APS - Progress within the IMI OrBiTo project predictive tools for oral biopharmaceutics GSK Stevenage Kristin Lacy, Alison Margolskee, Adam Darwich, Xavier Pepin, Amin Rostami Development of the compound database Tuesday 13th May 2014
Amin Rostami and Xavier Pepin 2014-05-13 WP4 Objectives within OrBiTo Refine existing in silico mechanistic tools to improve absorption modeling Define the gaps in the models Propose tools to calculate model performance Change the inputs to the models Change the algorithms to better reflect biology and drug formulation behaviour Monitor performance Use new models to increase biowaiver scope
Requirements for the OrBiTo compound database Amin Rostami and Xavier Pepin 2014-05-13 Be novel (capture EFPIA data) Be secure, accessible to all partners Data capture offline with standard tools Be searchable Feasibility to selectively blind some fields Allow interaction whilst maintaining full or partial anonymity Be flexible to allow new fields to be captured Be fast critical path for WP4
OrBiTo Database How does it work? Amin Rostami and Xavier Pepin 2014-05-13 XML File 1. Data gathering using Excel Plug-in developed by Simcyp 2. An XML file is generated containing the data collected in the Excel plug-in 3. XML file is imported into the OrBiTo database which is stored within the Cloud 4. Users query the database using the OrBiTo Query Language 5. Data is downloaded from the database and the Excel plug-in is re-populated with the downloaded data.
OrBiTo Database Backups Backups of the database are taken once a week so if any failures were to happen then we would be able to rebuild the database. All backups are stored within the Cloud and are kept for 90 days.
OrBiTo Database How/when was it built? Amin Rostami and Xavier Pepin 2014-05-13 August 2012 : Pre-project discussions started between Sanofi & Simcyp to decide what type of data, metadata was to be captured and in what units and format. A decision about how the data would be collated was also made during this time (Excel plug-in). Data type Metadata Value Unit Format Comments For what purpose Minimal dataset Main challenge with the drug Indicate the main challenge from picklist and also free text Text Drop down menu 0 Important n / 3 matrix [n values rounded to Indicate basic"b" of acidic "A" nature + method of obtention Calculate ionization, distribution, pka no unit +/- 0.1 / "A" for acid or "B" base/ Indicate all the pkas of the drug Compulsory (calculated or measured, indicate batch number used) permeability vs ph Method of obtention] For blinding purposes the use of ranges could be made. To calculate l the diffusion i coefficient i Molecular weight of active moiety None g.mol-1 1 value +/- 1 g/mol Compulsory Proposal to round it to +/- 1 g/mol (dissolution rate) +/- 1 g/mol API Log P Shake flask, ph-metric, ph-uv, HPLC, CE no unit (log ratio) API Log D = LogP @phn Shake flask, ph-metric, ph-uv, HPLC, CE no unit (log ratio) Apparent API permeability through membrane (Bi-directional) Indicate membrane nature (strain of cells or artificial) + drug concentration in donor+ direction or transport + ph in donor + ph in acceptor + recovery + inhibitor nature + inhibitor concentration + BSA concentrations in donor + value nm/s [1 value +/- 0.1 / Method of obtention = DDM5] n / 3 matrix (n ph values / n Log D values +/- 0.1/Method of obtention = DDM5] n/14 matrix [n values Papp (+ - 0.1 nm.s-1) / membrane nature = DDM1 / drug concentration in donor (µm)/ "A2B" or "B2A"/ ph in donor + - 0.1/ ph in acceptor + -0.1/ recovery (+ - 1%)/ inhibitor name / inhibitor concentration (µm)/ BSA concentration in donor (%)/BSA concentration in acceptor (%)/ Agitation rate (rpm)/ type of agitation = free text/ Size of Unstirred Water Layer(µm)/other info = free text/ Cross reference to method PxMn] Anticipation of many partition properties with/through biological membranes. Anticipation of log P = Log D of the unionized species Compulsory solubility in micellar systems, use in the ACAT model for absorption scaling factors QSAR modeling, anticipation of Choose as many log D as different types of ionized species based many partition properties on the pkas. Choose ph rationnaly, i.e. ph1 = pka1-2 ; ph2 = with/through biological membranes. Compulsory (pka1+pka2)/2...etc or best reconstruct full lipophilicity profile Anticipation of solubility in micellar systems Need to define a list of reference drugs to be tested at a certain concentration in the apical compartment to serve as reference for the Papp measurements coming from different labs, provide the Papp of these "reference drugs" in a separate table with open name for the drug but similar format indicated in format column Use to run/ validate in silico tools and biopharmaceutical models Compulsory
OrBiTo Database How/when was it built? Amin Rostami and Xavier Pepin 2014-05-13 November 2012: 1 st Excel plug-in was developed. Based on the structure of the Excel plug-in and the type of data collated, the database architecture was designed.
OrBiTo Database How/when was it built? Amin Rostami and Xavier Pepin 2014-05-13 December 2012: Agreement on numbering and blinding strategy January 2013: First usable Excel plug-in 1.1 distributed to EFPIA partners for T4.2, T4.3 & T4.4 Jan 2013-Sept 2013: Data capture from EFPIA
OrBiTo Database How/when was it built? Amin Rostami and Xavier Pepin 2014-05-13 Oct 2012 - June 2013: Development of the OrBiTo web database by SimCYP/Certara May 2013: Sanofi to pay the web cloud services to cover project duration 6 th June -Sept 2013: Data Upload open to EFPIA Sept 2013-March 2014 Data gap analysis & update (T4.6) Data gap identification feed-back to EFPIA Gap filling if possible SimCYP/ Manchester Uni + EFPIA
What is currently in the OrBiTo database API database characteristics: 86 compounds Considerable proportion of solubility limited compounds comparable to other databases.
What is currently in the OrBiTo database - Formulations Database mainly consisting of immediate-release solutions and tablets with a small proportion of controlled-release 4.4 3.62.0 1.6 0.6 IR crystalline 7.8 37.0 43.1 IR solution aqueous Prolonged release IR solution non aqueous IR amorphous Undefined Delayed release IR emulsion
OrBiTo database PK studies Studies = 489 Arms = 1576 1214 in fasted 274 in fed state 88 no mention of prandial state Subjects = 9231
OrBiTo database Administration routes IV 215 Oral 1278 Other (specify) (p 29 SC 4 Unspecified 50 Total 1576
OrBiTo database Administration routes Other administration routes Human or canine administration to the lower segments of the intestine Other route\apis A2853 A3837 A5766 A6598 A6646 A6939 Total Ascending colon 3 3 1 2 1 1 11 dog (beagle) 1 1 human 3 2 2 1 8 rat (specify strain) 1 1 2 Descending colon or rectal 4 4 human 4 4 Distal Small Bowel 1 1 1 1 4 dog (beagle) 1 1 human 1 1 1 3 Jejunum 1 1 2 dog (beagle) 1 1 human 1 1 Stomach 1 1 human 1 1 Total 4 4 3 3 7 1 22
Data gap analysis Task 4.6 1- Objective of the gap analysis : Improve data quality through gap analysis and active participation of EFPIA 2- Select dataset for task 4.9 : evaluation of performance of existing PBPK tools using blinded human dataset : Bottom-up anticipation of human PK and systematic evaluation of model failures Run by 3 independent partners comparison of selections and final recommendation
API data files Database query-tool API summary data
Data gap analysis Example of data quality Apparent permeability 5 APIs with no permeability data (some are drugs of pro-drugs) Large disparity in # of datapoints Majority with 2 or 1 datapoint In Feb 2014 : Only 35 API with reference compound Papp Issue for scaling to human Peff # of permeability data # of APIs 0 5 1 16 2 30 3 4 4 8 5 2 6 5 7 1 8 5 9 1 10 4 12 2 14 3
Filling the gaps: Strategy Commenting on missing parameters through orbitodatabase.eu Email updates twice per week on the status of parameter information for key parameters Parameters of interest: LogP/D Solubility Permeability Clearance Particle size fup BP Human oral PK data Human i.v. PK data
Filling the gaps: Strategy Commenting on missing parameters through orbitodatabase.eu Unique feature of info communication through database whilst keeping anonymity
Data gap analysis T4.6 Regular interaction with EFPIA increase data quality
Filling the gaps: Progress
Data gap analysis T4.6 Other criteria for clustering Formulations types Type of studies
Task 4.6 : gap analysis : Final selection Final selection included 43 APIs In order to expand M&S API set some deficiencies were allowed, including: Lack of blood-to-plasma ratios, lack of fumic and clearance informed via allometric scaling.
Modelling and simulation task: Plan and progress The final 43 APIs were allocated based on available man months with a considerable overlap for 10 APIs in order to test operator differences between sites. Deadline is currently set for end of September 2014. Based on Selection Criteria Academic contributors and EFPIA EFPIA API database Gap analysis API selection M&S Task Statistical evaluation ADAM ACAT GI- Sim June 2013 Feb 2014 Dec 2013 Feb 2014 March 2014 March September 2014 November 2014
Uses of the OrBito compound database T4.9 (Feb-Nov 2014) Gap analysis Bottom up anticipation of human PK from in vitro and animal data Systemic evaluation of model performance based on quality indicators (T4.8) Evaluation of gaps depending on biopharmaceutical space/type of formulation T4.7 (March Dec 2014) Regular updates Completion of missing data / Incomplete data (EFPIA) Creation of new fields in the database in support of other WPs work (SimCYP) WP1-3 : Identification of drugs of interest for performing tests in vitro or in vivo T4.20 : testing of various improvement options for in silico models Continuous improvement (new fields, new data, new simulations) monitor goodness of fit to increase performance
OrBiTo database creation and improvement Acknowledgements Amin Rostami and Xavier Pepin 2014-05-13 Kristin Lacy Steve Andrews Philip Hayward Ian Gledhill Shriram Pathak Adam Darwich Alison Margolskee All the EFPIA partners Sanofi (sponsor)
Planning 2013 2014 2015 2016 2017 2014 : Biggest year for WP4 Completion of T4.6, T4.9, T4.7, T4.5 Start of T4.10-T4.19
Status WP4 Amin Rostami and Xavier Pepin 2014-04-23 Overall status Tasks Milestones/ Deliveries People Comm. Budget Important updates Issues Mitigations Review of available PBPK models and gaps in physiology is accepted and available online Data gap analysis is completed (T4.6) T4.9 is started T4.7 Increase of budget in SimCYP to perform this task (for maintaining database and adding new fields) T4.5 not started T4.5 proposition to integrate WP3 data in vivo instead since RIVM database unavailable Completed On track Issues At risk Change from last report