7 replies to this topic

[Simon Davis]

Here is a copy of a poster presented at AAPS in 2007 that can shows an example of using a custom IVIVC model Simon [file name=AAPS2007_IVIVC_Poster.pdf size=312413]http://www.pharsight.com/extranet/media/kunena/attachments/legacy/files/AAPS2007_IVIVC_Poster.pdf[/file]

Attached Files

•  AAPS2007_IVIVC_Poster.pdf   305.09KB   1834 downloads

Edited by Simon Davis, 02 March 2015 - 06:16 PM.

[Simon Davis]

There is a short discussion of the above poster on the LearnPKPD blog;

http://learnpkpd.com...e-tip-for-pkpd/

together with a workaround tip if your custom model ever freezes.

I'd invite any readers to let us know of other publications using the IVIVC toolkit so we can add to the library of

Simon

Tips for custom IVIVC models in Phoenix
February 18, 2016 By Bayo Olowoyeye 2 Comments

Author: Simon Davis
Original Poster Link

The IVIVC tool within Phoenix is a convenient and powerful tool to build an in vivo-invitro correlation model using dissolution experiments and real in vivo profiles. This module can support your organisation in formulation development and even in applying for Biowaivers.

The tool has a flexible library of built-in models that can be modified to better fit more complicated, non-linear relationships as demonstated in the attached AAPS poster presentation.

To briefly summarise the problem was to develop an In Vitro / In Vivo Correlation (IVIVC) model for a compound whose release rate specific bioavailability would not initially seem to support an IVIVC.

The formulation under evaluation is a capsule containing four different types of delayed release beads. Each bead is are characterized by a mean dissolution time (MDT) and a time lag (Tlag) and each of the seven formulations under study is comprised of a different ratio of each bead type.



Mean plasma profiles for 4 formulations (10-20 subjects per formulation) were deconvolved using the PK parameters that would relate to the intravenous kinetics of the compound. The generated series of fraction absorbed (Fabs) profiles were then overlaid with the dissolution profiles. Reviewing this plot suggested that the slower formulations had greater bioavailability than the fastest release (CR01) and the slower the release -the greater the bioavailability.

raw_deconvolution

This meant that a simple library model could not describe the data well. Since the Levy plot indicated two time scales a bi-linear model was proposed where an initial lag (TL1) to be consistent with a stomach emptying effect together with a Tcut representing intestinal transit time switching between phases.

In this case reviewing the original exploratory plots together with the diagnostic plots an initial library model was refined to consider some physiological mechanismse.g. first pass metabolism greatest in the small intestine, thereby creating an acceptable custom IVIVC model.

customIVIVCresult

However occasionally when running a custom model you may find that IVIVC model the object freezes (you can tell when this has happened because the status bar fails to advance)

ivivc_freeze_img3

Should this happen

CTRL+SHIFT+Escape to bring up Windows Task Manager
Navigate to the Processes tab.
Try ending the wnlpk32.exe process
It should automatically restart and then complete without error
Take it as a reminder to save you Phoenix project!

Edited by Simon Davis, 15 July 2020 - 02:16 PM.

[Cornelia Preusse]

There is a short discussion of the above poster on the LearnPKPD blog;

 

http://learnpkpd.com...e-tip-for-pkpd/

 

together with a workaround tip if your custom model ever freezes.

 

I'd invite any readers to let us know of other publications using the IVIVC toolkit so we can add to the library of

 

 Simon

 

Hi Simon,

 I am also having some troubles with the Syntax of the user-specified correlation coding in ASCII Format. It would be great if you could share the ASCII model script for the user-defined correlation mentioned in the poster above as an example so that I could start working on from there (the Version below doesn't work)

 

I'm not shure if that was provided under the link above since the page cannot be found anymore.

 

Many thanks and best regards,

Conny

 

DISS = IVINTERP(T)

IF X < Tcut THEN

F(X)=AS1*FDISS*(TS1*X-TL1)

ELSE

F(X)=F(Tcut)+AS2*(FDISS*(TS2*X-TL2)-FDISS(Tcut)

[Simon Davis]

Hi Conny,  I used the wayback when engine to recover it ... https://web.archive....e-tip-for-pkpd/

I will edit the above post and have a look at the code later.

 Simon.

[Cornelia Preusse]

^{Hi Simon thanks for recoevring the poster !!}

^{Did you already had a look on the ASCII model script for the user-defined correlation, I'm having still trouble with the Syntax.}

 

^{Many thanks, Conny}

[harishkaushik]

Hello Simon,

It would be great if you can help us by sharing example code or phx project for our reference

 

Regards

Harish

[harishkaushik]

Hello Simon/PHX Expert

 

Iam trying to build IVIVC model and came across the above issue of Nonlinear correlations between Tvivo vs Tvitro from the Levy plot

 

I have used the default user defined ASCII model and tried building the correlation it didnt work

then i tried methodology mentioned in the below paper

https://link.springe...2248-019-0378-y

Tvitro={(A1*Tvivo)/(A2+Tvivo)}-B2

I have attached screenshot of the ascii code also herewith

 

I am not able to build correlation, if any one can help it will be very useful

there is no information even on how to build nonlinear IVIVC any where using WNL

 

Also attaching the data which iam trying

 

Regards

Harish

Attached Thumbnails

• 
• 
• 

Attached Files

•  DATA RUNNING EXCEL.xlsx   14.01KB   279 downloads

[Simon Davis]

The challenge is that the ASCII code is not used much except within IVIVC object nowadays.

Here is an example of a semi-physiological model where absorption does not occur until after entering small intestine.  The important part in this case is setting up a temporary time variable and the use of IF ____ THEN

 

(this example is worked through in the IVIVC course https://certaraunive...enix-winnonlin)
 

Ttemp=X

IF Ttemp < TStom THEN

  Ttemp = 0

  FAbsModelG1 = 0

ELSE

 

TVitro =  Ttemp - TStom

 

DISS = IVINTERPT(TVitro)

 

FabsModelG1 = ( 1 - FPGut )* Diss

 

IF ( Ttemp > Tintes ) then

    Remark Ffp is total eliminated in small intenstine

    Ffp = FPGut * IVInterpt(TIntes)

    FAbsModelG1 = Diss - Ffp

End if

 

End If

 

F = FabsModelG1*FA

 

 

Parameter Names and Initial Estimates:

 

Tstom    0.5

TIntes     4

FPgut    0.1

FA          4