Priors
Building population pharmacokinetic (PopPK) models can be hard when the data is limited, such as when only trough concentrations (Ctrough) are available. In these cases, prior PopPK models from earlier studies can help.
We can use parameters from previous models as a starting point. By allowing some flexibility, this method improves predictions. It also stabilizes the results, reduces variability between individuals, and fits the new data better than fixing parameter values. However, if the new population differs from the one in the earlier study, accuracy may be affected.
PopPK analysis estimates how drugs are absorbed, distributed, and eliminated in the body, along with differences between individuals. When the data is too sparse, it may not be possible to estimate all the parameters. The $PRIOR control record in NONMEM allows us to use parameters from prior models to estimate some or all parameters, instead of fixing them or adding more data. This is a common and efficient approach, especially in special populations.
When using $PRIOR, it is important to check how closely the results match the earlier model and how sensitive the estimates are to the prior values. Covariates (factors like age or weight) can either come directly from the prior model (a priori) or be added later for parameters estimated without prior input (a posteriori).
Adding a prior to a Maximum Likelihood Estimation would technically convert these into a mode a posteriori (MAP) estimation of the population parameters, even though this term does not show up on the NONMEM report.
The main advantages lay in the rapidity and stability of the runs.
| Abbreviation | Description |
|---|---|
| NWPRI | Normal Inverse-Wishart Prior |
| OP | Penalty function on the OFV |
| OS | OFV on the sparse data |
| TNPRI | Twice Normal (“Normal-Normal”) Prior |
References
- Chan Kwong AHP, Calvier EAM, Fabre D, Gattacceca F, Khier S. Prior information for population pharmacokinetic and pharmacokinetic/pharmacodynamic analysis: overview and guidance with a focus on the NONMEM PRIOR subroutine. J Pharmacokinet Pharmacodyn. 2020;47(5):431-46. DOI: 10.1007/s10928-020-09695-z