The Influence of Normalization Weight in Population Pharmacokinetic Covariate Models

Abstract

In covariate (sub)models of population pharmacokinetic models, most covariates are normalized to the median value; however, for body weight, normalization to 70 kg or 1 kg is often applied. In this article, we illustrate the impact of normalization weight on the precision of population clearance (CL pop ) parameter estimates. The influence of normalization weight (70, 1 kg or median weight) on the precision of the CL pop estimate, expressed as relative standard error (RSE), was illustrated using data from a pharmacokinetic study in neonates with a median weight of 2.7 kg. In addition, a simulation study was performed to show the impact of normalization to 70 kg in pharmacokinetic studies with paediatric or obese patients. The RSE of the CL pop parameter estimate in the neonatal dataset was lowest with normalization to median weight (8.1%), compared with normalization to 1 kg (10.5%) or 70 kg (48.8%). Typical clearance (CL) predictions were independent of the normalization weight used. Simulations showed that the increase in RSE of the CL pop estimate with 70 kg normalization was highest in studies with a narrow weight range and a geometric mean weight away from 70 kg. When, instead of normalizing with median weight, a weight outside the observed range is used, the RSE of the CL pop estimate will be inflated, and should therefore not be used for model selection. Instead, established mathematical principles can be used to calculate the RSE of the typical CL (CL TV ) at a relevant weight to evaluate the precision of CL predictions. Key PointsNormalization to a weight outside the observed weight range (e.g. 70 kg normalization in a paediatric study) can increase the uncertainty of parameter estimates in pharmacokinetic covariate models.The predictive performance of pharmacokinetic models and their covariate submodels is unaffected by weight normalization.When normalizing outside the observed covariate range, the RSEs of the corresponding population estimates should generally not be used for model evaluation. The RSE of the typical parameter at a relevant covariate value can be used instead.Sebastiaan Goulooze and Swantje Völler contributed equally to this work.

Population pharmacokinetics of treosulfan in paediatric patients undergoing hematopoietic stem cell transplantation

“…However, evaluation with a bootstrap procedure with 1000 bootstrap replicates showed estimates that are in line with the estimates of the PK parameters and their random variability of the final model. Standardizing to the median weight might be more appropriate, because standardizing to a weight outside the observed weight range can increase uncertainty of parameter estimates . By contrast, comparison with other models is more difficult when standardizing to the median weight, so in the final model the PK parameters were standardized to 70 kg.…”

“…12 Mohanan Standardizing to the median weight might be more appropriate, because standardizing to a weight outside the observed weight range can increase uncertainty of parameter estimates. 31 By contrast, comparison with other models is more difficult when standardizing to the median weight, so in the final model the PK parameters were standardized to 70 kg. As we compared the prediction-corrected VPCs of the current model vs the models of Ten Brink et al., 17 Danielak et al 12 and Mohanan et al, 16 it is clear that the current model has superior predictive performance both in the high and low concentration range.…”

Population pharmacokinetics of vancomycin in obesity: Finding the optimal dose for (morbidly) obese individuals

“…Covariates that were present in the dataset included TBW, LBW (calculated using the Janmahasatian formula), adjusted body weight (ABW, calculated with correction factor 0.4 as described elsewhere), BMI, ideal body weight (using the Devine formula), sex, age, GFR (based on collection of 24‐h urine) and serum creatinine‐based estimations of GFR such as CG‐TBW, CG‐LBW, MDRD or CKD‐EPI (the latter 2 both normalized for BSA 1.73 m 2 and de‐indexed for BSA by multiplying the original value by BSA/1.73). Covariates were implemented in the model using linear and power functions, standardized for a typical individual of 70 kg or median value of the covariate . Inclusion was considered when step‐by‐step inclusion resulted in a drop in OFV of at least −3.84 ( P < .05) and backward deletion gave an OFV increase of at least 10.8 points ( P < .001).…”

“…Values below LOD were analysed using the M3 method as described elsewhere. 30 35 Inclusion was considered when step-by-step inclusion resulted in a drop in OFV of at least −3.84 (P < .05) and backward deletion gave an OFV increase of at least 10.8 points (P < .001). Furthermore, the contribution of a covariate was judged based on the reduction in IIV and diagnostics described earlier.…”

Mechanistic and Quantitative Understanding of Pharmacokinetics in Zebrafish Larvae through Nanoscale Blood Sampling and Metabolite Modeling of Paracetamol

“…Therefore, an empirical time-dependent function was tested, with a sigmoidal time-dependent metabolic formation rate resulting in the best fit but with relatively high RSE values of the estimated parameters. Normalization of the metabolic formation rate for sulfation at time point 0, k PS,f,0 , at a different time point reduced its RSE value below 50%, indicating no overparameterization (Goulooze et al, 2019). Supplemental Fig.…”