modelling lag time and absorption

From: "R.A.A. Mathot" r.mathot@erasmusmc.nl Subject: [NMusers] modelling lag time and absorption Date: Wed, 15 Nov 2006 17:07:34 +0100 Dear NM-users, Currently I am working on the PK data of a drug that is taken once daily orally. I have data from 38 individuals with 8 samples per individual over an 8-hour period. Individual plots demonstrate that a 1-compartment PK model adequately describes the data. Furthermore, the plots show that there is considerable between-patient varibility in Ka, Tlag, V/F and CL/F. For instance, some patient had detectable drug levels at the first sample taken wheras in 1 patient the 8 hour (=last) sample was the first sample in which the drug could be detected. In the dataset concentrations <LOD were excluded with the exception of the last sample before concentration > LOD or the first sample after concentration > LOD. These concentrations were set at half the value of the LOD. ADVAN2 was used with FOCE+interaction and a combined error model. The results were dissapoining: 1. biased PRED vs DV plots 2. High WRES values 3. Parameter estimates were dependent on the number of significant digits and intitial values 4. Individual plots of IPRED vs time demonstrated that Tlag was underestimated in 25% of the patients I tried to improve the model by application of log-transformation (FOCE, additive error model, all concentrations <LOD excluded). Diagnostic plots slightly improved (1 and 2 above) whereas parameter estimates remained unstable and POSTHOC's of lag-time were underestimated (3 and 4). What could be the possible reasons for these observations? Thanks in advance for your time. Ron Matht ----------------------------------------------------- Ron A. A. Matht, PharmD PhD Department of Clinical Pharmacy and Clinical Pharmacology Erasmus MC University Medical Center Rotterdam P.O. Box 2040 3000 CA Rotterdam The Netherlands Phone: +31 10 4633202 Fax : +31 10 4636605 Email: r.mathot@erasmusmc.nl ----------------------------------------------------- _______________________________________________________