From structure to strategy: chemometric modeling for the prediction of terminal half-life of pharmaceuticals and its role in future therapeutics.

Abstract

The terminal half-life ( $t_{1/2}$ ) is a crucial pharmacokinetic parameter for estimating the dose regimen and duration of action of a drug. Previously, few research papers have been published on the pharmacokinetic parameters that correlate with the chemical structure of pharmaceuticals, but these are time-consuming and costly. The main goal of the current study is to generate a quantitative read-across structure-activity relationship (q-RASAR) for terminal half-life estimation of diverse pharmaceuticals. The dataset of 895 pharmaceuticals has been used for 2D descriptor computation and model development. Herein, the combinatorial (q-RASAR) approach of read-across and QSAR has been employed for model generation. Finally, the Partial Least Squares-based q-RASAR model is developed and validated based on the various validation parameters as per the OECD principles. The final q-RASAR model is statistically more significant, reliable, and robust than the corresponding QSAR model based on different statistical parameters (R² = 0.617, Q²_(Loo) = 0.601, error-based predictions = 0.221) and external parameters (Q²_F1 & Q²_F2 are 0.635). It has been concluded that the presence of the RA function and the presence of 6-membered rings are accountable for the long terminal half-life. Similarly, the presence of the phenol/enol/carboxyl OH group, the presence of positively charged N, solubility, and average molecular weight contribute negatively to the terminal half-life. Additionally, the DrugBank database was screened and predicted the terminal half-life of new and untested pharmaceuticals using the model, which further helped in the prediction of the dosing frequency and accumulation profile of new pharmaceuticals. This study further helps to formulate and optimize safe and eco-friendly pharmaceuticals.