Physics of Cardiac Arrhythmias Sitabhra Sinha Institute of

Physics of Cardiac Arrhythmias Sitabhra Sinha Institute of Mathematical Sciences (IMSc) Chennai 600 113, INDIA Collaborators: R Pandit, A Pande, T K Shajahan, A Sen (IISc, Bangalore) D J Christini and K M Stein (WMC- Cornell University, NYC)

Outline Motivation n Cardiac arrhythmias: tachycardia & fibrillation n Reentrant waves and spiral turbulence in excitable medium : a model for VF & VT n Models: Luo-Rudy I & Panfilov models n Spiral formation and breakup: VT and VF n Control of reentry and spiral chaos n Implications for cardiac pacing & defibrillation n Summary n

Motivation: Why Study Fibrillation ? n n Sudden cardiac death due to ventricular fibrillation (VF) is the leading cause of death in the industrialized world. One-third of all deaths in the USA are due to cardiac arrest - one out of six due to VF. Understanding VF is an essential prerequisite for improving current methods of defibrillation (massive electrical shocks ~ 600 Volts). Possible alternative: Controlling spatio-temporal chaos of VF through low-amplitude perturbations.

What is Ventricular Fibrillation ? n n Ventricular Fibrillation (VF): a disorganized electrical wave activity that destroys the coherent contraction of ventricular muscle. Underlying cause of VF: formation of “electrical vortices” - 2 -D (spiral)/3 -D (scroll) waves of action potential - creation of re-entrant pathways of electrical activity. Spiral/ scroll waves lead to abnormally rapid heart beat (t ~ 100 -200 ms) (Tachycardia). Ventricular tachycardia (if untreated) leads to VF in a few seconds through spiral/scroll wave break-up.

Characterizing Ventricular Fibrillation Normal sinus rhythm Tachycardia Ventricular Fibrillation Spiral wave on the surface of a canine ventricle. Color proportional to transmembrane potential. Image obtained with voltage sensitive dyes and CCD camera. http: //www. physics. gatech. edu /chaos http: //www. physics. gatech. edu/chaos

A Brief History of VF n 1874: “Fibrillation” (Alfred Vulpian) “. . . individual fibers contracting independently. . . a wad of writhing worms” (Tacker and Geddes, 1980) n 1888: “Sudden Cardiac Death” (J. A. Mac. William) “. . . the cardiac pump is thrown out of gear, and the last of its vital energy is dissipated in a violent and prolonged turmoil of fruitless activity in the ventricular wall. ” n n 1899: Electrical defibrillation in animals (Prevost and Batelli). 1914: Fibrillation induced through precisely timed electrical stimulus (G. R. Mines). 1947: Defibrillation of human heart (Claude Beck) 1960 s: Initial work in Internal Cardiac Defibrillator ( Mirowski).

Anatomy of the Heart

Anatomical vs. Functional Reentry Anatomically determined (Mines, 1913) Functionally determined (Allessie et al. , 1977) 1. Fixed length of circuit (determined by anatomical obstacle). 1. Circuit length dependent upon electrophysiological properties. (“Spiral waves”) 2. Usually excitable gap between head and tail of impulse. 2. No gap of full excitability. 3. Inverse relation between revolution time and conduction 3. Revolution time proportional to

George Ralph Mines (1886 -1914) Proposed theoretical basis for occurrence of reentrant arrhythmias. 1913: proposed a model for generatiing reentrant rhythms -a dual pathway with differing electrophysiologic properties; suggested that the twin conditions of unidirectional block and slow conduction may occur in abnormal myocardial tissue - allowing a circulating wavefront to be sustained as conductive tissue is always available for excitation. Cambridge University, 1912 Diagram from Mines(1913) demonstrating circulating rhythms in closed circuits in myocardial tissue: (a) Normal tissue; (b) Abnormal tissue with delayed conduction.

Implantable Cardioverter. Defibrillator (ICD) An ICD consists of a pulse generator and electrical leads. Endocardial leads are inserted through a vein and advanced to the right ventricle and/or atrium. The pulse generator is placed subcutaneously or submuscularly and connected to the leads. The ICD constantly monitors heart rhythm. Upon detection of VT/VF delivers a programmed treatment. Capable of applying variety of possible treatments: • Pacing - deliver a sequence of low-amplitude pulses. • Cardioversion - a mild shock (if pacing fails in terminating VT). • Defibrillation - a large shock to terminate VF. • Pacemaker - for slow heartbeat, can act as pacemaker.

Internal Defibrillation Example of an implantable cardiac defibrillator (ICD) Vol 54 cc Mass 97 gm Thickness 16 mm Longevity 9 yrs BOL Voltage 6. 4 V BOL Charge time 6. 0 sec

Pacing Algorithms Is there an optimal anti-tachycardia pacing algorithm ?

Spiral Waves in the Heart (Left) Spiral wave in anatomically correct model of the dog heart. Color code indicates calculated activation times (in milliseconds) of various regions of the heart muscle. (Right) Reentrant spiral wave excitation in a rabbit heart observed with a voltage-sensitive dye. Color code indicates measured activation time in milliseconds.

The Cardiac Cell Gap junction Myocardial fibers: contractile strand of cardiac muscle composed of many cells. Voltage-gated ion channels are pathways for charge movement (Na+, K+, Ca++ ions).

The Luo-Rudy (L-R) model Biologically realistic model for ventricular action potential proposed by Luo & Rudy (I: 1991, II: 1995): incorporates details of ionic currents. The L-R I model has 8 coupled ODEs describing the activity of each cardiac cell: the transmembrane potential (V ), the intracellular Calcium concentration (Cai) and six ionchannel gating variables (m, h, j, x, xi, d, f ). Action potential and ionic currents of a ventricular myocyte simulated with L-R model.

Luo-Rudy I model: Membrane potential equation The transmembrane potential V follows the reaction-diffusion equation where Cm = 1 F/cm 2 is the membrane capacitance, D (k -1) is the conductivity constant and ILR ( A/cm 2) is the instantaneous total ionic current through the cell : Inward currents: INa : Fast sodium current (Na+) Isi : Slow inward current (Ca++) Outward currents : IK : Time-dependent potassium current IK 1 : Time-independent potassium current IKp: Plateau potassium current Ib : Background current

Luo-Rudy I model: Formulation of Ionic Currents Fast sodium current (E Na = 54. 4 m. V): Slow inward current (E Si = 7. 7 -13. 0287 ln ([Ca] i)): Time-dependent potassium current (E K = -77 m. V): Time-independent potassium current: Plateau potassium current: Background current:

Luo-Rudy I model: Ca++ concentration and gating variable equations The intracellular calcium concentration (Ca i) satisfies the ODE: Each ion channel gating variable ( = m, h, j, x, xi, d, f ) is governed by ODEs of the form: The parameters and are functions of the rate constants and : and : normalized fraction of the population of ion channels in open state, : rate at which channels open; : rate at which the channels close. The rate constants and are complicated functions of the membrane potential V.

The Beeler-Reuter (B-R) model Biologically realistic model for ventricular action potential proposed by Beeler & Reuter (1977) - incorporating details of ionic currents. The B-R model has 8 coupled ODEs describing the activity of each myocardial cell - corresponding to the transmembrane potential (V ), the intracellular Calcium concentration (c) and six ion-channel gating variables (m, h, j, x, d, f ). Variation of the transmembrane potential during an action potential in the B-R model.

Beeler-Reuter model: Membrane potential equation The transmembrane potential follows the reaction-diffusion equation where IBR is the instantaneous total ionic current through the cell : IK : transient outward potassium current INa : fast sodium inward current Ix : time-activated outward current (mostly K+ ions) Is : slow inward calcium current

Beeler-Reuter model: Calcium concentration and gating variable equations The calcium concentration (c ) satisfies the ODE: Each ion channel gating variable ( = m, h, j, x, d, f ) is governed by ODEs of the form: The parameters and are functions of the rate constants and : and Note: represents the normalized fraction of the population of ion channels that is in an open state, is the rate at which channels open and is the rate at which they close.

Beeler-Reuter model: Rate constant equations

Excitable Media • Subthreshold stimulation perturbation decays • Suprathreshold stimulation • Feature: Conduction of propagating waves. • Pattern formation: wave of excitation can change the properties of excitable media and cause the formation of spatial patterns. Examples: • Aggregation of Dictyostelium Discoideum amoebae - the monolayer of the starving amoebae is an excitable medium which conducts excitation waves of the intra-cellular mediator, c. AMP.

The Heart as an Excitable Medium n n Excitable medium : a small but finite perturbation from equilibrium can lead to excitation (a large excursion away from equilibrium) before equilibrium is restored. Excitation in the Heart : the electromechanical wave inducing cardiac-muscle contractions which pump blood. Refractory period : once excited, the medium remains quiescent for a certain duration. Spiral waves associated with abnormal cardiac activity (Winfree).

The Panfilov Model for Ventricular Fibrillation n n Fitzhugh-Nagumo model for excitable media with `Puschino’ kinetics Describes an excitable medium with an absolute and a relative refractory period - the period after repolarization during which the membrane recovers its resting properties. The simplest model that shows spiral breakup qualitatively similar to that seen during VF. `Puschino’ kinetics shortens the relative refractory period. Shows a long chaotic transient, whose duration increases sharply with system size - agrees with the observation that the hearts of larger animals are more

Panfilov Model: Equations n n n n Described by two coupled partial differential equations. Variables: membrane potential, e (fast variable) and effective membrane conductance, g (slow variable). e / t = i dij j e - f(e) - g, g / t = (e, g) [k e - g]. dij : conductivity tensor - for isotropic medium, replaced by Laplacian. Diffusive term describes the coupling among cells. f(e) : nonlinear function - piecewise linear nature. (e, g) : information about refractory periods.

Panfilov Model: Parameters n n n n n (e, g) = 1, e < e 2 (e, g) = 2, e > e 2 (e, g) = 3, e < e 1 and g < g 1 Variables: membrane potential, e (fast variable) and effective membrane conductance, g (slow variable). e / t = i dij j e - f(e) - g, g / t = (e, g) [k e - g]. dij : conductivity tensor - for isotropic medium, replaced by Laplacian. Diffusive term describes the coupling among cells. f(e) : nonlinear function - piecewise linear nature.

Panfilov Model: Dynamics n n Dynamics in the absence of the diffusion term. e changes at a fast rate compared to g.

Spiral Turbulence in the 2 -D Panfilov model Pseudo-color plots of the e field at various values of 1 ( 3 = 0. 3). As 1 decreases, the pitch of the spiral decreases ultimately leading to spiral breakup. Local phase portraits at various values of 1. The increasing scatter of points indicates the onset of spatiotemporal chaos.

Spatiotemporal Chaos in the 2 -D Panfilov model The maximum Lyapunov exponent ( max) plotted as a function of time t : max approaches a positive constant (~ 0. 2) and then decays at large times to negative values The maximum value attained by the Kaplan. Yorke dimension DKY during spatiotemporal chaos - plotted as a function of linear system size L. The lifetime of the chaotic transient increases with L.

Panfilov Model: Controlling Spatiotemporal Chaos The model has non-conducting boundaries (no-flux or Neumann boundary conditions) - as the ventricles are electrically insulated from the atria. Observation : n n n Non-conducting boundaries absorb spiral defects. Spirals do not last for appreciable periods in small systems. Operating Principle for the Control Scheme : n n To divide the system ( L L ) into K 2 smaller blocks. Isolate the blocks ( of size L / K ) by stimulating the system along the block boundaries - driving them to refractory state. Each block is too small to sustain spiral activity - spirals absorbed by block boundaries. After the system is driven to the quiescent state, controlling stimulation is withdrawn - block boundaries recover from refractory

Control parameters in 2 -D Panfilov model L = 128 Pulse amplitude 57. 3 m. V / msec kept on for = 41. 2 msec. This implies a defibrillating current density of 57 A/cm 2. Beeler-Reuter model L = 200 Pulse amplitude 20 m. V / msec kept on for = 120 msec suffices. This implies a defibrillating current density of 20 A/cm 2.

Panfilov Model: Control in 3 -D Control algorithm as in 2 -D with the following modifications: • Control mesh only on one free face of a 3 -D domain ( L x L z). • With L = 256 control obtained for 4 L z. • For L z > 4, pulse control is necessary: • activate control mesh after msec • keep it on for ON msec • turn it off for OFF msec • keep it on for ON msec • continue n times We find ON = 0. 11 msec, OFF = 22 msec and n = 30 suffices. Note that OFF is the duration of one action potential.

Implications for Defibrillation n n n Current defibrillation techniques involve applying electrical shock to the fibrillating heart. Principle of operation: Simultaneous depolarization of all cells - so that the cardiac pacemaker can take over. External defibrillation ~ 5 k. V. Internal Cardiac Defibrillator (ICD) ~ 600 V. We propose using very-low-amplitude pulse (~ m. V) applied for a brief duration ( ~ 100 ms). Control over 2 -D surface is effective even for 3 -D control.

Summary n n Cardiac arrest due to VF is a spatio-temporally chaotic phenomenon. VF arises due to break-up of spiral/scroll waves induced by re-entrant activity. Panfilov model is the simplest one that shows spiral breakup similar to that in VF. We have controlled spiral break-up in 2 and 3 -D in the Panfilov model and the more realistic Luo -Rudy model of the heart.

Outline Motivation n Heart as excitable medium n Reentrant activity n Fitzhugh-Nagumo model for Ventricular Fibrillation (VF) n Spiral formation n Spiral breakup and VF n Control of spatio-temporal chaotic activity n Implications for defibrillation n

Motivation: Why Study Anti. Tachycardia Pacing ? n n n Sudden cardiac death is the leading cause of death in the industrialized world. One-third of all deaths in the USA are due to cardiac arrest. One out of six due to VF. Understanding VF is an essential prerequisite for improving current methods of defibrillation (massive electrical shocks ~ 600 Volts). Possible alternative: Controlling spatio-temporal chaos of VF through low-amplitude perturbations.

What is Ventricular Tachycardia ? n n Ventricular Tachycardia (VT): Abnormally rapid heart beat (t ~ 100200 ms). The heart is unable to pump blood efficiently for such rapid beating. Underlying cause of VT: creation of re-entrant pathways of electrical activity. VT (if untreated) may degenerate to Ventricular Fibrillation (VF) leading to death in minutes. Normal sinus rhythm Tachycardia Ventricular Fibrillation

Termination of Reentry by Pacing Each pacing wave splits into two branches while traveling around the reentry circuit : • Anterograde (along the direction of the rentrant wave) • Retrograde (against the direction of the reentrant wave). Pacing can result in : • No effect on the reentrant wave. • Resetting of the reentrant wave (the retrograde wave collides with the reentrant wave - the anterograde wave becomes the new reentrant wave). • Termination of reentry occurs by block in the anterograde direction since the retrograde branch of the wave will collide with the reentrant wave and annihilate each other.

Pacing Termination of Ventricular Tachycardia Several factors influence the ability of pacing to interact with VT: • VT cycle length. • The refractory period at the stimulation site and at the VT circuit. • The conduction time from the site of stimulation to the VT circuit. • The duration of the excitable gap. Why multiple stimuli ? Large number of conditions for reentry to be terminated single stimulus rarely sufficient. Double stimuli often used: first stimulus used only to “peel back” refractoriness, allowing the second stimulus to interact with the circuit more The response pattern of VT to the delivery of single prematurely. & double pacing stimuli (Josephson, 1993)

Pacing Termination of Reentry in the 1 -D Ring Pacing termination of reentry in 1 -D ring - a wellstudied problem. Termination occurs when the anterogarde branch of the reentrant wave is blocked in a region which is still refractory after the passage of the reentrant wave. Proper timing of the pacing wave is crucial. Glass (1995): From continuity arguments, there exists a range of stimuli phases and amplitudes that lead to successful reentry termination. When the pacing site is not located on the 1 -D ring itself (as is generally the case in any realistic pacing arrangement) this method of pacing termination of reentry fails.

Off-circuit Pacing of Reentry in 1 -D Ring Consider a homogeneous reentrant circuit (length= L). Pacing site located on ENTRY sidebranch distance z from circuit. • ENTRY sidebranch: x = 0. • Conduction velocity = c. • Refractory period = r. • t=0: Pacing stimulus applied. • t=z/2 c: Stimulus collides with reentrant wave branch propagating out through ENTRY sidebranch. • t=r: Pacing site recovers. • t=r+(z/c): 2 nd stimulus (applied at t=r) reaches the circuit but refractory tail of reentrant wave away from the stimulus ENTRY Whenatz distance > 0, it isx=z impossible for the sidebranch anterograde branchtail of in thea stimulus to catch up- with the refractory will not be blocked. homogeneous medium.

The Critical Role of Inhomogeneity Assuming inhomogeneity - e. g. , longer refractory period or slower conduction in narrow channel between non-conducting obstacles may lead to successful block of the anterograde branch of the pacing wave. (Abildskov & Lux, 1995) If an inhomogeneity (e. g. , a zone of slow conduction) exists in the circuit, the waves travel faster or slower depending on location in the circuit. As a result, stimuli may arrive at the circuit from the pacing site and encounter a region that is still refractory - leads to block of the anterograde branch of the stimulus successful termination.

L-R model simulation results: dimensional ring Parameter space dgm. of Coupling Interval (CI) vs. Pacing Interval (PI) at which termination occurs for 1 -D L-R ring of length 250 mm with a zone of slow conduction (25 mm) and VT period = 1303. 07 ms. (For top figure, CI = 899 ms & PI = 600 ms. ) One. Spatiotemporal propagation of a reentrant wave in a ring (L=250 mm)successfully terminated by two pacing stimuli applied at x=0 mm (at T 1 = 2600 ms and T 2=3200 ms). Zone of slow conduction has step boundary.

Two-dimensional Excitable Media Model Schematic diagram of anatomical figure-of-eight reentry - paced from stimulation site at the apex of the ventricles. Inset: Model used for 2 -D simulations. Square patches represent non-conducting scar tissue. Pacing site is at the ventricular apex - pacing waves propagating upward from the site represented as plane waves from the bottom. No-flux boundary conditions at top and bottom; periodic boundary conditions at

Panfilov model simulation results: Homogeneous Media 2 -D

Panfilov model simulation results: Inhomogeneous Media 2 -D

Effect of Anisotropy Cardiac tissue shows anisotropic propagation - the action potential propagates faster along the direction of the myocardial fibers than transverse to it. Axis of anisotropy rotates along the thickness of the myocardium (from the endocardial to the epicardial layer). For human ventricular myocardium, longitudinal conduction velocity ~ 50 cm/sec and transverse conduction velocity ~ 14 cm/sec. We used anisotropy ratio = 1: 0. 3. Our simulations of pacing in anisotropic models showed no qualitative difference from results in isotropic models.

Outlook Implications for pacing algorithms Limitations: • 1 - and 2 -Dimensional - instead of 3 -Dimensional (the anisotropy axis rotates along the thickness). • Another method of simulating ischemia - increasing K+ concentration (But this does not provide a chronic arrhythmogenic substrate). • Monodomain assumed instead of Bidomain (this is justified for the lowamplitude stimulus used in pacing).

Summary n n Cardiac arrest due to VF is a spatio-temporally chaotic phenomenon. VF arises due to break-up of spiral/scroll waves induced by re-entrant activity. Fitzhugh-Nagumo model with Puschino kinetics (Panfilov) is the simplest one that shows spiral breakup similar to that in VF. We have controlled spiral break-up in 2 and 3 -D in the Panfilov model (we are extending the control technique to more complex models of the heart).

Acknowledgements Collaborators: • Ashwin Pande • Prof. Rahul Pandit Computational Facilities: SERC, IISc Financial Support: JNCASR, Bangalore & CSIR