Nicolas Frey, Ronald Gieschke, Cornelia Weber, Johann Laurent, Carsten Hofmann, Sylvie Retout
Roche Pharma Research and Early Development, Clinical Pharmacology, Roche Innovation Center Basel, F. Hoffmann-La Roche Ltd, Basel, Switzerland
Objectives: A pre-specified futility analysis of SCarlet RoAD (SR), a Phase 3 of the anti-Aβ antibody gantenerumab (GAN) in prodromal Alzheimer’s disease (AD), estimated a low probability for trial success due to a lack of significant drug effect and dosing was discontinued. Further investigations highlighted an exposure-AD clinical scales relationship in a sub-group of fast progressor patients [1] and a signal of brain amyloid plaque removal activity in a PET sub-study (N=114, 14% of the SR patients). This signal was rather small (-10% PET decrease at 2 years (y) in the highest exposure group), suggesting an under-dosing. At the same time, aducanumab (ADU), a very similar drug investigating higher doses, reported up to -20% PET decrease at 1 y in the highest dose group (N=123), significantly associated with a slowing decline on AD scales [2]. Based on both SR data and ADU PET data, we proposed to build an exposure – PET model to identify new doses / dosing regimens for the GAN program.
Methods: Population pharmacokinetic/pharmacodynamic (PK/PD) analyses were performed on combined GAN and ADU data. The final PKPD dataset included 237 patients and 693 PET observations, collected at baseline, weeks 20, 60 and 100 for GAN and at baseline, weeks 26 and 54 for ADU. Direct and indirect responses were investigated, as well as different PKPD relationships from linear to sigmoid. In addition to goodness of fit tests, visual predictive checks (VPC) of the selected model were performed, per dose and per exposure groups, to test its appropriateness when predicting the GAN and ADU dataset, either pooled or separately. Last, combined with a safety model [3], simulations were conducted to investigate up-titration dosing regimens balancing both PET response (-20% at 2y) and safety criteria.
Results: The exposure-PET relationship was best described by a power model combined with an effect compartment. No bias in the predicted PET values was observed. The model captures both the central tendency and the between subject variability of both GAN and ADU PD in the target AD patients. Using simulations, up-titration dosing regimens balancing plaque removal and safety events were identified.
Conclusions: A robust model of exposure – amyloid brain plaque removal relationship was built using a meta-analysis of both internal and external data. It informed up-titration dosing regimens for higher Phase 3 success likelihood; they are currently investigated in GAN ongoing studies.
References:
[1] Retout, S et al. Abstr PAGE 25, Lisbon, Portugal (2016)
[2] Sevigny, J et al, 67th Annual Meeting of the American Academy of Neurology, Washington DC, US (2015)
[3] Gieschke, R et al. Abstr PAGE 25, Lisbon, Portugal (2016)
Reference: PAGE 25 (2016) Abstr 5795 [www.page-meeting.org/?abstract=5795]
Poster: Drug/Disease modeling - CNS