in silico modeling progress report for the cipa...
TRANSCRIPT
In Silico Modeling Progress Report for the CiPA Initiative
Zhihua Li, Ph.D.Division of Applied Regulatory Science
Office of Clinical Pharmacology
Dec 6, 2016
2
Challenges and Possible Solutions• Drug blocking assays are usually done at room temperature:
– Temperature-sensitive model to extrapolate data from room temperature to physiological temperature
• Drug binds to channels in a time- and state- dependent manner: – Modeling dynamic drug-channel interactions is preferable to IC50s
• In vitro hERG currents have some differences from native IKrcurrents: – Model needs to accurately replicate IKr current profiles and action
potential wave forms of human ventricular cardiomyocytes• The number of training and validation compounds is limited:
– Physiology/mechanism-based metric rather than statistics• Experimental data have variability:
– Incorporating experimental variability into the model to translate experimental uncertainty into metric/TdP risk uncertainty
3
Development of a Temperature Sensitive hERG Model
• Because O’Hara-Rudy model operates at physiological temperature, while industry-generated hERG data are often obtained at room temperature, a dynamic, temperature-sensitive hERG model is required
• We developed a modified hERG model that can reproduce temperature-induced changes in major channel gating processes
4
Modeling Dynamic drug-hERG Interactions
• A novel model was developed to capture dynamic drug-hERGinteractions, especially drug being trapped within closed channel (red arrows)
• Modeling shows that CiPA High TdP Risk compounds tend to have a higher probability of being trapped within hERG
Li Z et al. Improving the In Silico Assessment of Proarrhythmia Risk by Combining hERG-Drug Binding Kinetics and Multi-channel Pharmacology. (Under review)
O = OpenC = ClosedI = Inactivated
Channel gating Drug binding
5
Incorporating the dynamic hERG Model into the ORd Cardiomyocyte Model
www.fda.gov 5
O’Hara et al. PLoS Computational Biology 2011.
• The dynamic hERG model was integrated into the O’Hara-Rudy dynamic (ORd) cell model
• The modified ORd model was re-calibrated to reproduce experimentally recorded Action Potential Duration (APD) – Cycle Length (CL) relationship
refitting APD30
APD50
APD70APD90
Circles: experimentalLines: Simulation
6
Key Mechanism of TdP: imbalance of Inward and Outward Currents
Normal Action potential
Increased ratio between inward and outward currents
Inward Outward
ICaL IKr
INaL IKs
IK1
Major currents modulating plateau duration
Modified fromhttp://tmedweb.tulane.edu/pharmwiki/doku.php/cellular_basis_for_arrhythmias
plateau
7
Fold change over Cmax
A Candidate Metric: Change of Inward Currents at Slow Heart Rates
www.fda.gov 7
• Red: CiPA TdP High Risk• Blue: CiPA TdP Intermediate
Risk• Green: CiPA TdP Low/No Risk
: EAD induced
Change of Inward Currents: % Change of integral of inward currents (INaL and ICaL)
• hERG dynamic parameters from Dr. Wend Wu’s lab (FDA) and IC50s (INaL, ICaL, INa, IK1, IKs, and Ito) from Dr. Bill Crumb’s lab (Cytocentric) ; pacing at 2000 ms (0.5 Hz) to mimic bradycardia
• Three CiPA TdP categories were separated in a dose-dependent manner; three out of the four High Risk compounds induced Early After Depolarization (EAD, star in the plot)
• Separation is only good at high concentrations; Metric “ignores” outward currents
• A more physiological metric: change of Inet (difference between inward and outward current) ?
Chan
ge o
f Inw
ard
Curr
ent quinidine
bepridildofetilide
sotalol
cisaprideondansetronchlorpromazineterfenadine
verapamilranolazine
diltiazem
mexiletine
8
Improving the ORd Model Based on Experimental Data
IKr blocker (1 µM E-4031) ICaL blocker (1 µM nisoldipine)
INaL blocker (10 µM mexiletine)
Dutta S et al. Optimization of an In Silico Cardiac Cell Model for Proarrhythmia Risk. (In preparation)
experimentalORd simulationORd2 simulation
.
experimentalORd simulationORd2 simulation
.
experimentalORd simulationORd2 simulation
.
• ORd: original ORd simulation• ORd2: improved ORd model simulation
9
• Red: CiPA TdP High Risk• Blue: CiPA TdP Intermediate
Risk• Green: CiPA TdP Low/No Risk
Change of Inet: % change of the integral of Inet(inward currents – outward currents)
mexiletine
diltiazem
verapamilranolazine
Chan
ge o
f Ine
tAnother Candidate Metric: Change of Inet
chlorpromazineterfenadineondansetroncisapride
bepridilsotalol
dofetilidequinidine
• Simulation was done using ORd2 with integrated hERG dynamic model
• Drug separation is good along all concentrations from 1x to 25x Cmax
Fold change over Cmax
10
Metric Distribution Caused by Experimental Uncertainty
Change of Inet• A method combining MCMC (Dr. Gary Mirams) and bootstrapping was developed to capture uncertainty
• Simulation was done using ORd2 with integrated hERG dynamic model at 2x Cmax when pacing at 2000 ms (0.5 Hz)
• Each drug has a distribution of possible metric (change of Inet) values due to experimental uncertainty
• The distribution peaks (most probable metric values for each drug) are completely separated for the three categories
%Change of Inet
11www.fda.gov
• A temperature-dependent hERG model was developed and incorporated into the O’Hara-Rudy model to replicate normal myocyte electrophysiology and dynamic drug block
• Model calibration was done using manual patch clamp data and will continue using automated patch clamp data that meet data quality criteria (being determined)
• Two promising metrics identified; their performance to be assessed using independent validation drugs
• Method to incorporate experimental uncertainty established; Method to capture inter-subject variability also being considered in collaboration with Blanca Rodriguez’s group at Oxford
• The model/metric, along with supporting data and methods, are being regularly updated and made freely available to the public through publications and a proposed web portal system
Summary and Next Steps
12
AcknowledgementsFDA In Silico Working Group MembersDavid StraussSara DuttaKelly ChangKylie BeattieThembi MdluliWendy Wu (patch clamp)Phu Tran (patch clamp)Jiansong Sheng (patch clamp)Thomas Colatsky
CiPA Model Development Advisory GroupColleen Clancy (Cornell)Gary Mirams (Oxford)Blanca Rodriguez (Oxford)Yoram Rudy (WUSL) Tom O’Hara (JHU) Natalia Trayanova (JHU)Wayne Giles (Calgary)Gail Roberston (Wisconsin)
ICWG / Rapid Response TeamBernard Fermini (Pfizer)Najah Abi-Gerges (AnaBios)Adam Hill (Victor Chang CRI)Jamie Vandenberg (Victor Chang CRI)Jules Hancox (Bristol)William Crumb (Zenas)
FDA Research Colleagues Norman Stockbridge Rick Gray Pras Pathmanathan Ksenia Blinova Jose Vicente Lars Johannesen Maria Iacono
External Research Colleagues Alfonso Bueno (Oxford)