17 replies to this topic

[csheme]

Could somebody please help me create the model below in PHX for inhalation dosing.

 

Is the airway the absorption compartment (Aa) with absorption rate constants into the Lung and Plasma compartments ,followed by absorption into peripheral tissues from plasma and elimination.

 

 

My limited observations include Plasma, Lung, Liver, and Kidney Concentrations.

 

I am having trouble creating, its almost as if I need 2 central compartments, or would lung be the central compartment while plasma considered peripheral? (It seems I need 2 absorption rate constants).

 

I am attaching a link below of my current model.

 

https://www.dropbox....on.phxproj?dl=0

 

Thanks greatly!

 

 

Edited by csheme, 06 July 2017 - 04:49 PM.

[serge guzy]

Dear colleague

I made a basic model I believe can be identifiable if you have data in kidney(assuming kidney drug amount not removed) lung, plasma and liver.

The model you proposed is to me not identifiable as too many elimination rate constants that would not be identifiable when doing fitting.

If kidney level is not just cumulative, then add another elimiknation compartment that would mimic the elimination through unrine collection.

Best Regards

Serge

Attached Files

•  inhalation_simple_model.phxproj   118.65KB   463 downloads

[csheme]

Many Thanks Serge, I will try to fit the data with your model.

 

Kind Regards!

[serge guzy]

I added urine elimination compt which makes more sense(last model). If you fit, please first perform some initial estimate exploratory analysis not to put any number. If you can share, I will be happy to help.

Best Regards

Serge

Attached Files

•  inhalation_simple_model.phxproj   190.8KB   432 downloads

[csheme]

Hi Serge,

 

I did revise the model, could you please see the attached link below.

 

https://www.dropbox....el.phxproj?dl=0

 

Since the dosing is by a 30minute inhalation , is this treated as an IV infusion as opposed to extravascular bolus?

 

The data I have to fit are the Plasma, Lung, Liver, and Kidney concentrations, with PD measurements in lung, I would like to setup an indirect response model.

 

Any idea on how to go about obtaining good initial estimates, I have never used this type of model before.

 

 

Many Thanks!

Edited by csheme, 07 July 2017 - 06:11 PM.

[serge guzy]

Dear colleague

The 30 minutes (assuming you use minute foir time) shoukld be put as the duration of the extravascular infusion.

I did it with the interface ands the final code is below:

 

test(){
 deriv(Aa = - (Aa * K_dosing_to_lungs))
 deriv(Alungs = (Aa * K_dosing_to_lungs)- (Alungs * klung_plasma))
 deriv(A1 = (Alungs * klung_plasma)- (Q_C2_C3 * C)- (Q_C2_C4 * C)- (Q_C2_C5 * (C - C5)))
 deriv(Aliver = (Q_C2_C3 * C)- (CL_Cliver_A0liver * Cliver))
 deriv(Akidney = (Q_C2_C4 * C)- (CL_Ckidney_A0kidney * Ckidney))
 deriv(A5 = (Q_C2_C5 * (C - C5)))
 urinecpt(A0liver = (CL_Cliver_A0liver * Cliver))
 urinecpt(A0kidney = (CL_Ckidney_A0kidney * Ckidney))
 dosepoint(Aa, duration = (30), idosevar = AaDose, infdosevar = AaInfDose, infratevar = AaInfRate)
 Clung = Alungs / Vlungs
 C = A1 / V
 Cliver = Aliver / Vliver
 Ckidney = Akidney / Vkidney
 C5 = A5 / V5
 error(CEpslung = 0.1)
 observe(CObslung = Clung * (1 + CEpslung))
 error(CEpsliver = 0.1)
 observe(CObsliver = Cliver * (1 + CEpsliver))
 error(CEpskidney = 0.1)
 observe(CObskidney = Ckidney * (1 + CEpskidney))
 error(CEpsplasma = 0.1)
 observe(CObsplasma = C * (1 + CEpsplasma))
 deriv(E = Kin * (1 + Emax * C / (C + EC50)) - Kout * E)
 sequence{E = (Kin) / (Kout)}
 error(EEps = 1)
 observe(EObs = E + EEps)
 stparm(Vlungs = tvVlungs * exp(nVlungs))
 stparm(V = tvV * exp(nV))
 stparm(Vliver = tvVliver * exp(nVliver))
 stparm(Vkidney = tvVkidney * exp(nVkidney))
 stparm(V5 = tvV5 * exp(nV5))
 stparm(K_dosing_to_lungs = tvK_dosing_to_lungs * exp(nK_dosing_to_lungs))
 stparm(klung_plasma = tvklung_plasma * exp(nklung_plasma))
 stparm(Q_C2_C3 = tvQ_C2_C3 * exp(nQ_C2_C3))
 stparm(Q_C2_C4 = tvQ_C2_C4 * exp(nQ_C2_C4))
 stparm(Q_C2_C5 = tvQ_C2_C5 * exp(nQ_C2_C5))
 stparm(CL_Cliver_A0liver = tvCL_Cliver_A0liver * exp(nCL_Cliver_A0liver))
 stparm(CL_Ckidney_A0kidney = tvCL_Ckidney_A0kidney * exp(nCL_Ckidney_A0kidney))
 stparm(Emax = tvEmax * exp(nEmax))
 stparm(EC50 = tvEC50 * exp(nEC50))
 stparm(Kin = tvKin * exp(nKin))
 stparm(Kout = tvKout * exp(nKout))
 fixef(tvVlungs = c(, 1, ))
 fixef(tvV = c(, 1, ))
 fixef(tvVliver = c(, 1, ))
 fixef(tvVkidney = c(, 1, ))
 fixef(tvV5 = c(, 1, ))
 fixef(tvK_dosing_to_lungs = c(, 1, ))
 fixef(tvklung_plasma = c(, 1, ))
 fixef(tvQ_C2_C3 = c(, 1, ))
 fixef(tvQ_C2_C4 = c(, 1, ))
 fixef(tvQ_C2_C5 = c(, 1, ))
 fixef(tvCL_Cliver_A0liver = c(, 1, ))
 fixef(tvCL_Ckidney_A0kidney = c(, 1, ))
 fixef(tvEmax = c(, 1, ))
 fixef(tvEC50 = c(, 1, ))
 fixef(tvKin = c(, 1, ))
 fixef(tvKout = c(, 1, ))
 ranef(diag(nV5, nQ_C2_C3, nQ_C2_C5, nK_dosing_to_lungs, nVlungs, nklung_plasma, nVliver, nV, nVkidney, nCL_Cliver_A0liver, nCL_Ckidney_A0kidney, nEmax, nEC50, nKin, nKout, nQ_C2_C4) = c(1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1))
}

 

 

About how to find initial estimates, I need the data set to help you. There are different techniques like pseudo NCA in conjunction with the initial estimate tab and information from the biology but without data there is no one way to explain all this.

 

Please do something to your data in such a way I can get those and helping you fitting the model to the data.

 

You can change the headers names, the units. This usually is enough.

 

 

The reason I need the data is because then I can work on a strategy to find initial estimates and to upgrade the model if necessary.

 

Best Regards

​Serge

 

​poppharm@gmail.com  just in case if you need it one day
 

Attached Files

•  Revised Inhalation Model.phxproj   122.3KB   429 downloads

[csheme]

Hi Serge,

 

Could you please see the attached model below, I am having a hard time to get a fit.

 

I would like to focus on the plasma and lung PK and PD from lung.

 

The datasets and dosing sheet are in the project.

 

https://www.dropbox....es.phxproj?dl=0

 

I do have data for liver and kidney (less important), also most likely need to add an elimination compartment from lung.

 

As you can see the PD has a delayed response, The PD should be an indirect response.

 

The most simple and easy to understand model would be the best (fewest needed compartments while still having a nice fit). Graphical model is easy to comprehend.

 

Thanks greatly!

 

Study 1 Data (Duration of Action Study (single and multiple doses)) included within the dataset.

 

 

 

 

 

 

Dose Ranging Study 2 (Exposure Response) Also added into the dataset (evaluation only at a single timepoint after multiple doses).

 

Edited by csheme, 20 July 2017 - 11:54 PM.

[serge guzy]

Dear colleagues

I started working on that project but these ar enot easy data to handle. It will take sometime to find the best model.

I will update you.

Best Regards

Serge

[csheme]

Hi Serge,

 

Thanks greatly, I did make a few changes for Study 2 because the timepoints were not correct.

 

The dose schedule for both studies was inhalation on : Days 0, 3, 7, 10, 14 and 17.

 

I updated the dataset and dosing sheet for the Study 2 animals (20 of the last ones).

 

The correct project is in the link below.

 

https://www.dropbox....es.phxproj?dl=0

Edited by csheme, 20 July 2017 - 09:25 PM.

[csheme]

I am attaching some plots I have been working on for this study as reference for the pop PK and PD below.

 

Edited by csheme, 20 July 2017 - 11:57 PM.

[csheme]

[csheme]

I managed to get a 1-cmpt model to work for the dataset based on Lung exposure alone. I ran the inhalation model as an intravenous infusion (duration set to 30min), and used an Emax for PD using QRPEM for initial estimates, the diagnostics seemed OK, but I do not think this is the type of model I would like. Also the parameter estimates seemed odd, the IC50 seemed quite inflated. The simulations also did not look right, essentially when I simulated 5-fold higher doses the PD was not as expected and only marginally improved. This seems really odd, even though the gamma is so shallow for this drug (~0.2-0.3). Also because of the Emax there was no delay in response, essentially the PD dropped immediately with exposure for the Inhibitory response (this is not right) there should be a delay).

 

I know that I am after at least a 2-cmpt PK model with an indirect response PD to account for the delay observed in the effect (and am not sure the infusion setup for the model is correct for inhalation dosing). I assumed the 30-minute inhalation acts sort of as an IV-slow infusion as opposed to IV-bolus, or even setting as the true 'extravascular' route.

 

It would be nice to be able to predict what is going on in Lung based on Plasma exposure, but as you can see in the above plots in my correlation analysis with trough plasma data and lung concentrations the correlation could be better. 

 

Any assistance with a better model would be much appreciated!

 

Many Thanks.

Edited by csheme, 22 July 2017 - 04:15 AM.

[serge guzy]

Dear colleague

Can you please send me your model with the 1 compt for me to review.

I will be traveling next week to university of MARYLAND but at least I csn review your model.

Best Regards

Serge

[csheme]

Hi Serge,

 

Thanks for the reply and wish you a nice travel.

 

I am attaching the link: https://www.dropbox....on.phxproj?dl=0

[serge guzy]

Dear colleague

Here is what I suggest you to do. Summarize your data by dosing and consider each dosing group as an individual. At ;east you will have a better picture of the trend for each dosing. Then start the fitting procedure again.

This may increase stability.

hen send me your results.

Best Regards

Serge

[csheme]

Hi Serge, I did as you recommended and reran the model, the IC50 is way off. I think the entire structure of the model may need to be changed, and possibly the dosing input I am using.

 

https://www.dropbox....up.phxproj?dl=0

 

Many Thanks.

Edited by csheme, 23 July 2017 - 02:32 AM.

[csheme]

 

Simulations of the same 6-week regimen used in the study (0.5 mg/kg) in addition to 3-fold and 8-fold higher doses, the PK seems to be somewhat reasonable, but the PD (shows no delay in response and no difference in effect between cohorts), at these sustained lung exposures based on the PK, the PD should have a much higher inhibition response and also the different cohorts should be discernible.

Edited by csheme, 23 July 2017 - 02:47 AM.

[csheme]

After some tinkering around I was able to get the PD fit by indirect response. I am still left with the challenge of how to predict lung exposure and PD based solely on plasma observations (future clinical data).

 

Edited by csheme, 25 July 2017 - 07:42 PM.