The influence of sparse data sampling on population pharmacokinetics: a post hoc analysis of a pharmacokinetic study of morphine in healthy volunteers

Ariano, R.E.; Duke, P.C.; Sitar, D.S.

Clinical Therapeutics 34(3): 668-676


ISSN/ISBN: 1879-114X
PMID: 22381713
DOI: 10.1016/j.clinthera.2012.01.023
Accession: 056384385

Download citation:  

Article/Abstract emailed within 0-6 h
Payments are secure & encrypted
Powered by Stripe
Powered by PayPal

Intensive sampling of patients for drugs with complex pharmacokinetic profiles is difficult to perform in the clinic or hospitalized patient setting. We seek to address whether sparse sampling can obtain pharmacokinetic parameter values similar to those with traditional modeling from a post hoc analysis of 2 previous clinical trials. This study investigated whether population-guided, sparse-sampling pharmacokinetic analysis of morphine in 14 healthy volunteers allowed for optimal characterization of concentration-time profiles for a validation population of 5 young male patients receiving morphine. Data were analyzed using nonparametric adaptive grid (NPAG) population modeling to investigate optimal compartmental structure and the influence of sparse sampling (ie, 9 versus 3 samples per subject) on parameter identification. These results were compared with traditional standard 2-stage (STS) pharmacokinetic modeling. The coefficients of determination (R(2)), mean error (ME), and root-mean-square error were used to assess the predictive performance of the various sampling models against a validation population. Seventy-nine percent of the healthy volunteers were male, with a mean age of 36 (17) years and a mean weight of 68 (10) kg. NPAG modeling identified that intravenous morphine was best represented by a 3-compartment pharmacokinetic profile and that sparse sampling with a least 3 blood samples per subject resulted in virtually identical measures of central tendency as the more intensively sampled dataset. A validation cohort of 5 male patients undergoing elective surgery had a mean age of 26 (4) years and a mean weight of 80 (13) kg. Using mean parameter estimates generated from sparse sampling and the 3-compartment model structure, simulated profiles were compared against measured concentrations in this validation cohort. Sparse sampling using NPAG achieved similar values of predictive performance as mean parameter values from the more intensively sampled, with an ME of -1.0 ng/mL and precision of 26.2 ng/mL compared with 0.76 ng/mL and 25.8 ng/mL, respectively. Traditional (STS) modeling techniques resulted in the greatest degree of underprediction within the validation group (ME = 4.43 versus 0.76 ng/mL, STS and NPAG-9, respectively; P < 0.0001). This post hoc analysis suggests that intensive sampling for discerning complex, 3-compartment pharmacokinetic models, such as morphine, may not be necessary. Sparse sampling achieved accurate model structure recognition and parameter identification for predicting concentrations of very complex drug-dosage regimens.