Differences in effects of electrical therapy type for ventricular arrhythmias on mortality in implantable cardioverter-defibrillator patients

Transcription

1 Differences in effects of electrical therapy type for ventricular arrhythmias on mortality in implantable cardioverter-defibrillator patients Michael O. Sweeney, MD,* Lou Sherfesee, PhD, Paul J. DeGroot, MS, Mark S. Wathen, MD, Bruce L. Wilkoff, MD, FHRS From *Brigham and Women s Hospital, Boston, Massachusetts, Medtronic, Inc., Minneapolis, Minnesota, Vanderbilt University Medical Center, Nashville, Tennessee, and the Cleveland Clinic, Cleveland, Ohio. BACKGROUND Implantable cardioverter-defibrillator (ICD) shocks have been associated with an increased risk of death. It is unknown whether this is due to the ventricular arrhythmia (VA) or shocks and whether antitachycardia pacing (ATP) termination can reduce this risk. OBJECTIVE The purpose of this study was to determine whether mortality in ICD patients is influenced by the type of therapy (shocks of ATP) delivered. METHODS Cox models evaluated effects of baseline characteristics, ventricular tachycardia (VT; 188 bpm), fast VT (FVT; bpm), ventricular fibrillation (VF; 250 bpm), and therapy type (shocks or ATP) on mortality among 2135 patients in four trials of ATP to reduce shocks. RESULTS Over months, 24.3% patients received appropriate shocks (50.6%) or ATP only (49.4%), and 6.6% died. Mortality predictors were age (hazard ratio 1.07, 95% confidence interval , P.0001), New York Heart Association class III/IV (3.50 [ ]; P.0001), coronary disease (3.08 [ ]; P.01), and cumulative VA (VT FVT VF) episodes shocked (1.20 [1.13, 1.29]; P.0001). Beta-blockers (0.65, ; P.0001) and remote myocardial infarction (0.53, [ ] P.0004) predicted reduced risk. Since 92% of VT and all VF received a single therapy type (ATP and shocks, respectively), the effect of therapy on episode risk could not be established. For FVT (32% shocked, 68% ATP), episode and therapy effects could be uncoupled; ATP-terminated FVT did not increase episode mortality risk, whereas shocked FVT increased risk by 32%. Survival rates were highest among patients with no VA (93.8%) of ATP-only (94.7%) and lowest for shocked patients (88.4%). Monthly episode rates were 80% higher among shocked versus ATP-only patients. CONCLUSIONS Shocked VA episodes are associated with increased mortality risk. Shocked patients have substantially higher VA episode burden and poorer survival compared with ATP-onlytreated patients. KEYWORDS Implantable cardioverter defibrillators; Ventricular tachycardia; Ventricular fibrillation; Shocks; Antitachycardia pacing; Death ABBREVIATIONS ATP antitachycardia pacing; CAD coronary disease; CI confidence interval; CL cycle length; EF ejection fraction; EGM electrogram; HF heart failure; ICD implantable cardioverter-defibrillator; FVT fast ventricular tachycardia; MI myocardial infarction; NYHA New York Heart Association; SVT supraventricular tachycardia; VA ventricular arrhythmia; VF ventricular fibrillation; VT ventricular tachycardia (Heart Rhythm 2010;7: ) 2010 Heart Rhythm Society. All rights reserved. Introduction Implantable cardioverter-defibrillators (ICDs) reduce mortality by terminating ventricular arrhythmias (VAs) with shocks. Recent evidence indicates an association among appropriate shocks, inappropriate shocks, and increased risks of heart failure (HF) and death. 1 4 While VA episodes and shocks may be markers for higher risk patients, the possibility that shock-induced electrical trauma increases episode mortality risk has not been excluded. Since untreated VA is potentially lethal and shocks are lifesaving, Address reprint requests and correspondence: Michael O. Sweeney, M.D., Cardiac Arrhythmia Service, Brigham and Women s Hospital, Boston, Massachusetts address: mosweeney@partners.org. (Received August 3, 2009; accepted November 25, 2009.) uncoupling these interacting effects is difficult, particularly since almost all episodes in these studies were shocked. Comparing the effects of antitachycardia pacing (ATP) versus shocks could provide the opportunity to uncouple a harmful shock effect from the VA. We hypothesized that mortality in ICD patients is influenced by the type of therapy (shocks or ATP) delivered. Methods Predictors of mortality were analyzed in 2135 patients enrolled in four trials incorporating ATP to reduce shocks (PainFREE Rx, 5 PainFREE Rx II, 6 EMPIRIC, 7 and PREPARE 8 ). All patients had Medtronic (Minneapolis, MN) ICDs implanted less than 4 weeks before enrollment. Device programming was standardized in each study with some differences between /$ -see front matter 2010 Heart Rhythm Society. All rights reserved. doi: /j.hrthm

2 354 Heart Rhythm, Vol 7, No 3, March 2010 studies. Detection in the ventricular fibrillation (VF) zone required that 12/16 (PainFREE), 18/24 (PainFREE II, EM- PIRIC), or 30/40 (PREPARE) R-R intervals had a cycle length (CL) ms. If the last eight R-R intervals had CL 240 and ms, the episode was detected as fast VT (FVT). A slow ventricular tachycardia (VT) zone with CL ms and ms was required in PainFREE II and EMPIRIC, optional in PainFREE, and monitored in PREPARE. ATP before shocks was prescribed for VT and FVT episodes, except for in PainFREE II, where shocks versus ATP for FVT were randomized. VF episodes were prescribed shocks. Pre-onset storage of the ventricular electrogram (EGM) was required. Electrophysiologists used EGMs to adjudicate all detected episodes. Ventricular versus supraventricular origin was established by majority rule. Episodes without EGMs were excluded. Statistical methods Patient characteristics were summarized with standard descriptive statistics. Baseline characteristics associated with mortality risk were established using (1) a shared frailty model 9 (to account for possible differences in patient selection and treatment between studies) and a (2) stepwise Cox proportional hazards model. 10 Characteristics associated with mortality risk in both models were included in all subsequent models that included episode and therapy type covariates. Episodes were partitioned three ways: induced VA, spontaneous VA, and inappropriately shocked supraventricular tachycardia (SVT). Spontaneous VA episodes were subclassified as VT, FVT, or VF using median CL. Time-dependent covariates for each episode type and episode therapy type (no therapy/spontaneous termination, ATP only, shocked) were generated. Three mortality risk models were then created that added episode and/or therapy type covariates to the baseline predictors. The first predictive model (risk of death by clinical characteristics, arrhythmia episode type, and electrical therapy type) included all episode types with electrical therapy type time-dependent covariates (i.e., cumulative shocked VA episodes). The second model (risk of death by VA type) included VA episode covariates defined by CL (i.e., cumulative FVT episodes) to assess the effects of different VA types on mortality risk. The third model (risk of death by VA type electrical therapy type) included VA episode covariates segregated by therapy type (i.e., cumulative FVT receiving only ATP). Mortality rates were compared among three predefined episode/therapy type survival groups (no VA, ATP-onlytreated VA, and at least one shocked VA episode). VA occurrence rates, episode/therapy types, and durations were compared between ATP-only and shocked groups. The annualized rate for each VA episode type per patient month was calculated by dividing the total number of episodes by type by months of follow-up for each patient group. Algebraically, this is equivalent to taking a weighted average of the monthly episode rate per patient, where the weight assigned to each patient is the proportion of overall follow-up that the patient contributed to the aggregate followup. Episode rates for patients with longer follow-up therefore received more weight. Results Baseline characteristics of the study population (Table 1) Most patients were males with coronary disease (CAD), myocardial infarction (MI), moderately reduced ejection fraction (EF), and mild-moderate HF symptoms, representative of a typical ICD patient population. The reason for ICD was primary prevention in 67% of patients. Average follow-up was slightly 1 year. Table 1 Baseline characteristics by study Baseline characteristic PainFREE I, n 220 PainFREE II, n 634 EMPIRIC, n 900 PREPARE, n 381 Total, N 2135 Male 172 (78.2) 498 (78.6) 731 (81.2) 306 (80.3) 1707 (80) Age EF, % CAD 220 (100) 540 (85.2) 793 (88.1) 306 (80.3) 1859 (87.1) MI 176 (80) 405 (63.9) 625 (69.4) 263 (69.0) 1469 (68.8) Recent ( 6 months prior) 42 (19.1) 100 (15.8) 168 (18.7) 51 (13.4) 361 (16.9) 6 months prior 141 (64.1) 334 (52.7) 492 (54.7) 218 (57.2) 1185 (55.5) Ischemic cardiomyopathy 141 (64.1) 298 (47) 474 (52.7) 274 (71.9) 1187 (55.6) Beta-blocker usage 86 (39.1) 376 (59.3) 659 (73.2) 330 (86.6) 1451 (68.0) NYHA HF class : No HF/class I 58 (26.4) 295 (46.5) 454 (50.4) 97 (25.5) 904 (42.3) Class II 105 (47.7) 211 (33.3) 315 (35) 211 (55.4) 842 (39.4) Class III 49 (22.3) 114 (18.0) 120 (13.3) 67 (17.6) 350 (16.4) Class IV 6 (2.7) 10 (1.6) 11 (1.2) 6 (1.6) 33 (1.6) Unknown 2 (0.9) 4 (0.6) 0 (0) 0 (0) 6 (0.3) Prior atrial fibrillation 47 (21.4) 158 (24.9) 210 (23.3) 106 (27.8) 521 (24.4) Sustained VT/VF 120 (54.6) 225 (35.49) 345 (38.3) 9 (2.4) 699 (32.7) Prior CABG 122 (55.5) 300 (47.3) 398 (44.2) 167 (43.8) 987 (46.2) Note: Data are n (%) unless otherwise specified.

3 Sweeney et al Effects of Shocks versus ATP on Mortality % 6.2% Shocks Only 2000 ATP & Shocks Number of Episodes % 15.9% 13.1% 62.1% ATP Only No Therapies 0 VT (N=2291) 5.0% FVT (N=1470) 8.9% VF (N=173) 65.9% 33.5% % All Shocks (N=718) 24.7% 59.3% 16.0% Figure 1 Distribution of therapy types applied to VT, FVT, and VF. Spontaneous arrhythmia episodes A total of 5376 spontaneous episodes were analyzed. Adjudicated VA accounted for 3934 (73.2%); 1442 (26.8%) were inappropriately detected SVT. Device detection classification and response to therapy Most ventricular episodes were potentially ATP-terminable VT (2291/3934, 58.2%) or FVT (1470/3934, 37.4%; Figure 1). VF accounted for 173/3934 (4.4%). Overall, 7.7% (304/ 3934) episodes terminated spontaneously after detection but before therapy. Of the remaining 3630 episodes that were treated, 2912 (80.2%) were terminated by ATP. Of 2176 treated VT episodes, 177 (8.1%) received primary shocks (36/2176, 1.7%) or received shocks after failed ATP (141/ 2176, 6.5%). Therefore, ATP terminated 92.4% (1999/ 2140) of VT episodes attempted. Similarly, of 1339 FVT episodes, 426 (31.8%) received primary shocks (233/1339, 17.4%) or received shocks after failed ATP (193/1339, 14.4%). Therefore, ATP terminated 82.5% (913/1106) of all FVT episodes attempted. Models of mortality risk (Table 2) Risk of death by clinical characteristics, arrhythmia episode type, and electrical therapy type (model 1) There were 138 deaths (6.5%). Variables predictive of death were older age, worse New York Heart Association (NYHA) class, and CAD, whereas remote MI (as compared with MI 6 months prior) and -blocker use were associated with reduced risk. Adjudicated VA episodes (VT FVT VF) that received at least one shock were an independent predictor of death. The increased relative risk of death for each shocked episode was 20%. However, VT and FVT episodes treated with ATP only were not associated with increased risk. There was no evidence of an association between inappropriate shocks and increased risk in the mortality model. Of the 183 patients with one or more inappropriate shocks, five (2.7%) died. Conversely, 106/1761 (6%) patients who had neither appropriate nor inappropriate shocks died. Similarly, shocked-induced VA episodes and nonsustained VA episodes were not associated with increased risk. Risk of death by VA type (model 2) After adjusting for baseline predictors, spontaneous VT, FVT, or VF episodes, unqualified for electrical therapy type, were associated with increased mortality risk. Each episode of VT increased risk by 4%, FVT by 2%, and VF by 15%. Risk of death by VA type electrical therapy type (model 3) The purpose of this model was to uncouple the effect of ATP alone versus shocks from the effect of the ventricular episode after adjusting for baseline predictors. Each episode of VT treated with ATP was associated with an approximately 3% increased risk of death, whereas shocked FVT increased risk by 31% and shocked VF by 16%. In contrast, ATP-treated FVT did not increase episode mortality risk. Survival by VA episode therapy type (Figure 2) Observed survival rates among patients with VT or FVT that terminated spontaneously or with ATP and no shocks versus patients with no spontaneous VA were not different (94.7% vs. 93.8% at 12 months, respectively), whereas survival for patients who received one or more appropriate shocks was worse (88.4% at 12 months). The three episode/ therapy type survival groups (no VA, ATP only, and one or more shocked episodes) were similar with respect to baseline predictors, consistent with modeling (age , , ; NYHA III IV 18.2%, 17.9%,

4 356 Heart Rhythm, Vol 7, No 3, March 2010 Table 2 Mortality risk models Covariate Model 1 Model 2 Model 3 HR (95% CI) P HR (95% CI) P HR (95% CI) P Clinical characteristics, all models: Age 1.07 (1.04, 1.08) (1.04, 1.08) (1.05, 1.09).0001 NYHA class II, relative to class I 1.60 (1.04, 2.48) (1.07, 2.52) (1.14, 2.71).011 CAD 3.08 (1.31, 7.25) (1.20, 6.41) (1.31, 7.34).016 NYHA class III, relative to class I 3.50 (2.27, 5.41) (2.33, 5.59) (2.35, 5.63).0001 Remote MI 0.53 (0.38, 0.76) (0.40, 0.80) (0.37, 0.75) Beta-blocker use 0.65 (0.46, 0.92) (0.47, 0.94) (0.48, 0.95) Therapy, model 1: VT FVT VF episodes shocked 1.20 (1.13, 1.29).0001 Excluded Excluded Nonsustained VT FVT VF NA a.36 Excluded Excluded VT FVT episodes treated with NA a.70 Excluded Excluded ATP only Inappropriately shocked SVT NA a.17 NA a.18 NA a.15 episodes b Induced VT FVT VF shocked NA a.97 Excluded Excluded episodes Episode type, model 2 only: VT episodes Excluded 1.04 (1.02, 1.06).0001 Excluded FVT episodes Excluded 1.02 (1.01, 1.03).007 Excluded VF episodes Excluded 1.15 (1.05, 1.26).002 Excluded Episode and electrical therapy type, model 3 only: VT episodes treated with ATP only Excluded Excluded 1.03 (1.01, 1.05).002 FVT episodes treated with shocks Excluded Excluded 1.32 (1.23, 1.41).0001 VF episodes treated with shocks Excluded Excluded 1.16 (1.05, 1.29).006 Nonsustained VT episodes Excluded Excluded NA a.15 VT episodes treated with shocks Excluded Excluded NA a.15 Nonsustained FVT episodes Excluded Excluded NA a.15 FVT episodes treated with ATP only Excluded Excluded NA a.15 Nonsustained VF episodes Excluded Excluded NA a.15 Note: Model 1: risk of death by clinical characteristics, arrhythmia episode type, and electrical therapy type. Model 2: risk of death by VA type. Model 3: risk of death by VA type electrical therapy type. CI: confidence interval; NA: not applicable. a Final model results refit without nonsignificant predictors. b Inappropriate shocks were tested in all three models. 16.2%; CAD 87.7%, 85.8%, 83.9%; remote MI 54.4%, 57.7%, 61.6%; -blocker use 70.4%, 60.1%, 61.6%, respectively). A formal statistical comparison of survival between groups was not done owing to the retrospective nature of the grouping assignment, which cannot account for differences in VA episode therapy burden over time and because these states are time dependent (i.e., an individual patient begins with no VA, receives ATP for FVT, but later receives shocks for VF). That the survival curve for patients who experienced one or more shocks was below that of patients with no shocks illustrates the results of iterative risk models, which indicate that appropriate shocks are associated with increased risk. Progression of VA burden by shock burden and episode/therapy type survival groups Patients who received more appropriate shocks had higher occurrence rates of all VA types (Figure 3). VT episode rates were similar across the one-shock, two-shock, and three-shock groups, whereas FVT, and to a lesser extent VF, episode rates increased slightly across these three groups. In comparison, VT episode rates were substantially higher in the four-shock and five-shock group, but since most VT was ATP terminated, the higher shocked episode count was due to more FVT and VF episodes. By episode/therapy type survival group, patients who had one or more shocks, as compared with ATP-only-treated patients, had 1.7 times more VA-treated episodes per month (1.0 vs. 0.6). Shocked episodes accounted for 34% of the higher monthly treated episode burden, and 60% of all shocks were for FVT. Consequently, patients who had one or more shocks spent 5.2 times more time in treated VA per month (28.8 vs. 5.6 seconds), and 47% of this time was due to shocked episodes. VA episode durations were also longer in some shocked patients versus in those who received ATP only. The median duration of ATP-terminated VT and FVT episodes in both groups was similarly 10 seconds. However, 25% of ATP-terminated VT and FVT episodes in the shocked group were 18 and 13 seconds, respectively, and mean durations were correspondingly longer in the shocked versus ATP-

5 Sweeney et al Effects of Shocks versus ATP on Mortality Survival No VT/VF VT/VF, Shocked VT/VF, ATP, No Shocks Number at Risk No VT/VF (N=1671) VT/VF, Shocked (N=211) VT/VF, ATP, No Shocks (N=253) Months Figure 2 Survival rates by rhythm and therapy type. Survival among patients treated only with ATP was identical to that in patients with no VT, whereas survival among patients who received shocks was significantly worse. The three patient groups were similar with respect to baseline predictors of mortality, which is consistent with the modeling results only group (25.3 vs seconds for VT, 13.7 vs. 8.4 seconds for FVT). This difference is due to more than one ATP sequence before termination among some shocked patients. The longest durations were for ATP shocks episodes, where 25% of VT episodes exceeded 1 minute, and mean VT and FVT durations were 71.2 and 45.5 seconds, respectively. VA episode/therapy type rates by alive or dead All four studies showed higher observed annualized rates of appropriately shocked episodes (over 4 times higher in all four studies) and specifically of shocked VF episodes (at least 3.75 times higher in all four studies) among patients who died. Overall, more patients who died (27/ 138, 19.6%) had shocks versus survivors (184/1997, 9.2%), and the most prevalent shocked episode type was FVT. Shocks for VF or VT alone accounted for 41% or fewer of all shocks, alive or dead. Only 28% (59/211) of shocked patients had no FVT shocks. VA rates per patient follow-up month segregated by episode and therapy type are compared by survival status in Figure 4, which provides visual support for the mortality model results. Patients who died had 5 6 times more VA (VT FVT VF) than survivors. Patients who died had 6.2 times more ATP-only-treated VT than survivors, while the differential for FVT treated with ATP was only 2 times greater. Rather, for FVT, shocked episodes differed, where patients who died had a rate of FVT with shocks 12 times greater than survivors. The duration per follow-up month for all VA episode types (VT FVT VF) was 7.4 times greater among patients who died. Correspondingly, total durations for episodes treated with ATP only, ATP shocks, and shocks 3 3 Annualized Rate of Episodes per Patient Month Annualized Rate of Episodes per Patient Month No VT/VF (N=1671) VT/VF, ATP, No 1 Shock (100) 2 Shocks Shocks (N=39) (N=253) 3 Shocks (N=18) 4 Shocks (N=13) 5 or More Shocks (N=41) No VT/VF (N=1671) VT/VF, ATP, No Shocks (N=253) 1 Shock (100) 2 Shocks (N=39) 3 Shocks (N=18) 4 Shocks (N=13) 5 or More Shocks (N=41) Number of Appropriately Shocked Episodes Number of Appropriately Shocked Episodes True VT True FVT True VF VT/FVT/VF Untreated VT/FVT w/ ATP Only VT/FVT/VF w/ Shocks Figure 3 Progression of arrhythmia burden by appropriate shock groups. Patients are partitioned by number of appropriate shocked episodes over follow-up (horizontal axis) and corresponding annualized rate of VA by episode type and therapy applied (vertical axis). Left panel: Increasing annualized rates of all VA types with increasing shock burden. Right panel: Increasing annualized rates of all VA and therapy types with increasing shock burden.

6 358 Heart Rhythm, Vol 7, No 3, March 2010 Episodes per Patient Month Shocks Only ATP & Shocks 0.7 ATP Only Nonsustained Episodes VT: Survivors VT: Deaths FVT: Survivors FVT: Deaths VF: Survivors VF: Deaths Figure 4 Ventricular episodes and therapy types by survival status. VA episodes per patient month, segregated by type and therapy applied, are displayed among subjects who died compared with survivors. The monthly burden of all VA episode therapy type combinations is higher among patients who died. only were 5.5, 22.4, and 2.8 times greater among patients who died. Therefore, in addition to receiving more shocks, patients who died also spent more time in shocked VT FVT as compared with survivors. Discussion This experiment introduces the possibility that electrical therapy type may influence mortality risk in some ICD patients. The main findings are that (1) patients with VA episodes and shocks have higher mortality ( 20% increased risk per shocked episode) than patients with neither or patients with VA treated only with ATP; (2) patients with more VA episodes and more shocks have higher mortality than patients with less of both; (3) VA occurrence rates, durations, and electrical therapy burden of both types were highest among patients who were shocked and died; and (4) inappropriate shocked episodes were not associated with increased mortality risk. Three scenarios could explain these observations: (1) VA episodes do not increase mortality risk; electrical trauma from shocks, but not ATP, increases risk; (2) VA episodes increase mortality risk irrespective of terminating therapy type; neither shocks nor ATP add to this risk; (3) VA episodes shocks, but not ATP, increase mortality risk. Interactions between adverse shock effects are possible for scenarios (1) and (3) such that patients with more VA episodes may be more susceptible to harm from shocks. Determining the true pattern requires uncoupling the trigger (VA) from the therapy type (shock vs. ATP). We therefore exploited the theme of the four component studies that the vast majority of VA is VT or FVT, whereas de novo VF is uncommon in typical ICD patients. Since VT and FVT were targeted for ATP termination, a modeling opportunity for uncoupling therapy type from episode type was recognized. Adjusted for other mortality predictors, any occurrence of VT, FVT, or VF, unqualified for therapy type, increased the risk of death by 2% 15% and was greatest for VF episodes. When electrical therapy type was included, ATP-terminated VT and shocked VF remained significant predictors of death; however, the increased risk in either case was indistinguishable from the risk unqualified for therapy ( 4% for VT vs. 3% for VT ATP; 15% for VF vs. 16% for VF shocks). Because most VT and VF were terminated with a single therapy type (VT 91.9% ATP, VF 100% shocks), uncoupling the mortality effect of therapy type from episode type was impossible. Therefore, we cannot conclude using this data set and analytic approach that shocks for VT are harmful and that ATP is harmless or that shocks for VF increase episode risk. However, since 1/3 of FVT episodes were shocked and 2/3 ATP terminated, episode and therapy type mortality effects could be uncoupled. FVT treated with ATP only was not associated with increased risk of death, whereas similar FVT episodes that were shocked increased the risk of death by 32%. Since the majority of shocks ( 60%) overall were for FVT, and since 72% of shocked patients had at least one shock for FVT, shocked FVT was the most prevalent type of shocked episode and the dominant shock effect in the mortality models. Shocked patients had a substantially more active VA substrate, despite having indistinguishable baseline clinical profiles, and this difference was greatest for patients who were shocked and died. Shocked patients had greater VA burden (80% more episodes per month), and 1/3 of these were shocked. Most shocks were for FVT and occurred at a 12 times higher rate among patients who died, whereas shocks for VF occurred at a 8 times higher rate. The rate of shocked episodes preceded by failed ATP was 18 times higher in patients who died, whereas the differential for shock-only episodes was 6 times higher and was accounted for by the difference in FVT and VF rates between groups. Time spent in VA was 7 times higher per month among patients who died, and episode durations were higher for all episode/therapy combinations and greatest for

7 Sweeney et al Effects of Shocks versus ATP on Mortality 359 shocked episodes preceded by failed ATP ( 22-fold increase). Hence, in addition to receiving more shocks, patients who died had longer duration shocked episodes (including failed ATP) and spent more time in shocked episodes compared with survivors. It is possible that longer episode durations after failed ATP magnify the adverse effect of shocks. 11 It is also possible that the effect of episode duration may be enhanced by shocks in a way not observed for ATP termination. The possibility that VT and FVT were mechanistically less responsive to ATP in the shocked group cannot be directly addressed since by a priori definition the comparison group had 100% ATP success. It is possible that some VT and FVT episodes were less responsive to ATP at the substrate level in the shocked group, and perhaps ATPunresponsive VT and FVT are markers for a higher mortality risk patient. Additionally, since ATP response is a probabilistic phenomenon and VA episode burden was much higher in the shocked group, there was a greater chance for ATP failure. The findings that shocked VA episodes are associated with increased mortality risk and that this risk is lower for ATP-terminated episodes are not surprising. Lifesaving shocks contain an inherent paradox. Shock strengths sufficient to terminate VF cause myocardial damage, whereas weaker shocks cause less damage but do not terminate VF. 12 Shock-related myocardial injury has been extensively investigated. Large shock field strengths destroy cardiac myocytes causing biomarker release, which increases with shock strength and proximity to recent MI The severity of postresuscitation myocardial depression increases with shock strength, and survival duration is inversely related. 16 Repetitive shocks may cause cardiovascular collapse and death due to electromechanical dissociation. In contrast, there is no evidence that ATP has adverse cardiac effects. ATP termination of VT or FVT, unlike shocks, does not cause biomarker release 15 or reduce ventricular pump function. 17 Opportunities similar to ours to distinguish between effects of ATP versus shocks are nearly nonexistent since most ICD trials used shocks exclusively. Therefore, uncoupling shock effect from episode type is impossible. 4 However, ATP, unlike shocks, was not associated with increased risk of death in MADIT II, although it was used sparingly. 3 The idea that shocks are associated with risk of death and HF in ICD patients is not new. A commonly held interpretation is that VA is a marker for clinical deterioration, shocks are harmless, and the increased risk reflects an unnatural progression of disease granted by life-prolonging shocks. 4,18,19 An alternate and largely neglected interpretation is that shocks may causally increase risks of HF and death. In MADIT II, the risk of first and recurrent HF hospitalization increased by 90% and 74%, respectively, after appropriate shocks. 2 Survival after the first appropriate shock was 80% at 1 year, which was significantly less than the survival before the first shock, and the nonsudden cardiac death rate increased 17%. 19 Multivariable analysis demonstrated that appropriate shocks were associated with a more than threefold increased risk of death. 3 One appropriate shock in SCD-HeFT increased risk of death 5 times, was higher with additional shocks, and was 3 times greater in ischemic versus nonischemic patients. 4 In DINAMIT and IRIS, ICDs implanted immediately after MI failed to improve survival because the reduction in arrhythmic deaths with shocks was offset by parallel increases in nonarrhythmic deaths. 1,20 Subgroup analysis in DINAMIT demonstrated that total mortality and nonarrhythmic death rates were similar between control and ICD patients who were not shocked, whereas both were higher among shocked patients. 21 Similar to MADIT II and SCD-HeFT, mortality rates were substantially higher after shocks (10% within days after the first shock, 25% within 1 year, and 40% by 2 years). That a mortality difference was not observed between treatment arms (ATP vs. shocks for FVT) in PainFREE Rx II could be because only 25% of both groups had VA, and the percentage of patients with appropriate shocks was comparable between treatment arms (12.4% vs. 14.2%, respectively). Inappropriate shocks were also associated with a 2 times increased risk of death in MADIT II and SCD- HeFT. 3,4 Our analysis did not reveal increased mortality risk with shocked atrial episodes, unlike shocked ventricular episodes, despite similar differential in burden of inappropriate shocks between patients who died or survived. It seems reasonable to hypothesize that shocked episode type may precondition the myocardium to the adverse effects of shocks and that factors unique to spontaneous VA magnify these effects, particularly in ATP-unresponsive ventricular episodes. Clinical implications Investigations of shock-related myocardial injury have focused on acute effects that may be insufficient to account for reduced survival after appropriate shocks. Other mechanisms may be important. Shocks may activate signaling pathways in the molecular cascade of HF. The clinical consequence may manifest months after shocks. Nonetheless, despite any suggestion that shocks increase risk of death and HF, ICDs prolong survival. Near total reliance on shocks may have underestimated the ICD survival benefit in clinical trials. Since VF can only be terminated with shocks, uncoupling therapy from episode risk could be indirectly addressed with graded shock energies. Strategies to (1) minimize shocks with ATP 6,8 and shorten shocked episode durations without sacrificing ATP, (2) decrease shock energies, and (3) reduce VA burden (drugs, substrate modification) may further improve survival in ICD patients. Limitations This was a retrospective analysis with relatively short follow-up. Differences in detection and therapy programming between studies could have influenced the results. Postmortem ICD interrogations and mode of death were not recorded. Episodes without EGMs were excluded. Eventually,

8 360 Heart Rhythm, Vol 7, No 3, March 2010 all sustained VA receives therapy, and separating therapy effect from episode type becomes impossible. Neither the association between shocks and death in prior studies nor the association between shock-free termination and improved survival in this study prove that shocks are causal. Conclusions Shocked VA episodes are associated with increased mortality risk. Shocked patients have a substantially higher VA episode burden and poorer survival compared with ATPonly-treated patients. References 1. Hohnloser SH, Kuck KH, Dorian P, et al. Prophylactic use of an implantable cardioverter-defibrillator after acute myocardial infarction. N Engl J Med 2004; 351: Goldenberg I, Moss AJ, Hall WJ, et al. Causes and consequences of heart failure after prophylactic implantation of a defibrillator in the Multicenter Automatic Defibrillator Implantation Trial II. Circulation 2006;113: Daubert JP, Zareba W, Cannom DS, et al. Inappropriate implantable cardioverter-defibrillator shocks in MADIT II: frequency, mechanisms, predictors, and survival impact. J Am Coll Cardiol 2008;51: Poole JE, Johnson GW, Hellkamp AS, et al. Prognostic importance of defibrillator shocks in patients with heart failure. N Engl J Med 2008;359: Wathen MS, Sweeney MO, DeGroot P, et al. Shock reduction using antitachycardia pacing for rapid spontaneous ventricular tachycardia in patients with coronary artery disease. Circulation 2001;104: Wathen MS, DeGroot PJ, Sweeney MO, et al. Prospective randomized multicenter trial of empirical antitachycardia pacing versus shocks for spontaneous rapid ventricular tachycardia in patients with implantable cardioverter defibrillators. PainFREE Rx II Trial Results. Circulation 2004;110: Wilkoff BL, Ousdigian KT, Sterns LD, et al. A comparison of empiric to physician-tailored programming of implantable cardioverter-defibrillators: results from the prospective randomized multicenter EMPIRIC trial. J Am Coll Cardiol 2006;48: Wilkoff BL, Williamson BD, Stern RS, et al. Strategic programming of detection and therapy parameters in implantable cardioverter-defibrillators reduces shocks in primary prevention patients: results from the PREPARE (Primary Prevention Parameters Evaluation) Study. J Am Coll Cardiol 2008;52: Therneau TM, Grambsch PM. Modeling Survival Data: Extending the Cox Model. New York: Springer-Verlag; Cox DR. Regression models and life-tables. J R Stat Soc [B] 1972;34: Tang W, Weil MH, Sun S, et al. The effects of biphasic and conventional monophasic defibrillation on postresuscitation myocardial function. J Am Coll Cardiol 1999;34: Walcott GP, Killingsworth CR, Ideker RE. Do clinically relevant transthoracic defibrillation energies cause myocardial damage and dysfunction? Resuscitation 2003;59: Nikolski VP, Efimov IR. Electroporation of the heart. Europace 2005;7:S Hurst TM, Hinrichs M, Breidenbach C, Katz N, Waldecker B. Detection of myocardial injury during transvenous implantation of automatic cardioverterdefibrillators. J Am Coll Cardiol 1999;34: Runsiö M, Kallner A, Källner G, Rosenqvist M, Bergfeldt L. Myocardial injury after electrical therapy for cardiac arrhythmias assessed by troponin-t release. Am J Cardiol 1997;79: Xie J, Weil MH, Sun S, et al. High-energy defibrillation increases the severity of postresuscitation myocardial dysfunction. Circulation 1997;96: Stoddard MF, Labovitz AJ, Stevens LL, Buckingham TA, Redd RR, Kennedy HL. Effects of electrophysiologic studies resulting in electrical countershock or burst pacing on left ventricular systolic and diastolic function. Am Heart J 1988;116: Pacifico A, Ferlic LL, Cedillo-Salazar FR, Nasir N, Doyle TK, Henry PD. Shocks as predictors of survival in patients with implantable cardioverterdefibrillators. J Am Coll Cardiol 1999;34: Moss AJ, Greenberg H, Case RB, et al. Long-term clinical course of patients after termination of ventricular tachyarrhythmia by an implanted defibrillator. Circulation 2004;110: Steinbeck G, Andresen D, Seidl K, et al. Defibrillator implantation early after myocardial infarction. N Engl J Med 2009;351: Dorian P, Connolly S, Hohnloser SH, et al. Why don t ICDs decrease all-cause mortality after MI? Insights from the DINAMIT Study. Circulation 2004;110: III 502 (abstract 2356).