Development and experimental validation of a computational model ensemble to assist the search for new drugs for the treatment of refractory epilepsy

MELISA GANTNER; ROXANA PERONI; JUAN FRANCISCO MORALES; MARÍA LUISA VILLALBA; MARÍA ESPERANZA RUIZ; LUIS BRUNO-BLANCH; ALAN TALEVI

Cancún

Congreso; IX Congreso Latinoamericano de Epilepsia; 2016

ILAE (International League Against Epilepsy) - IBE (International Bureau for Epilepsy)

The transporter hypothesis holds that refractory epilepsy is a result of the local overexpression and hyperactivity of ABC (ATP-binding cassette) transporters such as P-glycoprotein (Pgp) and Breast Cancer Resistance Protein (BCRP) in the blood-brain barrier and/or the epileptic foci [1-2]. Early recognition of BCRP substrates is thus essential to find novel therapeutics for the treatment of refractory epilepsy and other central nervous system conditions linked to BCRP-mediated multidrug resistance issues. We present the development and experimental validation of an ensemble of nonlinear computational models capable of discriminating between BCRP substrates and non-substrates. Methodology: We generated a data set of 262 substrates and non-substrates of the human wild type BCRP. We applied J48 inducing decision tree algorithm and data fusion schemes to obtain the corresponding model ensemble. The everted rat intestinal sacs assay was used to evaluate 5 drugs with pre-clinical anticonvulsant activity classified as non-substrate by our ensemble. Results: According to the computational validation the average ranking of the best 12 decision tree models is the ensemble with the best capacity to discriminate between BCRP substrates and non-substrates. The experimental validation evidence demonstrates that the 5 drugs are BCRP non-substrates. Conclusion: The ensemble constitute a potentially valuable tool to be used as in silico filter to assist the search for new drugs for the treatment of pathologies with BCRP-mediated multidrug resistance issues like refractory epilepsy.