Pharmacokinetic Modelling of Antiretrovirals in HIV-Infected Patients Laura Dickinson NIHR Biomedical Research Centre, Royal Liverpool & Broadgreen University Hospital Trust & Department of Pharmacology, University of Liverpool PKUK, Bristol, UK. 04 November 20
Background HIV Therapy Therapeutic Failure Ritonavir Boosting Overview Examples Darunavir/Ritonavir PK Modelling Lopinavir/Ritonavir PK Modelling Nevirapine PK/PG Modelling Overall Summary
Global Distribution of Individuals Living with HIV in 2008 North America 1.4 million Caribbean 240 000 Western & Central Europe 850 000 North Africa & Middle East 3 000 Eastern Europe & Central Asia 1.5 million South & SE Asia 3.8 million East Asia 850 000 Latin America 2.0 million Sub-Saharan Africa 22.4 million Oceania 59 000 Global Estimation: 33.4 million (31.1-35.8) AIDS Epidemic Update 2009, UNAIDS/WHO 2009
Number of Drugs Approval of Antiretrovirals 30 25 20 15 Nucleoside reverse transcriptase inhibitor, NRTI Non-nucleoside reverse transcriptase inhibitor, NNRTI Protease inhibitor, Entry Inhibitor/CCR5 Antagonist Integrase Inhibitor Ritonavir Indinavir Nevirapine 3TC Saquinavir Nelfinavir Tenofovir Delavirdine Lopinavir/r Amprenavir Efavirenz Abacavir Enfuvirtide Atazanavir Emtricitabine Fosamprenavir Darunavir Maraviroc Raltegravir Tipranavir Etravirine 5 AZT ddl ddc d4t 0 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 Year
Combination Therapy NRTI/NNRTI Integrase Inhibitors Protease Inhibitors Entry Inhibitors Combination therapy: 2 NRTI + 1 NNRTI 2 NRTI + 1 boosted Highly Active AntiRetroviral Therapy - HAART Suppress viral replication and prolong life
Therapeutic Failure Despite access to combination therapy and new drugs emerging, many patients receiving HAART will fail therapy at some point Problematic PK High replication rate High mutation rate VIRUS DRUG Inadequate durability Toxicity Inconvenience Drug resistance PATIENT Toxicity Adherence <0% Cellular resistance (sanctuary sites; transport proteins)
Therapeutic Failure Despite access to combination therapy and new drugs emerging, many patients receiving HAART will fail therapy at some point Problematic PK High replication rate High mutation rate VIRUS DRUG Inadequate durability Toxicity Inconvenience Drug resistance PATIENT Toxicity Adherence <0% Cellular resistance (sanctuary sites; transport proteins)
Ritonavir (RTV) Boosting Pharmacoenhancement Second to be approved Originally dosed at 600 mg twice daily Associated with various adverse events including gastrointestinal disorders (e.g. nausea, vomiting), lipodystrophy, increased triglycerides As a pharmacoenhancer usually dosed at 0 or 200 mg twice or once daily In a boosted combination, RTV does not contribute to the antiretroviral effect
Mechanism of RTV Boosting Protease Protease inhibition antiviral activity
Mechanism of RTV Boosting P450 enzyme Protease Metabolised Protease inhibition antiviral activity Inactive metabolites Less drug available to inhibit HIV protease enzyme
Mechanism of RTV Boosting RTV RTV RTV RTV x P450 enzyme Protease Enzyme inhibition; RTV covalently binds? Protease inhibition antiviral activity Drug available to inhibit HIV protease enzyme
Boosted Protease Inhibitors Lopinavir/ritonavir; LPV/RTV LPV/RTV 400/50 mg Atazanavir/ritonavir; ATV/RTV ATV/RTV 300/0 mg LPV 400 mg Tipranavir/ritonavir; TPV/RTV ATV 400 mg Darunavir/ritonavir; DRV/RTV TPV/RTV 500/200 mg DRV/RTV Sham HL et al. AAC 1998 TPV 500 mg DRV Reyataz US Prescribing Information, 20 King JR & Acosta EP. Clin Pharmacokinet 2006 Hoetelmans R et al. th CROI 2003
Alternatives to RTV Investigational pharmacoenhancer: GS-9350 (cobicistat) No antiviral activity, reduced adverse effects Co-formulated QUAD tablet (1 tablet once daily) containing: elvitegravir/cobicistat/tenofovir/emtricitabine Effect of cobicistat & RTV on midazolam CL/F Mathias AA et al. Clin Pharmacokinet 20
Advantages of Boosted Regimens Improved bioavailability Reduced dosing frequency and pill burden Co-formulation of lopinavir and ritonavir Potential for once daily dosing Simplified regimens improved adherence
Atazanavir (mg/l) Why Use Population PK Incorporation of rich and sparse sampling Antiretrovirals, namely s, demonstrate high interand intra-individual variability in concentrations at a given dose 0 P P25 P50 P75 P90 1 0.1 0.01 ATV MEC; 0.15 mg/l ATV TDM, n=525; 300/0 mg qd 0 4 8 12 16 20 24 Time (h) Age Weight Ethnicity Sex Drug-Drug Interactions Co-morbidities Genetics
Population Pharmacokinetic Modelling of Once Daily Ritonavir-Boosted Darunavir in HIV-Infected Patients
Introduction Darunavir/ritonavir (DRV/RTV) has been approved for clinical use: Treatment naïve: 800/0 mg once daily Treatment experienced: 600/0 mg twice daily 1,2 Pharmacokinetics (PK) may be influenced by patient demographics and/or coadministered drugs Important to assess effect of age given the increasing ageing HIV-infected population Objective: Develop population PK model for once daily DRV/RTV. Investigate impact of patient demographics & co-medications on DRV apparent oral clearance (CL/F) 1 Janssen-Cilag Ltd. Prezista SPC 20; 2 Tibotec Inc. Prezista US Prescribing Information 20
Darunavir (mg/l) Darunavir (mg/l) DRV Data 3 PK studies n=51 HIV-infected patients 800/0 mg (n=32) or 900/0 mg qd (n=19) 1-3 PK profiles per patient (n=9) n=47 concentrations 15 12 0 1 Parameter Median 0.1 Sex (range)* 9 0.01 0 4 8 12 16 20 24 Time (h) Male [n (%)] Female [n (%)] 44 (86) 7 (14) 6 Age (yr) 39 (21-63) Weight (kg) 74 (57-5) BMI (kg/m 2 ) 24 (18-31) 3 RTV AUC 0-24 (mg.h/ml) 4.4 (2.3-11.0) BL CD4 cell count (cells/mm 3 ) 500 (227-1129) BL undetectable viral load [n (%)] 49 (96) * Unless stated otherwise 0 0 4 8 12 16 20 24 Time (h)
Methods 3 PK studies 1-3 n=51 HIV patients (7 female) n=47 concentrations Serial blood sampling (8-11 time points/patient) 1-3 PK profiles/patient HPLC-MS/MS PK Modelling NONMEM v. VI 2.0 Structural Model Covariate Model 2 compartment model: -Step-wise addition of mixed/random effects Potential covariates: RTV AUC 0-24, sex, ethnicity, age, weight, BMI, concomitant RGV Model fit & differences between models assessed: -Statistical methods (ΔOFV >3.84) -Diagnostic plots Retained if: -Statistically significant, ΔOFV >3.84 -Clinically relevant -Backwards elimination, ΔOFV >6.63 Simulations Visual Predictive Check 1 Jackson A et al. 12 th EACS 2009; 2 Garvey L et al. Antivir Ther 20; 3 Boffito M et al. Unpublished data. 20
Observed DRV (mg/l) Observed DRV (mg/l) Observed vs. Predicted Results RTV AUC 0-24 and patient age significant covariates Parameter CL/F (L/h) V2/F (L) Estimate (RSE) IIV (%) (RSE) IOV (%) (RSE) 12.5 (5) a 15.6 (6) b 12 (65) 20 (19) 125 (7) a 192 (9) b 39 (23) Q/F (L/h) 13.4 (9) 59 (45) V3/F (L) fixed 84 (n/a) Predicted DRV (mg/l) Observed vs. Individual predicted k a (h -1 ) 0.9 (3) 75 (30) Lag-time (h) 0.4 (5) Covariates θ RTV -0.4 (20) θ AGE -0.01 (22) Residual error Proportional (%) 27 (8) Individual predicted DRV (mg/l) RSE = (SE estimate /estimate) *0 a Study 1 & 2 b Study 3 IIV: interindividual variability IOV: interoccasion variability CL/F ij = θ 1 *((RTV ij /4.35) θ RTV *(1+ θage *(AGE-39)))*exp(η i +κ ij )
DRV (mg/l) Results RTV AUC 0-24 and patient age significant covariates 0 Visual Predictive Check Parameter CL/F (L/h) Estimate (RSE) IIV (%) (RSE) IOV (%) (RSE) 12.5 (5) a 15.6 (6) b 12 (65) 20 (19) V2/F (L) 125 (7) a 192 (9) b 39 (23) Q/F (L/h) 13.4 (9) 59 (45) 1 V3/F (L) fixed 84 (n/a) k a (h -1 ) 0.9 (3) 75 (30) 0.1 Lag-time (h) 0.4 (5) 0.01 0 4 8 12 16 20 24 Time (h) Covariates θ RTV -0.4 (20) θ AGE -0.01 (22) P2.5 P50 P97.5 Residual error Observed DRV concentrations; n=47 MEC1; treatment-naïve; 0.055 mg/l MEC2; treatment-experienced; 0.550 mg/l Proportional (%) 27 (8) RSE = (SE estimate /estimate) *0 a Study 1 & 2 b Study 3 IIV: interindividual variability IOV: interoccasion variability CL/F ij = θ 1 *((RTV ij /4.35) θ RTV *(1+ θage *(AGE-39)))*exp(η i +κ ij )
DRV (mg/l) DRV (mg/l) DRV (mg/l) P2.5 P50 P97.5 MEC1; treatment-naïve; 0.055 mg/l MEC2; treatment-experienced; 0.550 mg/l Simulations 0 1 0.1 25 years Effect of age: Increase in 1yr fractional decrease DRV CL/F of 0.014 (1.4%) Correspond to 14% decrease in DRV CL/F per yr increase in age Simulations (P2.5-P97.5): All 3 age groups above MEC1 At 25 & 55 years P2.5 fell below MEC2 At 75 years 95% prediction interval remained above MEC1 and MEC2 0.01 0 1 0.1 0.01 0 1 0.1 0.01 0 4 8 12 16 20 24 55 years 0 4 8 12 16 20 24 75 years 0 4 8 12 16 20 24 Time (h)
Summary Population model describing the PK of once daily RTVboosted DRV developed RTV AUC 0-24 and age significantly associated with DRV CL/F Only n=8 >50 years, n=1 >60 years in the dataset Despite outcome of simulations for older age group (75 years) further investigations of the impact of age on DRV PK warranted.
Sequential Modelling of Lopinavir and Ritonavir in Healthy Volunteers 72 h After Drug Cessation
Lopinavir/ritonavir (Kaletra ) Durable efficacy in treatment-naïve and experienced HIVinfected patients 1 400/0 mg twice daily & 800/200 mg once daily doses approved 2,3 New tablet formulation: diminished food effect reduced variability increased heat stability 4 Kaletra soft-gel capsules Kaletra tablets 1 Oldfield & Plosker. Drugs 2006; 2 Kaletra US Prescribing Information 20; 3 Kaletra SPC 20; 4 Klein et al. JAIDS 2007
TAIL Study 1 AIM Determine plasma drug concentrations over the dosing interval and up to 72 hours following drug cessation Evaluate persistence of drug in plasma if a dose is forgotten or delayed 1 Boffito et al. Antivir Ther 2008
TAIL Study: Design 1 Healthy volunteers (n) 16 16 16 Session 1 2 3 Regimen ATV/RTV qd LPV/RTV bid LPV/RTV qd Dose 300/0 mg qd 400/0 mg bid 800/200 mg qd NOTE: Lopinavir/ritonavir tablets Screening ATV/RTV qd Washout LPV/RTV bid Washout LPV/RTV qd Washout -28 0 1 5 11 17 18 22 27 28 34 35 39 44 45 51 Witnessed intake 72 hour PK Witnessed intake 72 hour PK Witnessed intake 72 hour PK Follow up Blood sampling: pre-dose (0 h), 0.5, 1, 2, 3, 4, 6, 8,, 12, 16, 20, 24, 30, 36, 48, 60, 72 post-dose 1 Boffito et al. Antivir Ther 2008
Lopinavir (ng/ml) Ritonavir (ng/ml) Atazanavir (ng/ml) Ritonavir (ng/ml) TAIL Study: Results 1 Atazanavir (300 mg qd) 000 Ritonavir (0 mg qd) 000 Geometric mean 00 0 MEC 150 ng/ml 00 0 1 0 9 18 27 36 45 54 63 72 Time (h) Lopinavir (400 mg bid) 0000 1 0 9 18 27 36 45 54 63 72 Time (h) Ritonavir (0 mg bid) 0000 000 00 MEC 00 ng/ml 000 00 0 0 1 0 9 18 27 36 45 54 63 72 Time (h) 1 0 9 18 27 36 45 54 63 72 Time (h) 1 Boffito et al. Antivir Ther 2008
Lopinavir (ng/ml) Ritonavir (ng/ml) Atazanavir (ng/ml) Ritonavir (ng/ml) TAIL Study: Results 1 Atazanavir (300 mg qd) 000 Ritonavir (0 mg qd) 000 Geometric mean 00 0 MEC 150 ng/ml 00 0 1 0 9 18 27 36 45 54 63 72 Time (h) Lopinavir (800 mg qd) 0000 1 0 9 18 27 36 45 54 63 72 Time (h) Ritonavir (200 mg qd) 0000 000 00 MEC 00 ng/ml 000 00 0 0 1 0 9 18 27 36 45 54 63 72 Time (h) 1 0 9 18 27 36 45 54 63 72 Time (h) 1 Boffito et al. Antivir Ther 2008
Objective Develop and validate a population pharmacokinetic model that integrates the relationship between lopinavir and ritonavir over 72 hours following drug cessation
NONMEM v. VI 2.0 Basic model LPV & RTV Methods N=16 healthy volunteers 14-17 samples/volunteer N=252 concentrations (LLQ: LPV 0.005mg/L; RTV 0.002mg/L) Covariate model LPV & RTV RTV Individual PK parameters Competitive inhibition model CL = CL0/(1+C RTV /K i ) LPV (combined model) Competitive inhibition model (1 st pass) CL INT = CL0/(1 + C RTV /K i ) CL = CL INT * QR/(QR + CL INT ) Linear function I = 1 (SLOPE * C RTV ) Direct response model CL = CL0 * I Maximum effect function I = 1 ((I MAX * C RTV )/ (IC 50 + C RTV ))
Observed LPV (mg/l) Observed LPV (mg/l) 20 Observed vs. Predicted Results RTV AUC as a covariate on LPV CL/F (power relationship) 15 Parameter Estimate (RSE) IIV (%) (RSE) 5 0 r² = 0.813 Unity 0 5 15 20 CL/F (L/h) 4.5 (5) 9 (37) V/F (L) 15.9 (9) 23 (54) k a (h -1 ) 0.3 (6) 27 (44) Predicted LPV (mg/l) Lag-time (h) 0.7 (8) Observed vs. Individual predicted Covariates 20 RTV AUC (power) -0.4 (12) 15 Residual Error 5 r² = 0.856 Unity Proportional (%) 40 (12) Additive (mg/l) 0.002 (52) 0 0 5 15 20 RSE = (SE estimate /estimate) *0 IIV: interindividual variability Individual predicted LPV (mg/l)
Observed LPV (mg/l) Observed LPV (mg/l) 20 Observed vs. Predicted Results RTV concentrations at each time point on LPV CL/F (direct response model) 15 Parameter Estimate (RSE) IIV (%) (RSE) 5 0 r² = 0.870 Unity 0 5 15 20 CL0/F (L/h) 21.6 (14) 11 (36) I MAX 0.9 (1) IC 50 (mg/l) 0.06 (22) Predicted LPV (mg/l) V/F (L) 55.3 () 19 (42) Observed vs. Individual predicted k a (h -1 ) 0.6 (0.4) 20 Lag-time (h) 0.4 (4) 15 Residual Error 5 r² = 0.918 Unity Proportional (%) 25.6 (18) Additive (mg/l) 0.004 (34) 0 0 5 15 20 RSE = (SE estimate /estimate) *0 IIV: interindividual variability Individual predicted LPV (mg/l)
RTV (mg/l) LPV (mg/l) Results RTV concentrations at each time point on LPV CL/F (direct response model) P2.5 P50 P97.5 Observed LPV or RTV concentrations; n=252 1.0E+02 1.0E+01 1.0E+00 1.0E-01 1.0E-02 1.0E-03 1.0E-04 1.0E-05 0 9 18 27 36 45 54 63 72 Parameter Estimate (RSE) IIV (%) (RSE) CL0/F (L/h) 21.6 (14) 11 (36) I MAX 0.9 (1) IC 50 (mg/l) 0.06 (22) V/F (L) 55.3 () 19 (42) Time (h) k a (h -1 ) 0.6 (0.4) 1.0E+01 Lag-time (h) 0.4 (4) 1.0E+00 Residual Error 1.0E-01 1.0E-02 1.0E-03 Proportional (%) 25.6 (18) Additive (mg/l) 0.004 (34) 1.0E-04 1.0E-05 0 9 18 27 36 45 54 63 72 RSE = (SE estimate /estimate) *0 IIV: interindividual variability Time (h)
Summary A maximum effect model best described the relationship between ritonavir concentrations and inhibition of lopinavir CL/F Gives a better understanding of the interaction between LPV and RTV Allow for better prediction of lopinavir concentrations compared to a model with RTV AUC as a covariate Of note: Model does not include inhibition of drug transporters or the induction effect of LPV on RTV
Population Pharmacokinetic and Pharmacogenetic Analysis of Nevirapine in Hypersensitive and Tolerant HIV- Infected Patients from Malawi
Introduction Approximately % of the population of Malawi are infected with HIV Nevirapine (NVP)-based therapy underpins first-line treatment of HIV in Africa Generic fixed-dose combination (Triomune ) 1 pill twice daily NNRTI: NVP 200 mg NRTI: Lamivudine (3TC) 150 mg NRTI: Stavudine (d4t) 30 or 40 mg
Introduction NVP associated with risk of developing hypersensitivity (HS) Metabolised by CYP2B6 and CYP3A4 1 Expression of CYP3A5 (90% sequence homology with CYP3A4) is higher in African populations compared to Caucasians 2,3 CYP2B6 and CYP3A5 are polymorphic with various single nucleotide polymorphisms (SNP) identified 1 Riska P et al. Drug Metab Dispos 1999; 2 Kuehl P et al. Nat Genet 2001; 3 Lui YT et al. Drug Metab Rev 2007
Objective Relationship between NVP exposure and HS is unknown but could be influenced by polymorphisms in CYP2B6 and CYP3A5 Develop population PK model for NVP measured in serum and assess impact of patient demographics, HS and genetics
Patient Cohort HIV-infected, treatment-naïve patients starting NVPbased therapy (200 mg twice daily) were prospectively recruited (Mar 2007-Sept 2008) Part of a larger cohort study evaluating NVP HS at the adult outpatients HIV clinic, Queen Elizabeth Central Hospital, Blantyre, Malawi Following a documented HS reaction, treatment with NVP was stopped: Maculopapular eruption & fever Severe SJS or TEN Hepatotoxicity (e.g. jaundice)
Nevirapine (mg/l) Nevirapine (mg/l) NVP Data Parameter n=180 HIV-infected patients n=40 patients rich sampling (0, 1, 3, 5, 9 h) n=140 patients sparse sampling 1-7 samples/patient, 1-3 occasions n=383 concentrations Median (range)* 30 25 20 0 1 0.1 0 4 8 12 16 20 24 28 Time (h) Sex Male [n (%)] Female [n (%)] 79 (44) 1 (56) 15 Age (yr) 34 (21-62) Weight (kg) 54 (35-94) BMI (kg/m 2 ) 20 (15-38) Wks on therapy at time of PK (wks) 6 (1-26) 5 BL CD4 cell count (cells/mm 3 ) 156 (1-812) Confirmed hypersensitivity [n (%)] 25 (14) Hepatitis B/C co-infection [n (%)] 23 (13) * Unless stated otherwise 0 0 4 8 12 16 20 24 28 Time (h)
Methods NONMEM v. VI 2.0 Basic Model 180 patients 383 serum concentrations Covariate Model Patient demographics HS Hepatitis co-infection Genetic Model Arab-Alameddine, M et al. Clin Pharmacol Ther 2009; 85: 485-94 Genotyping: Sequenom iplex SNP1: CYP3A5*6 SNP2: CYP3A5*3 SNP3: CYP2B6 983T>C SNP4: CYP2B6 516G>T SNP5: CYP2B6 785A>G Final Model
NVP CL/F (L/h) NVP CL/F (L/h) NVP CL/F (L/h) Results Patient demographics, HS, hepatitis co-infection and CYP3A5 genotype were not associated with NVP CL/F 15 CYP2B6 983T>C p<0.001 15 CYP2B6 516G>T p<0.05 5 5 0 TT TC 0 GG GT TT 15 CYP2B6 785A>G p<0.05 Allelic Frequencies CYP2B6 983T>C TC: 17%; CC 0% 5 0 AA AG GG CYP2B6 516G>T GT: 49%; TT: 16% CYP2B6 785A>G AG: 44% GG: 15%
Observed NVP (mg/l) Observed NVP (mg/l) Results 30 25 Observed vs. Predicted 20 15 5 0 30 25 20 15 5 0 r² = 0.2 Unity 0 5 15 20 25 30 Predicted NVP (mg/l) Observed vs. Individual predicted r² = 0.966 Unity 0 5 15 20 25 30 Individual predicted NVP (mg/l) Parameter Estimate (RSE) IIV (%) (RSE) IOV (%) (RSE) CL/F (L/h) 3.0 (4) 33 (27) 32 (24) V/F (L) 114 (23) k a (h -1 ) 0.6 (26) Covariates θ CYP2B6 983TT/516TT -0.7 (37) θ CYP2B6 983TC/516GG or GT -1.1 (22) Residual error Proportional (%) 13 (17) RSE = (SE estimate /estimate) *0 IIV: interindividual variability IOV: interoccasion variability CL/F ij = θ 1 + θ 2 * CYP2B6 983TT/516TT + θ 3 * CYP2B6 983TC/516GG or GT * exp(η i +κ ij )
NVP (mg/l) Results 0 Visual Predictive Check Parameter Estimate (RSE) IIV (%) (RSE) IOV (%) (RSE) CL/F (L/h) 3.0 (4) 33 (27) 32 (24) V/F (L) 114 (23) 1 0.1 0 5 15 20 25 30 Time (h) P2.5 P50 P97.5 Observed NVP concentrations; n=383 983TT/516GG or GT 983TT/516TT k a (h -1 ) 0.6 (26) Covariates θ CYP2B6 983TT/516TT -0.7 (37) θ CYP2B6 983TC/516GG or GT -1.1 (22) Residual error Proportional (%) 13 (17) RSE = (SE estimate /estimate) *0 IIV: interindividual variability IOV: interoccasion variability 983TC/516GG or GT Missing genotype MEC; 3.0 mg/l CL/F ij = θ 1 + θ 2 * CYP2B6 983TT/516TT + θ 3 * CYP2B6 983TC/516GG or GT * exp(η i +κ ij )
Results Compared to reference genotype (CYP2B6 983TT/516GG or GT): NVP CL/F decreased by 30% with CYP2B6 983TT/516TT NVP CL/F decreased by 56% with CYP2B6 983TC/516GG or GT Individual predicted trough concentrations (C12h): 22/219 (%) <NVP MEC (3.0 mg/l) The majority (n=15) were associated with the reference genotype None were CYP2B6 983 heterozygotes (TC)
Summary Available patient demographics and HS were not associated with NVP CL/F in this Malawian population The CYP3A5 polymorphisms assessed did not impact NVP CL/F A combination of CYP2B6 983T>C/516G>T genotype significantly reduced NVP CL/F but only described a fraction of parameter variability (1-3%) Further studies warranted to determine the mechanisms of NVP HS
To Summarise Rich PK data Sparse PK: TDM database >20,000 samples PGx data Drug-drug interactions Two key Issues: Dose optimisation patients on antiretrovirals for life Impact of older age and other covariates (e.g. other drugs)
Acknowledgements University of Liverpool Prof David Back Prof Saye Khoo Dr Gerry Davies Dr Alessandro Schipani Dr Mas Chaponda Dr Laura Else Dr Dan Carr Victoria Watson Chelsea & Westminster Hospital Dr Marta Boffito Dr Akil Jackson Dr Anton Pozniak St Mary s Hospital/Imperial College Dr Alan Winston Dr Akil Jackson Dr Anton Pozniak University of Manchester Prof Leon Aarons University of Strathclyde Dr Alison Thomson