Annotation Of FDA Labels For The Understanding Of Drug Induced Liver Injury

Transcription

1 Annotation Of FDA Labels For The Understanding Of Drug Induced Liver Injury Weida Tong, Ph.D Division of Bioinformatics and Biostatistics, NCTR/FDA 1

2 2

3 Some Drugs Are More Likely to Cause DILI HO O Ibuprofen OTC drug On the market >30 yrs with not much hepatotoxicity HO O Ibufenac withdrawal Marketed in 1966 and withdrawn in Feb, 1968 due to hepatotoxicity (no facts given). Late study demonstrated elevated ALT in 12/36 patients and jaundice in 5/400 cases Zolpidem safe drug Approved in 1992 and only one case related to hepatotoxicity has been reported Alpidem withdrawal Withdrawn because of causing death or requiring liver transplantation 3

4 Total R&D Expenditures ($B) R&D Cost per NME ($B) The R&D investment required to bring a new drugs to market has tripled, from $770M per NME in 1999 to $2.3B in 2010 GEN, 32(9), 18, 2012

5 Basic research innovations not reaching consumers 5

6 What is the problem? Drug discontinuation and withdrawal Landis & Kola, 2004 Nature reviews / Drug Discovery Failure to manage attrition during development Enforced withdrawals from the market place Efficacy and safety still the main reason for drug discontinuation From Tim Hammond, PhD, AstraZeneca 6

7 Safety related reasons of drug attrition Phase Preclinical Phase I-III Phase-IV #Drugs Cardiovascular: 24% 27% 35% 21% 45% Hepatotoxicity: 15% 8% 29% 21% 32% Haematology/BM: 3% 7% 3% 4% 9% Nervous system: 12% 14% 2% 21% 2% Immunotox; photosensitivity: 7% 7% 10% 11% 2% Gastrointestinal: 5% 3% 2% 5% 2% Reprotox: 9% 13% 5% 1% 2% The drug attrition in other toxicity domains not mentioned above are less than 9% 1-9% 10-19% >20% Adapted from Redfern WS et al. The Toxicologist 2010; 114 (S-1),

8 Liver Toxicity Knowledge Base (LTKB) Preclinical Premarket Postmarket - Predictive models - Biomarkers DILI Liver toxicity in humans LTKB Components A broad range of data associated with marketed drugs An array of predictive models that can be used individually or in combination for DILI assessment 8

9 - Chemical structure (SAR/QSAR) - Daily dose (Cmax) - Lipophilicity - Reactive Metabolites - P450 activities - Genetic variants - Sex-biased effects - Adverse events - Japanese TGX Project ~170 chemicals 4 testing systems Multi-doses and times >20,000 arrays - DrugMatrix ~700 chemicals ~5000 arrays - Others (e.g., in-house) >40K arrays (>500 drugs) In vitro assay (>200 drugs) Therapeutic uses Side effects DILI risk - DILI annotation - Severe DILI - In house in vitro data Human primary hepatocytes Rat primary hepatocytes HepG2 - Tox21 and ToxCast - Others - DILI types - DILI ontology Chen et al, Liver Toxicity Knowledge Base (LTKB) A Systems Approach to a Complex Endpoint, Clinical Pharmacology & 9 Therapeutics, 93(5): , 2013

10 LTKB TM Publicly Available Assess DILI risk of new chemicals Chemical structure & similarity search Link to external resources, e.g., PubMed, Wiki, LiverTox Excel-like spreadsheet Query interface Drug specific data 10

11 Increased Biological Complexity Liver Toxicity Knowledge Base (LTKB) - General Approaches PRECLINICAL DATA Drug Properties Genomic Data Cellular Response Predictive Models CLINICAL USES DILI Severity DILI Types Mechanisms Animal Data 11

12 Need a drug list with accurate DILI annotation!!! DILI Positives DILI Negatives Translational Biomarkers 1. Genomic biomarkers 2. In silico biomarkers 3. Mechanistic biomarkers 4. Integrated biomarkers 5..

13 How to Assess DILI Risk for a Drug? Three attributes of a drug are important for its DILI assessment: Causality: was liver injury caused by drug or other cause Incidence: how many case reports are considered significant Severity: elevated ALT; Hy s law; disability and hospitalization, liver failure; liver transplantation or death Risk = (How likely) x (How many) x (How severe) This is opinion based!!!

14 Drug Labeling As an Alternative Means for DILI Classification Drafted by manufacturers and approved by FDA Adverse drug reactions are obtained from clinical trial data and postmarketing case reports, and are included in three main sections: Boxed Warning (commonly known as black box warning): serious warnings, particularly those that lead to death or serious injury Warning and Precautions: clinically significant adverse reactions (indicating any that are potentially fatal, are serious even if infrequent, or can be prevented or mitigated through appropriate use of the drug) Adverse Reactions: overall adverse reaction profile of the drug based on the entire safety database does not include all adverse events observed during use of a drug, only those adverse events for which there is some basis to believe there is a causal relationship between the drug and the occurrence of the adverse event The relative severity of adverse effects is apparently in descending order of importance: Boxed Warning > Warning and Precautions > Adverse Reactions 14

15 Drug Labeling Defined by Code of Federal Regulations (21CFR201.57) The regulations are treated by the courts as being as legally binding as statutory law General requirements: The labeling must contain a summary of the essential scientific information needed for the safe and effective use of the drug. The labeling must be informative and accurate and neither promotional in tone nor false or misleading in any particular. the labeling must be updated when new information becomes available that causes the labeling to become inaccurate, false, or misleading. The labeling must be based whenever possible on data derived from human experience. No implied claims or suggestions of drug use may be made if there is inadequate evidence of safety or a lack of substantial evidence of effectiveness. 15

16 Guidance to Help Preparation of Drug Labeling According to 21CFR

17 Drug Labeling Decisions Should the drug be withdrawn because it is judged too likely to cause liver injury in recipients? If not, should there be a special black box warning up front in the labeling to warn prescribing physicians? Are warnings adequate, or should some action be taken, such as testing for liver dysfunction before starting drug treatment, or monitoring periodically afterward? What monitoring frequency and interval should be advised, and for how long? Is simple mention of precautions enough? How should prescribing physicians interpret them? Even less, should possible liver injury just be buried among other adverse effects? 17

18 Drug Discovery Today, 16(15-16): , 2011

19 LTKB Benchmark Dataset (LTKB-BD) - A subset of LTKB drugs with reliable DILI labels LTKB-BD contains 137 most-dili-concern drugs, 85 less-dili-concern drugs and 65 no-dili-concern drugs. Approved for at least 10 years Ensure drug s label is reliable and stable Commercially available May serve as a benchmark dataset for study of DILI in the research community 65 no-dili-concern drugs (little or no DILI concerns) Over 90% of the non-dili drugs have been approved for over 25 years The annotation is transparent and can be reproduced by others to augment the list

20 20

21 21

22 Drug Labeling Is Not Perfect Not according to a scientifically justified master plan Evolved over the years based on the prevailing knowledge Opinion based, not based on a well defined criteria Negotiation between manufactory and FDA due to difference in interpretation of the scientific data Guilty by Association Warnings and Precautions: the labeling must be revised to include a warning about a clinically significant hazard as soon as there is reasonable evidence of a causal association with a drug; a causal relationship need not have been definitely established. Adverse Reactions: This section must list the adverse reactions that occur with the drug and with drugs in the same pharmacologically active and chemically related class, if applicable 22

23 DILIrank Dataset: Improved Causality Largest reference dataset with 1036 drugs Less-DILI-concern category was improved 192 drugs 254 drugs 278 drugs Chen et al. Drug Discovery Today, 2016, online 312 drugs 23

24 24

25 Increased Biological Complexity Liver Toxicity Knowledge Base (LTKB) - Predictive Modeling PRECLINICAL DATA Therapeutic uses Side effects Drug Properties >40K arrays (>500 drugs) In vitro assay (>200 drugs) DILI risk CLINICAL USES Genomic Data Cellular Response Predictive Models DILI Severity DILI Types Mechanisms Animal Data 25

26 LTKB DILI Models Simple rules: The rule-of-two model high daily dose (DD>100mg) and high lipophilicity (logp>3) predict DILI Chen et al. Hepatology, 58(1): 388, 2013 Toxicogenomics model: Assess the utility of translational genomics biomarkers for DILI Zhang et al, Chem Res Tox, 25 (1), pp , 2012 In vitro model: the ROS/ATP ratio for DILI using cultures of primary human hepatocytes Zhang et al., (2015, submitted) Computational models DILI prediction systems (DILIps) by Liu et al. PLoS Computational Biology, 2011 QSAR model by Chen et al., Toxicol Sci, 2013, 136(1),

27 Rule-of-Two (RO2): High Lipophilicity (logp>3) + High Daily Dose (DD>100 mg) Predicts severe DILI Most DILI-concern No DILI-concern 4 2 logp Observed in 164 drugs - Verified by 179 drugs - Demonstrated on 5 drug pairs - Applied to co-medication Daily dose (mg) Chen et al. Hepatology, 58(1): 388,

28 FDA Questions about DILI Preclinical Pre-market Post-market IND Phase I Phase II NDA approval Q1: Mitochondrial toxicity was found from in vitro study; can you assess DILI risk before human test? (preclinical) Q2: Liver enzymes we elevated in some volunteers, and we are concerned about its DILI risk (phase I) Q3: This drug will be given to patients with liver disease; can you assess its DILI risk (phase II) Q4: Skin rash and hypersensitivity symptoms found in clinical trials were also found in trials of another withdrawn drug, and therefore we have high concern of DILI risk (NDA phase) 28

29 29

30 Can We Use Drug Labeling to Identify More Dangerous Drugs? Despite imperfections, limitations, still reflects serious and considered thought by expert consultants, and wide consensus, application in practice Based on 1) data from controlled trials; 2) published literature reports; 3) spontaneous reports to FAERS; 4) updated over time Not perfect, but probably the best and most consistent information we have, although other systems have been tried. 30

31 Acknowledgements LTKB Team: Minjun Chen (Predictive models) Shraddha Thakkar (LTKB and QSARs) Jie Zhang (in vitro) Yuping Wang (micrornas) Zhichao Liu (DILIps) Leihong Wu (ABC-DILI) Hong Fang (FDALabel) Kristen McCune (Model integration) FDA DILI Interest Group ~30 members Tox21 and ToxCast NTP: Rick Paules, Scott Auerbach, Steve Ferguson EPA: Rusty Thomas, Ann Richard and Richard Judson NCATS: Chris Austin, Menghang Xia, Ruili Huang External collaborators: Jurgen Borlak (Hanover Medical School, Germany) Ayako suzuki (UAMS) Xiaohui Fan (Zhejiang Univ, China) Bob Water (Leiden University) Ruth Roberts (ApconiX Ltd) In vitro service/collaboration: In Vitro ADMET Laboratories Promega Hepregen (Ascendance) Biotranex 31