Obesity and heart failure

Obesity has reached epidemic proportions globally, with over 2.6 million people dying annually as a result of being overweight or obese [1, 2]. The ‘overweight’ and ‘obese’ states are defined in terms of body mass index (BMI), with overweight being 25–30 kg/m2, class I obesity 30–35 kg/m2, class II obesity 35–40 kg/m2, and class III obesity >40 kg/m2. The 2007–2008 National Health and Nutrition Examination Survey (NHANES) indicates 34.2 % of U.S. adults aged ≥20 years are overweight, 33.8 % are obese, and 5.7 % meet the criteria for class III obesity [3, 4]. Obesity is an important determinant of cardiovascular health and is associated with widespread alterations in cardiovascular structure and function [1, 3]. Elevated body mass index (BMI), waist circumference and waist-hip ratio have all shown association with incident heart failure (HF) in several cross-sectional and prospective studies [5, 6, 7•]. Amongst 59,178 adults followed for mean 18.4 years, the adjusted hazard ratios of HF at BMIs <25, 25–29.9, and ≥ 30 kg/m2 were 1.00, 1.25, and 1.99 (P < 0.001) for men and 1.00, 1.33, and 2.06 (P < 0.001) for women, respectively [8••]. Recent publications have also implicated possible etiologic mechanisms via adipokines, which are cell-signaling proteins secreted by adipose tissue [9]. However, there is a puzzling survival paradox in advanced systolic HF that raises challenging questions regarding weight management in obese HF patients.

Obese individuals without cardiac pathology have increased circulating volume, enhanced cardiac output, lower systemic vascular resistance, a greater degree of sympathetic activation, and a slightly higher resting heart rate than lean counterparts, but a lower maximal oxygen uptake (VO2max) [10]. Various pathophysiological mechanisms have been postulated to explain the relationship between obesity and incident HF, such as increased blood volume or chronically elevated intra-thoracic pressure leading to chamber dilatation, hypertension causing left ventricular (LV) hypertrophy, the presence of epicardial fat and myocardial fatty infiltration, or a direct cardiotoxic effect of adipose tissue mediated by hormones and inflammatory proteins. The pediatric literature provides some of the most compelling evidence for a direct associated between obesity and cardiac dysfunction. Obese children have a low prevalence of overt diabetes or obstructive coronary artery disease and, therefore, display a less confounded expression of obesity cardiomyopathy than obese adults. Adiposity beginning in childhood is a consistent predictor of LV mass in young adults, independent of the effect of elevated blood pressure [11]. The association of obesity per se, rather than lipid and glycemic abnormalities, with LV hypertrophy has also been demonstrated in the pediatric literature [12]. Obese children and adolescents show early diastolic dysfunction in comparison with normal-weight controls [13] . The Strong Heart Study built on this observation in obese adolescents, who had four-fold higher probability of carrying an LV mass exceeding values compensatory for their cardiac workload. This observation was associated with a lower ejection fraction, poorer myocardial contractility, and greater LV diastolic dysfunction [14]. Overt failure of systolic function in obese children not been reported, although LV circumferential strain appears to be abnormal as early as mean age of 14 years, independent of any other cardiovascular risk factors [15]. This supports the hypothesis that a long period of LV hypertrophy, diastolic dysfunction, and subclinical systolic dysfunction precede any clear reduction in left ventricular ejection fraction (LVEF), or onset of symptomatic, usually congestive, HF. The concentric LV hypertrophy and diastolic dysfunction seen in another cohort of 38 patients aged 13 to 19 years significantly improved at a mean of 10 ± 3 months after gastric bypass surgery, implying reversibility of these early obesity-induced cardiac abnormalities [16]. Improvements in baseline LV hypertrophy and adenosine-induced sub-endocardial ischemia were also observed by cardiac magnetic resonance imaging (MRI) in a small cohort of adolescents who underwent bariatric surgery [17].

In obese adults, the diagnosis of obesity-associated HF can be challenging. The cardinal symptoms of HF—dyspnea, fatigue, orthopnea, lower extremity edema, and increased abdominal girth—can all be caused by obesity in the absence of cardiac pathology. The clinical examination may be misleading in obese individuals, due to indistinct heart sounds and difficulties discerning the jugular venous pressure. The primary imaging tool for determining left ventricular dysfunction, the transthoracic echocardiogram, also has reduced sensitivity in this population, although the use of echocontrast agents can aid visualization of the endomyocardial borders in challenging cases [18]. Likewise, cardiac MRI, which is increasingly used to define cardiomyopathy etiology, is often inaccessible to obese patients who exceed equipment weight limits. Laboratory parameters are also affected by adiposity; brain natriuretic peptide (BNP) is commonly used as a diagnostic aid, but is negatively associated with BMI in both healthy subjects and patients with cardiovascular diseases [19]. Furthermore, excess fluid because of decompensated HF is indistinguishable from excess adiposity when calculating BMI; hence, BMI categorization and trends over time can misrepresent the degree of obesity in HF patients. The combination of these limitations may impact the accuracy of a HF diagnosis, both in research settings and in clinical practice, and also the clinician’s ability to accurately follow the obese HF patient’s response to interventions.

Adipokines, gut hormones, and cardiac dysfunction

Recent research has revealed the role of adipose tissue in mediating obesity-related metabolic and inflammatory abnormalities. Furthermore, the location of this adipose tissue has been shown to be key to its degree of metabolic activity, with central or visceral (as opposed to subcutaneous) adipose cells being particularly active. It has been recognized for some time that the relationship between excessive weight and an abnormal metabolic profile is not absolute; obesity does not always cause a clinical metabolic disease, and not all patients with insulin resistance, hypertension, and hypertriglyceridemia are obese. This concept was reinforced by the recent 6-year study of HF incidence in a population without diabetes or baseline macrovascular complications. The metabolic syndrome (for which central adiposity is a criteria) is a risk factor for incident HF and is a stronger predictor of HF incidence than BMI alone [20, 21]. Obese individuals without the metabolic syndrome actually displayed a lower incident HF rate compared with nonobese individuals with a metabolic syndrome diagnosis [22••]. Therefore, it may not be simply the presence of adiposity that confers risk, but rather the degree of metabolic dysfunction of that adipose tissue. Hence, the concept of ‘adiposopathy’ [23, 24] has developed, with interest currently focusing on several adipokines synthesized and released by this metabolically active tissue. Of particular relevance to the cardiovascular system are adiponectin, resistin, leptin, tumor necrosis factor alpha (TNF-α), interleukin-6 (IL-6), angiotensinogen, and plasminogen activator inhibitor-1 (PAI-1) [25].

Adipokines with positive or negative myocardial effects are summarized in Table 1. Adiponectin is produced by both adipocytes and myocytes and is markedly reduced in obesity [26]. The effects of adiponectin include promotion of insulin sensitivity, fatty acid breakdown, normal endothelial function, and weight loss. It is also anti-inflammatory and anti-atherogenic; circulating levels inversely correlate with LV mass [23]. Cardiac derived adiponectin and cardiac adiponectin receptor expression is significantly reduced in human cardiomyopathic hearts [27]. This myocardial adiponectin resistance likely explains the reported positive association between adiponectin levels and mortality in HF [28, 29] and the paradoxically elevated levels seen in the setting of greatest insulin resistance during acutely decompensated HF [30]. In a small cohort of left ventricular assist device (LVAD) patients, adiponectin receptor expression was suppressed in the failing myocardium compared with control subjects, and increased after mechanical unloading [31].

Table 1 Adipokines with proposed cardiovascular effects
Full size table

Resistin appears to have opposing effects to adiponectin, as it is proinflammatory, reduces insulin sensitivity, and has been associated with incident HF [32, 33]. Circulating resistin levels correlate with HF symptom severity and higher levels are predictive of rehospitalization or cardiac mortality in advanced HF [34]. There is also evidence for a direct cardiotoxic effect in animal studies, with adenovirus-mediated resistin overexpression in rat cardiomyocytes being negatively inotropic [35]. Leptin is a 167-amino acid protein that strongly stimulates satiety and promotes insulin sensitivity. Deficiency due to leptin gene mutations causes insulin resistance and obesity, but in obese states not associated with this rare gene mutation, circulating leptin levels are high but activity is reduced due to leptin resistance [36]. Leptin is structurally related to inflammatory cytokines and is increased in patients with HF, with particular elevation in the subset with exercise intolerance [37], and is predictive of incident HF [9]. Levels have been observed to fall with mechanical unloading of the failing heart [38]. Leptin receptor isoforms are expressed in myocardium and leptin demonstrates in vitro induction of myocyte hypertrophy [39, 40].

Adipocytes generate angiotensinogen, which has roles in vasoconstriction, endothelial dysfunction, and insulin resistance [41]. The inflammatory cytokines IL-6 and TNF-α, also secreted by adipose tissue, are increased in obesity. It is hypothesized that a state of low-grade inflammatory upregulation stimulated by release of these proinflammatory proteins from dysfunctional adipose tissue may mediate the relationship between visceral adiposity, insulin resistance and HF [42, 43].

Additional contributors to the relationship between obesity and cardiovascular dysfunction may be the gut hormones, as outlined in Table 2. Ghrelin is an appetite-stimulator produced preprandially by the stomach. In obese individuals, total circulating ghrelin levels are lower than normal-weight controls, and not suppressed by food intake; diet-induced weight loss restores ghrelin plasma levels [44]. Reduced circulating levels of ghrelin in obesity could contribute to myocardial dysfunction, because effects of this peptide include anti-inflammatory activity, peripheral vasodilation, enhanced cardiomyocyte contractility, and inhibition of myocyte apoptosis. In a rat model of HF, ghrelin administration improves LV dysfunction and attenuates cardiac cachexia development, possibly due to its promotion of growth hormone secretion [45]. Ghrelin has even been used experimentally in humans as a therapy for HF. Twelve patients received either a single infusion of human ghrelin (at a pharmacological level, 43 times baseline value) or placebo, with the ghrelin group demonstrating a significant increase in cardiac index within 60 minutes [46]. The same group performed a longer nonrandomized study of 10 HF patients who received intravenous ghrelin twice daily for 3 weeks, with significant enhancement of LVEF, increased LV mass, and decreased LV end-systolic volume, although the actual numeric change in LVEF was marginal at 27 % ± 2 % to 31 % ± 2 %, P < 0.05 [47]. Ghrelin has recently been implicated in an interesting relationship with BNP and appetite. Intravenous BNP administration reduced ghrelin levels and induced satiety in health men, compared with placebo controls [48]. BNP itself has a role in systemic metabolic pathways. BNP1–32 exerts strong lipolytic effects in humans, independent of cyclic adenosine monophosphate production, which can contribute to excessive fatty acid mobilization and cachexia in advanced HF [49].

Table 2 Gut hormones with proposed cardiovascular effects
Full size table

The incretins, including glucagon-like peptide 1 (GLP-1) and gastric inhibitory polypeptide (GIP), are a family of gut hormones that stimulate postprandial insulin release, inhibit glucagon, slow the rate of gastric empting, and promote weight loss. They are found at lower-than-normal levels in patients with T2DM, and obese subjects show a blunted postprandial GLP-1 response compared with lean subjects. GLP-1 analogs such as exenatide, and sitagliptin, a dipeptyl peptidase-4 inhibitor that deters the degradation of GLP-1 and GIP, are new pharmacological agents for T2DM and have attracted attention regarding potentially favorable cardiovascular effects. GLP-1 receptors are found on cardiomyocytes, and animal models of HF have shown positive responses to GLP-1. Improvements in LVEF were reported in ten patients who received a 72-hour GLP-1 infusion after successful primary angioplasty for myocardial infarction, in comparison with 11 controls [50]. The same group also studied 18 patients with NYHA class III–IV HF, of which 11 received a 10-week subcutaneous infusion with an escalating dose of GLP-1 and seven received a placebo infusion [51]. Significant improvements in the GLP-1 group were reported for LVEF, oxygen consumption, the 6-minute-walk test, quality of life, and NYHA class. However, neither of these pilot studies was randomized.

Therefore, it can be seen that although no single adipokine or gut hormone has strong evidence as the sole mediator of myocardial dysfunction in obesity, there are biologically plausible mechanisms linking adiposopathy to abnormalities in ventricular mass and contractility that could explain the development of obesity-associated diastolic or systolic HF.

The obesity survival paradox

Although obesity is a risk factor for incident HF, excess adiposity has been associated with more favorable survival rates in multiple systolic HF cohorts [43, 52, 53••, 54, 55]. Amongst 7,767 patients in the DIG trial, crude all-cause mortality rates decreased in a near linear fashion across successively higher BMI groups, from 45.0 % in the underweight group to 28.4 % in the obese group (P for trend <0.001) [56]. In multi-variable analysis, overweight and obese patients remained at lower risk of all-cause mortality: overweight hazard ratio [HR], 0.88; 95 % confidence interval [CI], 0.80–0.96, obese HR, 0.81; 95 % CI, 0.72–0.92) compared with normal weight subjects. Among 108,927 decompensated HF patients, each incremental 5 kg/m2 BMI increase was associated with 10 % lower in-hospital mortality [57]. This ‘obesity paradox’ persists with body anthropometric measurements of obesity, such as a waist circumference [58]. It is not universally seen, but in cohorts not demonstrating a survival paradox there is no significant negative effect of increasing BMI [59]. HFPEF patients demonstrate a similar obesity survival paradox [54, 60]. Survival studies to date for chronic HF patients demonstrate a progressively protective effect of increasing BMI, whereas a U-shaped relationship between obesity and mortality is seen in some CAD cohorts [61]. Interestingly, the HF obesity paradox seems to be strongest in patients with a nonischemic HF etiology [62].

The mechanisms behind the obesity paradox are still unclear. Several potential confounders have been proposed, including the tendency for smokers to have lower body weights and higher mortality rates. However, the HF obesity paradox has persisted in several studies that were risk-adjusted for smoking status [63]. The ‘healthy survivor effect’ could also be at play, meaning that the sickest obese individuals die before they have the opportunity to develop HF, therefore, giving the surviving obese HF patients a more favorable prognosis than would be expected. However, it also seems plausible that a degree of excess adiposity is genuinely beneficial to the HF patient because it affords some protection from the negative effects of cardiac cachexia. Malnutrition is a well-established negative prognostic factor in HF [64]; loss of more than 6 % of bodyweight during the study duration was an independent mortality predictor in the SOLVD and V-HeFT II trials [65]. It is possible that being overweight or mildly obese may improve a HF patient’s metabolic reserve when challenged by the catabolic cytokines, which provoke anorexia, and weight loss in advanced HF. In addition, the increased abundance of cardioprotective adipokines, such as adiponectin, may medicate a direct cardioprotective effect of adiposity. This presents a challenging clinical conundrum: to optimize individual patient outcomes, should we advise our overweight and obese HF patients to engage in intentional weight loss, or just to stay at their current weight?

Treatment of obesity in heart failure

There are three major weight loss strategies: diet and lifestyle, pharmacological, and surgical. Dietary and lifestyle choices must always be addressed, regardless of any additional therapies. However, clinically significant changes in dietary and exercise habits can be challenging for many overweight and obese patients to achieve and maintain. Although best practices should be followed to engage the patient in a healthy eating plan and exercise regimen appropriate to their HF status, large reductions in weight are often unattainable. A recent systematic review demonstrated no conclusive evidence for sustainable weight loss utilizing dietary and lifestyle options [66]. Pharmacologic options have evolved over the past 5 years, but most of the agents with current FDA approval are relatively contraindicated in HF patients. Four anorectic sympathomimetics remain on the US market for short-term use—phentermine, diethylpropion, benzphetamine, and phendimetrazine—but their tendency to raise blood pressure and heart rate make them undesirable for the HF population. Lorcaserin, a selective 5-HT2C receptor agonist, was approved in June 2012. It is intended to promote satiety in individuals with a BMI >30 kg/m2 or >27 kg/m2 plus at least one obesity-related comorbidity. It is has approximately 60-fold greater selectivity for the 5-HT2C-receptor (which produces appetite suppression) than for the 5-HT2B-receptor (associated with valvulopathy and pulmonary vascular remodeling); nevertheless, due to the possibility of 5-HT2B-receptor upregulation during HF, caution is advised when considering lorcaserin use in patients with HF [67]. An extended-release combination of phentermine and topiramate was approved in July 2012. Contraindications include unstable cardiac disease within the prior 6 months. An alternate medication option is orlistat, which is probably the only currently approved weight-loss drug with an acceptable safety profile for HF patients. Orlistat inhibits pancreatic lipases, and achieved a 2.9-kg (95 % CI 2.5–3.2) mean weight reduction beyond placebo at 1 year in a meta-analysis of 15 studies [68].

Bariatric surgery is now established as one of the most successful and long-lasting treatment strategies for obesity. Restrictive procedures such as gastric banding or sleeve gastrectomy promote satiety and encourage decreased food intake, whereas malabsorptive operations such as biliopancreatic diversion decrease nutrition absorption; Roux-en-Y gastric bypass (RYBG) combines restrictive and malabsorptive features. A 22,000-patient meta-analysis demonstrated that an average postbariatric surgery excess weight loss of 61 % was accompanied by significant improvements in type II diabetes mellitus, hypertension, dyslipidemia, and obstructive sleep apnea [69, 70]. Follow-up data from the nonrandomized Swedish Obese Subjects trial has recently demonstrated a 50 % cardiovascular morality reduction in individuals who received bariatric surgery, compared with the nonsurgical controls, at a median of 14.7 years [71•]. Dramatic changes in the key adipokines and gut hormones are seen in the weeks and months after RYGB. Circulating levels of CRP and IL-6 fall in parallel with the improved insulin sensitivity that emerges almost immediately after gastric bypass or sleeve gastrectomy [72••, 73]. These inflammatory and glycemic changes long precede the nadir of weight loss. Both RYGB and sleeve gastrectomy are also associated with reductions in circulating leptin concentrations by almost half as early as 1 week postoperatively, with ongoing decreases until 12 months postoperatively. Adiponectin progressively rises over this time frame [74, 75]. The sharp increases in postprandial GLP-1 and peptide YY (PYY) levels that occur post-RYBG also precede significant weight loss and are independent of caloric restriction [45, 72••, 74, 76–78]. Circulating resistin falls after gastric bypass or sleeve gastrectomy [79] but ghrelin responses to RYGB have been more heterogeneous [80]. It is widely postulated that the changes in inflammatory and gut hormone profiles that occur after gastric bypass are responsible for the subsequent improvement in insulin sensitivity and marked weight loss that follow. It seems reasonable to postulate that any positive changes in cardiac structure and function occurring after bariatric surgery could also be associated with one or more of these biochemical responses.

Pre- and postbariatric surgery imaging of obese patients without HF has demonstrated reductions in ventricular wall mass and normalization of LV geometry at time points ranging from 3 months to 3.6 years postoperatively [69, 72••, 74, 76, 78]. These structural improvements may be independent of blood pressure changes [81]. The reduction in LV mass with bariatric weight loss has been associated with reduced circulating leptin concentrations in both animal models and humans [82–85]. Regression of LV hypertrophy also correlates with improvements in diastolic function after bariatric surgery [86]. Echocardiographic imaging for strain and strain rate analysis has shown improvements in subclinical abnormalities of myocardial deformability after bariatric surgery [87].

Bariatric surgery has also been performed for obese individuals with established systolic HF. Case reports of HF recovery after bariatric weight loss have been dramatic and compelling [88–91]. However, these reports describe extremely obese, relatively young individuals, with advanced systolic HF; the results may not be generalizable to the entire obese HF population. These case studies nicely illustrate the two main clinical endpoints for obese advanced HF patients who undergo weight-loss surgery: either the weight loss achieves significant clinical improvements in their HF status, or the BMI falls below the threshold where listing for cardiac transplantation is permissible—most transplant programs will not list patients with BMI >25 kg/m2. Another option may be the concurrent open placement of an adjustable gastric band and a left ventricular assist device. Two young, morbidly obese patients with advanced systolic HF reportedly had good clinical outcomes with this “VAD-BAND” approach [92].

Further evidence of the potential for bariatric surgery to reverse HF in obese patients is limited to three small published cohorts. The first is a prospective analysis of fractional shortening performed pre- and postvertical band gastroplasty that incorporated 13 subjects with low preoperative systolic function [93]. There were modest improvements in fractional shortening (22 % ± 2 % to 31 % ± 2 % P < 0.01) at a mean of 4.3 months after weight loss plateaued, accompanied by reductions in LV end-diastolic diameter and mean arterial blood pressure. The same group published a study of fractional shortening pre-and postvertical band gastroplasty in 14 subjects with clinical diagnoses of HF and an average fractional shorting that lay just below the lower limit of normal [94]. This cohort showed improvements in NYHA functional class, but no statistically significant improvements in systolic function. However, these postoperative echocardiograms occurred at only 4.5 ± 1.2 months postoperatively, and the older procedure of vertical band gastroplasty is not associated with the same degree of metabolic recovery as malabsorptive bariatric surgery.

An overlapping cohort of HF patients that underwent bariatric surgery was reported on by McCloskey and Ramani [95, 96]. Ramani et al. matched 12 patients with a mean age of 41 years, BMI 53 kg/m2, LVEF 22 % ± 7 %, to 10 nonsurgical controls. At 1 year, hospital readmission in bariatric patients was significantly lower than controls (0.4 ± 0.8 vs. 2.5 ± 2.6, P = 0.04). There was a significant improvement in mean LVEF for the bariatric group (35 % ± 15 %, P = 0.005), but not for controls, and the NYHA class improved in bariatric patients (2.3 ± 0.5, P = 0.02), but deteriorated in controls. The third cohort was a subset of nine patients with LVEF ≤ 50 %, within a 57-patient cohort of obese subjects with mean BMI 49 kg/m2. Although there did appear to be a trend towards increased LVEF in these 9 patients (LVEF 44.8 ± 7 to 59.5 ± 10.1), there was a similar rise in mean LVEF in the nonsurgical controls with initial LVEF ≤ 50 % (44.9 ± 7.9 to 58.6 ± 14.1) [81]. Existing small case series do however indicate that bariatric procedures are safe for clinically optimized HF patients with experienced surgical teams [97].

Conclusions

Obesity is a well-established risk factor for incident heart failure (HF). Clinical HF associated with obesity may be predominantly systolic or diastolic. It is now recognized that a spectrum of subclinical cardiac changes occur in obese individuals, including increased left and right ventricular mass, increased myocardial fatty infiltration and fibrosis, decreased efficiency of myocardial metabolism and subclinical systolic dysfunction. It appears that in some obese individuals these changes may progress to overtly symptomatic systolic HF. The association between obesity and incident HF is supported by myocardial signaling pathways mediated by adipokines and gut hormones. Bariatric procedures have been safely performed in medically optimized obese patients with HF. Patients with overt systolic HF and severe obesity have been shown to experience improvements in ejection fraction and functional status post bariatric surgery. However, although excess adiposity is a risk factor for incident HF, elevated body mass index may actually be associated with lower mortality rates in advanced HF. These seemingly contradictory observations have provoked clinical uncertainty as to the most evidence-based strategy for managing obesity in the advanced HF population. As described in this review, there is evidence to suggest clinical benefits from bariatric surgery for individuals with BMI ≥ 40 kg/m2 aged <50 years with severely depressed systolic function and NYHA class III-IV symptoms. Overweight and mildly obese HF patients (25–35 kg/m2) may be somewhat protected from cardiac cachexia and weight loss is not expected to enhance survival, but a reduction in BMI may offer symptomatic benefits.