An explanation for failure to predict cyclosporine area under the curve using a limited sampling strategy: a...

Recently, limited sampling strategies have been developed to predict the area under a blood drug concentration-time curve (AUC). Multiple stepwise linear regression is the method used to develop limited sampling strategies. Several limited sampling strategies are used to predict cyclosporine AUC. However, recent findings demonstrated that, of all the developed limited sampling strategies, none can predict cyclosporine AUC correctly. Although several explanations have been given for this, the possible mathematical reason has not yet been fully explored. In the present study, we demonstrated that: (i) a steady-state blood drug concentration can be decomposed into the linear and non-linear components using the Fourier series; (ii) the non-linear component of the steady-state blood drug concentration leads to an AUC with a non-linear component, which affects the validation of limited sampling strategies; (iii) the correlation coefficient in a limited sampling strategy can only account for the linear association of linear components between a steady-state blood drug concentration and the AUC; (iv) the average steady-state blood drug concentration is a linear component of the steady-state blood drug concentration; and (v) when developing a limited sampling strategy, the non-linear component can be effectively ignored by: (a) choosing the sampling time around the middle point between the peak and the trough steady-state blood drug concentrations; and (b) increasing the trough steady-state blood drug concentration. However, the latter is restricted by clinical considerations.