when I get a parameter estimates from mice and wish to scale to human, what is the general consensus to do the scaling? such as scale both kng and kout according to allometry with exponet of -0.25 → 0.16x from mouse to human, also feed in Cp as human PK but keep the Imax and IC50 as the same as mouse.
This is a tough question and without data I personally use only the PK allometric scaling from which there is solid evidence about the power scaling factor ~(-0.25 for Cl and ~0 for volume when giving dose per kg weight). It would be interesting to collect data on different species and see how the exponential grow rate kng , the max inhibition, IC50 and first order elimination would change with species. I think that the dominant factor still will be the PK part as there is a huge shift in PK characteristics when gong from mice to humans.
I would use for IC50 the same scaling as for clearance if I have no knowledge. The reason being that IC50 has concentration units and the concentration profile are completely different in mice and humans.
Best would be an experiment in mice, rat and monkeys.
Thanks for your suggestion and I agree this is not an easy question and it may be highly case-by-case.
One of the limitation is that we don’t usually have a disease model/ tumor model across species, which makes it harder to do the correlation excercise.
Here was an interesting paper from the folks at Genentech attempting to correlate human response with animal TGI models. I would imagine for this if you do not have human PK as they had in this paper you could use a PBPK model (preferred) or allometry. They make no mention of scaling of the growth rates but use the model to derived a simulated %TGI based on human PK and bench mark this against a variety of compounds with known RECIST response rates.