INIFTA   05425
INSTITUTO DE INVESTIGACIONES FISICO-QUIMICAS TEORICAS Y APLICADAS
Unidad Ejecutora - UE
artículos
Título:
On Application of Carhart Atom Pairs to Predict Anticonvulsant Activity
Autor/es:
A. TALEVI; J.J. PRIETO; L.E. BRUNO BLANCH; EDUARDO ALBERTO CASTRO
Revista:
Am J Biochem Biotechnol
Referencias:
Año: 2006 vol. 2 p. 119 - 128
Resumen:
Motivation. About 50 million people in the world suffer from epilepsy, especially in childhood, adolescence and old age. Available treatment fails to control epilepsy in up to 30 % of affected people. In developing countries, however, the amount of patients that do not receive adequate treatment climbs up to 75 %. Moreover, the new generation of antiepileptic drugs (AEDs) causes important central and peripheral side effects, including ataxia, diplopia, dizziness, headache, nausea, allergies and sedation. Method. A mathematical model previously developed by Bagchi and Maiti, involving Carhart atom pairs and similarity measures, is applied in the prediction of anticonvulsant activity of two sets of compounds which have shown to be active in the Maximal Electroshock Seizure (MES) test, meaning that their mechanism of action can be at least partially explained through sodium channels blockade. The first set of compounds is defined by nine structurally heterogeneous molecules, with Carhart similarities to carbamazepine ranging from 0.005 to 0.593. The second set is defined by four benzodiazepines derivatives with Carhart similarities to THIQ-10c ranging from 0.533 to 0.570. A new, more specific, model is constructed based on the one from Bagchi and Maiti and a pharmacophore previously identified in our laboratory through an active analog approach. Results. Applied to both sets of compounds, our model shows smaller average percentage error and average absolute error in the prediction than the one form Bagchi and Maiti, and smaller SD as well. Accuracy and precision in the prediction also increases compared to those obtained when using bare similarity coefficients as relative activity indicators. Conclusions. Results seem to indicate the proposed model has an improved predictive capability. This model may be useful as an optimization tool, and in the search for new leaders as well.