A new approach for population pharmacokinetic data analysis under noncompliance
Pankaj Gupta, Matthew M Hutmacher, Bill Frame and Raymond Miller
Pfizer
Objectives: To systematically evaluate a reported method for analyzing outpatient pharmacokinetic (PK) data in the presence of noncompliance to the prescribed dosage regimen.
Methods: A pilot simulation study was designed to address the aforementioned objective. PK data were generated assuming a one compartment body model with first order absorption and first order elimination. Noncompliance, defined here as a subject not taking a scheduled dose, was simulated under three different rates, r=0, 0.1, and 0.2. These rates represent the probability of a subject not taking his/her scheduled dose at each nominal dose time. Data were simulated using a multiple dose design, which was sufficient under full compliance (r=0) to achieve steady state PK concentrations. These population PK data were first analyzed by the conventional methodology, in which the model assumes perfect compliance to the multiple dosage regimen. Next, an approach, based on the principle of superposition (Soy et al.), which separates estimating the elimination rate from the model based steady-state PK concentration, was used to estimate the fixed and random effect parameters. NONMEM VI was used to simulate the PK concentration data and estimate the parameters. The relative performance of the two methods was assessed by noting the bias (mean error) and imprecision (mean absolute error) in the parameter estimates. The two approaches were further contrasted in a follow-up simulation study for an existing drug, which incorporated a quasi-realistic dosing and sampling design.
Results: The analysis revealed that the conventional method of analyzing population PK data, which ignores noncompliance, yielded biased values for clearance, volume and the absorption rate constant, and overestimated the interindividual variability in these parameters. Additionally the random intraindividual (residual) variability was also inflated compared to the true value and increased with an increase in the degree of noncompliance. The conventional approach also fared poorly with regard to the imprecision in the parameter estimates across different compliance rates. In contrast, the performance of the new approach was consistent across the three simulated scenarios.
Conclusions: The robustness in parameter estimation in presence of possible noncompliance combined with the simplicity in implementation make the new approach an attractive alternative to the conventional method of analyzing outpatient population PK data. The decreased bias, especially in the residual intraindividual variability, can facilitate covariate analysis and allow meaningful data interpretation.
References:
[1]. Soy D, Beal SL, Sheiner LB. Population one-compartment pharmacokinetic analysis with missing dosage data.Clin Pharmacol Ther. 2004 Nov;76(5):441-51.
[2]. Mu S, Ludden TM. Estimation of population pharmacokinetic parameters in the presence of non-compliance. J Pharmacokinet Pharmacodyn. 2003 Feb; 30(1):53-81.
[3]. Wang W, Husan F, Chow SC. The impact of patient compliance on drug concentration profile in multiple doses. Stat Med. 1996 Mar 30; 15(6):659-69.
