In Silico Safety Pharmacology on Intersubject Variability Population of Models: A Regression Model Approach

Abstract

Safety pharmacology aims at detecting undesirable effect of drugs during its development. However, limitations are present at both in-vitro and in-silico level because of its low detection efficacy during this process. In this work, the effect of drugs at tissue level was studied and inducibility in a multivariable scenario including 127 models tested for two different tissue sizes (basal and dilated) and two conditions (no drug and isoproterenol) was obtained. From these models, maintenance duration (MD) of the reentry was calculated and a regression model based on Canonical Correlation Analysis (CCA) was implemented to evaluate the proarrhythmic effect of isoproterenol depending on model size. The number of models with AF maintenance was larger for dilated atria and isoproterenol. CCA analysis obtained 96% accuracy on an arrhythmogenicity test set for basal size and 100% on the dilated one. A new promising methodology was proposed for safety pharmacology including variability between patients, setting the base for personalized medicine.