In silico model of drug permeability across sublingual mucosa

The objective of this work was to develop an in silico model to predict the sublingual permeability of a drug based on physicochemical descriptors of a molecule. Fourteen model drugs with diverse physicochemical properties were selected for this study. Molecular volume, molecular weight, log P, log D (pH 6.8), pKa, total polar surface area, hydrogen bond acceptors and donors (HBD), number of rotatable bonds, solubility (pH 6.8), and melting point were used as molecular descriptors. Apparent permeability coefficients (Pe) of drugs across porcine sublingual mucosa were determined experimentally. Multiple linear regression (MLR) was used to develop the model with permeability as the response variable and various descriptors as the predictive variables. Q2, the cross-validated correlation coefficient, was used to assess the prediction ability of the model. MLR analysis showed that HBD and log D were the significant descriptors (P < 0.05, Q2 = 0.88) in the sublingual permeability model. The resulting model is expressed as the following equation: log   P e = − 5.08 − 0.24 ⋅ HBD + 0.53 ⋅ log   D An excellent fit with R2 of 0.93 was obtained between experimental and predicted permeabilities. The analysis of contributions of molecular descriptors to sublingual permeability revealed the molecular structure basis of permeation across sublingual mucosa. In conclusion, an in silico model was developed to predict sublingual permeability of drugs using known descriptors for evaluating the feasibility of sublingual drug delivery.