Abstract

FR.13.15

Sample size calculation for interventional trials in geographic atrophy due to AMD

Adrion C.¹, Fleckenstein M.², Bindewald-Wittich A.², Schmitz-Valckenberg S.², Göbel A. P.², Kampe P.¹, Scholl H. P. N.², Holz F. G.², Mansmann U.^1
1Institute for Medical Informatics, Biometry and Epidemiology, LMU, Munich; ²Department of Ophthalmology, University of Bonn

Objective: While effective therapeutic modalities are now available for neovascular AMD, there is yet no treatment for geographic atrophy (GA). Based on extensive natural history data from the FAM study we performed sample size calculations and addressed relevant factors for interventional clinical trials in GA aiming to test strategies to prevent progression of atrophic patches.
Methods: In the FAM study, longitudinal data of 224 patients are available; based on these data, modelling of the natural history of GA progression was performed by a hierarchical linear random effects model. Sample size calculation was performed for different study designs.
Results: The mean progression rate of GA was 1.75 mm²/year, 95% CI [1.49; 2.02]; the interindividual variation of the progression rate was 1.3 mm²/year, 95% CI [0.84; 2.01] while the intraindividual variation of the progression rate in between two eyes within a single patient was 0.55 mm²/year, 95% CI [0.33; 0.93]; the measuring error was 0.76 mm²/year, 95% CI [0.65; 0.89]. The relevant effect for an intervention was defined as a reduction of the progression rate of 0.5 mm²/year. The overall variability can be calculated for different study designs. In order to detect a relevant effect at a significance level of alpha <0.05 with a power of 80%, the following sample sizes are needed for a parallel group design: (1) Comparison of the progression rate while only one eye of a patient is included: n=178; (2) Comparison of the progression rate while in 60% two eyes of a patient are included: n=126; (3) Genetic validation study for the gene variant G+ (prevalence p=0.1), that implicates a faster progression of GA (0.55 mm²/year was assumed): n=350.
Conclusions: By modelling the natural history of GA progression, four important values for sample size calculation can be quantified: the effect, the measuring error, intraindividual, and interindividual variability. Prospective, interventional clinical trials are feasible with a reasonable sample size.

Copyright: Porstmann Kongresse GmbH. All rights reserved. Impressum, Disclaimer.