Jump to content


Photo
- - - - -

ln-transformed PK parameters in PML code


  • Please log in to reply
3 replies to this topic

#1 LLLi

LLLi

    Advanced Member

  • Members
  • PipPipPip
  • 92 posts

Posted 18 March 2017 - 11:41 PM

Hi all,

 

I am going through PK 33 in Dr. Dan's book (5th version) about the transdermal input and kinetics. There are several models in the WinNonlin file, including Poisson model and Poisson model MD. I have some questions and hope someone can help me here.

 

1) The Poisson model is similar to the WNL5 model in the 4th version (the codes are similar). However, the results are not the exactly the same. For example, the V=253 in Poisson model while V = 336 in WNL5 model. What is the possible reason for the difference?

 

2) The result of another model, Poisson model MD, is almost the same as that in Dr. Dan's book (4th version). There are two dosepoints (one bio is F and the other bio is 1-F). It looks like it is similar to multiple absorption route?  Furthermore, lCl and IV were used in the PML code as primary parameters (I guess that lCl is LnCl and lV is Ln since V=exp(lCl) and Cl = exp(lCl) in the secondary parameter statements). I tried to use Cl and V directly in the code, but the results were bad. Why use LnCl and LnV in this model?

 

Thank you and any input is appreciated!

 

LLLi



#2 bwendt@certara.com

bwendt@certara.com

    Advanced Member

  • Administrators
  • 282 posts

Posted 19 March 2017 - 06:19 PM

Llli,

 

I'll let Dan Weiner respond to this question when he will be back from vacation - assuming you don't need an answer immediately.

 

Bernd



#3 dlweiner

dlweiner

    Member

  • Members
  • PipPip
  • 23 posts

Posted 25 March 2017 - 10:47 PM

Hi all,

 

I am going through PK 33 in Dr. Dan's book (5th version) about the transdermal input and kinetics. There are several models in the WinNonlin file, including Poisson model and Poisson model MD. I have some questions and hope someone can help me here.

 

1) The Poisson model is similar to the WNL5 model in the 4th version (the codes are similar). However, the results are not the exactly the same. For example, the V=253 in Poisson model while V = 336 in WNL5 model. What is the possible reason for the difference?

 

2) The result of another model, Poisson model MD, is almost the same as that in Dr. Dan's book (4th version). There are two dosepoints (one bio is F and the other bio is 1-F). It looks like it is similar to multiple absorption route?  Furthermore, lCl and IV were used in the PML code as primary parameters (I guess that lCl is LnCl and lV is Ln since V=exp(lCl) and Cl = exp(lCl) in the secondary parameter statements). I tried to use Cl and V directly in the code, but the results were bad. Why use LnCl and LnV in this model?

 

Thank you and any input is appreciated!

 

LLLi

Dear LLLi,

 

As you noted one of the advantages of having PHX available is that I often provide solutions beyond those discussed in the textbook.  In reply to your specific questions above 

 

1) in the WNL ASCII code the differential equation was for Cp (conc and not amount).  Here the initial condition was set  to a constant = 2 (same as on Cp(0)).  However, in the PML code the differential condition was for Amount, not Cp.  Amount = Cp * V; thus the amount at time zero was set to 2*V.  Since V was a parameter this enabled us to make the initial condition a parameter vs. Cp(0)=2.  Letting the initial condition vary is arguably a better model.

 

2) In the prior models, there was no dosepoint statement used.  Rather we just fit the functional equations and estimated the two fractions of dose as secondary parameters.  I then extended this example by showing how to estimate the fraction(s) absorbed directly using the dosepoint statement.  Here essentially both F's are estimated as model parameters (one estimated and the other derived from the estimated vs. secondary parameters.  I extended this example further to show how to implement a solution where one assumes a parameter follows a ln-normal distribution vs a normal distribution.

 

Hope that helps,

Dan



#4 LLLi

LLLi

    Advanced Member

  • Members
  • PipPipPip
  • 92 posts

Posted 27 March 2017 - 05:23 PM

Hi Dr. Dan,

 

Thank you for your reply. Hopefully you had a wonderful vacation. :)

 

I still have a question. Is there any general rule for the usage of PK parameters or ln(PK parameters) when we do the PK modeling? When we do the other PK analysis, like BE, we transform the PK parameter and then compare the ln(PK parameters) between different groups. But for PK modelling, I am not sure to use PK parameters or ln(PK parameters). Would you please provide more information?

 

Thanks!

LLLi


Edited by LLLi, 27 March 2017 - 05:24 PM.





0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users