Information Sharing and Automation of PK Calculations for Efficient Project Support

Transcription

1 Information Sharing and Automation of PK Calculations for Efficient Project Support Tibco Spotfire UM, Basel, Nov 3 rd, 2016 Stefanie Bendels Quantitative Systems Pharmacology Pharmaceutical Sciences, Roche Innovation Center Basel Data storage alone is not sufficient Access makes the difference Efficiency Relevance Real-time Integrity 2

2 Workflow in Drug Discovery High quality data Knowledge generation Informed decisions Consistency & accessibility Find patterns & relationships Right compound & target For value generation efficient access to data is crucial Efficient access Structured data storage Combination of different data sources Automation of workflows Talk Visual exploration Panel of exploratory tools Catalogue of pre-defined analysis Flexibility by interactive visualizations PKPD and statistical analysis Interpretation of results Clear documentation and communication Replacements for manual workflows to enable exploratory analysis using Spotfire and Pipeline Pilot (Discngine) 3 Former Workflow for pharmacokinetic calculations Compilation of in vivo and in vitro data Manual collection from different sources: Internal databases with different user interfaces Excel sheets from assay owners Data management software user user user user user 4

3 Filtering, Pivoting, and Calculations Manual input and updates in Excel files Decrease in efficiency Different methods for calculations Dependent on user different results might be reported Potential to do mistakes or do not use different methods 5 New PKPD Module in SpotAPP Prototype enabling project-specific analyses and views Input Roche internal databases Spotfire Automated Project Data Processing User Processing PKPD calculation module Custom parameters PK profiles Output Visualizations with interactivity 6

4 SpotAPP Concepts for the PKPD Module Design Goal: Provide facilitated processing of PKPD data for DMPK experts and provide key visualizations for project teams Key principles: Automation using a central R script Implementation of different calculation methods Comparison and selection of most appropriate workflow Simple customization of script behavior per project Team file Data processing Main data table Process controlled by DMPK expert PKPD tables DEV: File used for DMPK expert work 7 SpotAPP Intrinsic Liver Clearance from in vitro Hepatocytes Results using different calculation methods Intrinsic clearance [ml/min/kg]: Measure for liver capacity to metabolize compound In vitro/in vivo extrapolation IVIVE: Confidence in clearance translation and dose estimate for human? For calculation: Parameter needed that corrects for binding effects in in vitro incubation medium in vitro intrinsic clearance Dilution method Estimate protein binding (from fraction unbound measured for different species) in vitro intrinsic clearance Houston method Estimate unspecific binding in vivo intrinsic clearance in vitro intrinsic clearance in vivo intrinsic clearance Assume no binding in vivo intrinsic clearance 8

5 SpotAPP Interactive Database Information Link for SDPK Data on Animal Level Dilution method Administration route and matrix in vitro intrinsic clearance Retrieve raw data for observed clearance in vivo intrinsic clearance Concentration Dose Matrix Concentration Animal and matrix Details 9 Storage and access of in vivo PK and PKPD data For PKPD analysis pharmacokinetic (PK) and pharmacological data (PD) have to be combined and have to be easily accessible Pharmacology (PD) Access PK and PKPD data Pharmacokinetic (PK) Database For DMPK experts and project teams 10

6 Discngine/Pipeline Pilot Concentration Time Profiles (PK) in Spotfire Overview on available data Result tables are loaded into predefined Spotfire templates Matrices Doses Filters Projects Compounds Concentrations Administrations Species Cross tables for selections Dose-normalized concentrations Details on demand 11 Discngine/Pipeline Pilot Concentration Time Profiles (PK) in Spotfire Linear/nonlinear concentration-dose dependency Linear PK in the tested dose range Concentration D norm. concentration Concentration Dose-normalized concentration 12

7 Discngine/Pipeline Pilot PKPD Data in Spotfire Overview on available data Enzyme inhibition by tested compound Readout over Time, Mean per dose group with SD Pos. Control Cmpd 1 Readout over Time, Animal level Pos. Control Cmpd 1 Cross table for selections Cmpd 2 Vehicle Cmpd 2 Vehicle Compound might have been tested in different experiments 13 Discngine/Pipeline Pilot PKPD Data in Spotfire Pharmacological response per experiment Readout over Time, Mean per dose group with SD Readout over Time, Animal level Vehicle Pos. Control 14

8 Discngine/Pipeline Pilot PKPD Data in Spotfire Pharmacological response over plasma concentration Readout over Cplasma, Mean per dose group with SD Readout over Cplasma, Animal level Cmpd 1 Cmpd 1 Cmpd 2 Cmpd 2 Plasma concentration Plasma concentration 15 Discngine/Pipeline Pilot PKPD Data in Spotfire Pharmacological response over plasma concentration Readout over Cplasma, Mean per dose group with SD Readout over Cplasma, Animal level Cmpd 1 Cmpd 1 Cmpd 2 Cmpd 2 Easy switch to semi-log plots Plasma concentration Plasma concentration 16

9 Impact Tools for efficient data access using flexible and customizable solutions Pipeline Pilot protocols can be implemented in Spotfire/Discngine and in other applications like SpotAPP Quick and informative overview on generated data with quality assessment (controls over time, extreme values) Easy evaluation of different calculation methods Consistent results due to same methods used for calculations and processing Time savings Calculations and updates without manual input Processed and formatted data for detailed data analysis Control of which data is exposed to project teams to avoid overloading Direct interaction with customers needed to ensure engagement!! 17 Acknowledgments Pharmaceutical Sciences Rubén Alvarez Dragomir Draganov Holger Fischer Rodolfo Gasser Michael Gertz Martin Kapps Nicole Kratochwil Thierry Lavé Andrès Olivares Axel Pähler Neil Parrott predi Christian Blumenröhr Valerie Chabrier Hansjörg Graesslin Torsten Schindler Daniel Wenger TMo CADD Nicolas Zorn TMo Medicinal Chemistry Giuseppe Cecere Wolfgang Haap DTA-NORD Karlheinz Baumann 18

10 Doing now what patients need next Graphical data exploration is the first step in data analysis It enables to Comprehend information quickly Display information in a way that is meaningful Find relationships, patterns, and understand data Identify areas that need attention or improvement Clarify which factors influence experimental results without using more elaborate PKPD methods

11 We need The combination of multiple data sources An automatic and up-to-date project data delivery The possibility for customized data processing (filtering, pivoting, calculations) Efficiency (spending less time preparing data) Evaluation of different calculation methods to identify most suitable for specific project to enable pharmacokinetic (PK) calculations e.g. in vitro-in vivo extrapolation (IVIVE), early dose estimates PK Calculations with Direct Access to PK Profiles in SpotAPP Implementation of PK calculations ready for all projects Scaling of in vitro hepatocyte/microsome clearance with IVIVE (different methods for estimation of fraction unbound in experiments) Estimation of human dose for best case scenario (100% absorption, hepatic clearance as main pathway, IVIVE established) Efficacy index EI Interactive visualization of concentration-time profiles from database (database information link) 22

12 Discngine/Pipeline Pilot Example 1: In vivo Concentration-Time Profiles (PK) Goals Overview on generated data (compounds, conditions) Download and processing for interactive, exploratory analysis (dose normalizations, harmonization of units) Quality of measured data Identification of compounds with linear/nonlinear PK properties Inter-individual variability for concentrations measured in different animals Formatting for further PK analysis in additional software packages Time saving and reduction of copy/paste errors Customers: DMPK experts, project teams 23 Discngine/Pipeline Pilot Example 2: In vivo PKPD project data Goals Overview on generated data (compounds, conditions, dose, time points) Download and processing for interactive, exploratory analysis (normalizations, mean and variability within dose groups) PD effect over time, plasma and brain concentrations Brain/plasma distribution Identification of animals with extreme values Quality control for reproducibility of negative and positive controls Formatting for further PKPD analysis in additional software packages Time saving and reduction of copy/paste errors Customers: Project teams, M&S scientists 24

13 Doing now what patients need next