudinal changes in SUVRs and Centiloids were estimated using linear mixed models. Differences in longitudinal change between PET radiotracers and between drug arms were estimated using paired and Welch two sample t -tests, respectively. Simulated clinical trials were conducted to evaluate the consequences of some research sites using 11 C-PiB while other sites use 18 F-florbetapir for Aβ PET imaging. Results In the placebo arm, the absolute rate of longitudinal change measured by global cortical 11 C-PiB SUVRs did not differ from that of global cortical 18 F-florbetapir SUVRs. In the gantenerumab arm, global cortical 11 C-PiB SUVRs decreased more rapidly than global cortical 18 F-florbetapir SUVRs. Drug effects were statistically significant across both Aβ radiotracers. In contrast, the rates of longitudinal change measured in global cortical Centiloids did not differ between Aβ radiotracers in either the placebo or gantenerumab arms, and drug effects remained statistically significant. Regional analyses largely recapitulated these global cortical analyses. Across simulated clinical trials, type I error was higher in trials where both Aβ radiotracers were used versus trials where only one Aβ radiotracer was used. Power was lower in trials where 18 F-florbetapir was primarily used versus trials where 11 C-PiB was primarily used. Conclusion Gantenerumab treatment induces longitudinal changes in Aβ PET, and the absolute rates of these longitudinal changes differ significantly between Aβ radiotracers. These differences were not seen in the placebo arm, suggesting that Aβ-clearing treatments may pose unique challenges when attempting to compare longitudinal results across different Aβ radiotracers. Our results suggest converting Aβ PET SUVR measurements to Centiloids (both globally and regionally) can harmonize these differences without losing sensitivity to drug effects. Nonetheless, until consensus is achieved on how to harmonize drug effects across radiotracers, and since using multiple radiotracers in the same trial may increase type I error, multisite studies should consider potential variability due to different radiotracers when interpreting Aβ PET biomarker data and, if feasible, use a single radiotracer for the best results. Trial registration ClinicalTrials.gov NCT01760005. Registered 31 December 2012. Retrospectively registered.