Obesity and heart failure with preserved ejection fraction: new insights and pathophysiological targets

Abstract

Obesity and heart failure with preserved ejection fraction (HFpEF) represent two intermingling epidemics driving perhaps the greatest unmet health problem in cardiovascular medicine in the 21st century. Many patients with HFpEF are either overweight or obese, and recent data have shown that increased body fat and its attendant metabolic sequelae have widespread, protean effects systemically and on the cardiovascular system leading to symptomatic HFpEF. The paucity of effective therapies in HFpEF underscores the importance of understanding the distinct pathophysiological mechanisms of obese HFpEF to develop novel therapies. In this review, we summarize the current understanding of the cardiovascular and non-cardiovascular features of the obese phenotype of HFpEF, how increased adiposity might pathophysiologically contribute to the phenotype, and how these processes might be targeted therapeutically.

This article is part of the Spotlight Issue on heart failure.

1. Introduction

Heart failure (HF) is an enormous public health problem, with worldwide prevalence of over 64 million.¹ In the USA alone, over 6 million adults have HF, and by 2030 this number is projected to swell to 8 million,² with similar trends evident worldwide. Over half of patients with HF have preserved ejection fraction (HFpEF), and the prevalence is increasing relative to HF with reduced ejection fraction.^3,4 The epidemic of HFpEF is related to increases in mean age and rising trends in the prevalence of hypertension, metabolic syndrome, diabetes, and atrial fibrillation; all comorbidities that are intimately tied to the development of excess body fat.⁵

Obesity has doubled in more than 70 countries since 1980. Prevalence in the USA now exceeds 40%, a staggering increase from 15% as recently as the 1970s,⁵ and it is now projected that 1 in 2 adults in the USA will be obese by 2030.⁶ Epidemiological studies have identified a strong, independent relationship between obesity and development of HFpEF.^7–10 In contrast to atherosclerotic vascular diseases, the link between obesity and HF is not explainable by traditional cardiovascular risk factors.¹¹ Despite increasing recognition, HFpEF remains too often neglected as a cardiovascular complication among patients with excess body fat, leading to underdiagnosis of obesity-related HFpEF, and therefore, undertreatment.¹²

New data have changed our thinking about the contribution of obesity to HFpEF—from being a mechanical cause of dyspnoea, to a comorbid bystander, to the primary, direct cause of HFpEF, mediated via multiple deleterious effects on myocardial structure, function, and metabolism, as well as effects on lung, skeletal muscle, kidney, and liver, collectively mediated by systemic inflammation, neurohormonal activation, autonomic dys-regulation, and altered haemodynamic load.^13–17

Patients with obese HFpEF are often excluded from clinical trials, either directly because of restrictive body mass index (BMI) eligibility cutoffs, or indirectly through the requirement for elevated plasma natriuretic peptide (NP) levels, which are characteristically depressed in obesity.^{12,15,17–19} The distinct characteristics of obese HFpEF suggest that it may require unique therapies that could differ from non-obese patients with HFpEF. In this review, we summarize the current understanding of how increased adipose tissue pathophysiologically contributes to HFpEF, and consider specific treatment approaches that merit study in future trials.

2. Biology of excess body fat

While BMI is traditionally used to define obesity (≥30 kg/m²), a high BMI may not correspond to the same degree of body fat, or cardiovascular risk, in different individuals. Indeed, beyond quantity of excess fat, the distribution and quality of the excess fat also influences cardiovascular risk. Excess body fat is always the result of positive energy balance. The ability of mammals to store energy as triglycerides in adipocytes provides an extremely efficient means of retaining large amounts of energy that serves as fuel. However, excess body fat, and especially upper body (central) obesity, is associated with greater risk for dyslipidaemia, hypertension, Type 2 diabetes, and of course, HFpEF.^7–10 Many investigators have found that visceral adipose tissue (VAT) is an even better predictor of abnormal metabolic profiles than upper body subcutaneous fat, leading to the hypothesis that VAT is directly causing insulin resistance, inflammation, accelerated senescence, and mitochondrial dysfunction.^20–23

Gains in body fat occur by adipocyte expansion (hypertrophy) and/or the generation of new adipocytes (hyperplasia), even in adults (Figure 1). These processes vary from person to person and depot to depot.²⁴ Mean adipocyte size increases as a function of obesity, up to a point.²⁵ The balance between hypertrophy and hyperplasia appears linked to the metabolic consequences of obesity; enlarged abdominal subcutaneous adipocytes are associated with an increased cardiometabolic risk.²⁶

Figure 1.

Patterns of fat gain through hypertrophy and hyperplasia in obesity, and association with free fatty acid (FFA) release. See text for details.

2.1. Distribution of body fat: regional fat depots

The major fat depots have heterogenous and distinct characteristics. Upper body subcutaneous fat, which includes superficial and deep truncal depots,^27,28 is usually the largest. Lower body subcutaneous fat includes gluteal, femoral, and calf adipose tissue, i.e. all adipose tissue caudal to the inguinal ligament anteriorly and the ileac crest posteriorly. Intra-abdominal fat includes mesenteric and omental (visceral) depots, both of which drain into the portal vein; perinephric fat is intra-abdominal, but because it drains into the systemic circulation, it is not considered visceral. Adipose depots differ in terms of the fat cell size,²⁹ responsiveness to pro- and antilipolytic hormones,³⁰ as well as the ability to store fatty acids in triglycerides.^31,32

2.2. Adipocyte hypertrophy vs. hyperplasia

In 2000, it was posited that the adipocyte hypertrophy in abdominal subcutaneous fat in those predisposed to Type 2 diabetes was due to a failure of preadipocytes to differentiate as needed to store extra body fat.³³ This is important because hypertrophic adipocytes are insulin resistant and have greater rates of lipolysis. Figure 1 depicts the relative extremes of potential fat gain, ranging from a fat storage solely into hypertrophic subcutaneous adipocytes (lower right), with substantial gain of VAT, to excess fat storage solely from hyperplasia, with maintenance of normal subcutaneous fat cell size and function (lower left) and little gain of VAT.

Much has been learned about the process of the creation of new adipocytes in humans.^34,35 Peroxisome proliferator-activated receptor gamma (PPARϒ) is the master regulator of adipogenesis and required for adipocyte differentiation/survival; bone morphogenic protein-4 (BMP-4) promotes processes that activates PPARϒ, allowing the creation of new adipocytes from adipose mesenchymal stem cells. However, this process of adipogenic commitment and early differentiation is regulated by both pro- and anti-adipogenic pathways. The pro-adipogenic signals involve secreted proteins such as Dickkopf 1 and the secreted inhibitory protein gremlin-1 (GREM-1), which inhibits BMP-4 signalling. Humans with hypertrophic obesity appear to produce excess GREM-1, which suppresses adipogenesis, thereby forcing adipose lipid accumulation via hypertrophy.^34,35 This may be due to epigenetic and genetic factors that regulate adipogenesis. This ability or inability may then determine the relative gain in VAT. Inherited or acquired defects in these processes may explain the tendency of some people to gain body fat predominantly by adipocyte hypertrophy rather than hyperplasia. Evidence for this theory comes from the study of volunteers who were overfed in order to study adipose responses to fat gain.³⁶ Adults who created larger numbers of new leg adipocytes had less gain of VAT than those who gained fat by increasing abdominal subcutaneous adipocyte size.

2.3. Functions of adipose tissue

The major functions of adipose tissue include the storage and release of fatty acids. Adipose tissue stores the circulating triglyceride fatty acids and free fatty acids (FFAs) that are not used by lean tissue for essential processes. Healthy adipose tissue releases FFA into the circulation at rates tightly linked to energy needs.³⁷ Increases in upper body/visceral fat in obesity is associated with excess release of FFA into the circulation,³⁸ and this may have important and highly relevant deleterious effects on the heart in obese HFpEF. In contrast, humans who preferentially gain fat in their lower body have a lesser risk of metabolic abnormalities,³⁹ and more normal FFA metabolism.³⁸

In humans, insulin-stimulated glucose disposal is strongly correlated with insulin-suppressed FFA concentrations/flux,^40,41 and nadir FFA concentrations in response to insulin independently predict insulin sensitivity.⁴² Because skeletal muscle is the dominant site of glucose disposal in humans,⁴³ muscle insulin resistance with respect to glucose uptake is a major metabolic defect in obesity. Mechanistically, excess FFA can create insulin resistance affecting glucose metabolism in skeletal muscle and liver, as well as stimulate hepatic very low–density lipoprotein-triglyceride production. Artificially elevating FFA in healthy adults induces insulin resistance,⁴⁴ while obese mice that are haploinsufficient for hormone-sensitive lipase in adipose tissue have reduced lipolysis.⁴⁵ Elevated FFA can decrease tissue glucose metabolism by forcing the preferential oxidation of fatty acids (the Randle cycle),⁴⁶ and can impair insulin signalling via lipid intermediates.⁴⁷

Endocrine properties of visceral fat may directly contribute to metabolic health and cardiac structure/function. Cytokine concentration gradients have been demonstrated from portal venous to arterial blood, suggesting release from visceral fat,⁴⁸ whereas others have measured cytokine concentrations in the venous effluent of abdominal subcutaneous fat.⁴⁹

Humans with defects in the ability of adipocyte precursors to proliferate and differentiate in response to the need for more fat storage develop dysfunctional, hypertrophic subcutaneous adipocytes, which expose peripheral tissues to excess FFA.⁵⁰ Over time, excess FFA can drive the storage of fat in ectopic depots such as visceral fat, pericardial fat, skeletal muscle, liver, and intramyocellular triglycerides.⁵¹ We suggest that the most likely cause of excess visceral fat gain is the forced storage of fatty acids, possibly from excess fasting and postprandial FFA release, as a result of insulin resistant subcutaneous fat. Humans with greater amounts of visceral fat are less, not more efficient in storing fatty acids. The greatest efficiency of fatty acid storage in omental fat is seen in lean adults with the smallest visceral fat depots.^32,52 The efficiency of fatty acid storage in omental fat decreases as a function of visceral fat mass and there is down-regulation of proteins and enzymes responsible for fatty acid transport into the cells in those with visceral obesity.⁵² This finding contrasts with the tendency of fat storage (and adipocyte fatty acid storage factors) to remain stable (abdominal fat) or actually increase (femoral fat) as a function of depot size.³²

In order for VAT to continue to increase in mass despite down-regulation of fatty acid storage pathways, there must be an even greater down-regulation of lipolysis pathways. Because visceral obesity is associated with failure to normally suppress lipolysis after meals^53,54 as well as higher postprandial chylomicron concentrations,⁵⁴ it is possible that the higher FFA and chylomicron-triglyceride concentrations drive fatty acids into ectopic depots, including visceral fat, and potentially epicardial fat, despite lesser expression of the proteins and enzymes that facilitate fat storage, and this may importantly contribute to development of HFpEF through abnormalities in cardiac and non-cardiac structure and function.

3. Diagnostic challenges of HFpEF in people with obesity

Obesity is an important cause of HFpEF but can also be associated with symptoms of dyspnoea that are not caused by cardiac dysfunction. Diagnosis of HFpEF requires objective demonstration of congestion, and this is most commonly evaluated by physician examination, echocardiography, and NP testing.^55,56 Each of these are more challenging in patients with obesity. Physical signs such as jugular distention may be masked by increased neck girth. NP levels are well known to be lower and often ‘normal’ (lower than traditional diagnostic thresholds) in patients with HFpEF who are obese,¹⁹ and recent data have revealed that even echo-Doppler indicators such as the E/e′ ratio underestimate circulatory congestion in patients with obesity.¹² Several factors appear to contribute to lower NP levels in obese HFpEF, including increased external constraint on the heart, mitigating increases in wall stress.^12,15 Sex hormones may also contribute. When compared with premenopausal women, postmenopausal women, and men display increasing testosterone levels, which are associated with progressively lower NP levels,⁵⁷ further masking the presence of circulatory congestion. The lower NP levels in obese HFpEF also raise the possibility that many of these patients may suffer from an ‘NP deficiency’. Invasive haemodynamic exercise testing is frequently necessary in equivocal cases to determine whether HFpEF is present or absent.^55,56

4. Diabetes, obesity, and HFpEF

As described, Type 2 diabetes and metabolic syndrome are causally tied to excess body fat, blurring the lines between these conditions such that many refer to the common syndrome of diabesity.⁵ Approximately 40–50% of patients with HFpEF have diabetes, and another 15–20% have prediabetes.^58,59 When compared with those without diabetes, patients with HFpEF and diabetes have poorer patient-reported quality of life, worse exercise capacity, increased markers of inflammation, fibrosis, and endothelial dysfunction, worse congestion, worse renal function, higher left ventricular (LV) filling pressures, and increased mortality and HF admissions.⁵⁸

5. Organ-level pathophysiology of obese HFpEF

Even among patients without HF, obesity is associated with preclinical alterations that are known to lead to HFpEF, including LV hypertrophic remodelling, increases in chamber stiffness, prolonged myocardial relaxation, altered substrate utilization, reduced myocardial ATP availability, and mild systolic ventricular dysfunction.^60–71 Obesity is also associated with volume expansion, neurohormonal activation, systemic inflammation, and dysfunction of multiple extra-cardiac organ systems of relevance to HFpEF, including skeletal muscle, liver, and kidney.

5.1. Ventricular remodelling and dysfunction

In the absence of HF, the higher cardiac output and total blood volume (TBV) resulting from increased fat mass that accompanies obesity leads to LV cavity dilatation and subsequent hypertrophy secondary to increased wall stress.^72,73 Although this accounts for an eccentric hypertrophic pattern, it is evident that concentric remodelling occurs commonly in obesity.^73–75 Concentric remodelling appears to be more strongly related to insulin resistance,⁷⁶ diabetes,⁷⁶ hyperleptinaemia,⁷⁷ myocardial steatosis,⁷⁸ as well as VAT expansion.^68,73,79 Patients with obese HFpEF have more LV concentric remodelling than those with non-obese HFpEF, with increased LV mass-to-volume ratio.¹⁵ Similar to the LV, obesity is associated with right ventricular (RV) remodelling in patients with and without HFpEF.^15,77,80

LV diastolic function assessed using echocardiography is impaired in patients with obesity,^60,68 and weight gain is strongly correlated with increasing diastolic chamber stiffness over time (Figure 2).⁶² These effects are even more pronounced with increases in central adiposity.^62,66 Despite preserved EF, patients with HFpEF display subtle impairment in LV contractility at rest and severe impairments in systolic reserve with exercise.^81–83 Impaired systolic reserve worsens diastolic reserve, since the failure to contract to a smaller end systolic volume reduces elastic recoil at the onset of diastole, further worsening the elevation in pulmonary capillary pressures (Figure 2). Community-based studies have reported that LV systolic mechanics decrease as BMI, waist circumference, and waist-to-hip ratio increase.^63,65,84 Interestingly, calcium-activated force in skinned cardiomyocytes from RV septal endomyocardial biopsies was recently shown to be substantially depressed in patients with obesity-related HFpEF, with a direct correlation between sarcomere dysfunction and fat mass estimated by BMI (Figure 2).⁸⁵

Figure 2.

(A) Increasing weight gain is associated with progressively greater increases in estimated left ventricular (LV) end diastolic chamber stiffness (ΔLV Eed) over time in community-dwelling adults, predisposing to HFpEF. (B) Impairments in the ability to enhance LV early diastolic relaxation velocity (ΔLV e′) with exercise are associated with greater increases in pulmonary capillary wedge pressure (PCWP), and enhancement in LV e′ with exercise is substantially blunted in obese HFpEF (C). The ability to augment LV e′ is strongly correlated with enhancement in LV systolic function (LV s′), demonstrating the cross talk between systolic and diastolic reserve (D). When compared with cardiomyocytes sampled from healthy hearts and hypertensive HFpEF, individuals with obese HFpEF display reduced Ca²⁺ activated maximal tension development (T_max) at the level of the sarcomere, which is inversely related to body mass index (BMI) (E and F). Panel A adapted from Wohlfahrt et al.,⁶² Panels B–D adapted from Borlaug et al.,⁸³ and Panels E and F are from Aslam et al.⁸⁵

5.2. Atrial pathophysiology

Patients with HFpEF display left atrial (LA) remodelling and dysfunction that leads to increased risk for atrial fibrillation, a triad referred to as LA myopathy.^86–88 Obesity is also associated with greater risk of atrial remodelling⁸⁹ and atrial fibrillation.⁹⁰ However, in studies comparing patients with obese HFpEF to those with non-obese HFpEF, atrial remodelling is less marked, and atrial fibrillation burden is less in those with obese HFpEF.^15,91 This may relate in part to the fact that the latter patients with obese HFpEF are roughly a decade younger than those with non-obese HFpEF. Mean BMI decreases sequentially from patients with HFpEF and no AF to those with paroxysmal AF, to those with permanent AF, though all groups display higher BMI than controls.⁸⁶ Thus, while LA myopathy is an important pathophysiological contributor in obese HFpEF, its role may be less pivotal in this group than among non-obese HFpEF.

5.3. Epicardial adipose and pericardial restraint

HFpEF is associated with increases in total heart volume, and patients with obese HFpEF display even greater cardiomegaly due to chamber dilatation, increased wall thickness, and greater increases in epicardial adipose tissue (EAT; Figure 3).^15,92,93 This increases coupling between the heart and pericardium, amplifying ventricular interdependence, such that intracavitary pressures are higher for any net distending pressure.⁹⁴ This increase in external constraint on the heart at least partially explains why NP levels and echo-Doppler estimates of filling pressures are much lower in patients with HFpEF with obesity than those without obesity (Figure 3).¹²

Figure 3.

(A) Gross pathology showing heart from an 82-year-old woman with HFpEF at autopsy revealing marked increase in epicardial fat. (B) When compared with healthy controls and non-obese HFpEF, patients with obese HFpEF display increased epicardial fat thickness, which is associated with flattening of the interventricular septum (C), increased ventricular interdependence reflected by higher LV eccentricity index (D), and higher pulmonary capillary wedge pressures (PCWP) (E). The increase in external constraint on the heart in patients with obesity results in a higher PCWP for any LV transmural distending pressure (LVTMP) (F). See text for details. Panel A reproduced from Borlaug and Maleszewski,⁹³ Panels B–D from Obokata et al.,¹⁵ Panel E from Koepp et al.,⁹² and Panel F from Obokata et al.¹²

In addition to mechanical effects, EAT may directly influence myocardial structure and function through secretion of soluble factors.^95–98 Epicardial adipose (like VAT) has been shown to release proinflammatory cytokines, such as interleukin (IL)-1β, IL-6, and tumour necrosis factor (TNF)-α, which may directly cause remodelling and dysfunction in both cardiac and skeletal muscle through mitochondrial dysfunction, capillary rarefaction, or other processes.⁹⁹ In a study where EAT samples were harvested from patients at cardiac surgery, incubating rat atria with the EAT secretome from patients induced fibrosis, coupled with increased secretion of the adipo-fibrokine activin A.⁹⁷

The importance of EAT has been emphasized by recent studies showing that increases in EAT in HFpEF are associated with more severe haemodynamic perturbations, greater abnormalities in RV–pulmonary artery (PA) coupling, more LV fibrosis, more severely impaired exercise capacity, and increased risk of HF hospitalization or death.^92,100–104 Interestingly, some of the same studies have shown thinning of EAT in HF with reduced ejection fraction (as opposed to thickening in HFpEF), and an association of EAT thinning with worse exercise capacity and outcomes in HF with reduced ejection fraction.^103,104 These contrasting associations of EAT with cardiac function and outcomes in HF with preserved vs. reduced ejection fraction may reflect divergent quality and/or quantity of EAT in the types of HF [proinflammatory in HFpEF, depleted in HF with reduced ejection fraction (HFrEF)].

5.4. Cardiac metabolism and substrate use

In the normal heart under fasting conditions, the majority (70–90%) of ATP derives from FFA (Figure 4), with a lesser (10–30%) contribution from glucose, and a much smaller proportion from ketones and lactate (<1%).^5,105,106 In the setting of insulin resistance, like obesity, the heart is exposed to increased circulating FFA by increasing PPARα. As a result, initially a compensatory increase in myocardial FFA uptake and oxidation is driven by increased PPARα expression.¹⁰⁷ However, despite this initial protective mechanism, cardiac lipotoxicity occurs. With an increasing intracellular pool of long-chain fatty acyl-CoA, which accompanies increases FFA uptake, comes a larger pool of fatty acid substrate for non-oxidative processes, including the synthesis of triacylglycerol, diacylglycerol, and ceramides.¹⁰⁸ In a murine model of HFpEF, cardiomyocyte steatosis is coupled with increases in the activity of the transcription factor Forkhead box protein O1,¹⁰⁹ a key transcription factor involved in cell metabolism promoting abnormal unfolded protein response, a recently demonstrated cellular mechanism of HFpEF.¹¹⁰

Figure 4.

(A) Normal cardiac metabolism where substrate selection produces ∼30% glucose oxidation and ∼70% fatty acid oxidation providing nicotinamide adenine dinucleotide and flavin adenine dinucleotide (NADH/FADH2) to the electron transport chain for adenosine triphosphate (ATP) synthesis, and a normal phosphocreatine (PCR) to ATP ratio. (B) Cardiac metabolism in obesity where the balance of substrate selection is towards overreliance on fatty acid oxidation, with decreased glucose uptake (from insulin resistance) and oxidation due to pyruvate dehydrogenase (PDH) inhibition. The increased FADH2/NADH ratio is highlighted with additional uncoupling protein expression (UCP), leading to reduced PCr/ATP ratio. Increased fatty acid flux may also increase production of reactive oxygen species and non-oxidative metabolism of fatty acids resulting in lipotoxicity.

Whether significant lipotoxicity occurs in humans is less clear but increased levels of intramyocardial lipid has been shown in patients with obesity and diabetes,¹¹¹ with the degree of steatosis related to concentric LV remodelling and diastolic impairment.¹¹² A recent study has further showed increased myocardial fat in human HFpEF.¹¹³ This would imply that ectopic myocardial triglyceride plays at least some role in cardiac dysfunction and remodelling in humans.

While no study has yet evaluated cardiac metabolism specifically in obesity-related HFpEF, the available evidence suggests that FFA use is likely increased in this disorder (Figure 4). For example, in obesity without HF,^71,114,115 as well as conditions associated with HFpEF such as diabetes^116,117 and LV hypertrophy due to aortic stenosis,¹¹⁸ there is increased cardiac FFA uptake and oxidation. Increases in circulating FFA levels, uptake, and flux due to excessive lipolysis in obesity have deleterious effects on metabolism,^38,53 especially in the heart.^106,117 In obese women without HF, higher BMI is correlated with increased myocardial FFA uptake and reduced efficiency.⁷¹ Higher FFA use is associated with decreased glucose oxidation, which is the more efficient fuel source (higher P/O ratio, or ATP produced relative to O₂ consumed) when compared with fat.^5,119 The importance of the increase in FFA metabolism lies in the fact that only 50% of the reducing equivalents produced in the process of β-oxidation are able to donate electrons at Complex I of the electron transport chain, whereas the remaining half are donated by FADH2 at the flavoprotein site further downstream at Complex II, resulting in this relative O₂ inefficiency (Figure 4). Elevations in FFA levels also increase mitochondrial uncoupling protein 3 expression, leading to reduced ATP production,¹²⁰ with impaired excitation-contraction coupling¹²¹ and reduced myocardial glucose uptake.¹²² The increase in FFA flux and myocardial VO₂ in obesity-related HFpEF may also promote generation of reactive oxygen and lipid species that further impair myocyte function.^123,124

The surfeit of FFA in obese HFpEF may play a fundamental role in causing myocardial dysfunction. In obese patients without HF, imaging studies have shown that weight loss is associated with a decrease in cardiac FFA utilization,^125–127 and with a reduction in myocardial steatosis.¹⁰⁷ A small study (n = 13) reported that myocardial glucose oxidation was improved following weight loss,¹²⁸ and another found increased insulin-stimulated glucose uptake,¹²⁷ but neither study was performed in patients with HFpEF. There may be important differences in HFpEF, as recent studies have shown that myocardial ischaemia and injury are common in HFpEF,¹²⁹ even in the absence of epicardial coronary disease.^130,131 This is important because FFA use is reduced during ischaemia as the heart preferentially shifts to glucose utilization as it requires less O₂.¹³² Thus, deficits in glucose uptake would be expected to further exacerbate cardiac dysfunction during exercise in obese HFpEF.

5.5. Myocardial energetics in obese HFpEF

As a continual ATP delivery to the myocardium is required for both contraction and relaxation, the increased ATP demand in the context of this metabolic inefficiency in obesity is a potential explanation obesity-related HFpEF. In line with this, using ³¹P magnetic resonance spectroscopy, the PCr/ATP ratio (a marker of energetic state of the myocardium) has been shown to be lower in patients with obesity, and correlated with diastolic function.¹³³ This is in line with the concept that any impairment in ATP metabolism will initially affect the function of the sarcoplasmic reticular calcium ATPase (SERCA) to lower cytosolic Ca²⁺ during diastole (as it is energetically most demanding of all enzymes in the contractile apparatus). However, in obesity creatine kinase activity is increased to compensate for the reduced PCr/ATP ratio and maintain ATP delivery at rest.⁷⁰ This compensatory mechanism is also seen in obese HFrEF.¹³⁴ Successful weight loss in the absence of HF is associated with a reduced CK flux⁷⁰ and with increased PCr/ATP.¹³⁵ In HFrEF, normalization of CK flux is also seen,¹³⁴ but no study has yet evaluated effects of weight loss on myocardial energetics in HFpEF.

5.6. Pulmonary vascular disease and right-sided HF

Pulmonary hypertension (PH) is common in HFpEF and associated with increased mortality,¹³⁶ especially when pulmonary vascular disease (PVD; i.e. elevated pulmonary vascular resistance) is present.¹³⁷ PA pressures are similar at rest in patients with obesity-related HFpEF and non-obese HFpEF, but with the increase in lung blood flow accompanying exercise, obese patients with HFpEF display impaired pulmonary vasodilation, and greater elevations in PA pressure.¹⁵ Recent studies in animal models have shed new light. Obese ZSF-1 leptin-receptor knockout rats had normal PA pressures at rest, but elevated pressures during exercise, similar to what is commonly observed in human HFpEF.¹³⁸ Cultured PA smooth muscle cells treated with FFA, glucose, and insulin showed increased mitochondrial reactive oxygen species and reduced cyclic guanosine monophosphate-dependent signalling, which favourably modified following treatment with the SGLT2i empagliflozin.¹³⁹ Recent studies in porcine banded models have identified novel pathways activated in PVD due to HFpEF, and further research is needed to further delineate how obesity and insulin resistance might affect these pathways.¹⁴⁰

Sustained exposure to PH and PVD leads to RV dysfunction in patients with HFpEF.^141,142 A recent meta-analysis has revealed that the higher BMI is associated with RV dysfunction in HFpEF.¹⁴³ In a study in which patients with obese and non-obese HFpEF were directly compared, RV systolic function and RV-PA coupling were more impaired in obesity-related HFpEF compared with non-obese HFpEF,¹⁵ and in a longitudinal study, higher body weight was associated with greater risk for new onset RV dysfunction in patients with HFpEF.¹⁴²

5.7. Effects beyond the heart

5.7.1. Neurohormone activation and volume distribution

Plasma volume expansion is typical of obesity, and elevation in cardiac filling pressures is directly correlated with estimated plasma volume in patients with obese HFpEF.¹⁵ Two mechanisms likely contribute to the TBV expansion in obese HFpEF: systemic vasodilation and sodium retention. Because of increased metabolic demands in obese subjects, excessive systemic vasodilation occurs which is associated with arterial underfilling, contributing to neurohormonal activation, and increases in sodium retention.¹⁴⁴ Perivascular adipose tissues also release hormones such as leptin to induce vasodilation.¹⁴⁵ The ensuing reduction in vascular resistance coupled with these changes may lead to the rarer entity of obesity-related high output HF in some patients,¹⁴⁶ but most patients with obesity-related HFpEF have normal output.¹⁵

Obesity, especially when associated with increased VAT, is associated with activation of the sympathetic nervous system (SNS) and renin–angiotensin–aldosterone system (RAAS), both through increased synthesis of adipokines that activate these systems (e.g. leptin, aldosterone) and through mechanical compression of the kidneys.^147,148 Sodium reabsorption, plasma volume expansion, SNS activation, and development of hypertension are even more strongly related to increases in visceral fat.^149,150 Hyperfiltration and activation of inflammatory pathways may lead to obesity-related glomerulopathy, further contributing to volume overload.¹⁵¹

In addition to TBV expansion, recent studies have shown that the distribution of blood volume is altered in patients with obesity. TBV can be divided into two functional compartments, the unstressed blood volume (UBV), defined as the volume of blood necessary to fill the entire vascular space, and the stressed blood volume (SBV), which is the volume in excess of unstressed volume that increases transmural pressure in the vasculature (Figure 5). The relative distribution of these volumes is importantly regulated by the SNS, whereby sympathetic venoconstriction increases SBV and the SBV/TBV ratio, a teleologically important adaptation to augment venous return to the heart during stress. Sorimachi et al.¹⁵² have recently shown that SBV is increased in patients with HFpEF compared with controls, and notably both SBV and SBV/TBV ratio were directly correlated with BMI, suggesting that abnormalities in venous capacitance play an important role in obesity-related HFpEF (Figure 5).

Figure 5.

Altered venous capacitance in obese HFpEF. (A) Total blood volume (TBV) can be functionally divided into an UBV, defined as the volume of blood necessary to fill the vascular space, and a SBV, which is defined as the volume in excess of UBV that increases wall tension and vascular pressure. Under normal conditions SBV constitutes only 20% of TBV, but with venoconstriction, blood is translocated from the unstressed to stressed functional compartments, increasing SBV and the SBV/TBV ratio. (B) When compared with controls, patients with HFpEF display an increase in both absolute SBV and SBV/TBV ratio, indicating reduced venous capacitance, both at rest and during exercise. (C) In patients with HFpEF and non-cardiac dyspnoea (NCD), increasing body mass index (BMI) is associated with greater impairment in venous capacitance during exercise, estimated by the SBV/TBV. Panel A modified from Fudim et al.¹⁵³ Panels B and C from Sorimachi et al.¹⁵²

Neurohormonal activation may contribute to cardiac structural remodelling by inducing cardiac fibrosis and sodium retention-mediated blood pressure rise and plasma volume expansion. Studies have shown that leptin is an important mediator of neurohormonal activation, and hyperleptinaemia contributes to obesity-related cardiovascular disease.^154–156 Leptin enhances both RAAS and renal SNS, and stimulates aldosterone synthesis from the adrenal glands. Leptin receptors have been reported to be expressed in the heart,¹⁵⁷ and experimental studies have demonstrated that leptin can promote cardiomyocyte hypertrophy and fibrosis by mediating angiotensin II, mineralocorticoid, and endothelin-1 receptors.^155,158

5.7.2. Systemic inflammation

A detailed review of obesity-related inflammation is beyond the scope of this review, but growing evidence suggest that metabo-inflammation plays an important role in the pathophysiology of HFpEF, especially when coupled with increases in visceral fat.^{110,159–161} Infiltration of macrophages causes adipose tissue inflammation,¹⁶² promoting up-regulation of proinflammatory adipokines (e.g. leptin, TNF-α, IL-6, and resistin) and the down-regulation of anti-inflammatory adipokines (e.g. adiponectin, omentin-1), leading to chronic and low-grade systemic inflammation.¹⁶³ Obesity may combine with hypertension to induce proinflammatory macrophage polarization (M1 phenotype) that contributes to the alterations in myocardial substrate availability and utilization described above.¹⁶⁴ Chronic activation of these proinflammatory pathways is believed to increase systemic inflammation and nitrosative stress, and lead to capillary rarefaction and mitochondrial dysfunction, and impairment endothelium-dependent vasodilation, contributing to abnormalities in the heart, vasculature, skeletal muscle, and other organs in HFpEF through a wide variety of pathways, as recently reviewed.^159,165

5.8. Regional adipose distribution and HFpEF

5.8.1. Abdominal visceral fat

In addition to increases in EAT described above, a number of recent studies have pointed to a pivotal role for VAT in obese HFpEF.^98,166,167 In a multinational prospective study, central adiposity, as measured by waist circumference and reflecting VAT, was directly related with worse outcomes in patients with HFpEF, whereas an inverse relationship was found with BMI.¹⁶⁸ A population-based study has shown that the risk of HFpEF is most strongly tied to increases in VAT rather than subcutaneous fat (Figure 6).¹⁰ This may relate to the higher degrees of adipocyte hypertrophy, insulin resistance, and FFA elevation among patients with increased VAT. In this regard, Wang and colleagues²³ recently showed in mice that a unique cell line present in adipose tissue (p21^Cip1 highly expressing cells) importantly contributes to insulin resistance, and treatment with senolytic drugs eliminates these cells in human fat while improving insulin sensitivity. Insulin resistance, accumulation of perivascular adipose tissue-derived adipokines and vascular inflammation interfere with vascular nitric oxide bioavailability to limit increases in blood flow during exercise.^169,170 Such abnormalities are commonly observed in patients with HFpEF in both the peripheral and coronary microcirculations,^14,131 potentially leading to impaired O₂ delivery and utilization in skeletal muscle^171–173 as well as myocardium to cause ischaemia and injury during stress.¹²⁹ VAT expansion may also influence myocardiac structure via secreted profibrotic factors. Recent data from mouse models have shown that VAT-secreted osteopontin may play an important role in driving cardiac senescence.¹⁷⁴ Osteopontin levels increased in wildtype animals during ageing, and VAT removal decreased circulating osteopontin, restored cardiac function, and decreased myocardial fibrosis and transforming growth factor β levels, collectively supporting a novel role for VAT-derived secretome in cardiac changes with ageing and HFpEF.

Figure 6.

(A) Hazard ratio (HR) plotted development of incident HFpEF in community-dwelling adults for measures of adiposity, showing greatest risk with increasing visceral rather than subcutaneous fat. (B and E) When compared with age and BMI-matched control women, women with HFpEF display 35% higher visceral adipose tissue (VAT) area, whereas there were no statistically significant differences observed in men. (C and F) Greater VAT in women was associated with higher exercise pulmonary capillary wedge pressure in women but not men, with significant sex interaction. (D) The estimated causal odds ratio (OR) for hypertension and diabetes with increasing VAT significantly exceeds unity in both men and women, but is substantially greater for women, particularly for diabetes. Panel A plotted from data in Rao et al.,¹⁰ Panels B, C, E, and F adapted from Sorimachi et al.,¹⁶⁷ and Panel D plotted from data in Karlsson et al.¹⁷⁵

VAT expansion may be even more detrimental for women (Figure 6), and may help explain the strong sex differential in prevalence of HFpEF.⁹⁸ A Mendelian randomization study showed that increases in VAT are associated with greater risk for diabetes, hypertension, and hyperlipidaemia in both sexes, but the risk was substantially greater in women.¹⁷⁵ A large community-based study in Taiwan showed that higher BMI and larger waist circumference were associated with subclinical LV contractile dysfunction (detected by myocardial strain imaging), more pronounced in women than men (P for interaction < 0.05), with detectable LV dysfunction at low anthropometric cutoffs in Asian women that are far lower than typically considered pathological in western countries (BMI > 23.4 kg/m², waist circumference > 83 cm).⁸⁴

Fat distribution is known to vary in women and men,¹⁷⁶ with more gluteofemoral fat in women and more VAT in men, but around the time of menopause, VAT increases in women, which may contribute to the greater risk of HFpEF in women compared with men.¹⁷⁷ This sex difference in the effects of ageing on body composition may be very important in HFpEF. A recent study showed that women with HFpEF display significantly greater VAT when compared with control women, even after matching for age and BMI; no such differences were present in men (Figure 6).¹⁶⁷ In addition to these quantitative sex differences in VAT, there were qualitative differences as well. Increases in VAT in women were directly related to greater elevation in pulmonary capillary pressures during exercise, even as the absolute volume of VAT was lower compared with men, whereas in men there was no significant relationship present.¹⁶⁷ A more recent study has confirmed this finding using waist circumference alone as a less robust, but more practical method to evaluate for central obesity, showing relationships with HFpEF severity in women but not men with HFpEF.¹⁷⁸

5.8.2. Effects in kidneys, liver, and skeletal muscle

Increased renal fat deposition has been shown to be associated with microvascular proliferation and hyperfiltration in the kidney, with increased inflammatory and oxidative stress.¹⁷⁹ Renal sodium avidity may be enhanced through compression by visceral, perinephric, and renal sinus fat, as well as neurohormonal activation.^180,181 Visceral adiposity is often accompanied by increased fat deposition in the liver, further promoting inflammation, dyslipidaemia, and insulin resistance.¹⁶¹ Obesity is associated with increased inflammation in skeletal muscle, in which immune cell infiltration and proinflammatory activation occur in inter- and perimuscular adipose tissue.¹⁸² Recent studies have identified several critical abnormalities in the skeletal muscle, including adipose infiltration, capillary rarefaction, and mitochondrial dysfunction that importantly lead to symptoms and exercise intolerance in HFpEF,^{165,173,183,184} and many of these may be related to excess body fat and its impact on skeletal muscle.

6. The obesity paradox in HFpEF

Like most chronic diseases, obesity paradoxically protects against adverse outcomes in patients with established HFpEF, despite being a well-established risk factor for incident HFpEF.^185–187 However, these findings come from cross-sectional studies, which are limited by survival bias (i.e. sicker obese patients die prior to assessment), and reverse causation (patients with the most severe HF have lost weight due to cachexia). A Mendelian randomization study, which overcomes the limitations of cross-sectional analyses, has raised questions with the obesity paradox in HF.¹⁸⁸ The hazard ratio for HF mortality per 1 kg/m² increase in BMI was 1.04 [95% confidence interval (CI), 1.03–1.06] in the observational cohort, suggesting a relatively minor risk. In contrast, the causal risk ratio for the same 1 kg/m² increase in BMI via Mendelian randomization revealed a potentially stronger causal relationship between high BMI and HF mortality (causal risk ratio 1.18; 95% CI, 1.00–1.38), though differences in population characteristics might contribute to the differential relationships observed.¹⁸⁸ Other studies have shown that when more specific measures of adiposity, such as weight circumference, are used in place of BMI, there is a continuous increase in risk of all-cause and cardiovascular mortality with greater body fat, with no paradox.¹⁸⁹ The importance of different adiposity parameters for the assessment of outcomes in established HF was also demonstrated in a longitudinal multinational study, where higher BMI was associated with better outcomes, yet larger waist circumference was associated with worse outcomes, in patients with HF.¹⁶⁸

6.1. Treatment

6.1.1. Diuretics

As described above, patients with obesity-related HFpEF often display volume expansion,¹⁵ and diuretics are effective to reduce filling pressures and congestion in these patients. However, patient with obese HFpEF may tolerate diuresis more poorly than non-obese HFpEF, with greater risk of worsening renal function during decongestion,¹⁹⁰ despite the presence of more marked volume expansion.¹⁹¹

6.1.2. Mineralocorticoid antagonists

There is evidence that mineralocorticoid receptor antagonists (MRAs) may be of greater benefit in obese HFpEF.¹⁹² In mice, treatment with MRA was shown to reduce expression of proinflammatory factors in adipocytes.¹⁹³ In the TOPCAT trial, treatment with spironolactone had no significant effect on the primary endpoint of HF hospitalization, cardiovascular death, or aborted sudden death in the overall trial cohort of patients with HF and EF ≥ 45%, but there was regional variation in treatment response, which was more favourable in patients with genuine HFpEF, and signal for benefit in patients with BMI above the median (interaction P = 0.11).^194,195 Post hoc analyses from TOPCAT showed that patients with lower NTproBNP levels, a group that typically displays more prominent central adiposity,¹⁷⁸ derived greater benefit from MRA,¹⁹⁶ and patients with HFrEF and greater waist circumference displayed more favourable responses to eplerenone than those without central obesity.¹⁹⁷ Further insight into the role for MRA in obese HFpEF will be provided by ongoing trials evaluating MRA in HFpEF (NCT02901184, NCT04727073, NCT04435626).

6.1.3. Neprilysin inhibitors

In the PARAGON-HF trial, treatment with the dual neprilysin/angiotensin receptor antagonist sacubitril valsartan tended to reduce the rate of HF hospitalizations or cardiovascular death, narrowly missing statistical significance (rate ratio, 0.87; P = 0.06).¹⁹⁸ Neprilysin degrades BNP and other vasoactive peptides, and is expressed on adipocytes, suggesting excessive breakdown by adipocyte-derived neprilysin could contribute to obese HFpEF.^192,199 There was no subgroup analysis in PARAGON-HF stratified by obesity status, but there was no significant treatment interaction between patients with NTproBNP levels above and below the median.¹⁹⁸ A subsequent analysis from PARAGON-HF reported on regional differences in HFpEF characteristics, showing highest rates of obesity in HFpEF in North America, but treatment effects did not vary by geographic region.²⁰⁰

6.1.4. Statins

Statins reduce systemic inflammation,²⁰¹ which is strongly associated with the pathogenesis of obese HFpEF.¹⁵⁹ Treatment with statins induces EAT regression in a lipoprotein-lowering independent manner, which may be related to anti-inflammatory effects.²⁰² Statins have been associated with lower event rates in observational studies in HFpEF,^203,204 though randomized trial data are lacking.

6.1.5. SGLT2 inhibitors

Epicardial fat thickening and inflammation may also play a role in HFpEF, and sodium-glucose co-transporter 2 inhibitors (SGLT2i) have been shown to increase adipocyte glucose uptake, reduce the secretion of proinflammatory chemokines, and improve the differentiation of EAT cells in adipose samples of patient undergoing heart surgery.²⁰⁵ One study showed that SGLT2i treatment for 6 months decreased EAT and TNF-α levels compared with usual therapy in patients with diabetes and coronary disease (n = 40),²⁰⁶ while another placebo-controlled trial (n = 56) reported that treatment with SGLT2i for 12 weeks did not affect EAT, intramyocardial fat, or PCr/ATP, but resulted in favourable reductions in VAT and liver fat.²⁰⁷ In patients with HFpEF (EF > 40%), treatment with SGLT2i has been shown to reduce risk of HF hospitalization or cardiovascular death in the EMPEROR-preserved trial,²⁰⁸ and to improve patient-reported health status in EMPEROR and other trials.^208–210

Submaximal functional capacity assessed by the 6 min walk test was improved in the PRESERVED-HF trial, conducted within the USA,²⁰⁹ whereas no such benefit was observed in the similarly sized EMPERIAL trial.²¹¹ The reasons for the discrepant results are not clear, but may relate to differences in baseline characteristics in the participants studied: patients in EMPERIAL were notably less obese (BMI 29.6 vs. 34.9 kg/m²), and less likely to be women (43 vs. 57%) when compared with PRESERVED-HF.^209,211 Because health status,²¹² functional capacity,^15,91 and haemodynamic abnormalities¹⁵² are so strongly tied to excess body fat in HFpEF (especially among women¹⁶⁷), this could explain the disparate results. There was no significant treatment effect interaction by obesity status in EMPEROR preserved, but this trial was largely conducted outside the USA, with mean BMI of <30 kg/m².²⁰⁸ The mechanisms responsible for benefits from SGLT2i in HFpEF remain unknown. Weight loss is typically observed following treatment with SGLT2i, but the magnitude is modest (1–2 kg) and unlikely to account for the clinical improvements observed in obese or non-obese HFpEF.

6.1.6. Weight loss

An analysis from 4 community-based cohorts showed that risk for HFpEF increased by 34% for each 1 standard deviation increase in BMI, suggesting that weight loss might reduce risk for HFpEF, or attenuate HFpEF severity among patients with prevalent disease.²¹³ Weight loss can be achieved through lifestyle interventions including diet and exercise training, pharmacotherapy, and bariatric surgery. Gepner et al.²¹⁴ recently showed that the combination of Mediterranean/low-carbohydrate diet mobilized specific fat depots including EAT and VAT, and exercise training also interpedently contributed to reduction in VAT. In a landmark trial, Kitzman and colleagues²¹⁵ demonstrated that either weight loss via caloric restriction (prepared meals) or aerobic exercise reduced LV mass and inflammatory markers, improved exercise capacity, and enhanced quality of life in patients with obese HFpEF, while the combination of training and weight loss was additive (Figure 7).²¹⁶

Figure 7.

(A) Pooled hazard ratio for cardiovascular outcomes following bariatric surgery vs. control in a meta-analysis of observational studies. (B) Changes in aerobic capacity (peak oxygen consumption, VO₂) from a randomized controlled trial in obese HFpEF with diet-induced weight loss, exercise training, or combination. In observational studies of patients without heart failure, weight loss is associated with improved haemodynamics (C) and myocardial energetics, assessed by phosphocreatine/ATP ratio (D). Panel A plotted from data in van Veldhuisen et al.,²¹⁷ Panel B adapted from Upadhya et al.,²¹⁶ Panel C from Reddy et al.,²¹⁸ and Panel D from Rayner et al.⁷⁰

Bariatric surgery leads to the most dramatic and sustained weight loss, and while no randomized trial has been performed to date, observational data strongly indicate a reduction in the risk for new onset HF.²¹⁷ One small, non-controlled study in patients with mild HFpEF showed improvements in symptom severity with signal for improved diastolic relaxation, despite no effect on intramyocardial fat.²¹⁹ Reddy et al.²¹⁸ showed in a meta-analysis of invasive studies that weight loss decreases total body O₂ consumption, heart rate, and blood pressure, along with 20–25% reductions in cardiac filling pressures and PA pressures (Figure 7). While these evaluations were not carried out in patients with HFpEF, each of these effects would be expected to improve clinical status in such patients. Three randomized clinical trials are currently underway evaluating the role for pharmacological weight loss therapies in HFpEF, two using the glucagon-like peptide-1 (GLP-1) receptor agonist semaglutide and one the dual glucose-dependent insulinotropic polypeptide (GIP), GLP-1 agonist tirzepatide (NCT04788511, NCT04916470, NCT04847557).

7. Future directions

Patients with obese HFpEF may be better positioned to respond to therapies under development and evaluation in HFpEF owing to key pathophysiological features. For example, the greater abnormalities in venous capacitance observed in obese HFpEF discussed above¹⁵² may position these patients to respond more favourably to novel treatments targeting volume distribution, as with splanchnic nerve ablation.¹⁵³ Because patients with obese HFpEF display increased EAT and ventricular interdependence, novel therapies to reduce pericardial restraint, such as minimally invasive pericardiotomy, may also be more efficacious in this patient population.^220,221

Other novel treatments may target fat reduction through different pathways, alter the consequences of excess body fat on organ function, modify the adipocyte secretome, or target cells in obesity that promote disease progression, such as senescent cells.^23,222 In addition to the GLP-1 and combined GIP/GLP-1 agonists, other novel treatments to facilitate weight loss through decreased appetite and/or increased satiety are at varying stages of development. Other potential treatments may target weight loss through increased catabolism. Novel classes of drugs may target proton transport across the inner mitochondrial membrane, essentially lead to increased fat breakdown by uncoupling oxidative phosphorylation from the electron transport chain, increasing resting metabolic rate.^223,224 The prototypic compound is 2,4-dinitrophenol, which was used in the 1930s as a highly effective weight loss drug, but was withdrawn because of the high risk of malignant hyperthermia. A related compound with wider therapeutic index (HU6) is under development. Mishra et al.²²⁵ recently demonstrated that inhibition of phosphodiesterase 9A in mice reduced body (and myocardial) fat and stimulated mitochondrial activity in a PPARα-dependent manner, without altering activity levels or food intake. Notably, favourable cardiometabolic and weight loss effects were exclusively observed in male animals and ovariectomized females, additional evidence indicating a strong sexual dimorphism in treatment response.

Other drugs may target the metabolic derangements in obesity that lead to cardiac and non-cardiac organ dysfunction, particularly the excess use of circulating FFA as fuel, such as partial fatty acid oxidation inhibitors, carnitine palmitoyltransferase 1 inhibitors, and mitochondrial-targeted antioxidants.²²⁶ In HFrEF and hypertrophic cardiomyopathy, treatment with the carnitine palmitoyltransferase-1 inhibitor perhexiline was shown to improve PCr/ATP ratio and functional capacity, presumably by reducing myocardial fat oxidation and shifting to more favourable glucose dependence.^227,228 Therapies targeting obesity-associated inflammation may also hold promise to improve clinical status in obesity-related HFpEF, such as the IL-6 inhibitor ziltivekimab,²²⁹ and warrant testing in obesity-related HFpEF given the evidence of systemic inflammation in this cohort.⁹¹

8. Conclusion

Obesity is the one of the most important causes of the clinical syndrome of HFpEF, leading to symptomatic HF through a wide variety of pathophysiological mechanisms. While there have been important advances in our understanding over the past decade, many knowledge gaps persist, and further study is needed to advance our understanding about this disease, ranging from cellular mechanisms to clinical trials, in order to improve outcomes for the ever expanding group of patients with obesity and HFpEF.

Funding

Supported in part by NIH grants: R01 HL128526 and U01 HL160226 (B.A.B.), R01AG045551; R01AG18915; P30AG021332; U24AG059624; 1U01HL160272 (D.W.K.), by the British Heart Foundation FS/16/70/32157 (O.J.R.), and by the US Department of Defense: W81XWH2210245 (B.A.B.).

Data availability

Data will be made available upon request to the corresponding author.