Ronald Gieschke, Sylvie Retout, Nicolas Frey, Cornelia Weber, Jean Eric Charoin, Johann Laurent, Valérie Cosson, Daniel Serafin, Carsten Hofmann
Roche Pharma Research and Early Development, Clinical Pharmacology, Roche Innovation Center Basel, F. Hoffmann-La Roche Ltd, Basel, Switzerland
Objectives: SCarlet RoAD (SR) was set up as a pivotal Phase 3 study to evaluate the effect of 105 and 225 mg subcutaneous every 4 weeks of anti-amyloid compound gantenerumab (GAN) in prodromal Alzheimer’s disease. Following a futility analysis, dosing was stopped as there was no effect on AD scales at 2 years (N = 312). No new safety signals were observed, especially amyloid-related imaging abnormality-vascular edema (ARIA-E) events were manageable. At the same time, results from a Phase 1b study of aducanumab (ADU), a compound similar to GAN, became available [1]. ADU doses up to 10 mg/kg IV Q4W were explored and showed significant effects on AD scales, but also a large percentage of ARIA-E events in APOE e4 carriers. This suggested that higher doses of GAN may be of value to explore for clinical efficacy and safety. Here we carried out time to event analyses to describe the relationship between drug concentrations and cumulative occurrence of ARIA-E.
Methods: Relationships between drug concentrations and ARIA-E events were modeled with a hazard function first applied to anti-amyloid antibody bapineuzumab [2]. This function is dependent on study time (steadily decreasing the hazard to zero), drug serum concentrations, and APOE e4 carrier-status. Drug serum concentrations resulted from a population PK analysis of SR data. ARIA-E events were judged on MRI images taken every three month. An external validation of the model was performed with the Phase 1b ADU data. Using that model together with a biomarker (PET) model [3], new dose/dosing regimens matching pre-specified criteria of ARIA-E rate and brain amyloid plaque removal were investigated by simulation.
Results: Only hazard parameters related to drug concentrations, EC50 and Emax, were estimated. Other parameters, e.g. baseline hazard for e4 carriers, were fixed to values derived from [2]. Parameters could be estimated although with 83% relative standard error for EC50. Keeping all parameters fixed to estimated values, cumulative ARIA-E events were predicted and compared for the dosing regimens applied in the Phase 1b study of ADU. Graphical analysis revealed no noteworthy deviations. For each e4 carrier status, GAN titration regimens could be worked out balancing ARIA-E cumulative risk and effects on plaque removal (PET).
Conclusions: Quantitative clinical pharmacology analysis provided a model for ARIA-E events currently being used to define titration regimens in ongoing GAN studies.
References:
[1] Sevigny J et al, 67th Annual Meeting of the American Academy of Neurology, Washington DC, US (2015).
[2] Hutmacher M et al, 7th American Conference on Pharmacometrics, Fort Lauderdale, US (2013)
[3] Frey, N et al. Abstr PAGE 25, Lisbon, Portugal (2016)
Reference: PAGE 25 (2016) Abstr 5878 [www.page-meeting.org/?abstract=5878]
Poster: Drug/Disease modeling - CNS