> Our Services

Brief Examples

> Epilepsy Example
> Anti-viral Example
> Population PK Example
> Population PK/PD/Disease Marker Example

Epilepsy Example

The aims of this project were:

• To develop a simulation model for weekly seizure count data based on in-house Phase IIb and Phase III data.
• To adapt the simulation model allowing simulations for a follow-up compound in development.
• Evaluate several trial design aspects for the initial Phase II trial of the follow-up compound allowing to aid in the optimization of this trial.

The data for this project were captured from four studies which included about 1000 patients. The final model described weekly seizure counts in individual patients over time as a function of the administered dose and important inclusion criteria. The predictive performance of the model was evaluated by Posterior Predictive Checks. Performance of the planned initial Phase II trial was evaluated by clinical trial simulations based on the existing draft protocol as well as for alternative trial design aspects. The total project was executed in 6 months.

Value to the client:

• Integration of all available in-house clinical data in epilepsy.
• The project team was focused on the importance of pre-clinical potency estimates relative to other trial design aspects.
• The M&S project results gave rise to suggestions for improving the planned statistical analysis methods of the initial Phase II trial.
• The model that was developed could be updated based on new information that would become available throughout further clinical development, leading to the potential to further aid in optimization of the clinical development program.

[ back to top ] [ back to home ]

Anti-viral Example

The aims of this project were:

• To develop models that related explanatory variables to virological response and incidence of a major side effect.
• To evaluate alternative dose regimens in order to increase the likelihood of response without substantially increasing the incidence of the side effect.
• To clarify the impact of specific patient covariates on virological response and the incidence of the side effect.

The data for this project were obtained from two Phase III studies which included more than 1700 patients. Models were developed for response rate and incidence of side effect, as a function of dose and other explanatory variables. Simulations for the suggested dose regimen as well as alternative dosing schemes were undertaken using bootstrapping techniques. Graphical analyses were undertaken to clarify the relationship between response rate and incidence of the side effect and patient covariates. The M&S project took about 3 months.

Value to the client:

• The simulations showed that the suggested dosing scheme was appropriate and the simulation results were submitted to the regulatory authorities.
• Relationships between specific patient covariates of interest and response rate and incidence of side effect were successfully characterized and graphically presented.
• The availability of the model facilitated a fast response to be made to the regulatory questions that were asked.

[ back to top ] [ back to home ]

Population PK Example

Work done previously by the client had identified, but not characterized, complex changes in pharmacokinetics with dose and population. Individual exposure predictions were needed for interim analysis of an ongoing Phase II study.

The aims of this project were:

• To develop a population pharmacokinetic model based on Phase I/IIa data.
• To use this model to predict individual exposure in Phase II patients where only sparse PK data were available, and to compare this with target exposure.
• To update the model with the Phase II data.

The data for this project were rich and came from seven studies in healthy subjects and patients. A model with two parallel absorption pathways was found to be needed to describe the data. Both dose and food affected the rate and extent of absorption. A posterior predictive check confirmed that AUC, Cmax and Cmin were well captured by the model. When the interim Phase II data became available, Bayesian (POSTOC) individual parameters were predicted for individuals in the Phase II study conditional on the model and on the measured concentrations. AUC, Cmax and Cmin were calculated/simulated from the individual parameters and compared to the desired target exposure. Finally, the model was updated with the Phase II data. Model building took 4-6 weeks. The predictions for the individual exposure could be performed within days of getting the data.

Value to the client:

• The model allowed integration of data collected in a number of different studies, using different formulations, doses, and populations.
• It could be confirmed that the target exposure was reached with the doses used in the Phase II study.
• This information was made available very quickly, in time to support an interim efficacy analysis and go/no-go decision.

[ back to top ] [ back to home ]

Population PK/PD/Disease Marker Example

This project was performed during, and using, Phase I data.

The aims of this project were:

• To develop a population PK model for the compound.
• To develop models that linked drug concentration to a PD marker and to a marker of disease state.
• To link the PD marker with the disease marker data.
• To simulate steady state effects.

Intravenous and oral data from were available from two single dose studies performed in healthy volunteers. A three-compartment model with nonlinear bioavailability was developed to describe the PK of the compound. The PD marker was related to drug concentrations by a directly acting Emax model. The disease marker was related to drug concentrations and to the PD marker by indirectly acting response models. Posterior predictive checks of the peak, trough and time of peak concentration or effect were performed and demonstrated that the models described the data well. Simulations were performed to predict concentrations and effects at steady state.

Value to the client:

• It was shown that at the current doses, a reasonable effect on the PD marker was predicted at steady state
• However, the predicted effect on the disease marker was very low, even for high levels of PD effects.
• These results formed part of the basis for a decision to stop development of this compound and mechanism of action.
• A ‘no go' decision was made early in development, without the need for repeat dose or patient studies.

[ back to top ] [ back to home ]