Insulin Resistance in Heart Failure: Widening the Divide Between Reduced and Preserved Ejection Fraction?

This article refers to ‘Insulin resistance in heart failure: differences between patients with reduced and preserved left ventricular ejection fraction,’† by N. Scherbakov et al., published in this issue on pages 1015–1021. With a worldwide prevalence of 38 million and a long-term prognosis comparable to many cancers, heart failure (HF) is among the most significant public health problems of modern times.1 Approximately 50% of these patients have HF with preserved ejection fraction (HFpEF).2 Although survival for patients with HF with reduced ejection faction (HFrEF) has improved in the last two decades with the discovery of drug- and device-based therapies, HFpEF management remains devoid of evidence-based therapy, and treatment remains limited to optimization of co-morbidities and use of diuretics. Attempts at reproducing benefits of neurohormonal antagonism seen in HFrEF in HFpEF cohorts have been consistently unsuccessful. Thus, while these two entities continue to be crudely defined by EF, it is clear that each represents a distinct pathophysiology.3 Diabetes is one of the most common co-morbidities amongst HF patients across the spectrum of EF, present in 25–44% of patients, and may confer heightened independent risk of death and hospitalization.4-7 Likewise, multiple measures of insulin resistance (IR), including haemoglobin A1c, fasting glucose, and insulin levels, have been associated with incident HF risk and progressive cardiac dysfunction.7, 8 These maladaptive changes to the heart may be mediated via multiple mechanisms, including metabolic alterations, deposition of advanced glycation end-products, microvascular dysfunction, and autonomic neuropathy, among others.7, 8 However, despite this well-recognized link between HF and IR, data regarding a differential relationship among patients with preserved and reduced EF are limited. Clarifying the role of IR could provide further biological grounds for stratifying the broad HF population, offering potential for further refinement of the conventional EF classification. Moreover, a nuanced understanding of these interactions could carry important prognostic and therapeutic implications, particularly for HFpEF where identification of treatment targets is more urgently needed. In this issue of the journal, Scherbakov et al. provide incremental insights into the correlation between IR and HF.9 The investigators examined fasting and dynamic measures of IR separately in patients with stable ambulatory HFrEF and HFpEF without diabetes. Findings were then compared between these groups, as well as with healthy controls. Compared with controls, the authors found both HF populations to exhibit significantly heightened IR as measured by the homeostasis model assessment (HOMA) index. While absolute HOMA index and basal insulin levels tended to be higher among HFrEF patients, there were no significant differences between HFrEF and HFpEF populations. However, when provocative IR testing was performed with the short insulin sensitivity test, HFpEF patients did not significantly differ from controls, even after adjustment for lean body mass. In contrast, HFrEF patients continued to display marked impairment in glucose clearance. The authors are to be congratulated for an elegant study that supports further differentiation between the HFrEF and HFpEF syndromes beyond EF. Despite sound execution and thoughtful design, there are several considerations that must be highlighted when interpreting the present study. First, the small overall sample size and inclusion of few patients within each subgroup limit these findings to hypothesis generating only. Moreover, the small numbers may offer limited power to fully differentiate IR between HFrEF and HFpEF patients, as absolute differences in HOMA index may have reached statistical significance with a larger population. Second, the specific characteristics of the study population warrant scrutiny. Notably, all patients within the HFrEF group were male, whereas males constituted only 41% and 30% of the HFpEF and control groups, respectively. This probably explains the surprising finding of HFrEF patients having the highest lean body mass. Generalizability may be further impaired by virtue of differing NYHA functional class between the HFrEF and HFpEF cohorts. Nearly 60% of HFpEF patients had class I symptoms, as compared with only 6% of HFrEF patients. The mild symptoms and lack of notable echocardiographic or exercise testing differences between HFpEF and control patients also questions the accuracy of the HFpEF diagnosis, although the modestly elevated natriuretic peptide level among included HFpEF patients does suggest a unique subgroup. Third, no haemoglobin A1c data are available, and diabetic status was assessed by the investigators. Although one-time fasting glucose levels are reported and not in the diabetic range, documentation of A1c levels may have better portrayed the baseline level of glycaemic control, confirmed the absence of diabetes, and facilitated applicability of the findings to other populations. As compared with type 1 diabetes where insulin is deficient, type 2 diabetes is characterized by a relative excess of insulin and IR. Despite this clear difference in biology, all patients with diabetes are usually lumped together within HF studies. Previous data established a strong relationship between IR and HFrEF, with IR shown to be an independent predictor of death among patients without pre-existing diabetes.10 Fewer data are available comparing differential levels of IR between HFrEF and HFpEF. In contrast to the present study, a previous small analysis found no difference in HOMA index values between the two conditions after accounting for body mass index (BMI), age, and other factors.11 However, the results of Scherbakov and colleagues are intriguing as the link between IR, obesity, and clinical outcomes continues to be poorly defined in HF populations. Obesity is associated with adverse health outcomes and is a well-established risk factor for IR, type 2 diabetes, and HF. In contrast, while obesity confers heightened risk of incident HF and cardiovascular disease, once HF is diagnosed, data are mixed regarding the potential influence on subsequent mortality and hospitalization. In predominantly HFrEF patients, the ‘obesity paradox’ has been well described whereby higher BMI has repeatedly shown a protective effect.6, 12 In comparison, a study from a large HFpEF clinical trial cohort found a U-shaped relationship between BMI and clinical outcomes where the hazard for death or cardiovascular hospitalization was similarly high for those with BMI <23.5 kg/m2 and >35 kg/m2.13 The current findings by Scherbakov et al. help provide a biological construct for this differing relationship between obesity and outcomes, and support distinct metabolic signatures for HFrEF and HFpEF. Specifically, in the context of known differences in syndrome pathophysiology,3 the magnitude of the difference in IR profile between the two HF groups seen in the present study suggests potential differing mechanisms for IR. For example, one could speculate that IR in HFpEF is substantially mediated by excessive fat content and associated humoral factors that impair insulin sensitivity. This hypothesis would be consistent with the recently described novel HFpEF paradigm, whereby co-morbidities, including obesity and diabetes, are implicated in fostering a pro-inflammatory state that drives the syndrome and associated poor outcomes.3 Previous works suggest a higher prevalence of obesity among HFpEF patients compared with HFrEF, and HFpEF patients had the highest total fat mass in the present study.6 In contrast, HFrEF patients may be comparatively more likely to experience IR from sarcopenia/cardiac cachexia. In these patients, a higher BMI may better correlate with superior nutritional status and the associated benefits, similar to the proposed explanation for the ‘cholesterol paradox’ in HFrEF.14 There is significant uncertainty regarding the role of IR as an effective target for HF therapy. Although not synonymous with IR per se, it is unclear if, or to what degree, pharmacological modification of hyperglycaemia and haemoglobin A1c in HF patients confers clinical benefit. Multiple reports, both within and outside HF populations, have demonstrated heightened clinical risk at lower haemoglobin A1c levels among those receiving diabetic therapy.15, 16 Moreover, type 2 diabetes medications have been repeatedly associated with increased HF event rates.7 Notably, the thiazolidinedione agents, which mechanistically work as ‘insulin sensitizers’, carry a black box warning against use in HF patients.17 Although the complex relationship between diabetes, IR, and pharmacotherapy warrants further research in HF patients, the present analysis supports future studies making an important distinction between HFrEF and HFpEF. In summary, Scherbakov et al. contribute an interesting analysis to the growing body of literature exploring interactions between HF and diabetes/IR. The next step is validation of these findings in a larger cohort. While much more work is needed before the field readily adopts the phrase ‘HF with high IR’, these data serve as another reminder that there is more to defining the HF syndrome than EF. Conflict of interest: none declared.