Regression Modeling Strategy for Prediction of AUC of Evogliptin, a Novel Dipeptidyl Peptidase IV Inhibitor in Humans, Using Single dose PK Data

Abstract

Aim: To develop a limited regression model of evogliptin for prediction of AUC data for internal (within study) and external studies. Method: Regression analyses (linear/power/polynomial) were performed in multitiered approach using paired peak plasma concentration (Cmax) versus AUC data of evogliptin. For all models, correlation co-efficient (r) and root mean square error (%RMSE) were used in predicting internal/external data. Bland–Altman analysis was performed for all the models. Results: Limited power model showed highest predictability (r = >0.98 and ≤15.5% RMSE), followed by linear model (r = >0.98 and ≤20.5% RMSE) and polynomial (r = >0.96 and ≤27.0% RMSE). Bland–Altman plots confirmed acceptable bias and precision. Conclusion: Limited regression models were successfully developed for prediction of AUC of evogliptin.