Author + information
- Received February 1, 2001
- Revision received May 16, 2001
- Accepted June 4, 2001
- Published online September 1, 2001.
- Tamara B Horwich, MD∗,
- Gregg C Fonarow, MD, FACC†,* (, )
- Michele A Hamilton, MD, FACC†,
- W.Robb MacLellan, MD, FACC†,
- Mary A Woo, DNSc‡ and
- Jan H Tillisch, MD∗
- ↵*Reprint requests and correspondence: Dr. Gregg C. Fonarow, Ahmanson-UCLA Cardiomyopathy Center, Division of Cardiology, 47-123 CHS, 10833 Le Conte Avenue, Los Angeles, California 90095-1679
The study aimed to evaluate the role of obesity in the prognosis of patients with heart failure (HF).
Previous reports link obesity to the development of HF. However, the impact of obesity in patients with established HF has not been studied.
We analyzed 1,203 patients with advanced HF followed in a comprehensive HF management program. The patients were subclassified into categories of body mass index (BMI) defined as: underweight BMI <20.7 (n = 164), recommended BMI 20.7 to 27.7 (n = 692), overweight BMI 27.8 to 31 (n = 168) and obese BMI >31 (n = 179). This sample size allows the detection of small effects (0.02), with a power of 0.80 and an alpha level of 0.05 for comparing one-year survival between BMI groups.
The four BMI groups had similar profiles in terms of ejection fraction (mean 0.22), sodium, creatinine and smoking. The obese and overweight groups had significantly higher rates of hypertension and diabetes, as well as higher levels of cholesterol, triglycerides and low density lipoprotein cholesterol. The four BMI groups had similar survival rates. Ejection fraction, HF etiology and angiotensin-converting enzyme inhibitor use predicted survival on univariate analysis (p < 0.01), although BMI did not. On multivariate analysis, cardiopulmonary exercise tests, pulmonary capillary wedge pressure and serum sodium were strong predictors of survival (p < 0.05). Higher BMI was not a risk factor for increased mortality, but was associated with a trend toward improved survival.
In a large cohort of patients with advanced HF of multiple etiologies, obesity is not associated with increased mortality and may confer a more favorable prognosis. Further studies need to delineate whether weight loss promotion in medically optimized patients with HF is a worthwhile therapeutic goal.
Heart failure (HF) is an increasingly important cause of morbidity and mortality. The annual incidence of HF in the U.S. has recently been estimated at 465,000 and is steadily rising, despite advances in medical therapy. Five-year survival rates are near 50% (1). Obesity is regarded as a significant risk factor for cardiovascular disease and has also been linked to the development of HF (2,3). A body mass index (BMI) <25 is generally considered healthy, yet over half of U.S. adults have a BMI >25 and over one-third have a BMI >27 (4,5). Elevated BMI and obesity have been associated with the cardiovascular disease risk factors of hypertension, insulin resistance and dyslipidemia (6). Furthermore, deleterious hemodynamic and morphologic cardiovascular changes have been attributed to obesity, even in the absence of clinical cardiac disease. Obesity-associated alterations include left ventricular (LV) hypertrophy, diastolic dysfunction, increased LV end-diastolic dimension (LVEDD), increased circulating blood volume, increased cardiac output and a decreased stroke work to LV end-diastolic pressure index (7,8). There have been reports of HF occurring in morbidly obese individuals, with reversal of HF on weight loss, and several investigators have suggested the concept of an obesity-related cardiomyopathy (9–11).
In light of evidence that obesity is harmful to cardiovascular health, weight loss is often recommended as a therapeutic goal to patients with HF and excess body fat (12). However, no published study has addressed the question of obesity’s impact on survival in patients with established HF. As pharmacologic weight loss measures with potentially adverse effects become increasingly available and more widely utilized, obesity’s role in the prognosis of patients with HF needs to be clarified. To investigate this issue, we examined the relationship between obesity and mortality in a large cohort of patients with advanced HF of multiple etiologies.
A total of 1,734 patients were referred to a university medical center for heart transplant evaluation between 1983 and 1999. All subjects were followed in a comprehensive HF management program, as previously described (13). Medical record review was approved by the Medical Institutional Review Board. Patients with LV ejection fraction (LVEF) >40% (n = 106) and those without adequate height, weight and initial follow-up data (n = 425) were excluded from the analysis. The final study group consisted of 1,203 subjects (76.6% male, age range 16 to 82 years). New York Heart Association functional class was class II in 5.5% of patients, class III in 33.3% and class IV in 60.6%. Mean LVEF was 22%. Etiologies of HF were ischemic (48%), idiopathic (40%) and valvular (4.5%); the remaining 7.5% had alcohol-induced, hypertrophic and postpartum cardiomyopathy.
Body mass index—weight in kilograms divided by square height in meters—strongly correlates with total fat mass and was used as the primary index of obesity in our study (2). Percent ideal body weight (PIBW) was used as an alternate index of total body fat. For PIBW calculations, ideal body weight was considered 48.2 kg for men 152.4-cm (5 feet) tall and 45.5 kg for women 152.4-cm tall, with 2.3 kg added for each additional 2.5 cm of height (14). Height data were recorded at the time of initial referral or at subsequent clinic visits. To eliminate edematous weight as a confounder of accurate body weight measurements, BMI and PIBW calculations were made using the patients’ weights recorded at the time of hospital discharge, at conclusion of pulmonary artery (PA) catheter-guided therapy to achieve optimal hemodynamic variables and euvolemia. The hemodynamic variables employed in the analyses were likewise the optimal values recorded after PA catheter-tailored medical therapy, as these hemodynamic measurements have been shown to correlate best with survival (15). The medical treatments recorded were those implemented after baseline hemodynamic evaluation. Laboratory testing, echocardiography and cardiopulmonary exercise tests all occurred within three months of the initial referral; later values were excluded from our analysis. A history of hypertension, diabetes and hypercholesterolemia was determined from a review of medical records.
Death was the primary end point in this study. Deaths were classified as sudden death, HF death or death secondary to myocardial infarction. Death was considered sudden if it occurred out of the hospital within 15 min of the onset of unexpected symptoms or during sleep. Death during the hospital period for worsening congestive symptoms was considered as HF death. Urgent heart transplants (status I) were analyzed as HF deaths, under the assumption that these patients would have died without a transplant. Nonurgent transplants (status II) were considered as nonfatal at the end of follow-up.
Patients were classified into four BMI categories, based on the National Center of Health Statistics categories: underweight (BMI <20.7), recommended weight (BMI 20.7 to 27.7), overweight (BMI 27.8 to 31) and obese (BMI >31) (4). The PIBW was divided into five categories: <80%, 80% to 100%, 101% to 120%, 121% to 140% and >140%. We calculated proportions and mean values of baseline characteristics for the four BMI groups and five PIBW groups. Actuarial survival curves for the four BMI groups and five PIBW groups were calculated using the Kaplan-Meier estimate, and differences between curves were calculated using the log-rank statistic. Univariate survival analyses were performed with the likelihood ratio test, using the Cox model for baseline variables of BMI, PIBW, age, gender, etiology, medicine use, echocardiographic, exercise testing and electrocardiographic results, initial blood tests and hemodynamic variables, Multivariate analysis was performed by Cox proportional hazards regression analysis, using the SPSS version 10.0.5 statistical package, to estimate adjusted odds ratios and 95% confidence intervals for potential predictors of survival. The Cox model included variables found to be significant predictors of survival on univariate analysis and retained all independent variables with a p value <0.10. Power analysis for a logistic regression model with 15 covariates indicated that a sample of 1,200 subjects would allow the detection of small effects (0.02), with a power of 0.80 for comparing one-year survival between BMI groups, at an alpha level of 0.05. For five-year survival, this same sample size would allow the detection of moderate effects (0.15) between the BMI groups, with a power of 0.80 and an alpha level of 0.05.
Almost one-third of the patients were above the recommended BMI range, with the overweight and obese groups comprising 14.0% and 14.9% of the cohort, respectively. Fifty-nine patients had a BMI >35 and 15 patients had a BMI >40. The majority of patients (57.5%) were in the recommended BMI category, and 164 patients (13.6%) fell below the recommended weight. When patients were stratified by PIBW categories (<80%, 80% to 100%, 101% to 120%, 121% to 140% and >140%), the groups comprised 2.1%, 21.8%, 35.7%, 24.9% and 15.5% of the total cohort, respectively.
Tables 1, 2 and 3⇓⇓⇓show the baseline characteristics of the four BMI groups. Obese patients were more likely to have hypertension and diabetes, but less likely to be smokers. Patients in the overweight and obese categories had significantly larger LVEDDs, yet significantly lower LVEDD indexes. Peak oxygen consumption (V̇o2) was higher in overweight and obese subjects, yet lower when corrected for weight. Analysis of final hemodynamic variables after PA catheter-guided medical therapy revealed higher cardiac output, blood pressure (BP) and right atrial pressure in patients with a higher BMI, but no significant difference in final PA pressure, pulmonary capillary wedge pressure (PCWP) or cardiac index. Serum sodium and creatinine levels were similar in the four groups, but cholesterol, triglycerides and low density lipoprotein cholesterol levels were significantly higher and high density lipoprotein cholesterol was significantly lower in the obese and overweight patients. When patients were classified by PIBW, a similar distribution of characteristics was seen, with the exception of LVEF, which was significantly higher in the highest PIBW group (23.0 ± 7.0 in patients with PIBW >140% vs. 20.8 ± 6.8 in patients with PIBW 80% to 100%), although this difference did not reach significance (p = 0.072) when patients were stratified by BMI.
During the five years of follow-up (2,111 patient-years), 537 deaths occurred, with 350 occurring in the first year and 438 in the second year. Of the 537 total deaths, 143 deaths were sudden, 174 were progressive HF deaths, 6 were myocardial infarctions and 42 were from unknown or other causes. During the follow-up, 172 patients received urgent transplants. Figure 1shows actuarial survival curves for the four BMI categories. To assess the effect of obesity in particular subsets of patients with HF, we recalculated survival curves in clinically significant subgroups. Survival curves for BMI categories were similar (p > 0.10) in subgroups of men, women, patients with and without coronary artery disease, and patients dying suddenly. When patients with transplants were eliminated from the analysis, survival rates of the four BMI groups were still statistically similar. There were no significant differences in the types of death among the four BMI categories (p > 0.10).
On univariate analysis, LVEF, peak V̇o2, PCWP and angiotensin-converting enzyme (ACE) inhibitor use were predictors of mortality, whereas BMI and PIBW were not. Table 4shows the association between patient characteristics and one-year survival. Table 5shows the results of multivariate analysis adjusting for age, gender, hypertension, diabetes, LVEF, hemodynamic variables, peak V̇o2, mitral regurgitation, tricuspid regurgitation, medicine use and serum sodium, creatinine and lipid levels. Multivariate analysis confirms that obesity did not increase mortality and demonstrates that obesity was associated with a statistically significant survival benefit at one and two years, but not at five years of follow-up. The association between elevated BMI and a reduced risk of death was not solely a function of impaired survival in underweight, cachectic patients with HF. Elevated BMI remained an independent predictor of improved survival, even after the underweight patients were eliminated from the analysis. Figure 2shows five-year survival curves by BMI adjusted for baseline differences.
Our initial hypothesis was that obese patients with HF would have impaired survival, compared with patients of recommended weight. Obesity is associated with an altered hemodynamic profile and cardiovascular disease risk factors that could be expected to increase both the risk of developing HF and the risk of mortality in patients with established HF (6–8). In community cohort and insurance life-table analyses, overweight status and obesity are risk factors for cardiovascular and overall mortality at long-term follow-up (16). Yet, we observed that obesity was not a significant risk factor for five-year mortality in patients with advanced HF and that excess body weight may, in fact, confer survival benefit to patients with HF.
Despite the wide prevalence of both HF and obesity, few previous studies have addressed the role of obesity in HF survival. A study of prognostic variables in 401 patients with HF did not find overweight status to be a risk factor for mortality, despite inclusion of >40% overweight patients (BMI >26) (17). Similar findings were also seen in a recent preliminary report involving a retrospective analysis of 589 patients with HF. This study reported that survival was not impaired in obese patients with HF and that mildly obese patients had the most favorable prognosis (18). A similar relationship has been observed between obesity, essential hypertension and mortality. Overweight status was associated with decreased stroke risk and decreased total mortality, compared with lean subjects in the Systolic Hypertension in the Elderly Program study (19).
Hemodynamic alterations in obesity
Several explanations for obesity’s potentially protective role in HF merit discussion and further investigation. Despite similar PCWPs and cardiac indexes, overweight and obese patients had higher BPs. Improved BP tolerance to afterload-reducing agents may explain why a larger proportion of obese and overweight patients were on ACE inhibitors, which are known to prolong the lives of patients with advanced HF (20).
The interaction between obesity and serum lipid levels is also a potential explanation for the survival advantage seen in obese patients. Our data show a significant positive correlation between higher cholesterol levels and improved survival, and this trend has been noted in previous studies (21). Other investigators have postulated that serum lipoproteins play a beneficial role in HF through downregulation of inflammatory cytokines (22).
Neurohormonal alterations in obesity
Altered cytokine and neuroendocrine profiles of obese patients may play a role in modulating HF progression. Changes in the cytokine tumor necrosis factor-alpha (TNF-alpha) system are observed in obese patients. Adipose tissue produces soluble TNF-alpha receptors, resulting in higher circulating levels of both type I and II receptors in obese subjects (23). Soluble TNF-alpha receptors are also elevated in HF and may play a cardioprotective role, as they neutralize the biologic effects of TNF-alpha. Tumor necrosis factor-alpha is elevated in HF and may contribute to cardiac injury through its pro-apoptotic and negative inotropic effects (24).
Obesity has also been associated with alterations in the sympathetic nervous system and renin-angiotensin system. A recent study comparing exercise responses in obese and lean subjects found that the lean ones had significantly higher increases in plasma epinephrine and renin levels during treadmill testing, despite similar baseline levels and history of hypertension (25). Because heightened sympathetic and renin-angiotensin activity are associated with a poor HF prognosis (26), diminished stress responses of these neurohormonal systems may provide insight into the favorable HF prognosis seen in obesity.
It could be rationalized that obese patients present at an earlier, less severe stage of disease as a consequence of more prominent symptoms of dyspnea and greater functional impairment associated with excess body weight. However, LVEF and other measures of the severity of illness were not different in obese patients, and after adjustment in multivariate analysis, obesity was not associated with a worse prognosis. In addition, patients with a higher BMI were less likely to receive heart transplants. Because a similar percentage of obese patients and those of recommended weight were accepted for transplantation, the lower rate of transplants in obese patients was likely secondary to reduced availability of larger-sized hearts for transplant. In this study, 21.8% of obese patients received heart transplants, compared with 39.9% of those of recommended weight. This disparity would be expected to adversely effect observed survival in the obese group of patients with advanced HF.
Even if obesity is not associated with worse HF survival, weight loss may be desirable if it results in improved functional capacity and reduced symptoms. Furthermore, preoperative obesity may increase morbidity and mortality with heart transplantation (14). However, if obesity is protective, then weight loss attempts could be associated with increased mortality risk. Further studies are needed to examine the risks and benefits of weight loss in patients with HF.
Although our observations do not directly address the role of obesity in the development of HF, the distribution of obesity in our study is similar to that reported for the U.S. adult population (4). Selection bias in referral to a transplant center may account for the relatively normal weight distribution seen in this cohort of patients with advanced HF.
Our study has several strengths. Its large sample size provides adequate power to detect true differences in survival. The study involves a single center, allowing for accurate and thorough follow-up data. We have eliminated the potential confounding variable of edematous weight gain in our calculation of BMI by using weights recorded after therapy, aimed at hemodynamic optimization and euvolemia. Our data base includes numerous demographic, laboratory, echocardiographic and hemodynamic variables, permitting a detailed and adequately powered survival analysis.
The potential limitations of our study include its retrospective nature and our selected group of patients with advanced HF disease referred for transplant evaluation. The study took place over a period when HF treatments were changing, although the rates of obesity were evenly distributed over the study period. Data on the use of beta-blockers and doses of ACE inhibitors are not available. We also do not have data on cytokine or neurohormonal levels, or direct measures of percent body fat. Furthermore, we have not accounted for the body distribution of adipose tissue, which may confer differential risks of cardiovascular disease. An additional limitation is that BMI was assessed at a single point in time. Cachexia has been associated with an unfavorable prognosis when defined by weight loss rather than absolute weight, and change in weight over time was not evaluated in our study (27).
In this cohort of patients with advanced HF followed in a comprehensive HF center, elevated BMI was not associated with increased mortality. Furthermore, elevated BMI was an independent predictor of improved survival at one and two years. The present findings suggest that promotion of weight loss in patients with HF may not lower the mortality risk, and may even be potentially harmful. Further investigations into the interaction between obesity and progression of HF are needed.
☆ This study was supported by the Ahmanson Foundation, Los Angeles, California.
- angiotensin-converting enzyme
- body mass index
- blood pressure
- left ventricle, left ventricular
- heart failure
- left ventricular end-diastolic dimension
- left ventricular ejection fraction
- pulmonary artery
- pulmonary capillary wedge pressure
- percent ideal body weight
- tumor necrosis factor-alpha
- oxygen consumption
- Received February 1, 2001.
- Revision received May 16, 2001.
- Accepted June 4, 2001.
- American College of Cardiology
- Willet W.C,
- Dietz W.H,
- Colditz G.A
- Hubert H.B,
- Finleiv M,
- McNamara P.M,
- et al.
- VanItallie T.B
- Kuczmarski R.J,
- Flegal K.M
- Krauss R.M,
- Winston M
- Divitiis O,
- Fazio S,
- Maurizio P,
- et al.
- Eckel R.H
- Fonarow G.C,
- Stevenson L.W,
- Walden J.A,
- et al.
- Davos C.H,
- Doehner W,
- Rauchhaus M,
- et al.
- Fonarow G.C,
- Chelimsky-Fallick C,
- Stevenson L.W,
- et al.
- Feldman A.M,
- Combes A,
- Wagner D,
- et al.
- Weber M.A,
- Neutel J.M,
- Smith D.H.G