490 likes | 731 Views
Quantitative Clinical Pharmacology: Applications of Modeling and Simulation in Clinical Development. Rajesh Krishna, PhD, FCP October 9, 2006 Program in Integrative Information, Computer and Application Sciences (PICASso), Princeton University. Prior Knowledge -Analogues -Disease
E N D
Quantitative Clinical Pharmacology:Applications of Modeling and Simulation in Clinical Development Rajesh Krishna, PhD, FCP October 9, 2006 Program in Integrative Information, Computer and Application Sciences (PICASso), Princeton University
Prior Knowledge -Analogues -Disease -Competitors -Patients -Discovery/Pre-clinical Drug Product -indication -patients -formulation -dose -safety/ efficacy Drug Molecule Clinical Development Plan Trial 1 Trial 2 Trial … Trial N Learn Learn / Confirm Confirm Efficacy Tox PK/PD Mechanistic Therapeutic Benefit PPK/PD Clinical Endpoint MTD, Efficacy D-R PK/PD, (P)PK/PD Biomarker, Surrogate Typical Drug Development Program
Reality of Present Day New Drug Development • High NME attrition • High failure rate before IND • NME IND = NDA <20% of time • Reported >50% failure rate in Phase 3 (Carl Peck, CDDS) • Decreased NME NDAs despite increased INDs • Cost per NME approved estimated at >$800M (Tufts) Adapted from: Kola and Landis, Nature Review Drug Discovery, 2004 (3):711-715
Probability of Success for New Mechanisms ~11% Adapted from: Kola and Landis, Nature Review Drug Discovery, 2004 (3):711-715
New Mechanisms Often Fail Because of Lack of Efficacy or Demonstrated Benefit Adapted from: Kola and Landis, Nature Review Drug Discovery, 2004 (3):711-715
Drivers For Change • Escalating R&D costs despite flat growth • R&D expenditure not proportional to number and quality of NMEs • Need to productivity (do more with same amount of investment) • Low POS of NME’s entering Phase I - ~11% (Kola and Landis, 2004) • Phase II as stage for decision making often the default today • High attrition given low POS for new mechanisms without evidence of pharmacological benefit in Phase I • R&D costs required to support decision-making for new mechanisms this late, given POS, prohibitive • Lack of resources to create tools to facilitate early decision-making • Knowledge management and integration • New tools to link outcomes, predict hazards, reduce uncertainty in risk/benefit • Quick win-quick kill paradigm
FDA Critical Path Initiative • Goals • Develop new predictive “tools” • Improve the productivity and success of drug development • Speed approval of innovative products Adapted from: http://www.fda.gov/oc/initiatives/criticalpath/
Areas for Change • Key objective: • Need to dramatically improve predictions of efficacy and safety in clinical development • Enablers: • Biomarkers ~ target validation • Biomarker qualification, qualifying disease specific biomarkers • M&S ~ effective knowledge management leveraging bioinformatics • Drug disease models, clinical trial simulation • Clinical trials ~ better decision making, improving efficiency • Adaptive trial designs, seamless trials
PK Mixed Effects Modeling 0.20 0.15 Drug Concentration 0.10 0.05 Safety Efficacy 0.0 0 1 2 3 4 5 6 Time (h) Pulse Check - Terminology
Model Based Drug Development • Hypothesis based drug development emphasizing integrating information and improving the quality of decision making in drug development • Preclinical and clinical biomarkers • Dose-response and/or PK-PD relationships • Mechanistic or empirical disease models • Novel clinical trial designs • Clinical trial simulations and probabilities of success • Baseline-, placebo- and dropout-modified models • Outcome models
Roadmap for Model Based Drug Development Capture Prior knowledge Model Clinical trial Simulate Optimize
Case Example 1: Meta-Analysis of Statin Efficacy • Accumulated data from 25 trials (~9500 patients) • 5 Pfizer sponsored trials for Lipitor • 7 AstraZeneca summary basis trials for Crestor • 9 Merck summary basis trials for Zetia • 4 Pfizer sponsored trials for an investigational non-statin • Epidemiology Trials • Wilson et al, Framingham risk equations, • (Prediction of Coronary Heart Disease Using Risk Factor Analysis, Circulation, 1998, 97:1837-1847) • Riker et al, C-Reactive Protein and LDL • (Comparison of C-Reactive Protein and Low-Density Lipoprotein Cholesterol Levels in the Prediction of First Cardiovascular Events, NEJM 2002, 347:1557-1565) Adapted from: Mandema and Hartman. Model-Based Development of Gemcabene a New Lipid-Altering Agent. AAPS Journal. 2005; 7(3): E513-E522.
Pharmacodynamic Model Development • Trials looked at the following alone or in combination with Ezetamibe or gemcabine: • Atorvastatin • Rosuvastatin • Simvastatin • Lovastatin • Pravastatin Adapted from: Mandema and Hartman. Model-Based Development of Gemcabene a New Lipid-Altering Agent. AAPS Journal. 2005; 7(3): E513-E522.
Dose vs LDL lowering Response: Population, Mixed Effects Response Meta-Analysis:
Statin Dose-Response Relationship: Absence (E 0) and Presence of 10 mg Ezetimibe (E 10) Adapted from: Mandema and Hartman. Model-Based Development of Gemcabene a New Lipid-Altering Agent. AAPS Journal. 2005; 7(3): E513-E522.
Dose-Response Relationship for Non-Statin Without and With Statin With Atorvastatin Alone Adapted from: Mandema and Hartman. Model-Based Development of Gemcabene a New Lipid-Altering Agent. AAPS Journal. 2005; 7(3): E513-E522.
Prediction of Simvastatin Risk Reduction vs Dose Using a Model Based Approach Adapted from D. Stanski
Case Example 2: Gabapentin Approval and Label • Gabapentin was approved by FDA for post-herpetic neuralgia • Approved label states under clinical studies: “Pharmacokinetic-pharmacodynamic modeling provided confirmatory evidence of efficacy across all doses” • Model and Data Provided with Submission • FDA reviewers used model to test various scenarios • Supported doses and conclusions of Pfizer • Provided confidence to eliminate need for replicate doses • FDA proposed language in the label on PK-PD modeling and clinical trials Adapted from: Miller, J Pharmacokinet Pharmacodynamics, Vol. 32, No. 2, April 2005
Gabapentin PHN Study Designs • Used all daily pain scores • Exposure-Response analysis utilized titration data for within-subject dose response Adapted from: Miller, J Pharmacokinet Pharmacodynamics, Vol. 32, No. 2, April 2005
Gabapentin PHN Data Fits Time Dependent Placebo Response, Emax Drug Response and Saturable Absorption Adapted from: Miller, J Pharmacokinet Pharmacodynamics, Vol. 32, No. 2, April 2005
Adapted from: Miller, J Pharmacokinet Pharmacodynamics, Vol. 32, No. 2, April 2005
Case Example 3: Optimizing Dose Selection for an ACE Inhibitor • A 2-compartment PK model with first order absorption and first order output • Daytime variation of ACE is described with a cosine function with time period tp, amplitude A and shift ACE(t)=ACEo+ A cos(2π (t+S)/tp) • An Emax model and a sigmoidal Emax model are tested to describe the relationship between concentrations and plasma ACE activity Adapted from: Pfister. Dose selection of M100240. J Clin Pharmacol 2004 Jun;44(6):621-31
Simulation Scenarios • Target: • > 90% inhibition of plasma ACE activity in at least 50% of patients • Simulations at steady state: • For comparison of oral daily doses ranging from 25 to 150 mg • PK and plasma ACE activity profiles (n=500) under these dosage regimens are simulated with parameters drawn from the population PK and PD distribution
Model Based Simulations of BID Regimens PK ACE activity Fraction of patients achieving target; horizontal lines denote 50 and 80% 24h 24h
Simulations at Steady State • Simulations are used to evaluate candidate QD and BID dose regimen to achieve >90% plasma ACE inhibition at 24 hours • For comparison of oral daily doses ranging from 25 to 250 mg, PK and plasma ACE activity profiles (n=500) under these dosage regimens are simulated with parameters drawn from the population PK and PD distribution
Case Example 4: Tipranavir (TPV) Approval and Label • Protease inhibitor for experienced patients or patients with viral resistance to other PIs • Plasma TPV levels are a major driver of efficacy and toxicity, boosted with ritonavir (RTV) • HIV-1 protease mutations represent a major driver of resistance and decreased efficacy • 500/200 TPV/RTV dose employed in Phase III • Plasma TPV levels > IC50 to suppress viral load and avoid development of resistance Adapted from: FDA Antiviral Drug Advisory Committee Meeting, May 19, 2005
Inhibitory Quotient (IQ) As A Predictor of Efficacy • Protein Binding Correction Factor (PBCF = 3.75x) • TPV is highly bound in plasma (99.96 - 99.98%) • Cell culture media only contains 6% fetal bovine serum (99.88%) • PBCF estimated using 2 methods: • Method 1: Equilibrium Dialysis: 0.120% free / 0.034% free = 3.5x • Method 2: Addition of 75% human plasma to antiviral assay resulted in a 4x shift • IQ = Cmin / (IC50 fold WT ● mean WT HIV IC50● 3.75) standardized TPV susceptibility PBCF susceptibility in patient isolate PK Adapted from: FDA Antiviral Drug Advisory Committee Meeting, May 19, 2005
Control+T20 Control TPV Cmin, IC50, T20 Parameters and Viral Response For IQ ≥ 100, 54% responded to TPV and 73% responded to TPV+T20For IQ < 100, 21% responded to TPV and 52% responded to TPV+T20 Adapted from: FDA Antiviral Drug Advisory Committee Meeting, May 19, 2005
Benefit: Viral Load Change From Baseline (log10) Risk: Grade 3-4 ALT, AST or Bilirubin Risk vs. Benefit: Impact of IQ on 24-Week Viral Load Response and Cmin on Liver Toxicity Adapted from: FDA Antiviral Drug Advisory Committee Meeting, May 19, 2005
TPV Label Statements • “Among the 206 patients receiving APTIVUS-ritonavir without enfuvirtide…..the response rate was 23% in those with an IQ value < 75 and 55% with an IQ value > 75.” • “Among the 95 patients receiving APTIVUS-ritonavir with enfuvirtide, the response rate in patients with an IQ < 75 vs. those with IQ > 75 was 43% and 84% respectively.” Adapted from: TPV Label, under “Pharmacodynamics”.
Case Example 5: Drug Disease Model • Mechanistic disease model for HIV/AIDS • Pharmacodynamic model incorporating dose, concentration, HIV viral load time course • Biomarkers of efficacy – viral RNA time course • Biomarkers of safety – GIT events time course • Dose response relationships or PK/PD model • Outcome analysis
Viral Dynamics Adapted from: Bonhoeffer (1997) Proc. Natl. Acad. Sci. USA 94, 6971-6976
p d2 PI Active Infected l fAbVT CD4+ Cells (N)NRTI Virus a + fLbVT Latent Infected (N)NRTI d1 c d3 Drug Disease Models l:production rate of target cell d1: dying rate of target cell c: dying rate of virus b: infection rate constant d2: dying rate of active cells d3: dying rate of latent cells p: production rate of virus Adapted from: J Acquir Immun Defic Syndr 26:397, 2001, FDA EOP2A Slides
Case Example 5: Applying Drug HIV Disease Model • Maraviroc (MVC;UK-427,857) • Novel CCR5 antagonist in development for the treatment of HIV • Blocks the CCR5 receptor, which is used by HIV to enter CD4+ cells • Simulate decline of HIV-1 RNA plasma levels for 400 patients per treatment arm • Dosing regimens simulated were as follows: 150 mg twice daily fed, 150 mg twice daily fasted, and 300 mg once daily fasted • HIV-1 RNA measurements were performed daily for 40 days after the start of treatment
HIV-1 RNA Log10 Time Course Adapted from Rosario. J Acquir Immune Defic Syndr, Volume 42(2). 2006.183-191
Measured and Predicted HIV-1 RNA log10 Adapted from Rosario. J Acquir Immune Defic Syndr, Volume 42(2). 2006.183-191
Measured and Model Simulated HIV-1 RNA log10 Adapted from Rosario. J Acquir Immune Defic Syndr, Volume 42(2). 2006.183-191
Predicted Inhibition Fraction in Function of Mean Viral Load Drop Adapted from Rosario. J Acquir Immune Defic Syndr, Volume 42(2). 2006.183-191
Prior Knowledge -Analogues -Disease -Competitors -Patients -Discovery/Pre-clinical Drug Molecule Clinical Development Plan Trial 1 Trial 2 Trial … Trial N Summary: QCP Enabled Drug Development Program Drug Product -indication -patients -formulation -dose -safety/ efficacy QCP Enablers: Reducing Uncertainty in Risk/Benefit Drug and Disease Modeling Dose Response, PK-PD and Dosing Targeted Label Information Optimal Use Adaptive Trial Design “The best way to predict the future is to create it” – Peter F. Drucker
Acknowledgements • John Wagner (Merck) • Gary Herman (Merck) • Marc Pfister (BMS) • Joga Gobburu (FDA)