Selection of exposure metrics AUC, Cmax or Cmin in exposure response analyses – a simulation study.
Matts Kågedal, Qi Liu, Jin Yan Jin
Genentech Inc
Objectives: The relation between drug exposure and response (ER) is often described based on summary metrics of exposure such as AUC, Cmax or Cmin. This analysis approach ignores time, but makes the analysis more efficient compared to longitudinal PK-PD analyses, and is commonly used in drug development and regulatory reviews. The aims of this simulation study were to understand the consequences of ignoring time in the simplified ER analysis and to understand in what situations AUC, Cmax or Cmin correlates better with response.
Methods: Longitudinal PK and PD data were simulated for three different doses based on a one compartment first order absorption PK model with 50% variability on PK parameters (CL, V and Ka) and an indirect response PD model [1] with drug effect either on the rate constants Kout or Kin. Three underlying models for the relation between drug concentrations and the effect on rate constants were tested: linear, E-max, and an E-max model with a high sigmoidicity factor (on/off like effect). AUC, Cmax and Cmin were derived and correlated with response at end of treatment
Results: The exposure metric that correlated best with response was dependent on the underlying relationship and also varied between doses for the same assumed relationship. AUC was the best metric (highest correlation with response) in the linear range. Cmin correlated best in the exposure range approaching saturation. Cmax correlated best when exposures were mostly below the EC50 for the on/off like effect. The correlation across doses covering a wide response range was highest for AUC in all cases. The ER relationships derived based on each dose group separately were in agreement only when the underlying relation was linear and AUC was used as exposure metric. In all other cases there was a discrepancy between the ER relations derived based on each dose separately, in spite of the overlapping individual exposure between doses. I.e. at the same exposure (e.g. high Cmin from low dose group and low Cmin from high dose group), the expected response differed between doses.
Conclusions: This simulation study suggests that AUC is the best metric overall for ER analyses, but the best exposure metric varies with dose and the underlying PK-PD relation. The ER relationship derived based on one dose group may not be predictive of the response of another dose. This complication may need special attention when ER assessment and dose justification is based on a single dose group.
References:
[1] Jusko WJ , Ko HC. Physiologic indirect response models characterize diverse types of pharmacodynamic effects. Clinical Pharmacology and Therapeutics [1994, 56(4):406-419]