The implanted cardioverter-defibrillator (ICD) improves. Heart Failure

Transcription

1 Heart Failure Causes and Consequences of Heart Failure After Prophylactic Implantation of a Defibrillator in the Multicenter Automatic Defibrillator Implantation Trial II Ilan Goldenberg, MD; Arthur J. Moss, MD; W. Jackson Hall, PhD; Scott McNitt, MS; Wojciech Zareba, MD, PhD; Mark L. Andrews, BBS; David S. Cannom, MD; for the Multicenter Automatic Defibrillator Implantation Trial (MADIT) II Investigators Background Implantable cardioverter-defibrillator (ICD) therapy may be associated with an increased risk for heart failure (HF). The present study evaluated the frequency, causes, and consequences of HF after ICD implantation. Methods and Results We performed a retrospective analysis of the clinical factors and outcomes associated with postenrollment HF events in 1218 patients enrolled in the Multicenter Automatic Defibrillator Implantation Trial II. The adjusted hazard ratios (HRs) of ICD:conventional therapy for first and recurrent HF events were 1.39 (P 0.02) and 1.58 (P 0.001), respectively. The risk was increased among patients who received single-chamber or dual-chamber ICDs. Development of HF was associated with an increased mortality risk (HR, 3.80; P 0.001). Among patients who received a single-chamber ICD, there was a similar survival benefit before and after the development of HF (HR, 0.59 and 0.61, respectively; P 0.92 for difference), whereas among patients with dual-chamber devices, there was a significant reduction in survival benefit after HF (HR, 0.26 and 0.83, respectively; P 0.01 for difference). Within the defibrillator arm of the trial, patients who received life-prolonging therapy from the ICD had an increased risk for first and recurrent HF events (HR, 1.90; P 0.01 and 1.74; P 0.001, respectively). Conclusions Patients with chronic ischemic heart disease who are treated with either single-chamber or dual-chamber ICDs have improved survival but an increased risk of HF. The present data suggest that ICD therapy transforms sudden death risk to a subsequent HF risk. These findings should direct more attention to the prevention of HF in patients who receive an ICD. (Circulation. 2006;113: ) Key Words: defibrillation heart failure mortality The implanted cardioverter-defibrillator (ICD) improves survival in appropriately selected high-risk cardiac patients, with a 30% to 54% reduction in the risk of mortality in several randomized trials. 1 4 In the Multicenter Automatic Defibrillator Implantation Trial (MADIT) II, an increase in the rate of hospitalization for new or worsened heart failure (HF) was observed in patients randomly assigned to defibrillator therapy compared with those assigned to conventional therapy. 3 A subsequent publication by Wilkoff et al 5 provided evidence that a high frequency of right ventricular pacing with dual-chamber defibrillator units was a contributing factor to increased HF. The present investigation is a retrospective analysis of clinical factors associated with HF progression after ICD implantation and its prognostic implications in patients enrolled in the prospective MADIT-II trial. Clinical Perspective p 2817 Methods MADIT-II The MADIT-II enrolled 1232 patients with myocardial infarction (MI) 1 month or more before entry and an ejection fraction of 30% or less. Patients were randomly assigned in a 3:2 ratio to receive either an ICD or conventional medical therapy. Details of the design, recruitment, data collection, mortality end point, and medical and data analyses are provided in the primary publication. 3 Device Type and Rate of Cumulative Ventricular Pacing Of the 742 patients allocated to the defibrillator arm, 718 patients with follow-up data (including records of device therapy) received an ICD. Information on ICD type was available for 715 patients: 402 received a single-chamber ICD with the back-up pacing rate set Received July 20, 2005; revision received March 17, 2006; accepted April 7, From the Cardiology Unit of the Department of Medicine (I.G., A.J.M., S.M., W.Z., M.L.A.) and the Department of Biostatistics and Computational Biology (W.J.H.), University of Rochester Medical Center, Rochester, NY, and Cardiology Associates (D.S.C.), Good Samaritan Hospital, Los Angeles, Calif. The online-only Data Supplement can be found at Correspondence to Ilan Goldenberg, MD, Heart Research Follow-up Program, Box 653, University of Rochester Medical Center, Rochester, NY Ilan.Goldenberg@heart.rochester.edu 2006 American Heart Association, Inc. Circulation is available at DOI: /CIRCULATIONAHA

2 Goldenberg et al ICD and Heart Failure 2811 mostly at VVI-40 to 50, and 313 received a dual-chamber ICD with the pacing set mostly at DDD-60 to 70. The implanted devices included the VENTAK AV series, the VENTAK Mini series, and the VENTAK Prizm series (Guidant Corp, St. Paul, Minn). No investigational devices were used. Information on the cumulative rate of right ventricular pacing in 152 patients was unavailable for data analysis (single-chamber ICDs, 98 patients [64%]; dual-chamber ICDs, 51 patients [34%]; unknown ICD type, 3 patients [2%]), of whom 67 patients (44%) had died and 58 patients (38%) had developed postenrollment HF. Because of possible selection bias (a high proportion of missing interrogation data among higher-risk patients in whom a single-chamber ICD was implanted), a comprehensive analysis of the association between cumulative ventricular pacing and risk of postenrollment HF was not feasible. However, available data demonstrated that 92% of patients who received single-chamber devices received little or no pacing throughout the study, whereas among 66% of patients in whom a dual-chamber ICD was implanted, the cumulative rate of right ventricular pacing exceeded 50% (mostly in the range of 90% to 100%). 6 Therefore, all study end points were also analyzed according to device type. Definition of Events and End Points Clinical follow-up of the enrolled patients averaged 20 months (range, 6 days to 53 months). Patients hospitalized for any cause after enrollment were identified, and pertinent information was abstracted from the hospital record. Postenrollment hospitalizations for new or worsened HF and for MI or unstable angina (UA) were categorized according to prespecified criteria, classified by each site investigator on case report forms, and confirmed at the data center by the study coordinator. In the analysis of cardiac events, time to the first nonfatal event was evaluated. Hospitalizations due to cardiac events that terminated in death during the same hospital stay were categorized as death events and not considered as postenrollment hospitalizations. Data concerning arrhythmias and device therapy were obtained at the time of device interrogation at each 3-month follow-up visit. The retrieved electrograms were reviewed by the 2 members of the ECG Core Laboratory. Defibrillator shocks were considered appropriate when administered during episodes of ventricular tachycardia or fibrillation, whereas all other ICD shocks were considered inappropriate. Defibrillator therapies terminating in death were not included as interim shocks. The end points of the study included (1) the occurrence of a first hospitalization for HF (nonfatal); (2) recurrent hospitalizations for HF; and (3) all-cause mortality. To reduce potential bias created by censoring HF-related deaths, analyses for end point 1 were repeated for the combined end point of a first hospitalization for HF or nonsudden cardiac death (NSCD). A modified Hinkle-Thaler system was used to classify deaths. 7 HF end points were analyzed in the total population and in ICD-specific models. In the latter models, postenrollment ICD shock therapy was added as a time-dependent covariate. To distinguish the effects of defibrillator shocks per se on outcome from the lifeprolonging effect of ICD therapy, inappropriate and appropriate shocks were incorporated as separate variables in the ICD-specific models. Allocation to treatment groups in the present study was made on an intention-to-treat basis. However, in the analysis of the association between ICD shocks and postenrollment HF, only ICD-arm patients in whom an ICD was implanted were included, beginning on the day of defibrillator implantation, with censoring at the end of follow-up or explantation of the device. Several important differences between the present and the original MADIT-II analysis 3 should be noted. First, in the reported MADIT- II, a sequential stopping rule was applied. That is, the data were analyzed repeatedly as the study progressed, with a predetermined rule about when the results warranted stopping the trial. In the final analysis, adjustment was made to remove bias resulting from the sequential stopping rule. In the present MADIT-II subgroup analyses, such adjustments were not made because an appropriate statistical methodology is unavailable. Second, in this study, patients who were lost to follow-up during the study period were censored at the time of inactivation owing to the lack of data on interim cardiac events, even when their subsequent survival status became known. Third, we included in the Cox regression models postenrollment cardiac events as time-dependent covariates, whereas no timedependent covariates were included in the original MADIT-II analysis. Fourth, the present study was based on more complete follow-up data, in which more nonfatal events were recorded (postenrollment HF hospitalization counts were 14% higher in the present study in both the ICD and the conventional groups compared with those in the original publication). Statistical Analysis The probability of a first hospitalization for HF was estimated and graphically displayed according to the Kaplan-Meier method, with a comparison of cumulative events by the log-rank test. The Cox proportional-hazards regression model was used to evaluate the independent contribution of baseline clinical factors and timedependent cardiac events to the development of the first occurrences of end points in the total population and in ICD-specific models. Because significant baseline clinical differences existed between ICD-allocated patients treated with single-chamber units and those treated with dual-chamber units, we performed an alternative analysis in which a propensity score was included in the proportionalhazards modeling of mortality to analyze the effectiveness of the 2 ICD types and the interim effects of hospitalization for HF by ICD type. 8 In the analysis of the association between ICD shocks and a first hospitalization for HF, time to first HF was analyzed with the first ICD shock as a time-dependent covariate, with and without other explanatory risk factors. In the former analysis, use of -blockers at baseline was also considered a covariate. In the analyses of the factors and outcomes associated with recurrent HF hospitalizations, the end-point intensity function was adjusted for covariate effects by the Anderson-Gill proportionalintensity regression model 9 ; alternative analyses of recurrent HF hospitalizations were performed by negative binomial regression, 10 allowing for estimation of the average (over patients) rate of occurrence of events along with the effects of covariates. The former method is analogous to the proportional-hazards method, but it deals with the risk of repeatable events instead of terminating observation after a first event. As with hazard functions for first events, intensity functions for recurrent events are not assumed to be constant over time, and thus, there is no single rate for a patient without reference to a common time span (in contrast to negative binomial regression, wherein each patient is assumed to have a constant event rate over time). The statistical software used for the analyses was SAS version 9.13 (SAS Institute, Cary, NC). A 2-sided probability value 0.05 was used for declaring statistical significance. The authors had full access to the data and take full responsibility for its integrity. All authors have read and agree to the manuscript as written. Results Follow-up data on postenrollment HF hospitalizations were incomplete for 14 study patients. Thus, 1218 patients were included in the present analysis. Of the study patients, 253 (21%) were hospitalized at some time for postenrollment HF: 169 of 736 ICD-allocated patients (23%) versus 84 of 482 conventional therapy allocated patients (17%; P 0.02). The clinical characteristics of the patients in the conventional-therapy arm and in patients in the defibrillator arm in whom an ICD was implanted, categorized by ICD type, are presented in Table 1. Compared with conventionally treated patients and those who received a single-chamber ICD, patients in whom a dual-chamber ICD was implanted had a higher proportion of baseline risk factors, including older age, more advanced baseline New York Heart Associ-

3 2812 Circulation June 20, 2006 TABLE 1. Clinical Characteristics of Patients in the Conventional-Therapy Group and in the Defibrillator Group by Device Type Conventional (n 482) Single Chamber (n 402) Defibrillator Dual Chamber (n 313) Age 65 y, % Female sex, % NYHA functional class 2, %* I II III IV Angina pectoris grade 2, %* Hypertension, % Diabetes, % Prior CABG surgery, % BUN 9 mmol/l (25 mg/dl), % QRS interval 0.12 s, % Left bundle-branch block, % Atrial fibrillation, % Ejection fraction 0.25, % Blood pressure Systolic 130 mm Hg, % Diastolic 80 mm Hg, % Heart rate 80 bpm, % BMI 30 kg/m 2,% Postenrollment events HF, % HF or death, % MI, % UA, % Appropriate ICD shocks, % NA Inappropriate ICD shocks, % NA Medical therapy ACE inhibitors, % Digitalis, % Blockers, % Amiodarone, % Lipid-lowering agents, % NYHA indicates New York Heart Association; CABG, coronary artery bypass graft; BUN, blood urea nitrogen; BMI, body mass index; and ACE, angiotensinconverting enzyme. All characteristics are given as percentages. *NYHA functional class and angina pectoris grade (Canadian Classification) represent highest class during the 3 months before enrollment. Includes patients who survived after receiving 1 or more appropriate or inappropriate shocks during follow-up. Statins comprised 95% of lipid-lowering agents. P 0.05 by 2 test for comparison between the conventional, singlechamber, and dual-chamber groups; 18 defibrillator-arm patients in whom an ICD was not implanted in the study and 3 patients with an unknown ICD type were omitted. ation functional class, wider QRS duration, and the presence of left bundle branch block on the baseline ECG, whereas the proportion of patients with a baseline heart rate 80 beats/ min was higher among patients who received a singlechamber ICD. Postenrollment hospitalization for HF occurred at a higher frequency both among patients who received single-chamber ICDs (22.4%) and those who received dualchamber ICDs (24.9%) compared with patients in the conventional-therapy group (17.4%; P 0.03). The frequency of NSCD was nonsignificantly higher among patients who received single-chamber (5.8%) and dual-chamber (6.5%) devices than among conventionally treated patients (4.6%), whereas the frequency of postenrollment hospitalization for MI or UA was similar in the ICD and conventional-therapy groups. Treatment with -blockers was less frequent in patients who were hospitalized for HF in the ICD group compared with those who were not (54% versus 63%, respectively; P 0.003), whereas the proportion of patients with and without HF who received angiotensin-converting enzyme inhibitors was similar in the ICD group (76% versus 78%, respectively; P 0.61). Kaplan-Meier estimates of all-cause mortality, all-cause mortality or postenrollment hospitalization for HF, and postenrollment hospitalization for HF by treatment group are presented in the Figure. Patients allocated to the ICD group demonstrated a significant survival benefit compared with the conventional group (Figure, A). However, the 2 treatment curves overlapped when the combined occurrence of allcause mortality and postenrollment hospitalization for HF was considered (Figure, B), a finding reflecting a significantly greater probability of occurrence of at least 1 HF hospitalization in the ICD group (26%) compared with the conventional-therapy group (21%) during a 2-year follow-up (Figure, C). Predictors of HF in the Total Population and in the ICD Group Baseline clinical factors, identified by the multivariate Cox proportional-hazards regression analysis as being independently associated with postenrollment hospitalization for HF, were similar in the total population and the ICD group and included atrial fibrillation, QRS duration 0.12 seconds, blood urea nitrogen levels 9 mmol/l (25 mg/dl), diabetes mellitus, heart rate 80 beats/min on the ECG, and female sex. Medical therapy with -blockers was significantly associated with a reduction in the risk of HF hospitalization in the ICD group only (hazard ratio [HR], 0.59; 95% confidence interval [CI], 0.43 to 0.80; P 0.001). Allocation to the ICD group was associated with a statistically significant 39% increase in the risk of a first HF hospitalization and a 58% increase in the intensity of recurrent HF hospitalizations compared with conventional therapy after adjustment for baseline clinical covariates and postenrollment MI or UA (Table 2). A negative binomial regression gave results similar to those derived from the Anderson-Gill model (60% increase in the risk of recurrent HF hospitalizations), with estimated average annual rates of HF hospital-

4 Goldenberg et al ICD and Heart Failure 2813 Kaplan-Meier estimates of the probability of (A) all-cause mortality, (B) all-cause mortality or first hospitalization for HF, and (C) first hospitalization for HF with censoring on death, by treatment group. izations of 25% in the ICD group compared with 16% in the conventional-therapy group (P 0.004). The risk of first and recurrent HF hospitalizations was increased both among patients who had received a singlechamber defibrillator unit (by 29% and 44%, respectively) and, to a greater extent, among patients in whom a dualchamber ICD was implanted (by 59% and 84%, respectively) compared with the conventional-therapy group (Table 2). The increased risk of HF among patients allocated to the ICD group was related to the life-prolonging effect of appropriate defibrillator shocks. Before appropriate ICD shock therapy, 147 HF events occurred during a total

5 2814 Circulation June 20, 2006 TABLE 2. Multivariate Analysis: Predictors of First and Recurrent Nonfatal Postenrollment HF Events in the Total Population and in the Defibrillator Group* First HF Event Recurrent HF Variable HR (95% CI) P HR (95% CI) P Total population (n 1218) Total defibrillator group vs conventional-therapy group 1.39 ( ) ( ) Single-chamber-device group vs conventional-therapy group 1.29 ( ) ( ) Dual-chamber-device group vs conventional-therapy group 1.59 ( ) ( ) Defibrillator group: first appropriate shock vs no appropriate shock (n 718)# All ICDs 1.90 ( ) ( ) Single-chamber devices 2.06 ( ) ( ) Dual-chamber devices 1.69 ( ) ( ) 0.07 *All findings were adjusted for the following additional covariates: (1) in the total population model: age 65 y, atrial fibrillation, BUN 9 mmol/l (25 mg/dl), NYHA class 2, QRS duration 0.12 s, diabetes mellitus, heart rate 80 bpm by ECG, female sex, and postenrollment MI or UA; (2) in the ICD-specific model: atrial fibrillation, BUN 9 mmol/l (25 mg/dl), NYHA class 2, QRS duration 0.12 s, diabetes mellitus, female sex, treatment with -blockers at baseline, and postenrollment MI or UA. In the analysis of the total defibrillator group vs the conventional-therapy group, 736 ICD arm allocated patients vs 482 conventional arm allocated patients were included; in the analysis of single- and dual-chamber devices, findings were analyzed among the respective 402 and 313 patients in whom these devices were implanted and for whom information on ICD type was available compared with the 482 patients in the conventional-therapy group. The end point was defined as postenrollment hospitalization for nonfatal HF, and similar results were obtained when the end point of first HF or NCD was used. Analysis of recurrent cardiac events was based on the Anderson-Gill model 9 ; similar results were obtained by negative binomial regression (see text). The test was repeated with NCD added to the end point, with nonconflicting results. Findings were analyzed for 718 patients in the defibrillator arm in whom an ICD was implanted, beginning on the day of defibrillator implantation with censoring at the end of follow-up or explantation of the device. #The P value for the difference in risk between single- and dual-chamber ICDs was 0.20 for first HF events and 0.04 for recurrent HF events. The P value for the ICD type appropriate shock interaction was 0.70 for first HF events and 0.59 for recurrent HF events. follow-up time of 919 person-years (unadjusted HF rate, 16.0% per 100 person-years of follow-up), whereas after appropriate ICD shock therapy, 18 HF events occurred during a relatively short total follow-up time of 80 person-years (unadjusted HF rate, 22.5% per 100 person-years of follow-up); the corresponding unadjusted HF rates before and after inappropriate ICD shock therapy were 16.4% and 18.3%, respectively (see onlineonly Data Supplement Table). In multivariate analysis, there was a significant 90% increase in the risk of a first HF hospitalization and a 74% increase in the intensity of recurrent HF hospitalizations after appropriate ICD shocks for ventricular tachycardia or fibrillation (Table 2), whereas inappropriate defibrillator shocks were not associated with a significantly increased risk of a subsequent first (HR, 1.30; 95% CI, 0.76 to 2.22; P 0.33) or recurrent (HR, 1.17; 95% CI, 0.66 to 2.05; P 0.60) HF hospitalizations. Similar results were obtained when the end point of HF or NSCD was analyzed. The association between appropriate ICD shocks and the risk of subsequent postenrollment first and recurrent HF hospitalizations was consistent for both single- and dualchamber ICD units (probability value for interaction between ICD type and the risk of first and recurrent HF after appropriate shocks was 0.70 and 0.59, respectively). Postenrollment HF and Mortality in the Total Population and by ICD Type After adjustment for baseline covariates and other postenrollment cardiac events, the occurrence of a postenrollment HF hospitalization was associated with a nearly 4-fold increase in the risk of subsequent all-cause mortality in the total population (Table 3). Hospitalizations for other cardiac causes (postenrollment MI or UA) were associated with a smaller and less significant increase in the risk of all-cause mortality (HR, 1.58; P 0.04). The efficacy of ICD therapy was maintained after HF (48% and 34% reduction in the hazard of all-cause mortality before and after first HF hospitalization, respectively); a treatment HF interaction was not significant (P 0.79) when the total defibrillator group was compared with the conventional-therapy group (Table 3). However, when the efficacy of ICD therapy was evaluated for each device type separately, survival benefit before and after HF was different for the 2 device types. There was a similar survival advantage with single-chamber ICDs before and after HF (P 0.92 for single-chamber ICD HF interaction; Table 3). In contrast, there was a meaningful difference in survival advantage with dual-chamber ICDs before and after HF (P 0.01 for dualchamber ICD HF interaction; Table 3). Similar results were obtained when a propensity score was included in the proportional-hazards modeling of mortality (single-chamber versus conventional: before or no HF, HR, 0.61; P 0.02; after HF, HR, 0.62; P 0.08; dual-chamber versus conventional: before or no HF, HR, 0.35; P 0.002; after HF, HR, 0.94; P 0.83). As supporting evidence for this latter finding, Kaplan- Meier survival curves (not shown) for dual-chamber patients and for conventional-therapy patients, with time origin at the first HF hospitalization, were nearly identical and demonstrated similar probabilities of survival (72% and 71%, respectively) at 1 year.

6 Goldenberg et al ICD and Heart Failure 2815 TABLE 3. Multivariate Analysis: Predictors of All-Cause Mortality in the Total Population and by ICD Type* Variable HR 95% CI P Total population Defibrillator vs conventional Defibrillator vs conventional before HF or no HF Defibrillator vs conventional after HF Postenrollment HF Postenrollment MI/UA Single-chamber devices Single chamber vs conventional Single chamber vs conventional before HF or no HF Single chamber vs conventional after HF Dual-chamber devices Dual chamber vs conventional Dual chamber vs conventional before HF or no HF Dual chamber vs conventional after HF *Findings were adjusted for the following additional covariates: age 65 y, BUN 9 mmol/l (25 mg/dl), systolic blood pressure 130 mm Hg, heart rate 80 bpm by ECG, BMI 30 kg/m 2, and the presence of left bundle-branch block on baseline ECG. In the analysis of the total defibrillator group vs the conventional-therapy group, 736 ICD arm allocated patients vs 482 conventional arm allocated patients were included; in the analysis of single- and dual-chamber devices, findings were analyzed among the respective 402 and 313 patients in whom these devices were implanted and for whom information on ICD type was available compared with the 482 patients in the conventionaltherapy group. The P value for the treatment HF interaction was The analyses by device type were based on the Cox proportional-hazards regression model with covariate adjustment; similar results were obtained by propensity score adjustment 8 (see text). The P value for the treatment HF interaction was The P value for the treatment HF interaction was Discussion Several clinical implications emerge from the present analysis of MADIT-II: (1) The development of HF is a major determinant of subsequent mortality in patients with ischemic heart disease and left ventricular dysfunction; (2) the survival benefit conferred by ICD therapy is associated with an increased risk of first and recurrent HF hospitalizations among patients who receive single-chamber or dual-chamber devices; and (3) the life-prolonging efficacy of ICD therapy is maintained after HF among patients who receive singlechamber devices, whereas there appears to be a significant reduction in ICD benefit after HF among patients who receive dual-chamber devices. Right ventricular pacing with a dual-chamber ICD has been shown to contribute to an increased risk of HF after ICD implantation. 5 In the present study, the risk of first and recurrent HF hospitalizations was increased to a greater extent in patients with dual-chamber ICDs compared with those with single-chamber devices. However, the increase in risk of the first (29%) and recurrent (44%) postenrollment HF hospitalization was also considerable among patients who had received single-chamber ICDs. Because the back-up pacing rate of single-chamber units was set at VVI-40 to 50 and analysis of the available data demonstrated that most single-chamber devices received little or no pacing throughout the study, it appears that the increased risk of postenrollment HF in the defibrillator arm of the trial was not due solely to the effect of right ventricular pacing with dual-chamber ICDs, and additional ICD-specific factors may be important. Myocardial damage induced by defibrillator shocks has been suggested to contribute to the HF risk among ICDtreated patients. 11,12 In the present analysis, we have shown that in contrast to the significant association between appropriate ICD shocks and subsequent risk of HF among patients who received both single- and dual-chamber ICDs, inappropriate ICD shocks were not associated with a significant increase in risk of subsequent HF. Thus, the effects of shock per se on the myocardium did not contribute to HF after ICD implantation. Rather, it seems that in the MADIT-II, high-risk patients whose lives were saved by appropriate ICD therapy were also at high risk for the development of subsequent HF, resulting in a significant reduction in the risk of sudden death but a concurrent increase in the risk of HF events among patients allocated to the defibrillator arm of the trial. The recently published Defibrillator in Acute Myocardial Infarction Trial reported a reduction in arrhythmic deaths in the ICD group, which was offset by a similar increase in nonarrhythmic cardiovascular deaths. 13 Further analysis by the Defibrillator in Acute Myocardial Infarction Trial investigators indicated that the increased risk of death from nonarrhythmic causes was confined to patients who had received appropriate ICD shocks. 14 The CABG-Patch Trial investigators 15 also reported that ICD therapy reduced the rate of death due to arrhythmias by 45% but did not reduce overall

7 2816 Circulation June 20, 2006 mortality, with the majority of deaths (71%) caused by nonarrhythmic cardiovascular events. The authors of both studies suggested that patients who survived an arrhythmiarelated event owing to appropriate ICD therapy were at increased risk for subsequent death from pump failure, and therefore a survival benefit in the ICD arm was not observed. 14,15 In contrast to the former 2 studies, the rate of sudden death in MADIT-II was markedly reduced, whereas the rate of death from nonsudden causes was increased only to an extent explainable by reduced arrhythmic mortality. 7 This MADIT-II finding may be explained by the fact that higherrisk patients with recent MI or recent coronary revascularization were excluded from the trial. However, in the present analysis, the 42% reduction in risk of all-cause mortality was associated with a 39% increase in the risk of hospitalization for HF. Therefore, it seems that in the MADIT-II, arrhythmia-related events were transformed into an increased risk for subsequent HF events. From the present study, HF emerges as a marker of increased risk of subsequent death among post-mi patients with left ventricular dysfunction. The development of HF was associated with a nearly 4-fold increase in the risk of all-cause mortality in the total MADIT-II population. Notably, the magnitude of risk associated with hospitalizations due to other interim cardiac events (including UA or MI) was smaller and only of marginal statistical significance compared with HF. Therefore, it is important to prevent HF progression after ICD implantation. One possible preventive strategy may be resynchronization therapy. A QRS duration 0.12 seconds was associated with an increased risk of HF in our study. Resynchronization therapy has been successfully applied to HF patients with this risk factor, treated with or without a defibrillator, and was shown to attenuate various parameters of HF progression and reduce the risk of HF hospitalization Thus, this mode of therapy might be considered an adjunct to ICD implantation in patients whose baseline clinical characteristics suggest an enhanced risk for HF. Optimization of adjunctive medical therapy should also be considered in ICD patients. In the present study, -blocker therapy was associated with a reduced risk of HF hospitalization in the ICD arm, whereas angiotensin-converting enzyme inhibitor therapy was not. Previous analysis of MADIT-II has consistently shown that -blocker therapy is associated with improved survival and a reduced risk for ventricular tachycardia or fibrillation in ICD-treated patients. 19 These adjunctive medical therapeutic modalities should be prospectively evaluated in ICD trials. The type of implanted device affected the outcome that patients experienced after the development of HF. Among patients who received single-chamber devices, the efficacy of ICD therapy before and after HF was similar (41% and 39% reduction in the risk of all-cause mortality, respectively). By contrast, in patients in whom a dual-chamber device was implanted, there was a 74% reduction in the risk of all-cause mortality before the development of HF, whereas no survival benefit was shown after the development of HF. Therefore, dual-chamber devices appear to be associated with both a significant increase in the risk of developing HF and reduced efficacy after developing HF. Study Limitations The present study is a retrospective analysis of postenrollment HF hospitalizations in the MADIT-II, and thus, it was not prospectively designed to evaluate predictors of HF progression after ICD implantation or therapeutic modalities that reduce HF progression nor to compare types of ICD devices. The present study results are applicable only to the MADIT-II population, which comprised patients with coronary heart disease and left ventricular dysfunction. Data on HF progression after ICD implantation for other indications, such as malignant arrhythmias or nonischemic cardiomyopathy, cannot be derived from the present analysis. In the present study, we have attempted to analyze complex data comprising time-dependent variables within different (and sometimes overlapping) subsets of study patients. This analysis was limited by the fact that censored data existed for every subgroup, and the follow-up times not only were of variable duration but also occurred with different start-up times (eg, postshock follow-up occurred later during total follow-up than did preshock follow-up). Moreover, these different periods of follow-up occurred in patients with very different risk factor profiles. For these reasons, we based our tests for statistical difference on proportional-hazards (or proportional-intensity) models with time-dependent covariates. Despite the limitations of these models, we believe that these analyses provide the best basis for statistical testing available, whereas alternative tests based on Poisson modeling are more restrictive. Conclusions Our findings show that the life-prolonging benefit of ICD therapy is associated with increased HF events. The augmented rate of HF after appropriate defibrillator shocks suggests that high-risk patients whose lives are saved by ICD therapy are at increased risk for subsequent HF events. Further prospective trials are needed to evaluate therapeutic modalities to reduce HF progression in ICD-treated patients. These may include cardiac resynchronization therapy or optimization of adjunctive medical therapy. Source of Funding MADIT-II was supported in part by a grant from Guidant Corp to the University of Rochester. Disclosures Dr Cannom has received honoraria from and acts as a consultant to the advisory board of Guidant Corp. The other authors report no conflicts. References 1. Moss AJ, Hall WJ, Cannom DS, Daubert JP, Higgins SL, Klein H, Levine JH, Saksena S, Waldo AL, Wilber D, Brown MW, Heo M, for the Multicenter Automatic Defibrillator Implantation Trial Investigators. Improved survival with an implanted defibrillator in patients with coronary disease at high risk for ventricular arrhythmia. N Engl J Med. 1996;335: The Antiarrhythmics versus Implantable Defibrillators (AVID) Investigators. A comparison of antiarrhythmic-drug therapy with implantable defibrillators in patients resuscitated from near-fatal ventricular arrhythmias. N Engl J Med. 1997;337:

8 Goldenberg et al ICD and Heart Failure Moss AJ, Zareba W, Hall WJ, Klein H, Wilber DJ, Cannom DS, Daubert JP, Higgins SL, Brown MW, Andrews ML. Multicenter Automatic Defibrillator Implantation Trial II Investigators. Prophylactic implantation of a defibrillator in patients with myocardial infarction and reduced ejection fraction. N Engl J Med. 2002;346: Bardy GH, Lee KL, Mark DB, Poole JE, Packer DL, Boineau R, Domanski M, Troutman C, Anderson J, Johnson G, McNulty SE, Clapp- Channing N, Davidson-Ray LD, Fraulo ES, Fishbein DP, Luceri RM, Ip JH; Sudden Cardiac Death in Heart Failure Trial (SCD-HeFT) Investigators. Amiodarone or an implantable cardioverter-defibrillator for congestive heart failure. N Engl J Med. 2005;352: Wilkoff BL, Cook JR, Epstein AE, Greene HL, Hallstrom AP, Hsia H, Kutalek SP, Sharma A; Dual Chamber and VVI Implantable Defibrillator Trial Investigators. Dual-chamber pacing or ventricular backup pacing in patients with an implantable defibrillator: the Dual Chamber and VVI Implantable Defibrillator (DAVID) Trial. JAMA. 2002;288: Steinberg JS, Fischer A, Wang P, Schuger C, Daubert J, McNitt S, Andrews M, Brown M, Hall WJ, Zareba W, Moss AJ; MADIT II Investigators. The clinical implications of cumulative right ventricular pacing in the Multicenter Automatic Defibrillator Trial II. J Cardiovasc Electrophysiol. 2005;16: Greenberg H, Case RB, Moss AJ, Brown MW, Carroll ER, Andrews ML; MADIT-II Investigators. Analysis of mortality events in the Multicenter Automatic Defibrillator Implantation Trial (MADIT-II). J Am Coll Cardiol. 2004;43: D Agostino RB Jr. Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Stat Med. 1998;17: Andersen PK, Gill RD. Cox s regression for counting processes: a large sample study. Ann Stat. 1982;10: Lawless JF. Regression methods for Poisson process data. J Am Stat Assoc. 1987;82: Runsio M, Bergfeldt L, Brodin LA, Ribeiro A, Samuelsson S, Resenqvist M. Left ventricular function after repeated episodes of ventricular fibrillation and defibrillation assessed by transesophageal echocardiography. Eur Heart J. 1997;18: Hurst TM, Hinrichs M, Breidenbach C, Katz N, Waldecker B. Detection of myocardial injury during transvenous implantation of automatic cardioverter-defibrillators. J Am Coll Cardiol. 1999;34: Hohnloser SH, Kuck KH, Dorian P, Roberts RS, Teck M, Hampton JR, Hatala R, Fain E, Gent M, Connolly SJ on behalf of the DINAMIT investigators. Prophylactic use of an implantable cardioverterdefibrillator after acute myocardial infarction. N Engl J Med. 2004;351: Dorian P, Connolly S, Hohnloser SH. Why don t ICDs decrease all-cause mortality after MI? insights from the DINAMIT study. Circulation. 2004;110(suppl III):III-502. Abstract. 15. Bigger JT Jr, Whang W, Rottman JN, Kleiger RE, Gottlieb CD, Namerow PB, Steinman RC, Estes NA 3rd. Mechanisms of death in the CABG Patch Trial: a randomized trial of implantable cardiac defibrillator prophylaxis in patients at high risk of death after coronary artery bypass graft surgery. Circulation. 1999;99: Young JB, Abraham WT, Smith AL, Leon AR, Lieberman R, Wilkoff B, Canby RC, Schroeder JS, Liem LB, Hall S, Wheelan K; Multicenter InSync ICD Randomized Clinical Evaluation (MIRACLE ICD) Trial Investigators. Combined cardiac resynchronization and implantable cardioversion defibrillation in advanced chronic heart failure: the MIRACLE ICD Trial. JAMA. 2003;289: Bristow MR, Saxon LA, Boehmer J, Krueger S, Kass DA, De Marco T, Carson P, DiCarlo L, DeMets D, White BG, DeVries DW, Feldman AM; Comparison of Medical Therapy, Pacing, and Defibrillation in Heart Failure (COMPANION) Investigators. Cardiac-resynchronization therapy with or without an implantable defibrillator in advanced chronic heart failure. N Engl J Med. 2004;350: Cleland JG, Daubert JC, Erdmann E, Freemantle N, Gras D, Kappenberger L, Tavazzi L; Cardiac Resynchronization-Heart Failure (CARE-HF) Study Investigators. The effect of cardiac resynchronization on morbidity and mortality in heart failure. N Engl J Med. 2005;352: Brodine WN, Tung RT, Lee JK, Hockstad ES, Moss AJ, Zareba W, Hall WJ, Andrews M, McNitt S, Daubert JP; MADIT-II Research Group. Effects of -blockers on implantable cardioverter defibrillator therapy and survival in the patients with ischemic cardiomyopathy (from the Multicenter Automatic Defibrillator Implantation Trial-II). Am J Cardiol. 2005;96: CLINICAL PERSPECTIVE In the Multicenter Automatic Defibrillator Implantation Trial II, an increase in the rate of hospitalization for new or worsened heart failure (HF) was observed in patients assigned to implanted cardioverter-defibrillators (ICDs) compared with those assigned to conventional therapy. Subsequently, it was suggested that a high frequency of right ventricular pacing with dual-chamber ICDs might have been a contributing factor to the increased HF. The present analysis shows that life-prolonging ICD therapy in the Multicenter Automatic Defibrillator Implantation Trial II was associated with a respective 39% and 58% increase in the risk of first and recurrent HF hospitalizations during follow-up. Notably, the risk of HF was increased both among patients who received single-chamber ICDs and those who received dual-chamber ICDs. Using time-dependent analysis, we have demonstrated that after appropriate ICD shock therapy, the risk of first and recurrent HF events was increased by 90% and 74%, respectively, whereas inappropriate ICD shocks were not associated with a significant increase in the risk of HF events. The development of HF was associated with a 4-fold increase in the risk of death. However, the life-saving benefit of ICD therapy was maintained after the development of HF. Our findings suggest that ICD therapy transforms the risk of sudden death to a subsequent HF risk. More attention should be directed to therapeutic modalities to reduce HF progression in ICD-treated patients. These may include cardiac resynchronization therapy or optimization of adjunctive medical therapy.