Carl Peck, Tom Ludden
Leiden University, The Netherlands, and CDER, FDA, USA
Retrospective population analysis in search of concentration-effect (C-E) relationships in data from randomized, concentration-monitored, dose-response trials has been advocated (1,2). While this may sometimes lead to discovery of important C-E relationships, there are pitfalls. Finding a weak, downwards biased, or absent apparent relationship between observed concentrations and effects when a true relationship exists may be due to (a) achieved concentrations being in a flat region of concentration-response relationship, (b) excessive pharmacokinetic (PK) and/or pharmacodynamic (PD) variability, (c) or strong positive correlations between PK and PD determinants, especially in the face of high PK and PD variability (3,4). On the other hand, an apparent relationship may be stronger than warranted or upwards biased when a strong negative correlation exists between PK and PD. It is even possible to observe a fictitious C-E relationship if there is unrecognized correlation between the baseline response measure and the PK, if no placebo responses are available. Retrospective analysis for C-E relationships should be accompanied by diagnostic procedures to uncover conditions where pitfalls can operate, and therefore should be interpreted with caution.
References:
1. Levy, G. Concentration-controlled versus concentration-defined clinical trials. Clin. Pharmacol. Ther. (1993), 53:385.
2. Peck, CC. Concentration-controlled versus concentration-defined clinical trials: A reply. Clin. Pharmacol. Ther. (1993), 53:385-386.
3. Sanathanan, L, Peck, CC. The randomized concentration-controlled trial: an evaluation of its sample size efficiency. Cont. Clin. Trials (1991), 12:780-794.
4. Peck, CC, Ludden, T. Bias and efficiency effects of correlation between pharmacokinetics and pharmacodynamics in post-hoc concentration-effect analysis of parallel dose-response trials. Clin. Pharmacol. Ther. (in press, 1994).
Reference: PAGE 3 () Abstr 867 [www.page-meeting.org/?abstract=867]
Poster: oral presentation