Tool for in vitro-in vivo correlation

• US 20030078761A1
• Filed: 10/18/2002
• Published: 04/24/2003
• Est. Priority Date: 10/19/2001
• Status: Active Grant

• Current Assignee: GloboMax LLC (ICON PLC)
• Original Assignee: Globomax Holdings LLC
• Inventors: Gillespie, William R., Young, David, Gibiansky, Leonid V., Farrell, Colm B., Shepard, Theresa A., Bigora, Sian F.
• Granted Patent: US 7,027,970 B2
• Time in Patent Office: Days
• US Class Current: 703/11
• CPC Class Codes: G16B 40/00 ICT specially adapted for biostatistics; ICT specially adapted for bioinformatics-related machine learning or data mining, e.g. knowledge discovery or pattern finding
• CPC Class Codes: G16B 40/10 Signal processing, e.g. from mass spectrometry [MS] or from PCR

Abstract

A biological modeling system and method for enhanced computer-aided analysis of biological response data provides information synthesized from immediate and extended release in vivo data and in vitro data. An executable model of a biological system is developed from information and structures based on the data. In a preferred embodiment, a two stage approach to modeling is used in the development of an IVIVC. The first stage of the procedure is deconvolution, where the percentage of drug absorbed is determined. In the second stage, the in vivo percentage absorbed data is correlated to the in vitro fraction or percentage dissolved data. This correlation then represents a point-to-point relationship between the in vitro dissolution and the in vivo input rate of the drug from the dosage form. In such a correlation, the in vitro dissolution and in vivo absorption profiles are either directly superimposable or may be made to be superimposable by the use of a scaling factor. Prior to the deconvolution stage, a unit impulse response function can be determined from immediate-release concentration-time data. This impulse response function is used in the deconvolution process to determine the in vivo percent absorbed for the extended release formulations. A nonlinear IVIVC model is developed that can incorporate time-scaling and time-shifting into the model if needed. After the two-stage modeling is completed, the predictability of the developed IVIVC model is evaluated by both internal and external validation.

Claims

1. A computer executable model of a biological system in combination with a computer system including a memory and a processor, said computer executable model comprising:

a plurality of biological representations stored in said memory and having a plurality of chemical level data points, each said chemical level data point representing a level of a chemical within a particular time period in a biological component;

a means for determining unit impulse response from a first collection of said plurality of biological representations;

a means for performing deconvolution having as inputs said unit impulse response and a second collection of said plurality of biological representations and producing as an output in vivo mean absorption data;

a means for developing an in vitro in vivo correlation model having as inputs a third collection of said plurality of biological representations, each of said third collection of said plurality of biological representations chemical level data points representing in vitro ER data and said in vivo mean absorption data; and

a means for validation of said in vitro in vivo correlation model.

2. The computer executable model of a biological system in combination with a computer system of claim 1, wherein said means for developing an in vitro in vivo correlation model further comprises a means for time-scaling at least on of said third collection of said plurality of biological representations and said in vivo mean absorption data to enable said third collection of said plurality of biological representations and said in vivo mean absorption data to be superimposed upon a graph.

3. The computer executable model of a biological system in combination with a computer system of claim 1, wherein said means for developing an in vitro in vivo correlation model further comprises a means for time-shifting at least one of said third collection of said plurality of biological representations and said in vivo mean absorption data to enable said third collection of said plurality of biological representations and said in vivo mean absorption data to be superimposed upon a graph.

4. The computer executable model of a biological system in combination with a computer system of claim 1, wherein said means for developing an in vitro in vivo correlation model further comprises a means for adjusting said developing means to compensate for bioavailability differences between said third collection of said plurality of biological representations and said in vivo mean absorption data to enable said third collection of said plurality of biological representations and said in vivo mean absorption data to be superimposed upon a graph.

5. The computer executable model of a biological system in combination with a computer system of claim 1, wherein said means for developing an in vitro in vivo correlation model further comprises a means for adjusting said developing means to compensate for release rate differences between said third collection of said plurality of biological representations and said in vivo mean absorption data to enable said third collection of said plurality of biological representations and said in vivo mean absorption data to be superimposed upon a graph.

6. The computer executable model of a biological system in combination with a computer system of claim 1, wherein said means for developing an in vitro in vivo correlation model further comprises a means for imposing a limit on a gastrointestinal absorption duration to enable said third collection of said plurality of biological representations and said in vivo mean absorption data to be superimposed upon a graph.

7. The computer executable model of a biological system in combination with a computer system of claim 1, wherein said means for determining unit impulse response further comprises:

means to select a treatment from the group that is identified as having faster release than the ER treatments to be used in the deconvolution;

means to identify whether a dosage regimen is a single or multiple dose regimen;

means to choose one of a mean approach or an individual approach;

means to execute a polyexponential function estimated to describe pharmacokinetic characteristics; and

means to save results from said executing means to a database.

8. The computer executable model of a biological system in combination with a computer system of claim 7, wherein said polyexponential function estimated to describe pharmacokinetic characteristics is approximated by:

cδ

(t)=Σ

i ai exp(α

i(t−

t0)), t≧

t0 

 

(1);

cδ

(t)=0, t<

t0 

 

(2)

where t0 is a common lag time.

9. The computer executable model of a biological system in combination with a computer system of claim 1, wherein said means for developing an in vitro in vivo correlation model further comprises a means to fit a linear regression model to said in vitro ER data and said in vivo mean absorption data of the following form:

10. The computer executable model of a biological system in combination with a computer system of claim 1, wherein said means for developing an in vitro in vivo correlation model further comprises a means to plot exploratory plots to estimate initial conditions in the IVIVC model.

11. A method for processing biological profile data for in vitro in vivo correlation comprising the steps of:

storing a plurality of biological profiles in a computer memory in computer readable form, each biological signal profile in ones of said plurality of biological signal profiles comprising a plurality of data points, each data point representing a measurement of a chemical level;

determining a unit impulse response from an at least first one of said biological profiles;

performing deconvolution on an at least second one of said biological profiles using said unit impulse response and producing in vivo mean absorption data;

developing an in vitro in vivo correlation model using said in vivo mean absorption data and an in vitro ER data biological signal profile; and

validating said in vitro in vivo correlation model.

12. The method for processing biological profile data for in vitro in vivo correlation of claim 11 wherein said step of developing an in vitro in vivo correlation model further comprising the steps of:

inputting said in vitro ER data biological signal profile and said in vivo mean absorption data;

running analytics to generate correlation data representative of said in vitro ER data biological signal profile and said in vivo mean absorption data;

plotting said correlation data;

selecting IVIVC model initial default values;

executing said IVIVC model to produce results;

reviewing said IVIVC model results; and

determining whether said IVIVC model is within FDA guidelines.

13. The method for processing biological profile data for in vitro in vivo correlation of claim 12 wherein said plotting step further comprises forming a plot of exploratory plots to estimate initial conditions in the IVIVC model.

14. The method for processing biological profile data for in vitro in vivo correlation of claim 12 further comprising the step of deciding whether a weighting factor is added to said IVIVC model.

15. The method for processing biological profile data for in vitro in vivo correlation of claim 12 further comprising the steps of:

saving IVIVC model outputs to a computer memory; and

inputting an entry into a history log.

16. The method for processing biological profile data for in vitro in vivo correlation of claim 15 further comprising the step of refining said IVIVC model initial default values responsive to said FDA guidelines determining step.

17. The method for processing biological profile data for in vitro in vivo correlation of claim 11 wherein said validating step further comprises a comparison of prediction errors for bioavailability parameters derived from mean observed and predicted in vivo data.

18. The method for processing biological profile data for in vitro in vivo correlation of claim 11 wherein said validating step further comprises a comparison of prediction errors for geometric mean bioavailability parameters derived from individual observed and predicted in vivo data.

19. The method for processing biological profile data for in vitro in vivo correlation of claim 11 wherein said validating step further comprises a comparison of prediction errors for individual bioavailability parameters derived from individual observed and predicted in vivo data.

20. The method for processing biological profile data for in vitro in vivo correlation of claim 11 further comprising the step of predicting in vivo characteristics from in vitro data.

21. The method for processing biological profile data for in vitro in vivo correlation of claim 11 wherein said step of performing deconvolution further comprises the step of choosing whether one or more of said at least second one of said biological profiles are to be selected.

22. The method for processing biological profile data for in vitro in vivo correlation of claim 21 wherein said at least first one of said biological profiles comprises individual data for an individual subject, and said at least second one of said biological profiles comprises individual in vivo data for an individual subject.

23. The method for processing biological profile data for in vitro in vivo correlation of claim 22 wherein said at least first one of said biological profiles further comprises at least one study of multiple subject data, and said at least second one of said biological profiles further comprises at least one study of multiple subject data, and step of performing deconvolution further comprises selecting between said multiple subject data and said individual subject data.