Short-Term Memory and Restitution During Ventricular Fibrillation in Human Hearts An In Vivo Study

Short-Term Memory and Restitution During Ventricular Fibrillation in Human Hearts An In Vivo Study Satish C. Toal, MD; Talha A. Farid, MD; Raja Selvaraj, MD; Vijay S. Chauhan, MD; Stephane Masse, MASc; Joan Ivanov, PhD; Louise Harris, MD; Eugene Downar, MD; Michael R. Franz, MD; Kumaraswamy Nanthakumar, MD Background Action potential duration (APD) variation is an important determinant of wave break and reentry. The determinants of APD variability during early ventricular fibrillation (VF) in myopathic human hearts have not been studied. The objective of this study was to study the role of APD restitution and short-term cardiac memory on variation in human VF. Methods and Results The study consisted of 7 patients (67 9 years old) with ejection fraction 35%. Monophasic action potentials were recorded from the right and/or left ventricular septum during VF. APD 60/90 was measured in sinus beat preceding induction of VF, and its amplitude was used to define 60%/90% repolarization in VF. The monophasic action potential upstroke (dv/dt max ) was used to characterize local excitability. Simple linear regression showed that variability in APD n60 was determined by APD/diastolic interval restitution (R 2 0.48, P 0.0001) and short-term memory (APD 60 n 1, n 2, n 3, n 4; R 2 0.55, 0.40, 0.33, and 0.27 respectively; P 0.001). Using multiple stepwise regression, short-term memory and restitution accounted for 62% of variance in APD 60 (P 0.001). Individually, memory effect had the greatest contribution to APD variability (R 2 0.55, P 0.0001). Conclusions In early human VF, short-term memory and APD/diastolic interval restitution explain most of the APD variability, with memory effects predominating. This suggests that in early human VF, short-term cardiac memory may provide a novel therapeutic target to modulate progression of VF in myopathic patients. (Circ Arrhythmia Electrophysiol. 2009;2:562-570.) Key Words: fibrillation action potentials sudden death In myopathic human hearts, during early ventricular fibrillation (VF), we recently demonstrated that wave fronts emanating from a few rotors are responsible for the significant degree of organization of electric activity on the endocardium and epicardium. 1,2 Conduction block is responsible for wave front fractionation and reentry, an important mechanism in proliferation of wave fronts and rotors during VF. 3 Oscillations in action potential duration (APD) could increase in a fashion in which a site becomes refractory to depolarization and thus lead to conduction block and reentry. 4 Variation in APD is probably determined by multiple factors in VF. In experimental models of VF, APD restitution (relationship of APD n to immediately preceding diastolic interval (DI n 1 ) and short-term memory (relationship of APD n to preceding APD n 1, n 2, n 3, n 4) have been shown to play a part in APD variability. 5 8 However, these studies in VF were conducted in modeling experiments, 6 tissue preparations, 7 animal Langendorff models, 8 and recently in an in vivo porcine model. 5 The role of APD restitution and memory in APD variability during early in vivo human VF, in which modulating VF could have a significant impact on outcome, has not been studied and is largely unknown. Clinical Perspective on p 570 The concept that the previous DI determines most of the variability in APD is at the core of the APD/DI restitution hypothesis, which suggests that when the slope of the APD/DI relationships is 1, conduction block results, 9 leading to VF. Koller et al 10 have demonstrated that initial restitution curve slope on the right ventricular (RV) endocardium of the myopathic human heart is greater than that of a normal heart. We have shown in myopathic patients that those who are prone to arrhythmia have a steeper curve than those who are not prone to development of arrhythmias. 11 It Received November 5, 2008; accepted June 1, 2009. From the Division of Cardiology, Toronto General Hospital (S.C.T., T.F., R.S., V.S.C., S.M., J.I., L.H., E.D., K.N.), Toronto, Canada; and Veterans Affairs Medical Center (M.R.F.), Washington, DC. Correspondence to Kumaraswamy Nanthakumar, MD, Hull Family Cardiac Fibrillation Management Laboratory, Division of Cardiology, Toronto General Hospital, 150 Gerrard St West, Gerrard Wing, 3-522b, Toronto, Ontario, Canada M5G 2C4. E-mail k.nanthakumar@uhn.on.ca Drs Toal and Farid contributed equally to this study. 2009 American Heart Association, Inc. Circ Arrhythmia Electrophysiol is available at http://circep.ahajournals.org DOI: 10.1161/CIRCEP.108.833442 562

Toal et al Memory and Restitution During Human VF 563 Patients undergoing delayed defibrillation threshold test (DFT) 6 weeks after implantable cardioverter-defibrillator (ICD) implantation were studied. Patients with an acute ischemic event, decompensated heart failure, or hemodynamic and/or respiratory instability within 1 month of the DFT were excluded from the study. None of the patients were chronically ventricularly paced or dependent on ventricular pacing. DFT was done under conscious sedation with Propofol without intubation. The patient s serum electrolytes were all within the normal range within 7 days of the DFT. Figure 1. A radiographic cine image showing Franz MAP catheter positioned at the RV septum to record MAP potentials during VF at DFT in a patient with a dual-chamber ICD system. would be important to know if this concept is applicable during early VF to prevent its progression in myopathic patients. The restitution hypothesis has been established in situations in which cycle length and activation sequences have been constant. 12,13 In VF in which cycle length could change beat to beat, it is unclear what role APD restitution plays in determining APD. 14 Short-term memory is the concept in which APD is determined not only by previous DI but also by APDs from earlier cycles. 15,16 Short-term memory may contribute to APD variance during VF in which the cycle length is not constant. To test these hypotheses, we studied the relative contribution of APD restitution relationship and short-term memory effects on APD variation during early VF in myopathic hearts using the monophasic action potential (MAP) catheter in patients. Methods Patients The University Health Network ethics committee approved the study protocol, and informed consent was obtained from each patient. Experimental Protocol Before induction of VF for determining DFT, an 8F sheath was inserted into the right femoral artery and/or vein. A 7F Franz MAP catheter (Boston Scientific) was used for recording monophasic action potential signals. The MAPs were recorded in areas of healthy tissue in the left ventricular (LV) wall away from regions of infarct and were recorded in the RV apex 2 cm away from the RV electrode, where there was healthy myocardium. We ensured that it was in healthy regions as the ICD lead sensing was appropriate (Figure 1). VF was induced using T-wave shock in 6 patients (Figure 2). In 1 patient, VF could not be induced with T-wave shock but was induced with 50-Hz pacing. VF was allowed to continue until the defibrillator charged and defibrillated the patients. If defibrillation was not successful, it was further recorded up until the second shock cardioverted the patient. External rescue was not required in any patient. Signal Recordings Using MAP Catheter The MAP catheter used in our study has previously been validated against transmembrane action potential recordings from microelectrodes and have been found to have very close correlation. 17 Importantly, during VF, the MAP recordings previously have been shown to represent the underlying voltage time course. 18 In addition, it has been used to characterize in vivo human VF 19 21 and specifically used in myopathic human hearts in vivo for the study of APDs in relation to fibrillation dynamics. 10,19 The MAP catheter was positioned as described by Franz 22 and Sager 23 (Figure 1). The MAP catheter was advanced to make contact with the endocardium until MAP signals qualified for adequacy as previously described by Franz et al 22 (typical MAP morphology with flat baseline, rapid upstroke, plateau, and recovery with amplitude of 10 mv or greater). After a rest period of 5 minutes and verification of MAP catheter position and signal stability, VF was induced (Figure 2). Amplification and sampling rate was performed as described by Franz 22 and Sager. 23 Signals were recorded using the previously described method by Gilmour et al, 10 using 1-kHz sampling with a bandwidth of 0.05 to 500 Hz. Data were transferred to DVDs for offline analysis. Adequacy of MAP for analysis during VF was determined by using criteria proposed by Franz (see Figure 2). 23 Recorded MAP signals were analyzed with a custom semiautomated program written in Matlab (Matlab 6.5 for windows, The Mathworks Inc). APD Figure 2. Top, Surface ECG recording of lead II, revealing a pacing train. Two ECG leads and the MAP recording during and immediately after induction of VF are shown. The 2 panels from the top, representing lead II and lead V 1, respectively, show 1 sinus beat followed by a pacing train for 5 beats terminated with a T-wave-shock (large shock artifact in the ECG leads), which results in the induction of VF. Bottom, Simultaneous MAP recording from the RV; 2 solid lines at the bottom represent the repolarization level for measuring APD 90 and APD 60. Dashed line at the top represents 60% of the MAP amplitude in sinus rhythm, which was used as the criterion to determine adequacy of the MAP recordings during VF.

564 Circ Arrhythmia Electrophysiol October 2009 Table 1. Clinical Characteristics of Patients Patient Age, y Sex EF, 5 IHD -Blockers Antiarrhythmics 1 57 M 27 Y Y N 2 55 M 20 Y Y N 3 65 M 24 to 30 Y Y N 4 69 M 20 to 29 Y Y N 5 72 M 32 Y N N 6 60 M 20 to 39 Y Y N 7 80 M 20 N N Y EF indicates ejection fraction; IHD, ischemic heart disease; M, male; Y, yes; N, no. amplitude was determined from the last paced beat before the induction of VF. This served as a reference level for determining APD n60 or APD n90 during VF (Figure 2). Maximum dv/dt on the upstroke was used to determine the time of activation. Recovery was measured at levels of 60% and 90% repolarization. APD of a beat was derived as the interval from activation to recovery. The preceding DI was the interval from the previous recovery to activation. Data Analysis and Statistics A series of simple linear regression analyses were performed, to determine the relationship of APDs and DIs to APD n. The independent predictors of APD n were determined by a multiple linear regression using stepwise selection. The variables tested included 4 APDs (n 1, n 2, n 3, and n 4) and 5 DIs (n 1, n 2, n 3, n 4, and n 5). The independent variables were entered one at a time in the order in which they most improved the model R 2. The level for retention of a variable in the model was 0.15. Analysis of covariance was performed to study the temporal and LV, RV differences within the model and in clinical patients. The differences between different time segments were tested by 1-way analysis of variance. Specific differences between times (each VF episode was divided into 5-second segments for analyzing effects of time) were identified by the post hoc Scheffe test. Results All patients were male. The mean age of the 7 patients was 67 9 years (range, 55 to 80 years). Patient characteristics are defined in Table 1. Of the 9 episodes, in 3, MAP recordings were obtained from the LV and in 6 MAPs were recorded in the RV. In patients 3 and 4, both RV and LV VF data were obtained but not simultaneously. In patient 7, only LV VF recordings were obtained. In all other patients, the VF recordings were from the RV only. The mean VF episode lasted 11.4 5 seconds. Simple Regression Models Table 2 details the individual simple regression relationship between APD n, short-term memory, and APD restitution. APD Restitution and APD Variability The restitution relationship obtained by plotting DI n 1 to APD n90 for the entire VF episodes showed a scattered cluster of points with a negative linear slope (Figure 3A; also see Table 2). Thus, as seen in Figure 3B, paradoxically, a short DI was followed by a long APD and a long DI was followed by a relatively short APD. For all VF episodes, APD n90 correlated significantly with DI n 1 (R 2 0.51, P 0.0001) and with APD n60 (R 2 0.48, P 0.0001). Individually in each patient, the restitution relationship obtained by plotting DI n 1 to APD n90 showed a scattered cluster of points with a negative linear slope in each patient (Figure 4) (slope 0.54, 0.3, Table 2. Simple Linear Regression Analysis of All VF Episodes and RV and LV VF Episodes at APD n60 and APD n90 APD 90 DI n 1 DI n 2 DI n 3 DI n 4 DI n 5 APD n 1 APD n 2 APD n 3 APD n 4 All VF R 2 0.51 0.40 0.21 0.20 0.27 0.62 0.40 0.35 0.39 P 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 RV VF R 2 0.61 0.50 0.24 0.20 0.31 0.68 0.46 0.39 0.41 P 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 LV VF R 2 0.13 0.04 0.07 0.15 0.09 0.33 0.09 0.11 0.20 P 0.0001 0.0269 0.0057 0.0001 0.0011 0.0001 0.0010 0.0003 0.0001 APD 60 DI n 1 DI n 2 DI n 3 DI n 4 DI n 5 APD n 1 APD n 2 APD n 3 APD n 4 All VF R 2 0.48 0.20 0.20 0.06 0.05 0.55 0.40 0.33 0.27 P 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 RV VF R 2 0.54 0.19 0.21 0.28 0.13 0.59 0.50 0.42 0.34 P 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 LV VF R 2 0.38 0.26 0.20 0.05 0.06 0.47 0.22 0.18 0.18 P 0.0001 0.0001 0.0001 0.0180 0.0086 0.0001 0.0001 0.0001 0.0001

Toal et al Memory and Restitution During Human VF 565 Figure 3. A, DI n 1 of all sustained VF episodes is plotted to the corresponding APD n90. The resultant slope defining APD/DI restitution shows that DI n 1 contributes to APD n90 duration in early in vivo human VF, with an R 2 of 0.5134. Additionally, the APD/DI restitution slope is negative. This means that a short DI is followed by a long APD and vice versa, as indicated in B. B, Snapshot of APD and DI readings during a VF episode explaining the negative APD/DI restitution relationship. Actual values of APD and DI are given for each cycle, APD values at the top and DI values at the bottom. Red crosses indicate the time of activation; vertical blue lines indicate the time of recovery. A long DI (*) is followed by a relatively short APD, whereas a short DI (**) is followed by a longer APD, leading to a negative slope of APD/DI restitution. 0.3, 0.69, 0.28, 0.67, 0.62, 0.93, and 0.27 in patients 1 to 7, respectively). Memory Effects and APD Variability Short-term cardiac memory defined as electric events preceding a certain APD, a concept in which APD variability is determined by both APDs from earlier cycles and also DIs from previous cycles, is shown in Table 2. APD n60/90 correlated significantly to previous APDs, that is, APD n 1 to APD n 4 and to previous DIs. The APD n of all early VF episodes correlated most significantly with APD n 1 (R 2 0.552), diminishing throughout the previous APDs: APD n 2, APD n 3, APD n 4 (R 2 0.339, 0.329, 0.273, respectively; P 0.0001). The relationship for each, APD n 1 to APD n 4, to APD n shows a positive slope, indicating that an increase in APD n 1 to APD n 4 leads to Figure 4. Separate graphs of DI n 1 plotted to APD n90 in each episode of sustained VF. The APD/DI restitution relationship reveals a negative slope in all patients for sustained VF episodes, as seen in Figure 3A.

566 Circ Arrhythmia Electrophysiol October 2009 Table 3. Multiple Linear Regression Analysis of All VF Episodes at APD n90 APD 90 DI n 1 DI n 2 DI n 3 DI n 4 DI n 5 APD n 1 APD n 2 APD n 3 APD n 4 All Step 3 5 2 1 6 4 R 2 0.01 0.01 0.04 0.63 0.003 0.01 P 0.0004 0.0089 0.0001 0.0001 0.0829 0.0097 Individual patients Patient 1 Step 1 3 2 R 2 0.30 0.04 0.31 P 0.0021 0.0981 0.0001 Patient 2 Step 1 R 2 0.29 P 0.0001 Patient 3 R 2 0.46 0.056 P 0.0001 0.0153 Patient 4 Step 2 1 3 R 2 0.038 0.65 0.0156 P 0.0249 0.0001 0.1372 Patient 5 Step 4 1 3 2 5 R 2 0.043 0.40 0.04 0.09 0.038 P 0.0635 0.0001 0.0975 0.0140 0.0721 Patient 6 Step 1 R 2 0.54 P 0.0001 Patient 7 Step 1 R 2 0.17 P 0.00070 All variables left in the model significant at the 0.1500 level. an increase in APD n and vice versa. Table 2 shows that distant events (APD n 4 ) had a lesser contribution to APD variability compared with recent events (APD n 1 ). Multiple Regression Model The variables that entered into the model in a stepwise linear regression, provided they had a significance of P 0.05, are detailed in Table 3. Memory and APD restitution together formed important determinants of APD variability for both APD n90 (R 2 0.70, P 0.001) and APD n60 (R 2 0.62, P 0.001) (also see Table 4). equation, with an R 2 of 0.01 (P 0.0004), and for APD 60 it entered the equation second, with an R 2 of 0.03 (P 0.0001). In individual patients DI n 1 entered the equation in 3 patients for APD n90 and in 4 patients with APD 60. Memory Effects and APD Variability Memory accounted for most of the variance in APD 90, with APD n 1 being the first variable entering the equation with an R 2 of 0.63 (P 0.0001) (Tables 3 and 4). Also for APD 60, the greatest contribution to the APD was from APD n 1, which entered the equation first, with an R 2 of 0.55 (P 0.0001). APD/DI Restitution and APD Variability APD/DI restitution as a determinant of the variance in APD, though significant in the multivariate model, accounted for only 1% of the variability in early human VF (Tables 3 and 4). For APD 90,DI n 1 was the third variable entering the Temporal Effects The change in APD n and DI n 1 with time was studied. As VF progressed, the mean APD n90 decreased from 137.21 29.70 in the first 5 seconds to 126.62 27.19 at 5 to 10 seconds, falling further to 111.25 16.75 at 10 to 15 seconds.

Toal et al Memory and Restitution During Human VF 567 Table 4. Multiple Linear Regression Analysis of All VF Episodes at APD n60 APD 60 DI n 1 DI n 2 DI n 3 DI n 4 DI n 5 APD n 1 APD n 2 APD n 3 APD n 4 All Step 2 5 6 1 4 3 R 2 0.03 0.006 0.003 0.55 0.004 0.01 P 0.0001 0.0274 0.1349 0.0001 0.0820 0.0002 Individual patients Patient 1 R 2 0.59 0.105 P 0.0001 0.0160 Patient 2 Step 2 1 R 2 0.04 0.34 P 0.0569 0.0001 Patient 3 R 2 0.24 0.067 P 0.0004 0.0407 Patient 4 R 2 0.77 0.08 P 0.0001 0.0001 Patient 5 Step 2 1 R 2 0.041 0.37 P 0.1076 0.0001 Patient 6 R 2 0.67 0.04 P 0.0001 0.0779 Patient 7 Step 2 1 R 2 0.040 0.325 P 0.0562 0.0001 All variables left in the model significant at the 0.1500 level. (P 0.001). The DI n 1, on the other hand, increased as VF progressed from 69.68 33.13 in the initial 5 seconds to 78.43 28.32 at 5 to 10 seconds and further to 94.10 16.06 at 10 to 15 seconds (P 0.001). Similar changes were seen with APD 60 and DI 60. When each VF episode was divided into 5-second time segments for each segment, the APD restitution relationship still showed a cluster with a negative slope, but the slope was more negative for the first 5-second segment than the latter (Figure 5). The slope was 0.51 for all APD n90 VF. In the first 5 seconds, the slope was 0.63; from 5 to 10 seconds, it was 0.47, which further flattened to 0.10 at 10 to 15 seconds (Figure 5). These changes were significant by analysis of covariance (P 0.001). Similar results were seen when the RV and LV were analyzed separately (P 0.001). Similar results were seen when the restitution curves were drawn for each patient individually except for patient 7, in whom the slope was positive (0.03) at 0 to 5 seconds. Positive slope was seen in patients 3 and 5 after 5 seconds of VF and in patient 2 after 10 seconds of VF. Discussion We have shown that during human VF, short-term memory has a greater contribution to APD variation than the restitution relationship. These findings have not been demonstrated during human VF previously. Though restitution is thought to play an important part before the onset of VF, our findings suggest that after the onset of VF, short-term memory effect is a greater determinant of APD variation than restitution dynamics. In addition, the slope of the restitution curve was never 1. Taken together, this suggests that a strategy of altering the restitution relationship alone is not adequate to modulate progression of VF after its onset.

568 Circ Arrhythmia Electrophysiol October 2009 Figure 5. Temporal effects on APD/DI restitution relationship analyzed by plotting DI n 1 to corresponding APD n90 for each 5-second segment of sustained VF. The negative slope of APD/DI restitution relationship is seen flattening as time progresses with R 2 of 0.6317 for the first 5 seconds diminishing to R 2 of 0.1003 for VF after 10 seconds. The shortening of APD and prolongation of DI as VF progressed could explain this (see text for discussion). Short-Term Memory During VF In experimental models, it has been shown that the usual effect of short-term memory is to shorten APD progressively as cycle length decreases, especially at rapid heart rates. 24 Although the DIs were short at VF initiation in this study, the APDs during early VF followed long APDs (of the paced beat immediately preceding VF induction), and as a function of memory, 25 the APDs at VF initiation are therefore longer than predicted by the preceding DI. A previously developed memory model 25 considers memory to accumulate and dissipate during depolarized and repolarized states and predicts that an APD after a long APD would be longer than that predicted by preceding DI and vice versa. Our multivariate analysis showed that the greatest contribution to the APD was from APD n 1, entering the equation with an R 2 of 0.63 (P 0.0001) for APD 90 and R 2 of 0.55 (P 0.0001) for APD 60. It is likely that the increase in APD after a short DI (Figure 3) is a result of accumulation of memory due to longer APDs during preceding activations. 16 In a previous porcine study, 5 memory together with restitution showed only 30% contribution to APD variability. That study used a microelectrode on the epicardium of an openchest intubated-ventilated porcine model, and the level for retention of a variable in the model was 0.25. The fact that our study was a closed-chest, unintubated-ventilated, human model using MAP technique on the endocardium during the onset of VF is important to consider, and may partly explain the differences in findings. Another important observation in our study models is that cardiac memory consistently appeared first in the multiple regression model followed by restitution. The fact that our observations are different from porcine model 5 may suggest that the relative contribution of cardiac memory and restitution may be species-specific. Banville et al 26 studied conduction block during early VF in porcine heart in a Langendorff model after 16 beats of VF in the RV, using optical mapping technique. They found that a decrease in cycle length resulted in APD DI pairs above the restitution curve. They had suggested that the acute memory effect acted to damp out proarrhythmic alternans and action potential shortening and hoped that this was true of human hearts. Indeed, data from our study suggest that memory in early human VF leads to APDs that are longer than otherwise expected during shorter DIs. This fact may provide a potential explanation for the organized wave fronts seen during early VF by us and others in human VF on both the epicardium 1,2,27,28 and endocardium, 1,2,28 that is, the memory effect during early VF serves to prevent wave break and thus allows for organization into large coherent wave fronts. Indeed, our findings are further supported by the fact that as the memory effects wane over 15 seconds, there is change in the restitution slope, suggesting that the influence of memory on restitution wanes, resulting in a trend toward a positive slope or flattening of the restitution curve (Figure 5). Restitution During VF The slope of the restitution curve is negative in our in vivo study. Huang et al 5 demonstrated in an animal model that the slope of the restitution curve is negative in early VF, gradually becoming positive, and the value becomes positive only after 30 to 40 seconds of VF. All of our VF episodes were recorded during first 30 to 40 seconds after the onset, hence our results support the observations by Huang et al. Huang et al 29 tested maximum rate of depolarization (V max ) as a predictor of APD variability and compared it with cardiac restitution and memory. In their study, V max was the first or second variable entered into the regression equation in 75% of the VF episodes, whereas DI n 1 and APD n 1 were entered second and third most frequently in 58% and 47% of the VF episodes, respectively. However, when they performed stepwise linear regression without V max, using only previous DIs and APDs to predict APD n, APD n 1 appeared most frequently in the regression equation for 97% of the episodes, followed by DI n 1, which entered for 75% of the episodes, supporting our findings in human VF.

Toal et al Memory and Restitution During Human VF 569 Clinical Implications Our study evaluating the memory and restitution relationship during early human VF has important therapeutic implications. The results indicate that the mechanisms that maintain VF immediately after its onset are more complex than can be explained by the restitution hypothesis alone. Thus, interventions of only flattening the restitution curve may not be enough to terminate VF after its onset. In human hearts in vivo, our study confirms the potential of short-term cardiac memory on fibrillation dynamics, as suspected in experimental models. 30 Unlike conduction velocity restitution, which is mainly governed by the sodium channel and gap junctions, the ionic mechanisms that underlie APD restitution and memory during VF are probably multifactorial. However, if the memory responses are related to mechanisms such as intracellular calcium cycling, further understanding of the specific ionic mechanisms behind memory may provide a target to prevent spiral wave breakup and thus potentially be useful in modulating human VF in patients with myopathic hearts. Limitations MAP recording may get bombarded by activation wave fronts from any nearby tissue. This may produce gaps in activations as well as crowding of activation fronts during VF. This possibility exists if wave fronts are small and fractionated during early VF. However, multiple studies from our laboratory have shown in humans, using optical mapping studies, 1 high-density plaque studies, 3 socks, and balloons, 2 that wave fronts during early VF are coherent and sweep out regions of myocardium as large as 8 cm. The MAP technique has previously been validated to reflect local APD during pacing and in VF. 17,20,31 MAP recordings were considered acceptable only if they strictly satisfied previously described criteria during sinus rhythm 20,21 and VF 21 and the sampling rate as defined previously by Koller et al 10 was used. However, we did not evaluate the role of conduction velocity restitution and excitability on APD variation as it relates to VF. Conclusion We have demonstrated that during human VF, short-term memory is a greater determinant of APD variability than APD/DI restitution. In addition, during early VF, APD restitution slope was never 1. These findings underscore the dynamicity of APD changes and adaptation (short-term memory) and suggest that static restitution analysis anchored to a steady cycle length cannot adequately describe the mechanisms that maintain VF after its onset in humans. Sources of Funding This study was supported by Canadian Institutes of Health Research grant NA 777687. Dr Nanthakumar is a recipient of the Clinician- Scientist Award from the Canadian Institutes of Health Research. None. Disclosures References 1. Masse S, Downar E, Chauhan V, Sevaptsidis E, Nanthakumar K. Ventricular fibrillation in myopathic human hearts: mechanistic insights from in vivo global endocardial and epicardial mapping. Am J Physiol Heart Circ Physiol. 2007;292:H2589 H2597. 2. Nanthakumar K, Jalife J, Masse S, Downar E, Pop M, Asta J, Ross H, Rao V, Mironov S, Sevaptsidis E, Rogers J, Wright G, Dhopeshwarkar R. Optical mapping of Langendorff-perfused human hearts: establishing a model for the study of ventricular fibrillation in humans. Am J Physiol Heart Circ Physiol. 2007;293:H875 H880. 3. Jalife J. Ventricular fibrillation: mechanisms of initiation and maintenance. Annu Rev Physiol. 2000;62:25 50. 4. Weiss JN, Garfinkel A, Karagueuzian HS, Qu Z, Chen PS. Chaos and the transition to ventricular fibrillation: a new approach to antiarrhythmic drug evaluation. Circulation. 1999;99:2819 2826. 5. Huang J, Zhou X, Smith WM, Ideker RE. Restitution properties during ventricular fibrillation in the in situ swine heart. Circulation. 2004;110: 3161 3167. 6. Qu Z, Weiss JN, Garfinkel A. Cardiac electrical restitution properties and stability of reentrant spiral waves: a simulation study. Am J Physiol. 1999;276:H269 H283. 7. Xie F, Qu Z, Yang J, Baher A, Weiss JN, Garfinkel A. A simulation study of the effects of cardiac anatomy in ventricular fibrillation. J Clin Invest. 2004;113:686 693. 8. Zaitsev AV, Guha PK, Sarmast F, Kolli A, Berenfeld O, Pertsov AM, de Groot JR, Coronel R, Jalife J. Wave break formation during ventricular fibrillation in the isolated, regionally ischemic pig heart. Circ Res. 2003; 92:546 553. 9. Garfinkel A, Kim YH, Voroshilovsky O, Qu Z, Kil JR, Lee MH, Karagueuzian HS, Weiss JN, Chen PS. Preventing ventricular fibrillation by flattening cardiac restitution. Proc Natl Acad Sci U S A. 2000;97: 6061 6066. 10. Koller ML, Maier SK, Gelzer AR, Bauer WR, Meesmann M, Gilmour RF Jr. Altered dynamics of action potential restitution and alternans in humans with structural heart disease. Circulation. 2005;112:1542 1548. 11. Selvaraj RJ, Picton P, Nanthakumar K, Chauhan VS. Steeper restitution slopes across right ventricular endocardium in patients with cardiomyopathy at high risk of ventricular arrhythmias. Am J Physiol Heart Circ Physiol. 2007;292:H1262 H1268. 12. Nash MP, Bradley CP, Sutton PM, Clayton RH, Kallis P, Hayward MP, Paterson DJ, Taggart P. Whole heart action potential duration restitution properties in cardiac patients: a combined clinical and modelling study. Exp Physiol. 2006;91:339 354. 13. Riccio ML, Koller ML, Gilmour RF Jr. Electrical restitution and spatiotemporal organization during ventricular fibrillation. Circ Res. 1999;84: 955 963. 14. Ideker RE, Rogers JM, Gray RA. Steepness of the restitution curve: a slippery slope? J Cardiovasc Electrophysiol. 2002;13:1173 1175. 15. Gilmour RF Jr, Otani NF, Watanabe MA. Memory and complex dynamics in cardiac Purkinje fibers. Am J Physiol. 1997;272: H1826 H1832. 16. Wu R, Patwardhan A. Restitution of action potential duration during sequential changes in diastolic intervals shows multimodal behavior. Circ Res. 2004;94:634 641. 17. Franz MR, Burkhoff D, Spurgeon H, Weisfeldt ML, Lakatta EG. In vitro validation of a new cardiac catheter technique for recording monophasic action potentials. Eur Heart J. 1986;7:34 41. 18. Franz MR. Monophasic Action Potentials: Bridging Cell and Bedside. Armonk, NY: Futura Publishing Co; 2000. 19. Kirchhof PF, Larissa Fabritz C, Franz MR. Phase angle convergence of multiple monophasic action potential recordings precedes spontaneous termination of ventricular fibrillation. Basic Res Cardiol. 1998;93: 412 421. 20. Moubarak JB, Karasik PE, Fletcher RD, Franz MR. High dispersion of ventricular repolarization after an implantable defibrillator shock predicts induction of ventricular fibrillation as well as unsuccessful defibrillation. J Am Coll Cardiol. 2000;35:422 427. 21. Swartz JF, Jones JL, Fletcher RD. Characterization of ventricular fibrillation based on monophasic action potential morphology in the human heart. Circulation. 1993;87:1907 1914. 22. Franz MR. Method and theory of monophasic action potential recording. Prog Cardiovasc Dis. 1991;33:347 368. 23. Sager P. Monophasic Action Potential: Bridging Cell and Bedside. Armonk, NY: Futura Publishing Co; 2000.

570 Circ Arrhythmia Electrophysiol October 2009 24. Goldhaber JI, Xie LH, Duong T, Motter C, Khuu K, Weiss JN. Action potential duration restitution and alternans in rabbit ventricular myocytes: the key role of intracellular calcium cycling. Circ Res. 2005; 96:459 466. 25. Vinet A, Chialvo DR, Michaels DC, Jalife J. Nonlinear dynamics of rate-dependent activation in models of single cardiac cells. Circ Res. 1990;67:1510 1524. 26. Banville I, Chattipakorn N, Gray RA. Restitution dynamics during pacing and arrhythmias in isolated pig hearts. J Cardiovasc Electrophysiol. 2004;15:455 463. 27. Nanthakumar K, Walcott GP, Melnick S, Rogers JM, Kay MW, Smith WM, Ideker RE, Holman W. Epicardial organization of human ventricular fibrillation. Heart Rhythm. 2004;1:14 23. 28. Nash MP, Mourad A, Clayton RH, Sutton PM, Bradley CP, Hayward M, Paterson DJ, Taggart P. Evidence for multiple mechanisms in human ventricular fibrillation. Circulation. 2006;114:536 542. 29. Huang J, Cheng KA, Dosdall DJ, Smith WM, Ideker RE. Role of maximum rate of depolarization in predicting action potential duration during ventricular fibrillation. Am J Physiol Heart Circ Physiol. 2007; 293:H2530 H2536. 30. Baher A, Qu Z, Hayatdavoudi A, Lamp ST, Yang MJ, Xie F, Turner S, Garfinkel A, Weiss JN. Short-term cardiac memory and mother rotor fibrillation. Am J Physiol Heart Circ Physiol. 2007;292:H180 H189. 31. Qu Z, Karagueuzian HS, Garfinkel A, Weiss JN. Effects of Na( ) channel and cell coupling abnormalities on vulnerability to reentry: a simulation study. Am J Physiol Heart Circ Physiol. 2004;286:H1310 H1321. CLINICAL PERSPECTIVE During ventricular fibrillation, substantial variation in activation times and action potential duration contribute to conduction block and wave break that are thought to be important for the generation of new excitation wave fronts. Manipulating the relation between the preceding diastolic interval and action potential duration (restitution relationship) is a potential drug target for preventing or terminating ventricular fibrillation. In humans, especially during early ventricular fibrillation, in which interventions may have significant impact on outcomes, determinants of action potential duration have not been evaluated. Using monophasic action potential recordings in humans, we showed that although the preceding diastolic interval is a determinant of action potential duration (restitution relationship), the history of preceding action potential durations (short-term memory) appeared to have a greater influence on action potential variability. These findings suggest that electrophysiological interventions that flatten the restitution slope may have limited effects on ventricular fibrillation soon after its onset. The mechanisms that maintain ventricular fibrillation are complex and do not appear to be adequately described by the restitution hypothesis alone. Understanding the ionic mechanisms that are responsible for memory effects may provide a target for preventing wave breakup to prevent or facilitate termination of ventricular fibrillation.