Q: Do you plan to have any webinar showing PK/Target engagement/Occupancy models specifically?¬
A: As you may know, we have covered TMDD models in our 6th session (https://support.cert...ug-disposition/), but we have no further Target/Occupancy models in our current schedule. Our schedule is published here:
https://www.certara....ion/pml-school/
We might pick up more specific models when we plan an extension of this series.
Q: In this dataset, you had maximum effect and so you could freeze the Imax to 1, but if you do not have enough high dose in your study to see maximum inhibition, do you recommend to estimate that parameter from fitting?
A: No, as both the IMAX and IC50 values would be biased.
Q: To compare the values of Kon and Koff found from the PK/PD model to say in vitro lab-derived receptor off rate estimates should you then use unbound (free) concentrations as PK input? What do you think?
A: Yes, you should use unbound concentrations.
Q: What would be the recommended method to 'smooth' the curves after model fitting for presentation purposes? Do a simulation with large number of time points simply?
A: If you wanted to smooth PD curves you must interpolate the Cp values for the intermediate time points.
Q: Would there be a limit of PK samples needed for this approach?
A: Not really. But it is important to collect samples at time points that fully show the shape of the PD curve.
Q: What if we have very limited PD data (For. e.g. 3-4 points) vs PK data (7-8) points..? do you recommend fitting the PK model ?
A: Fitting the PK model will not “add” PD time points. If you can only get 3-4 PD time points but can treat many animals then randomize times across the animals and do a population model approach.
Q: What if we have bell shaped PD (stimulatory or buildup of response) type curve ? in that case also Koff initial estimate process works ?
A: There you would use Emax vs Imax models for the link model. You would also use either a stimulatory of input function or inhibition of output function for the turnover model. For the binding model, I believe you only need to change E = Bmax - RC to E = Bmax + RC.
Q: Could you please demonstrate how to do the model comparison?
A: Because we are using a population model, you can use the model compare object. (Insert Phoenix Model/Model Compare). Of course AIC can be uses to compare these models as well.
