Certara, Inc. · NASDAQ:CERT

Certara — Technical Presentation (February 2015)

44 slides from Certara's technical presentation, Feb 17, 2015

1. Slide 1: Improve Your Success Rate in Costly Bioequivalence Studies with IVIVC

   Improve Your Success Rate in Costly Bioequivalence Studies with IVIVC Dr. Terry Shepard, Pharmacokinetics Assessor, MHRA Dr. Venkateswari Muthukrishnan, Senior Scientific Consultant

2. Slide 2: Terry Shepard

   Terry Shepard • Joined MHRA (UK regulatory agency) in 2010 • Responsible for Assessment of Marketing Authorisation Applications and Providing Scientific Advice (MHRA, EMA) • Chair EMA Modelling and Simulation Working Group, contributed to numerous guidelines including Modified Release • Prior to MHR

3. Slide 3: Venkateswari Muthukrishnan

   Venkateswari Muthukrishnan • Joined Certara in Apr 2012 • Responsible for Training and Pre-sales Support in South Asia Pacific and Middle East • Provides Training and Support on Phoenix Products: Phoenix WinNonlin, NLME, Connect and IVIVC Toolkit • Prior to Certara ‒ AAI International , Pharmacokine

4. Slide 4: Agenda For Today

   Agenda For Today European Regulatory Guidance Key Points Illustrated by MHRA Experience Introduction and Top-Level Overview The IVIVC Wizard, Step-by-step using Reference IR data Individual Steps in Further Detail 1. Modeling dissolution 2. Estimating UIR from Immediate Release Formulation 3. Estima

5. Slide 5: Why do companies develop IVIVCs? Why do regulators encourage IVIVCs?

   Why do companies develop IVIVCs? in vitro dissolution = surrogate for BE study save time and money Why do regulators encourage IVIVCs? Approval Process Post-approval Reassurance that positive benefit/risk balance is maintained throughout life of product Slide 5

6. Slide 6: MR Product Variations: Example

   MR Product Variations: Example 2011: 9/year 54 variations 53 54 52 51 50 49 48 47 II Label/Leaflet 46 1b 45 44 43 42 41 1a 40 39 38 37 36 35 34 33 32 Pharmaceutical variations: 31 30 pharmaceutical variations 29 30 ·27 28 Manufacturer, batch release site x 3 26 24 25 · Batch size 22 23 · In process

7. Slide 7: MR Product Variations: Example (cont'd)

   MR Product Variations: Example (cont’d) 2011: 9/year Supporting evidence: 53 54 · Quality Overall Summary 52 51 50 · Batch data 48 49 47 · Stability data 45 46 · IVIVC (Level A) 43 44 42 41 40 39 38 37 36 34 35 Purpose of Variation: To update the dissolution test 33 32 31 29 30 method in line with t

8. Slide 8: MR Product Variations: Example (cont'd)

   MR Product Variations: Example (cont’d) 100 100 Upper limit Proposed Low(UL) and High 80 upper limit target % Released in vivo 80 60 Target (Pivotal Batch) lower limit +  40 % Released in vitro 60 20 Lower limit (LL) 0 0 20 40 60 80 100 % Released in vitro 40 IVIVC Predicted C(t) 20 Relative Bioava

9. Slide 9: MR Product Variations: Example (cont'd)

   MR Product Variations: Example (cont’d) 100 100 Upper limit Proposed Low(UL) and High 80 upper limit target % Released in vivo 80 60 Target (Pivotal Batch) lower limit +  40 % Released in vitro 60 20 Lower limit (LL) 0 0 20 40 60 80 100 % Released in vitro 40 IVIVC Predicted C(t) 20 Relative Bioava

10. Slide 10: MR Product Variations: Example (cont'd)

    MR Product Variations: Example (cont’d) 100 100 Upper limit Proposed Low(UL) and High 80 upper limit target % Released in vivo 80 60 Target (Pivotal Batch) lower limit +  40 % Released in vitro 60 20 Lower limit (LL) 0 0 20 40 60 80 100 % Released in vitro 40 IVIVC Predicted C(t) 20 Relative Bioava

11. Slide 11: MR Product Variations: Example (cont'd)

    MR Product Variations: Example (cont’d) 100 100 Upper limit Proposed Low(UL) and High 80 upper limit target % Released in vivo 80 60 Target (Pivotal Batch) lower limit +  40 % Released in vitro 60 20 Lower limit (LL) 0 0 20 40 60 80 100 % Released in vitro 40 IVIVC Predicted C(t) 20 Relative Bioava

12. Slide 12: MR Product Variations: Example (cont'd)

    MR Product Variations: Example (cont’d) 100 100 Upper limit Proposed Low(UL) and High 80 upper limit target % Released in vivo 80 60 Target (Pivotal Batch) lower limit +  40 % Released in vitro 60 20 Lower limit (LL) 0 0 20 40 60 80 100 % Released in vitro 40 IVIVC Predicted C(t) 20 Relative Bioava

13. Slide 13: MR Product Variations: Example (cont'd)

    MR Product Variations: Example (cont’d) 100 100 Upper limit Proposed Low(UL) and High 1. Anchor to clinical batches. 80 upper limit target % Released in vivo 80 60 Target (Pivotal Batch) lower limit +  40 % Released in vitro 60 Lower limit (LL) 2. Relationship to BE limits. 20 0 0 20 40 60 80 100 %

14. Slide 14: Dissolution Limits in Product Specifications: Relationship to BE Limits

    Dissolution Limits in Product Specifications: Relationship to BE Limits 100 IVIVC  in vitro = surrogate for in vivo Upper limit Proposed Low(UL) and High (BE) study 80 Target (Pivotal Batch) % Released in vitro 60 Lower limit (LL) 40 20 0 0 4 8 12 16 20 24 Time (h) Time (h) 80% 125% Test/Ref (%) Di

15. Slide 15: Impact of IVIVC Validation Range on Justification of Dissolution Limits

    Impact of IVIVC Validation Range on Justification of Dissolution Limits 100 IVIVC batches 100 80 80 Dissolution limits 60 60 %Released in vitro %Released in vitro 40 40 20 20 3 0 0 0 4 8 12 16 20 24 0 4 8 12 16 20 24 Time (h) Time (h) Time (h) Time (h) Choice of formulations: widest possible range o

16. Slide 16: European Guidance relating to IVIVC – revised 2014

    European Guidance relating to IVIVC – revised 2014 Slide 16

17. Slide 17: European Guidance relating to IVIVC

    European Guidance relating to IVIVC Quality Clinical IVIVC Topic Covered Section Section Levels of IVIVC: definitions   Levels of IVIVC: advantages, disadvantages of different levels   Role of IVIVC • establishing discriminatory power of in vitro dissolution test (Quality) • clinical relevance o

18. Slide 18: European Guidance relating to IVIVC

    European Guidance relating to IVIVC Quality Clinical IVIVC Topic Covered Section Section Study design: in vitro study design (dissolution media, sampling times)  - Study design: in vivo study design (number of subjects, sampling times) -  IVIVC data analysis: acceptable methodology for model devel

19. Slide 19: Key Messages and Opportunities

    Key Messages and Opportunities Discriminatory Power of ICH Q8 Dissolution Tests Anchor to Clinical Batches Use IVIVC to interpolate within a validated range of dissolution Align dissolution Q10 specification to average BE limits Opportunities for product specific advice (EU): · National Scientific A

20. Slide 20: What is IVIVC?

    What is IVIVC? • IVIVC: In Vitro – In Vivo Correlation • In Vitro: Dissolution • In Vivo: PK Study • Definition: An IVIVC is a mathematical relationship that is satisfactorily predictive of some key pharmacokinetic parameters (Cmax, AUC) from in vitro dissolution data © Copyright 2015 Certara, L.P.

21. Slide 21: Goals of IVIVC Exercise

    Goals of IVIVC Exercise • Determine when an IVIVC is likely to succeed or fail • Understand the data requirements • Learn to use the IVIVC Object • Model dissolution data • Estimate UIR from IR Reference formulation Cp data • Use Deconvolution to estimate Fraction absorbed • Learn to use Levy plots

22. Slide 22: The overall Workflow and the IVIVC Wizard

    The overall Workflow and the IVIVC Wizard IVIVC Wizard 2. Model Dissolution Profiles 1. Explore 4. Construct 5. Predict Data IVIVC Model PK 3. Estimate Absorption using UIR from Reference © Copyright 2015 Certara, L.P. All rights reserved. Slide 22

23. Slide 23: Workflow of the wizard: Some Details

    Workflow of the wizard: Some Details IN-VITRO ANALYSIS 1. Partition formulations 2. Model Dissolutions CORRELATION ANALYSIS PREDICTION OF PK 1. Obtain scaling and 1. Model new dissolution data Use of common shifting parameters for 2. Predict PK using formulation names both time and absorption correl

24. Slide 24: Dataset Requirements

    Dataset Requirements • In Vitro tab – 2 internal formulations recommended – No external formulations are required • In Vivo tab – 1 reference for generation of UIR (e.g. IV Bolus, IR) for each individual – Formulation names must match in vitro formulation names – In vivo data for Target formulation

25. Slide 25: Exploratory Analysis: Plot in-vitro data

    Exploratory Analysis: Plot in-vitro data © Copyright 2015 Certara, L.P. All rights reserved. Slide 25

26. Slide 26: Exploratory Analysis: Plots of In-Vivo Data

    Exploratory Analysis: Plots of In-Vivo Data © Copyright 2015 Certara, L.P. All rights reserved. Slide 26

27. Slide 27: Exploratory Analysis: Plot In-Vitro data

    Exploratory Analysis: Plot In-Vitro data © Copyright 2015 Certara, L.P. All rights reserved. Slide 27

28. Slide 28: IVIVC Wizard: In-Vitro Modeling

    IVIVC Wizard: In-Vitro Modeling Locate data and partition formulations ‒ Locate worksheet and define data columns ‒ Define Formulations for: • Internal validation and modeling • External validation Model the dissolution data – Using new sigmoidal models: Weibull, Hill, Double Weibull, Makoid- Banaka

29. Slide 29: Plots of the available dissolution models

    Plots of the available dissolution models © Copyright 2015 Certara, L.P. All rights reserved. Slide 29

30. Slide 30: IVIVC Wizard: In-vivo Analysis to estimate absorption

    IVIVC Wizard: In-vivo Analysis to estimate absorption Locate Data and map data fields – Map formulations to patients Specify “Reference” Drug Input – Dosing route can by IV or Oral – This is what is used to find the model of distribution/elimination Obtain a model of distribution/elimination ( “UIR”

31. Slide 31: Use PK modeling to establish UIR

    Use PK modeling to establish UIR We’ll use a “reference” where we know: – Input rate vs. time – Response vs. time Get back the Unit Impulse Response (UIR), same as system disposition function gives Ai and alphai depending upon number (i) of exponents required to describe the system. INPUT UIR OUTPUT

32. Slide 32: Determining the Input: Deconvolution

    Determining the Input: Deconvolution If we know: – How the system responds to input – Response vs. time Get back the required input to describe observations INPUT UIR OUTPUT Fabs(t) Plasma Concentration ? X = time © Copyright 2015 Certara, L.P. All rights reserved. Slide 32

33. Slide 33: Concentration in terms of input (absorption) and output (elimination) component

    Concentration in terms of input (absorption ) component and output (elimination ) component Concentration = Absorption  Distribution/Elimination © Copyright 2015 Certara, L.P. All rights reserved. Slide 33

34. Slide 34: Estimation of UIR from IR as Reference

    Estimation of UIR from IR as Reference • Using IV bolus or infusion for estimating Unit Impulse Response (UIR) may be desirable to avoid questions on flip-flop with extravascular • Most often, it is not possible to get plasma Concentration data following IV route  May not be available in public dom

35. Slide 35: IVIVC Wizard: Correlation Analysis

    IVIVC Wizard: Correlation Analysis • Do Exploratory Plots (Levy Plots) – Examine Fabs vs. Fdiss or Levy Plot (Tvivo vs. Tvitro) • Account for Tscale, Tshift, AbsScale and AbsBase if required • Obtain scaling parameter for time and shifting parameter for absorption – Use built-in equation or custom m

36. Slide 36: Fabs vs Fdiss Plot

    Fabs vs Fdiss Plot Can be used to determine if AbsScale or AbsBase should be accounted for in the Correlation model. Since Fdiss will usually go to 1, the maximum Fraction Absorbed would be the Abs scale. © Copyright 2015 Certara, L.P. All rights reserved. Slide 36

37. Slide 37: Levy Plot

    Levy Plot Can be used to determine if Tscale or Tshift should be accounted for in the Correlation model. © Copyright 2015 Certara, L.P. All rights reserved. Slide 37

38. Slide 38: Correlation Models that are Linear in terms of the equations

    Correlation Models that are Linear in terms of the equations 1. Fabs=AbsScale*Diss(TScale * Tvivo) 2. Fabs=AbsScale*Diss(TScale * Tvivo - TShift) 3. Fabs=AbsScale*Diss(TScale * Tvivo - TShift) – AbsBase Fabs vs. Fdiss Plots and Levy Plots can be used to help determine which of the three linear model

39. Slide 39: IVIVC Wizard: Correlation Analysis

    IVIVC Wizard: Correlation Analysis Build and Validate Correlations – Uses partitioning of data formulations done in first Wizard tab – “Internal” ones used to build model; “external” used to validate – Obtain results using a “best-fit” approach using AIC or SSR criteria Examine Correlation Grid – Sh

40. Slide 40: Validation of the Correlation Model

    Validation of the Correlation Model What is the FDA recommended %Prediction Error (%PE) in order for an IVIVC correlation to be valid? • % PE should be less than 15% for any one formulation • However, on average of the internal prediction errors must be less than 10% • The Absolute errors are used ©

41. Slide 41: Predicting PK behavior of new dissolution profiles

    Predicting PK behavior of new dissolution profiles 1. Select a test formulation dissolution dataset 2. Define dissolution model (same as the one used in IVIVC model) and fit 3. Select the innovator for comparison 4. Predict PK for test formulation based IVIVC model 5. Compare with innovator’s PK dat

42. Slide 42: Q&A

    Q&A © Copyright 2015 Certara, L.P. All rights reserved.

43. Slide 43: Training Courses

    Training Courses • Certara offers introductory, intermediate, and advanced courses for Phoenix products • New Courses added in Europe – Sample listing: • IVIVC Toolkit for Phoenix WinNonlin – In Vivo-In Vitro Correlations • April 21-22,2015 in Dusseldorf, Germany © Copyright 2015 Certara, L.P. All r

44. Slide 44: Contacts & Follow-Up: Thank you!

    Contacts & Follow-Up: Thank you! Americas Telephone (US): 888-708-7444 Telephone (Intl): +1-919-852-4685 Email: sales@certara.com Europe Telephone: +33 1 56 95 17 10 Email: sales@certara.com India Telephone: +91 984 500 6625 Email: sales@certara.com Japan Telephone: +81 (0)50-3786-9898 Email: sales@

Certara — Technical Presentation (February 2015) on 6ix