1RICHARD N. VEST III, MD, 2RADHIKA R. GADESAM, MD, 5SAMEERA RUKSHAN WIJAYAWARDANA, MS,3,4RAVINDER R. VALADRI, MD, 3,4MAZIAR ZAFARI, MD, PhD, 5KIRK A. EASLEY, MS and 3,4HEATHER L. BLOOM, MD
1Medical University of South Carolina, Charleston, SC
2Morehouse School of Medicine, Atlanta, GA
3Emory University School of Medicine, Atlanta, GA
4Atlanta Veterans Affairs Medical Center, Decatur, GA
5Emory University Rollins School of Public Health, Atlanta, GA
KEYWORDS. heart failure, implantable cardioverter-defibrillator, obesity, ventricular arrhythmia.
The authors report no conflicts of interest for the published content.
Manuscript received August 22, 2011, final version accepted September 9, 2011.
Address correspondence to: Heather L. Bloom, MD, FACC, Atlanta VAMC, Atlanta GA. E-mail: firstname.lastname@example.org
Obesity is a rapidly growing epidemic in the United States with 66% of United States adults estimated to be overweight (body mass index (BMI) >25 to <30) or obese (BMI ≥30) according to the 2003–2004 National Health and Nutrition Examination Survey.1 Previous studies have reported an association of obesity and increased arrhythmogenesis, including higher incidence of atrial fibrillation, change in QTc duration and sudden cardiac death in obese patients.2–4 In a retrospective analysis on 476 non-diabetic patients with ischemic left ventricular dysfunction from MADIT-II (Multicenter Automatic Defibrillator Implantation Trial) trial, BMI ≥30 kg/m2 was found to be an independent risk factor for ventricular arrhythmias and appropriate implantable cardioverter-defibrillator (ICD) therapies/shocks.5 However, patients enrolled in major ICD trials may not represent an unscreened general population. Bunch et al6 reported lower long-term survival in normal or underweight individuals and no weight-based difference in shocks in 226 survivors of out-of-hospital cardiac arrest due to ventricular fibrillation.
We conducted this study to evaluate the association of obesity and appropriate ICD therapy for ventricular tachycardia (VT) and ventricular fibrillation (VF) in a general unselected heart failure patient population, to represent our real-world experience at a center with a high prevalence of comorbidities: a VA hospital. No study has examined the relationship of elevated BMI to ventricular arrhythmias in patients with both ischemic and non-ischemic cardiomyopathy, nor has this relationship been examined outside of a clinical trial patient population.
This study was a cross-sectional evaluation of patients who received ICD care at a VA hospital from 1998 to 2008. Patients were included if they were ≥21 years of age, had systolic heart failure with left ventricular ejection fraction (LVEF) ≤40%, and had previous ICD placement including single-chamber, dual-chamber, or cardiac resynchronization therapy. Patients were excluded if they had less than 3 months of follow-up. A total of 398 patients were screened, and 369 met inclusion and exclusion criteria for analysis.
Definitions and endpoints
The primary outcome studied was the first occurrence of appropriate ICD shock or anti-tachycardia pacing (ATP) for VT or VF. Appropriate ICD therapy was determined by reviewing medical records from the arrhythmia clinic. Patients receiving routine ICD care at the arrhythmia clinic were typically evaluated in 3-month interval visits, and appropriate ICD therapy was confirmed with intracardiac electrogram analysis. ICD programming was not standardized, as this was a retrospective review. All patients had a minimum of one zone set, and the majority had two or three zones set, which is standard for the implanting physicians at the facility. All ICD therapies were reviewed and adjudicated for appropriateness by an attending electrophysiologist. All patients received height and weight measurements at the time of treatment as well as at subsequent visits. Weight and height measurements at the initial clinical evaluation were used for BMI determination, and patients' weights were followed over time to determine change in weight over the study period. Obesity was defined by a standard clinical definition of BMI ≥30 kg/m2. The secondary endpoint considered was mortality. In addition to examining medical records, mortality was screened and confirmed for all patients with use of the social security death index.
Baseline clinical characteristics were compared across the prespecified BMI categories using the chi-square test or the Fisher exact test for categorical variables and the Student's t test for independent samples for continuous variables. Kaplan–Meier estimates for endpoints, stratified by BMI category, were determined and statistically evaluated for a difference in the cumulative mortality or cumulative appropriate ICD therapy using the log-rank test. The cumulative estimates of the rate of each endpoint are reported for the first 5 years of follow-up at yearly intervals. Additionally, rates of appropriate ICD therapy per 100 patient follow-up years by BMI quartile group were estimated and compared using methods based on the Poisson distribution.
Univariable Cox proportional-hazards regression models were used to evaluate the independent contributions of each of the baseline clinical factors to the development of endpoints. All variables identified as being significant in the univariable analyses were then used to fit a multivariable model using stepwise selection of significant variables. The dichotomized as well as the continuous forms of the variables (BMI, age, and LVEF) were considered in the model fitting and both forms led to similar results. All predictors with univariable significance at or below p = 0.20 were considered as predictors in all multivariable analyses. Stepwise selection of covariates was then used to select the most parsimonious set of predictive variables using a p-value ≤0.10 cut-off. The hazard ratio (HR) and its 95% confidence intervals were calculated for each factor in the presence of others in the final model. All p-values considered were two-sided, and unless otherwise mentioned a p-value ≤0.05 was considered statistically significant. All analyses were performed with the use of SAS software (version 9.2).
Subjects were 98.9% (n = 365) male with mean age of 66.2 (SD = 10.1) years. Median follow-up time was 2.8 years (range 0.3–11.6) for subjects not dying, and the overall mortality by the end of the studied period was 26.6%. There was no difference in mean follow-up time between obese and non-obese patients (2.8 versus 3.1 years, p = 0.58). Prevalence of appropriate ICD therapy for VT/VF was 36.9%. Mean BMI was 29.4 (SD = 5.9) kg/m2, and there was no significant difference in patients' BMI values at the end of the study period. Compared with patients with BMI <30 kg/m2, patients with BMI ≥30 kg/m2 were younger, had more diabetes, and had more β-blocker use (Table 1). Obese patients had a lower but not significant incidence of appropriate ICD therapy (33.8% versus 43.9%, p = 0.053) (Table 2). Inappropriate ICD shocks were observed in 14% of patients without a difference between obese and non-obese patients. When divided into 4 equal groups for evaluation by BMI quartiles (BMI 32.6 kg/m2), there was a marginally significant difference in the rates of appropriate ICD therapy per 100 patient follow-up years (p = 0.07) (Figure 1). Patients with BMI 25.4–28.8 kg/m2 had the highest therapy rates, but this was before controlling for confounding variables.
Figure 1: Incidence of implantable cardioverter-defibrillator therapy (per 100 patient follow-up years) by body mass index quartile categories.
Kaplan–Meier estimates for cumulative appropriate ICD therapy demonstrated no significant difference between obese and non-obese patients (Figure 2). The 1–5 year cumulative ICD therapy rates were 16.1%, 28.4%, 39.2%, 39.2% and 51.3% in obese patients and 19.5%, 32.3%, 43.0%, 53.8% and 70.6% in non-obese patients. Cox univariable analysis was performed evaluating clinical factors associated with increased risk of appropriate ICD therapy, and a marginally significant greater risk was observed in obese patients (HR 0.73, p = 0.07) compared with non-obese patients (Table 3). The presence of LVEF <25% demonstrated significantly increased risk for appropriate ICD therapy, but no significant risk was observed for ischemic cardiomyopathy, diabetes, renal insufficiency, New York Heart Association (NYHA) class ≥3, age ≥65 years, CRT, or β-blocker use. When Cox multivariable regression was performed with stepwise selection to control all factors studied in the univariable analysis, risk of appropriate ICD therapy was found not to be significantly associated with obesity in the presence of the effect of LVEF (Table 4).
Figure 2: Kaplan–Meier estimate for cumulative appropriate implantable cardioverter-defibrillator therapy between obese and non-obese patients.
Kaplan–Meier estimates for cumulative mortality did not demonstrate a significant difference between obese and non-obese patients (Figure 3). The 1–5 year cumulative mortality rates were 4.4%, 11.1%, 20.8%, 32.8%, and 52.5% in obese patients and 3.3%, 9.5%, 22.2%, 30.8%, and 40.7% in non-obese patients. Cox multivariable regression demonstrated no significant difference in risk of mortality between obese and non-obese patients (HR 0.76, 95% CI 0.54–1.07, p = 0.11) when results were controlled for ischemic cardiomyopathy, diabetes, renal insufficiency, age ≥65 years, LVEF <25%, NYHA class ≥3, and β-blocker use. Patients with LVEF <25% had significantly increased risk of mortality (HR 1.81, 95% CI 1.15–2.87) as did patients with NYHA class ≥3 heart failure (HR 6.04, 95% CI 3.44–10.58) (Table 5). Patients who used β-blockers had significantly decreased risk of mortality (HR 0.45, 95% CI 0.27–0.73).
Figure 3: Kaplan–Meier estimate for cumulative mortality between obese and non-obese patients.
To our knowledge this is the first study performed evaluating the association of obesity with appropriate ICD therapy in patients with both ischemic and non-ischemic cardiomyopathy. This study of patients with systolic heart failure and previous ICD placement demonstrates no significant association between appropriate ICD therapy and obesity. No significant difference in ICD therapies was observed between highest, intermediate, and lowest BMI categories. There was also no difference observed in time to first event. These results remained unchanged when controlled for potentially confounding factors.
Previous studies have suggested increased risk of supraventricular arrhythmias in obese patients,2 but studies of obesity and ventricular arrhythmias are limited. A single study performed by Pietrasik et al5 demonstrated increased risk of appropriate ICD shocks and ATP in obese patients without diabetes in the MADIT II trial. The increased risk for ICD therapy in obese patients observed by Pietrasik et al5 was postulated to result from increased sympathetic activity caused by leptin. Studies have demonstrated increased leptin levels in obese patients, and animal studies have shown higher heart rates and mean arterial pressures in rats with elevated leptin levels.7 Leptin has also been associated with elevated heart rates in denervated hearts in recipients of orthotopic heart transplants.8
Patients in a large clinical trial may not represent an unscreened patient population in clinical practice given inclusion and exclusion criteria necessary for a clinical trial. Our patients represent an unscreened heart failure population that is likely representative of other predominantly male populations in clinical practice. The differences found when comparing our patient population with the most recently published major ICD trial were multifold. Patients with LVEF <35%, ischemic or non-ischemic cardiomyopathy, and ICD placement in the Sudden Cardiac Death in Heart Failure trial (SCD-HeFT) had a 22% overall prevalence and 5.6% average yearly incidence of appropriate ICD therapy.9,13 This is significantly lower than our patients' 36.9% overall prevalence and 13.9% average yearly incidence. Our patients had similar LVEF and NYHA class as the SCD-HeFT patients but had significantly higher rates of ischemic cardiomyopathy (69.6% versus 52.0%), atrial fibrillation (34.8% versus 15.5%), diabetes (50.3% versus 21.7%), and higher mean creatinine levels (1.7 versus 1.1 mg/dl). The higher ICD therapy rates and incidence of comorbidities suggest significant clinical differences between our patients and those studied in the major clinical trials. The annual mortality seen in our patient population is quite striking: overall mortality in our patients was 26.6% with no difference observed in mortality between obese and non-obese patients. The 1–5 year cumulative mortality rates were 4.4%, 11.1%, 20.8%, 32.8% and 52.5% in obese patients and 3.3%, 9.5%, 22.2%, 30.8% and 40.7% in non-obese patients. We note the higher cumulative rates of death in obese patients, despite lower rates of appropriate ICD therapy with interest.
Previous studies have described the risk of increased mortality in patients with obesity,10 and the lack of mortality difference observed between obese and non-obese patients in this study is noted with interest. Wu et al11 reported no difference in mortality between obese and non-obese patients who developed heart failure due to acute coronary syndrome. In a meta-analysis that examined nine studies of heart failure patients and obesity, Oreopoulos et al12 reported lower all-cause and cardiovascular mortality. A possible explanation for what is explained as the “obesity paradox” includes the observance of a catabolic state in heart failure.12 Patients with more severe disease may develop loss of muscle, bone, and fat and may become underweight. Selection bias has also been suggested as a possible explanation.
Although the mechanism of obesity as a potentially protective factor in heart failure is not known, we suggest that it could be an explanation for the lack of difference in appropriate ICD therapy observed in obese patients in this study. Poole et al13 demonstrated significantly higher all-cause mortality in patients in the SCD-HeFT trial who received appropriate or inappropriate ICD therapy. Considering these findings as well as the previously observed protective effect of obesity on heart failure mortality,12 one could hypothesize that obese patients with heart failure would actually experience a lower incidence of ICD therapy. Regardless, while the “obesity paradox” in obese heart failure patients has been acknowledged in a number of studies, it must be viewed as a hypothesis to explain our findings. Health dangers of obesity are well-known, and the “obesity paradox” should not discourage heart failure patients from pursuing an ideal body weight.
Patients in this study had significantly lower risk of mortality when using β-blockers. This is consistent with numerous studies which have resulted in β-blocker therapy as a standard of care in patients with systolic heart failure.14,15 Despite advances in medical device therapy for improvement of functional status and prevention of sudden death in patients with heart failure, medical therapy must remain a primary objective.
Despite the interesting findings in this study, the authors acknowledge several limitations. This study was a retrospective analysis, and further prospective study is needed to evaluate obesity and ventricular arrhythmias. Details of ICD programming were not available in this patient population, and different VT or VF zone rates could have impacted detection of ventricular arrhythmias. Finally, it is also possible that some patients with ICD therapies could have sought care at a different medical center with a resulting difficulty in our ability to document these episodes.
In conclusion, obesity was not associated with appropriate ICD therapy or higher mortality in this study, but further study is needed in larger patient populations to confirm our findings. We hypothesize the lack of difference in ICD therapy incidence between groups could be related to the previously reported “obesity paradox” observed in obese patients with heart failure. Although our findings are consistent with previously reported lower or unchanged mortality in obese patients with heart failure, the mechanism of these findings is not known. Health dangers of obesity are well described,10 and clinicians should remain vigilant to encourage proper lifestyle modifications in obese patients.