Hi NMusers, I am trying to develop a population model of a drug that was given at a single dose using the PK data from two studies with different dosage. In one study with the higher dose (n=26, four-way cross-over), it seems to have three different patterns of kinetics: 1. the first measured concentration is Cmax (fast absorption); 2. a secondary peak; 3. more common oral PK with 1st-order Ka(normal absorption). In the other study with lower dose (n=46, two-way cross-over), the overall kinetics are less variable within the study. But there are about 30% subjects have about 50% higher Cmax than others.
I tried one-compartment, two-compartment and ALAG models to fit the PK from the first study with higher dose and also played with the introduction of BOV and BSV on Ka and/or TLAG. All the models generally cannot capture the absorption very well with an underprediction of Cmax. And None of the complex models really gave improved fitting compared to simple one-compartment model. I also tested transit model and it didn't improve the fitting either. Then, I focused on the PK with fast and normal absorption only and exclude the PK with the secondary peak. I found that two-compartment model can give good fitting for PK with fast absorption and normal absorption,respectively. The estimated Ka for fast and normal absorption PK are three times different, and then I tried mixture model on Ka to fit the two PK dataset together. But it suffered boundary failure and the percentage of the subpopulations can not be estimated. For the second study with lower dose, I have similar problem. The majority of Cmax is underpredicted, but I have good fitting of elimination phase in both case. It seems that I exhausted the possibilities that I can think of. Can anyone give some comments or suggestions ? Thanks a lot! Best, -- Xu, Claire Ph.D Candidate Division of Clinical Pharmacology, Wishard Hospital Indiana University School of Medicine 1001 West 10th Street, Myers W7122 Indianapolis, IN 46202 T - 317/7558242
