Clinical and Genetic factors affecting Alzheimerís disease progression in subjects on stable acetylcholinesterase inhibitor therapy: a comparison between mechanistic and empirical disease progression modelling approaches
Monica Simeoni1, Michael Gold2, Marina Zvartau-Hind3, Michael Irizarry4, Daren Austin1, Roberto Gomeni5
1Clinical Pharmacology and Discovery Biometrics (Stockley Park, UK), 2Neurosciences Medicine Development Centre (RTP, USA), 3Neurosciences Medicine Development Centre (Stockley Park, UK),4WW Epidemiology (RTP, USA), 5Pharmacometrics (Verona, Italy), GlaxoSmithKline
Objectives: The objectives of the analysis were to compare empirical and mechanistic AD progression models based on subjects on stable Acetylcholinesterase inhibitors (AChEIs) therapy using an individual patient meta-analysis approach and to quantify the impact of MMSE, age, gender, and APOE ε4 status as covariates.
Methods: Longitudinal ADAS-Cog scores collected from 926 subjects with mild to moderate AD from the placebo arms of two 54-week clinical trials were included in the analysis. These studies investigate the effects on cognition and overall clinical response of rosiglitazone as adjunctive therapy to donepezil (REFLECT-2), or to galantamine, rivastigmine and donepezil (REFLECT-3) 
Empirical Model: Two empirical modeling approaches were explored:
ADASij=ADAS0j·e- tijlhj +Kj·tij+eij (4)
where: ADAS0 is ADAS-Cog at baseline, Kel and Keq are the rate constant for the offset and onset rate of the placebo effect, A is the magnitude of the placebo effect, h is the time to response of placebo, ε is the residual error, i and j are the time and subject suffixes. One (Models 1, 2 and 3) characterized the ADAS-Cog longitudinal change from baseline with an additive placebo transient effect [2,3], while the other (Model 4) with a multiplicative placebo effect .
Mechanistic Model: Using a novel Disease System Analysis approach , the loss of cognitive functions (ADAS) can be described by:
where: kin is a time-dependent deterioration of ADAS-cog, pi are parameters characterising the degenerative process, covj is a set of covariates and Kout is the first order constant characterizing the compensatory regulatory response by the homeostatic control systems.
Results: Model-fitting was performed using a population-analysis approach (NONMEM Version VI). Only Models (3), (4) and (5) always successfully converged. Model (5) with a linear time-varying disease progression rate (kin) adequately fitted the data. The inclusion of covariates for K and kin provided a statistically significant improvement in the data fitting.
Conclusions: Among the empirical models, model (4) better described AD progression in placebo-treated patients on stable AChEI therapy. A mechanistic model for AD fits the observed data as precisely as model (4). Baseline MMSE severity, Age, and APOE ε4 genotype were relevant predictors of AD progression. These findings support the mechanistic-modelling approach for AD as the reference for developing and implementing a disease-drug-trial model strategy.
 GSK Clinical Data Register, http://ctr.gsk.co.uk/medicinelist.asp
 Ito K, Ahadieh S, Corrigan B, French J, Fullerton T, Tensfeldt T and AD Working Group. A disease progression meta-analysis model in Alzheimer's disease. Alzheimer's & Dementia (In press: Accepted May 13, 2009)
 Holford NH, Peace KE. Methodologic aspects of a population pharmacodynamic model for cognitive effects in Alzheimer patients treated with tacrine. Proc Natl Acad Sci U S A 1992;89:11466-11470.
 Dua P.,Berges A.,Chen C.,Gomeni R. ADAS-Cog Placebo Modelling in Alzheimer's Disease. Abstracts of the Annual Meeting of the Population Approach Group in Europe, 2009; [http://www.page-meeting.org/default.asp?abstract=1515].
 Post, T.M., Freijer, J.I., DeJongh, J., Danhof, M. Disease system analysis: Basic disease progression models in degenerative disease Pharmaceutical Research 2005; 22-7: 1038-1049.