1. Munir Pirmohamed David Weatherall Chair of Medicine and NHS Chair of Pharmacogenetics Department of Molecular and Clinical Pharmacology University of Liverpool Genomics of Adverse Drug Reactions: The Need for a Multi-Functional Approach

2. Adverse Drug Reactions: Classification ON TARGET REACTIONS  Predictable from the known primary or secondary pharmacology of the drug  Clear dose-dependence relationship within the individual OFF TARGET REACTIONS  Not predictable from a knowledge of the basic pharmacology of the drug and can exhibit marked inter- individual susceptibility  Complex dose-dependence

3. Outline Phenotyping Sample sizes Genomic approaches The path to clinical translation Genetic exceptionalism

4. Genetic Contribution Many factors predispose to adverse drug reactions, many of which are environmental and clinical We do not know the overall genetic contribution to the occurrence of adverse reactions The genetic effect will vary according to drug and reaction ADRs account for: 6.5% of all hospital admissions 15% rate in in-patients 8000 NHS Beds in the UK ADRs account for: 6.5% of all hospital admissions 15% rate in in-patients 8000 NHS Beds in the UK

5. Deep Phenotyping ADRs can affect any organ system, can be of any severity – MIMIC OF DISEASE Important to be aware of the phenotypic heterogeneity – link between clinicians and genomics experts Although overall burden of ADRs is high, the incidence of individual ADRs may be low or rare in many instances – so patient identification can be difficult (cf. Type 2 Diabetes)

7. Power of Studies Many pharmacogenetics studies in the past had small sample sizes, compunded by poor phenotype Led to low effect sizes with lack of replication in independent cohorts But since ADRs may be uncommon, it will never be possible to attain samples sizes seen in complex diseases  International consortia  Electronic medical records Toxic epidermal necrolysis 1 in million per year

8. InTernational Consortium on Drug Hypersensitivity (ITCH) EU Australia CanadaUSBrazil CroatiaNorwayEUDRAGENE Sponsored by the International Serious Adverse Event Consortium (iSAEC) 12 international centres 50 UK centres 1500 patients

9. Electronic Medical Records: Clinical Practice Research Datalink Previously GPRD 12 million patient records (March 2011) Increased to 52 million with the transition to CPRD Feasibility study using statin myopathy as paradigm 641,703 patients prescribed a statin 127,209 with concurrent CPK measurement

10. The R&D Governance Burden Statin myopathy Identified via CPRD Link to DNA samples 132 R&D approvals

11. 1.Implicated SNP is in the SLCO1B1 gene (transporter) 2.Shown with simvastatin 40mg and 80mg

12. Genotype Frequency nT/TT/CC/Cp Per C-allele OR (95%CI) All Statins (n=448) Tolerant3720.700.270.03-- All Myopathy760.530.390.080.0052.08 (1.35-3.23) Severe Myopathy230.350.440.210.00034.47 (1.84-10.84) Simvastatin Only (n=281) Tolerant2220.660.320.02-- All Myopathy590.490.420.090.0142.13 (1.29-3.54) <40mg/day240.630.370.000.9971.03 (0.45-2.36) ≥40mg/day350.400.460.140.00023.23 (1.74-5.99) Severe Myopathy180.280.500.220.00044.97 (2.16-11.43) <40mg/day50.400.600.000.7781.84 (0.34-9.86) ≥40mg/day130.230.460.310.00046.28 (2.38-16.60)

13. Statin Myopathy GWAS All myopathy (n=128) vs. WTCCC2 (unimputed) SLCO1B1

14. Severe myopathy (n=32) vs. WTCCC2 (unimputed) SLCO1B1

15. Carbamazepine Hypersensitivity More complicated than abacavir hypersensitivity Different phenotypes  Skin (mild → blistering)  Liver  Systemic (DRESS) Predisposition varies with ethnicity and phenotype  HLA-B*1502 (Chinese)  HLA-A*3101 (Caucasian)

16. CPT, 2012 HLA-B*1502

17. Liverpool 22 patients with HSS Replicated in Japanese, Chinese, South Korean, Canadian and EU populations NNT = 47 SmPC/drug label changed (for information) Patient and clinician preferences Cost effectiveness 55% likelihood Cluster RCT being planned Replicated in Japanese, Chinese, South Korean, Canadian and EU populations NNT = 47 SmPC/drug label changed (for information) Patient and clinician preferences Cost effectiveness 55% likelihood Cluster RCT being planned

18. Whole Genome Sequencing in CBZ Hypersensitivity N= 48 (28 CBZ-induced severe hypersensitivity and 20 tolerant controls)

19. HLA-A* Loci Using NGS data 30 HLA-A* loci typed 18 HLA-A* alleles identified 40% CBZ hypersensitive patients are A*31:01 positive

20. Rare Variant Pathway Analysis Namep-value#Genes#Variants#Cases#ControlsGene Name Graft-versus-Host Disease Signaling3.96E-04475270 HLA-A, HLA-DRB1, HLA-DRB5, KIR2DL1/KIR2DL3 Antigen Presentation Pathway7.75E-04361260HLA-A, HLA-DRB1, HLA-DRB5 Crosstalk between Dendritic Cells and Natural Killer Cells1.08E-03475270 HLA-A, HLA-DRB1, HLA-DRB5, KIR2DL1/KIR2DL3 Mitotic Roles of Polo-Like Kinase3.55E-03382280ANAPC5, CDC27, SLK Type I Diabetes Mellitus Signaling4.57E-03482260 HLA-A, HLA-DRB1, HLA-DRB5, MAP2K3 Autoimmune Thyroid Disease Signaling7.50E-03361260HLA-A, HLA-DRB1, HLA-DRB5 B Cell Development9.20E-03245230HLA-DRB1, HLA-DRB5 Cytotoxic T Lymphocyte-mediated Apoptosis of Target Cells1.25E-02361260HLA-A, HLA-DRB1, HLA-DRB5 Inhibition of Matrix Metalloproteases1.28E-02223200MMP24, TIMP2 OX40 Signaling Pathway1.47E-02361260HLA-A, HLA-DRB1, HLA-DRB5 Allograft Rejection Signaling1.69E-02361260HLA-A, HLA-DRB1, HLA-DRB5 Estrogen Receptor Signaling2.11E-02388280CTBP2, MED13L, NCOR1 Communication between Innate and Adaptive Immune Cells2.37E-02361260HLA-A, HLA-DRB1, HLA-DRB5 IL-17 Signaling4.56E-02262270MAP2K3, MUC5B IL-4 Signaling4.84E-02245230HLA-DRB1, HLA-DRB5

21. T Cells in Carbamazepine Hypersensitivity: HLA-A*31:01+ patient GenderAgeTime to reaction (days) Details of reactionTime since reaction (years) RechallengeHLA-A genotype Comments female746Generalized rash, raised liver enzymes, fever, eosinophilia, lymphocytosis →Hypersensitivity syndrome 22NoA*11:01/ A*31:01 Previously experienced allergic reaction to Cotrimoxazol Clinical data Lymphocyte transformation test

22. Carbamazepine-Responsive T-cell clones Clones tested (n) Specific clones (n) Proliferation (cpm)CD phenotype (%) control CBZ (25μg/ml) CD4+CD8+ CD4+ CD8+ 94767 5,525.8 (±18,928.0) 34,418.8 (±43,632.5) 353728 Specificity and Phenotype IFN γ IL-13PerforinGranz.B FasL 0 CBZ a) CD4+ TCC IFN γ IL-13PerforinGranz.B FasL 0 CBZ b) CD8+ TCC Secretion of cytokines and cytolytic molecules

23. HLA Restriction of CBZ-Specific TCC MHC restriction of CD4+ (a) and CD8+ (b) TCC * p = 0.03 ns b) CD8+ (n=3) * p = 0.03 a) CD4+ (n=3) ns HLA class II restriction of CD4+ TCCHLA A31 restriction of CD8+ TCC ** p = 0.004 p = 0.008 n = 3 * p = 0.03 n = 3

24. Hierarchy of Evidence What type of evidence is required for demonstration of clinical utility?

25. Technology-Based Reduction in the Burden of ADRs: The Case of Abacavir Hypersensitivity Clinical phenotype Association with HLA-B*5701 Clinical genotype Incidence before and after testing for HLA-B*5701 CountryPre testingPost testingReference Australia7%<1%Rauch et al, 2006 France12%0%Zucman et al, 2007 UK (London)7.8%2%Waters et al, 2007

26. Uptake of HLA-B*5701 in Different Continents

27. Drug label changed before prospective study Two prospective studies did not contradict previous data from retrospective studies

28. Evidence standards differ between non-genetic and genetic tests 3 examples given:  Drug exposure  Prevention of adverse drug reactions  Health technology assessment

29. Drug Exposure: Differential Evidential Standards Example: Aztreonam SmPC  “after an initial usual dose, the dosage of aztreonam should be halved in patients with estimated creatinine clearances between 10 and 30 mL/min/1.73 m 2 ” Many different examples in hepatic and renal impairment with dose instructions based on PK studies and occasionally PK-PD modelling No need for RCTs – in fact, would be impractical However, a genetic polymorphism leading to same degree of change in drug exposure is often ignored and/or RCT data are required for implementation

30. Differential Evidence Standards Unfamiliarity with genetic tests Lack of experience in interpretation Perceived cost of genetic testing Lack of availability of tests Poor turnaround time recommendations on dosing evaluation in patients with polymorphisms in known metabolic pathways

31. Summary Prediction of adverse drug reactions (safety biomarker) Insights into mechanisms of the adverse drug reaction Poste, Nature, 2011

32. “Hierarchies of evidence should be replaced by accepting—indeed embracing—a diversity of approaches........It is a plea to investigators to continue to develop and improve their methods; to decision makers to avoid adopting entrenched positions about the nature of evidence; and for both to accept that the interpretation of evidence requires judgment.” “Hierarchies of evidence should be replaced by accepting—indeed embracing—a diversity of approaches........It is a plea to investigators to continue to develop and improve their methods; to decision makers to avoid adopting entrenched positions about the nature of evidence; and for both to accept that the interpretation of evidence requires judgment.”

33. Acknowledgements Ann Daly (Newcastle University) Panagiotis Deloukas (Sanger Institute) SERIOUS ADVERSE EVENT CONSORTIUM EPIGEN EU-PACT FDA Funders: Dept of Health (NHS Chair of Pharmacogenetics) MRC, WT, DH, NIHR, EU-FP7 The University of Liverpool B Kevin Park Ana Alfirevic Maike Lichtenfels Dean Naisbitt Ben Francis Dan Carr