This case study presents the modeling of the tobramycin pharmacokinetics, and the determination of a priori dosing regimens in patients with various degrees of renal function impairment. It takes advantage of the integrated use of Datxplore for data visualization, Mlxplore for model exploration, Monolix for parameter estimation and Simulx for simulations and best dosing regimen determination.
The case study is presented in 5 sequential parts, that we recommend to read in order:
- Part 1: Introduction
- Part 2: Data visualization with Datxplore
- Part 3: Model development with Monolix
- Part 4: Model exploration with Mlxplore
- Part 5: Simulations for individualized dosing with Simulx and Monolix
Part 2: Data set overview and data visualization with Datxplore
Data set overview
Tobramycin bolus doses ranging from 20 to 140mg were administrated every 8 hours in 97 patients (45 females, 52 male) during 1 to 21 days (for most patients, during ~6 days). Age, weight (kg), sex, height and creatinine clearance (mL/min) were available as covariates. Because height information was missing for around 30% of the patients, this covariate was ignored. The tobramycin concentration (mg/L) was measured 1 to 9 times per patients (322 measures in total), most of the time between 2 and 6h post-dose.
Below is an extract of the data set file:
Data set visualization with Datxplore
We first start with the exploration of the data set in Datxplore. After having opened Datxplore, we create a new project and load the data set. Based on the header, most columns are automatically recognized. The column-type must be set manually for the observed concentration (CP column set to column-type OBSERVATION), and the creatinine clearance (CLCR column set to column-type CONTINUOUS COVARIATE for continuous covariate):
The spaghetti plot is displayed (figure below). It is possible to split or filter the plot using the covariates, which in the present case does not bring meaningful insights.
To have a closer look at the elimination kinetic, it is possible to visualize the data individual per individual. Most patients have only 2 measures between two doses, which is not informative for the elimination kinetics. Yet, for a few individuals, more measures are available. For individual 50, one can see that there are 9 observationsI we plot the y-axis in log-scale and zoom on the last dose, we observe the following kinetics, which may hint towards a 2 compartment model:It is also interesting to display one covariate against another one. We, in particular, observe a negative correlation between age and creatinine clearance (correlation at -.7):
On the opposite, the creatinine clearance is similar for both sexes:
In summary, from the data exploration in Datxplore, we have gained the following insights:
- Some individuals may not be properly described by a 1 compartment model with linear elimination. A 2 compartment model may be more appropriate.
- Older patients have a lower creatinine clearance.