In Vitro Safety Profiling During Lead Optimisation Murray Brown Manager, Data Interpretation and Business Process Screening and Compound Profiling Drug Discovery Process Target Selection Basic Research Years Candidate Selection Lead Discovery 3 Preclinical Development 1 IND filing NDA filing Clinical Development 6 FDA Filing 1.5 Once a candidate is selected the pharmacological properties of the molecule are fixed High Throughput Screening approaches apply within Lead Discovery to: – Reduce the cycle time from target selection to candidate selection – Increase the number of candidates per program – Increase the quality of candidates selected Strategies to Improve the Quality of Lead and Candidate Generation Robust screening infrastructure – Automation quality control – Statistical methods for hit selection Quality “drug-like” compound libraries – Low molecular weight and cLogP New Screening Paradigms – Fragment Screening – Encoded Library technology Relevant assay biology – Native cellular systems – Biophysical Screening Pharmacology of lead series – Ability to assess on-target and off-target activity Causes of Attrition – Safety & Efficacy Reasons For NME Termination By Stage 2005-2009 Industry Portrait KMR Terminology Early Dev: Preclin up to Ph III Start Late Dev: Start Ph III to Launch Termination Reasons EARLY LATE Strategic 243 18 Resources 19 2 Efficacy 229 25 Safety 457 8 Technical 93 0 Unknow n 42 3 Total 1083 56 Strategic Resources Efficacy GSK EARLY 29 0 25 LATE 3 1 2 Safety 65 4 Technical 7 0 Unknow n 5 0 Total 131 10 KMR Group R&D General Metrics Study Final Report July 1, 2010 Addressing the Challenge of Drug Safety in Early Discovery Vision • To increase candidate quality and probability of clinical success through the identification and mitigation of safety hazards in chemical series prior to candidate selection Strategy • Using high throughput techniques, implement panels of assays for use in early discovery that identify likely safety liability in hits, leads and candidates Attrition Reduction Activities in GSK Strategic Intent – Implement assays during H2C to identify and manage compound series likely to cause toxicity in preclinical or clinical studies eXP Cardiotoxicity Hepatotoxicity Genotoxicity A bi-weekly panel of 50 molecular target assays with known clinical liability A panel of Ion Channel assays enabled by high throughput electrophysiology Cell Health assay detects 70% of known hepatotoxicants GreenScreen assay licensed to identify Genotoxicants Assays configured during 2009-2010 with capacity to screen 1200 hits, leads and candidates/year eXP (enhanced cross screen panel) Neuronal: Cardiac/vascular: KCNQ1/minK L-type Ca VR4 M2 Adenosine 2a β2 adrenergic α1b adrenergic α2a adrenergic a1b3g2 GABA SERT DAT NET M1 MAO-B μopioid κopioid Dopamine 1 Dopamine 2 Histamine 1 NK1 Hepatic: OATP1B1 PXR Gastro-intestinal: 5-HT3 PDE4B GSK3b PI3k Immunological: LCK Cannabinoid 2 Nav1.5 Kv1.5 5-HT1B 5-HT2A 5-HT2C COX2 V1a Neuromuscular: a1bgd nAChR 11 point dose response curve Functional/activity assays Interpharma Safety Profiling Knowledge-Sharing - Secondary Pharmacology Screening AstraZeneca Outcomes: Joanne Bowes GlaxoSmithKline 2009-present • Poster at SPS meeting: Andrew Brown Arun Sridhar • What • Rational design of an inTargets? vitro safety profiling panel to reduce undesired secondary pharmacology of drug candidates. • What Technologies? Steven Whitebread, Joanne Bowes, Andrew Brown, Jacques Hamon, Wolfgang G. Jarolimek, Gareth Waldron and Arun Sridhar Journal of Pharmacological and Toxicological Methods; 64(1), July-August 2011, Page e18. • What Process? • Shared Case Studies • Manuscript - in preparation Pfizer Gareth Waldron Pharmaxis Wolfgang Jarolimek Novartis Steven Whitebread Jacques Hamon High Throughput Electrophysiology Assays to predict Functional Cardiotoxicity Cardiac Ion Channels CNS (GABAA) (Barracuda) hERG NaV1.5 CaV1.2 KV1.5 KCNQ1 (IonWorks® and PatchXpress®) See posters by Metul Patel et al on hERG IonWorks® population patch clamp and Joanna Taylor et al on stem cell derived Cardiomyocytes Genetic Toxicology: The GreenScreen Assay Genotoxic agents either react directly with DNA or disrupt the cellular apparatus which regulate the fidelity of the genome Regulations require a minimum of 3 GLP tests: – a test for gene mutation in bacteria (for example, the Ames test), – a test for chromosomal aberrations in vitro or the MLA – an in vivo test for chromosomal damage in rodent haematopoietic cells. The GADD45a gene is upregulated in response to DNA damage in the GreenScreen assay (Gentronix) Reporter transfected into Human p53 competent lymphoblastoid cells Genetic Toxicology: GreenScreen Assay 15% of pre-candidates are terminated due to Genotoxicity GSK has licensed the GreenScreen HC genotoxicity assay for profiling of hits, leads and candidates 110 3.5 3 90 80 2.5 70 60 2 50 1.5 40 30 Relative GFP Induction 31 of 34 known genotoxic agents induced GADD45a reporter Relative Suspention Growth 100 1 20 10 0.5 Blank 0.03 0.06 0.13 0.25 0.50 1.00 2.00 4.00 8.00 µg/ml Etoposide 110 3 90 2.5 80 70 2 60 50 1.5 40 30 1 Relative GFP Induction 100 Relative Suspention Growth 41 of 41 nongenotoxic agents did not induce the GADD45a reporter 20 10 0.5 Blank 7 14 28 57 114 228 µg/ml D-Mannitol 456 911 1822 Early stage (HitID and SoC) identification of GreenScreen HC actives enable LO chemistry to focus on molecules without this liability BlueScreen HC has now been implemented to allow higher throughput See poster by Kate Simpson et al on BlueScreen assay validation Cell Health Assay to Detect Hepatotoxicants Drug-Induced Liver Injury (DILI) is a recurrent problem in pharmaceutical development Idiosyncratic hepatotoxicity is one of the leading causes of drug withdrawals, non-approvals and warnings (Kaplovitz 2005) Can we identify hepatotoxicants prior to candidate selection and reduce attrition due to pre-clinical or clinical hepatotoxicity? • 96well assay using HepG2 (Human liver carcinoma) cells GSK Cell Health Assay Description Measures cytotoxic effect of compounds in human liver-derived HepG2 cells in 384-well format 3 parameter automated imaging assay Using fluorescent staining, the key parameters measured in this assay are : Nuclear Condensation Hoechst 33342 Cell permeable DNA binding dye Mitochondrial membrane potential TMRM Accumulates in healthy Mitochondria but leaks out when mitochondrial membrane potential is discharged Membrane permeability TOTO-3 Cell membrane impermeable nuclear stain Impaired mitochondrial function is an early indicator of cell injury whereas loss of membrane integrity and changes in nuclear morphology are indicators of acute or late stage cytotoxicity. Quantification is carried out using the InCell Example images Nuclei Mitochondrial Potential Membrane Permeability Negative Control Postive Control Typical dose response curves Correlation between Cell Health readouts Compounds usually show very similar IC50s in all 3 readouts, but there are exceptions where toxicity is specific to a single readout Cell Health Assay to Detect Hepatotoxicants The Cell Health assay for profiling of hits, leads and candidates Negative compounds Human and rat hepatotoxicants Cytotoxicants Idiosyncratic human hepatotoxicants 28/28 29/30 4/4 3/18 Early stage elimination of Cell Health actives enable LO chemistry to focus on molecules without this liability Compound A failed due to liver toxicity. A B C 150 125 100 No reported hepatotoxicity for Compounds B and C. 75 50 25 0 -25 -50 5E-7 1E-6 5E-6 1E-5 5E-5 0... 0.0... 5E-7 1E-6 5E-6 1E-5 Concentration 5E-5 0... 0.0... 5E-7 1E-6 5E-6 1E-5 5E-5 0... 0.... Frequency of toxicity in Cell Health for marketed drugs and failed clinical candidates 100% Failed Candidates twice as 90% 80% likely to be active in Cell Health than marketed drugs 70% 59 Inactive in Cell Health 226 60% Significant proportion of marketed drugs show toxicity in Cell Health 50% 40% 55 30% 20% 71 10% 0% Drug Drugs FC Failed Development Candidates Active in Cell Health Toxic compounds? Toxic dose Phillippus Aureolus Theophrastus Bombastus von Hohenheim (1493-1541) Paracelsus Alle Ding' sind Gift, und nichts ohn' Gift; allein die Dosis macht, daß ein Ding kein Gift ist. "All things are poison and nothing is without poison, only the dose permits something not to be poisonous." Cell Health cytotoxicity vs normal and toxic exposure levels Therapeutic or ‘normal’ blood concn Toxic blood concn Exposure level 100 uM 1 uM 10 nM 100 uM Cell Health pIC50 Therapeutic and toxic blood concentrations of more than 800 drugs and other xenobiotics M. Schulz, A. Schmoldt Pharmazie 58(7) 2003 447-474 10 uM http://fscimage.fishersci.com/webimages_FSC/downloads/winek.pdf What drives cytotoxicity? i) Physchem properties - clogP What drives cytotoxicity? ii) Physchem properties – rotatable bonds What drives cytotoxicity? iii) chemical series Cell Health cytotoxicity by chemical cluster Cell Health pIC50 clogP distribution for each cluster clogP vs Cell Health pIC50 Cell Health Cytotoxicity is SAR-able in lead optimisation Physicochemical properties can be manipulated to reduce likelihood of Cell Health cytotoxicity – ↓clogP, ↓# aromatic rings, ↓heavy atom count (or ↑ >50!), ↑ heteroatoms, ↓ rotatable bonds Cell Health cytotoxicity is a feature of chemical series beyond their physicochemical properties (toxicophores) – Early screening in Cell Health assay at HitID allows selection of series with lower likelihood of cytotoxicity Cell Health Assay Related to Promiscuity in eXP 1 100% 3 5 90% 80% 36 70% 60% 156 17 50% 19 24 40% 30% 19 20% 10% 28 0% x ≤ 0.10 0.10 < x ≤ 0.20 0.20 < x ≤ 0.30 0.30 < x ≤ 0.40 Fraction assays with pXC50 >5 Green = Inactive in Cell Health Red = Active in Cell Health 0.40 < x In vitro cross screening profile of selected drugs Low promiscuity, mechanism based toxicity detected by eXP, non-toxic in Cell Health Non-promiscuous, highly tolerable, non-toxic in Cell health a1B Moderately promiscuous and highly tolerable, non-toxic in Cell health CB2 Ag Assay a2C b2 Highly promiscuous, low tolerability, toxic in Cell Health Attrition Reduction Toolkit: An annotated one-stopshop for all attrition reduction assays at GSK Attrition Reduction Toolkit: An annotated one-stopshop for all attrition reduction assays at GSK Compound Structure Profile similarity in eXP Conclusions We have implemented a panel of assays in early discovery to assess toxicity hazards in hit and lead series Assays are annotated according to likely clinical effect Each assay is not decision making in isolation – Data enables comparative decisions between chemical series – Activity in multiple assays needs to be considered Assays can be used to drive SAR Cell Health assay in HepG2 cells concords well with physicochemical properties shown to be important in clinical attrition and clinical in vivo tolerability Exposure levels are key for toxicology (as well as efficacy) expected dose is important factor to be included in interpretation of early Safety Profiling data Acknowledgements Steve Rees (formerly Screening and Compound Profiling) Andrew Brown (Screening & Compound Profiling) Dave Morris (Screening & Compound Profiling) Wolfgang Jarolimek (formerly Screening and Compound Profiling) Joanna Taylor (Screening & Compound Profiling) Kate Simpson (Screening & Compound Profiling) Metul Patel (Screening & Compound Profiling) Rob Jepras (Screening & Compound Profiling) Rob Eagle (Screening & Compound Profiling) Darren Green (Computational & Structural Chemistry) Cerys Lovatt (Safety Assessment) Julie Holder (Safety Assessment/Stem Cells) Nick McMahon (Safety Assessment) Paul Hastwell (Safety Assessment) Patrick Wier (Safety Assessment) Steve Clarke (DMPK) Bob Hertzberg (Screening & Compound Profiling) Many other GSK scientists responsible for generating the assays and data