Predicting blood-brain barrier penetration of drugs by polar molecular surface area and molecular volume
Abstract
Aim: To predict the blood-brain barrier penetration by polar molecular surface area and molecular volume.
Methods: Polar molecular surface area and molecular volume are calculated by Monte Carlo method from the lowest energy conformation obtained using the semiempirical self-consistent field molecular orbital calculation AM1 method. The stepwise multiple regression analysis is used to derive the correlation equations between the ratios of the steady-state concentrations of the training compounds in the brain to in the blood (logBB)and their structural parameters.
Results: For a training set of 56 compounds, logBB values are well correlated with the sums of surface areas of oxygen and nitrogen atoms (SO,N, A2, excluding the nitrogen atoms in nitrogen molecule or in nitro) and molecular volumes (V, A3). The regression equation is logBB = -1.331 x 10(-5)V2 + 9.228 x 10(-3)V -0.02439 SO,N -0.4318 (n = 56, r = 0.9043). The calculated logBB values of a test set of 10 compounds from the model agree well with their experimental logBB values.
Conclusion: The model is simple and effective. It can be used to predict the logBB values of candidate molecule in drug design.
Keywords:
Methods: Polar molecular surface area and molecular volume are calculated by Monte Carlo method from the lowest energy conformation obtained using the semiempirical self-consistent field molecular orbital calculation AM1 method. The stepwise multiple regression analysis is used to derive the correlation equations between the ratios of the steady-state concentrations of the training compounds in the brain to in the blood (logBB)and their structural parameters.
Results: For a training set of 56 compounds, logBB values are well correlated with the sums of surface areas of oxygen and nitrogen atoms (SO,N, A2, excluding the nitrogen atoms in nitrogen molecule or in nitro) and molecular volumes (V, A3). The regression equation is logBB = -1.331 x 10(-5)V2 + 9.228 x 10(-3)V -0.02439 SO,N -0.4318 (n = 56, r = 0.9043). The calculated logBB values of a test set of 10 compounds from the model agree well with their experimental logBB values.
Conclusion: The model is simple and effective. It can be used to predict the logBB values of candidate molecule in drug design.