Methods

Recruitment details and procedures have been described elsewhere.13 Briefly, the ALOFT study randomised 302 patients with New York Heart Association (NYHA) class II–IV heart failure treated with either ACE inhibitor or ARB and β-blocker (in 95%) to either treatment with aliskiren or placebo. For the purposes of this report, subjects taking mineralocorticoid receptor blockers (n = 101) were excluded from further analysis.

At the initial baseline visit, 186 subjects underwent 24-hour urine collection for measurement of aldosterone, total protein and sodium excretion. Additionally, early morning fasting plasma samples were taken from subjects (supine) for plasma glucose, glycosylated haemoglobin (HbA_1c), HOMA-IR (fasting insulin (mU/ml) × fasting plasma glucose (mmol/l)/22.5 as previously described14) as well as aldosterone, plasma renin activity (PRA) and plasma renin concentration (PRC).

As all patients had been on a stable dose of ACE inhibitor or ARB for at least 4 weeks before analysis, we used a plasma aldosterone cut-off level of >144 pg/ml to define plasma aldosterone escape in keeping with previous studies.15 Twenty-four-hour urinary aldosterone level of >12 μg/24 hours was used to define “high” urinary aldosterone again in agreement with previous studies and with well recognised criteria for the diagnosis of primary aldosteronism (PA).16

As not all biochemical data were normally distributed, non-parametric methods of analysis were used to directly compare markers of insulin sensitivity according to aldosterone status (Mann-Whitney test using Minitab, State College PA, version 14). Adjustment for potential confounders (age, sex, body mass index, diabetes status and blood pressure) was performed using binary logistic regression. A p value <0.05 was deemed significant.

Results

General

Summary statistics of the subjects subdivided according to plasma and urinary aldosterone levels are demonstrated in tables 1 and 2, respectively. Of the patients studied, 20% (37) demonstrated aldosterone escape and 34% (63) had high urinary aldosterone excretion. There was some overlap between these two groups with 26 subjects demonstrating aldosterone escape in both urine and plasma.

Aldosterone levels positively correlated with fasting plasma insulin at baseline. This was seen with both plasma aldosterone (r = 0.216, p<0.01; fig 1) and urinary aldosterone excretion rates (r = 0.188, p<0.03).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Correlation between early morning fasting insulin and plasma aldosterone levels.

Insulin sensitivity and aldosterone status

Subjects on insulin (n = 23) were excluded from this analysis. Early morning fasting insulin was significantly higher in aldosterone escape and high urinary aldosterone groups in comparison to subjects without aldosterone escape and lower levels of urinary aldosterone excretion (fig 2). This relation was consistent using other markers of insulin sensitivity; HOMA-IR levels and insulin-glucose ratios (IGR) were higher in aldosterone escape and high urinary aldosterone cohorts (fig 2). Patients who demonstrated both high plasma and urinary aldosterone levels were most insulin resistant (p<0.01, data not shown). All relations remained significant when adjusted for potential confounders known to influence insulin resistance (age/sex/blood pressure/BMI).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Relation between (A) fasting insulin, (B) homeostasis model assessment of insulin resistance (HOMA-IR) and (C) insulin/glucose ratio and plasma and urinary aldosterone. (Data compared by Mann-Whitney non-parametric tests).

When plasma and urinary aldosterone levels were divided into tertiles, it was found that as plasma levels of aldosterone increased fasting insulin also increased (fig 3) as well as HOMA-IR (p<0.007) and IGR (p<0.0001).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Relation between plasma and urinary aldosterone tertiles and fasting plasma insulin. Data analysed by one-way ANOVA. Plasma aldosterone (pg/ml): tertile 1, 34.5–116, tertile 2, 119–266. tertile 3, 269–1795. Urinary aldosterone (μg/24 h); tertile 1, 2.65–16.2, tertile 2, 16.2–38.2, tertile 3, 38.2–255.2.

Diabetes status

Patients with non-insulin-treated diabetes were not excluded from this study. This was because exclusion of diabetics resulted in a substantial reduction in subject numbers particularly in the aldosterone excess groups and a loss of study power, meaning that although there was still a tendency to increased insulin resistance in the aldosterone excess groups, this did not always reach statistical significance. Additionally, there was a trend to higher prevalence of both diabetes and insulin use in subjects with biochemical evidence of aldosterone excess although again this did not reach statistical significance (table 3). However, the presence of diabetes was incorporated into binary logistic regression analysis and the relations between insulin resistance and biochemical aldosterone excess remained significant (p<0.008 aldosterone escape and insulin, p<0.03 high urinary aldosterone and insulin).

Discussion

These data demonstrate in a stable heart failure population that subjects with relative aldosterone excess have evidence of increased insulin resistance, a relation which persists after correction for diabetes status. In addition, there is a trend to an increased prevalence of diabetes in the high aldosterone subjects although this does not reach statistical significance.

High aldosterone levels and insulin resistance are both associated, independently, with greater risk of cardiovascular disease and heart failure. Subjects with hypertension and aldosterone excess have an excess of adverse vascular events (particularly stroke) when compared to subjects with equivalent essential hypertension.2 It is noteworthy that serum levels of aldosterone (and the glucocorticoid hormone, cortisol) have recently been shown to be independent predictors of increased mortality risk in patients with chronic heart failure,17 while the role of aldosterone in heart failure has been highlighted by large-scale clinical trials demonstrating additional benefit of mineralocorticoid blockade in conjunction with conventional therapy.5

Aldosterone levels are positively correlated with poor outcome after myocardial infarction, including increased risk of death, cardiac arrest and dysrhythmia. An association between essential hypertension and insulin resistance is also supported by a large body of evidence and is consistent with the notion that insulin-resistant subjects have greater long-term cardiovascular risk.18 19 20

Insulin resistance is associated with vascular endothelial dysfunction and, in particular, with reduced availability of endothelial nitric oxide21; it is possible that this contributes to the associated increased risk of cardiovascular events in patients with hypertension and insulin resistance. Insulin resistance is independently associated with risk of heart failure and adverse outcome in that circumstance.12 However, despite the associations between aldosterone and insulin resistance in cardiovascular disease, the nature and potential mechanism of the relation between the two remains unclear. In a large prospective study, Catena et al demonstrated that patients with PA were more insulin resistant than normotensive controls and that medical or surgical treatment of the aldosterone excess improved insulin sensitivity.9 Nevertheless, PA subjects were less insulin resistant than those with essential hypertension. The same group subsequently demonstrated that plasma aldosterone levels correlated with markers of insulin resistance and hyperinsulinaemia in hypertensive patients.22 Indeed, the majority of clinical evidence supports an association between aldosterone excess and increased insulin resistance in hypertension, although there are conflicting reports as to whether this is a stronger23 or weaker9 association than in essential hypertension.

Our data suggest a positive correlation between aldosterone and insulin resistance in heart failure. This finding is consistent with other observational findings. Thus, chronic heart failure and diabetes often co-exist, an association noted in the Framingham Study, where 19% of patients with chronic heart failure had diabetes; the risk of heart failure increased twofold to eightfold in the presence of diabetes.24 This relation between chronic heart failure and diabetes may also extend to insulin resistance. In one large prospective study of 1187 elderly men there was a clear inverse relation between the risk of heart failure and insulin sensitivity and there are other groups who have confirmed such an observation.25 This notion is further supported by recent evidence from the CHARM study illustrating that, in heart failure patients, glycosylated haemoglobin A_1c is an independent predictor of morbidity and mortality.26

The mechanism of the link between aldosterone and insulin resistance remains unclear. It is, of course, possible that the association is not causal; subjects with more severe heart failure (who may be more likely to be insulin resistant) may have greater activation of the neuroendocrine system and, consequently, higher aldosterone levels. Alternatively, it is possible that insulin resistance, which in its early stages is associated with hyperinsulinaemia, might directly lead to increased aldosterone production. This seems plausible as we observed a positive correlation between plasma insulin and aldosterone and so a possible explanation is that activation of the RAAS contributes directly to insulin resistance. An association between insulin resistance and the renin-angiotensin-aldosterone system has recently been strengthened by several large-scale clinical trials, which demonstrate that blockade of the RAAS with ACE inhibitors or angiotensin receptor blockers results in a reduction in new cases of diabetes mellitus.27 However, this link remains controversial with a recent robust clinical trial failing to show a significant effect of ramipril on the subsequent development of diabetes in subjects with impaired glucose tolerance or impaired fasting glucose.28

A simple haemodynamic explanation linking the RAAS to the development of insulin resistance is possible, where increased angiotensin II and aldosterone levels may be associated with reduced blood flow to skeletal muscle and, as a result, diminished tissue glucose uptake.29 However, there is controversy about the contribution of flow to the efficiency of glucose extraction and more direct effects on insulin release and action seem more likely. This is supported by work by Mosso et al who demonstrated that impaired glucose metabolism in PA patients was associated with decreased markers of pancreatic β cell function (C-peptide and HOMA-βF), which improved in response to spironolactone treatment.30 Another potential mechanism, aldosterone-induced electrolyte abnormalities; hypokalaemia, is known to alter insulin secretion and modulate insulin receptor function,31 and recent in vitro experiments on isolated pancreatic islets suggest that extracellular potassium stimulates insulin secretion.32 In addition, magnesium deficiency can also reduce insulin production and magnesium supplementation prevents diabetes in rat models.33 Finally, mineralocorticoid receptors are present on adipoctyes and their activation affects insulin signalling and adipocytokine release.34 Subsequent experiments have demonstrated that adipocytokines can stimulate production of adrenal corticosteroids in vitro.35 These effects may directly alter tissue insulin sensitivity, and offer a plausible direct association between aldosterone, insulin resistance and obesity.36

Limitations of this study are mainly those imposed as a result of post-hoc analysis of data where the primary outcome was not to examine the relation between aldosterone and insulin status. In addition, diabetic subjects would, ideally, not have been included in this study but were included to provide sufficient study power. However, diabetes status and insulin use were incorporated into a binary logistic regression model of aldosterone and insulin resistance and did not alter these relations. The final limitation is our use of simple indirect measurements of insulin sensitivity which, although robust and in common clinical use, are less sensitive than the gold standard measure of insulin sensitivity, the euglycaemic-hyperinsulinaemic glucose clamp. Strengths of the study include the fact that this was a large, well phenotyped heart failure cohort with reliable biochemical measurements of both insulin and aldosterone in both plasma and urine, allowing accurate determination of mineralocorticoid production.

Conclusions

We have shown that increased aldosterone levels associate with biochemical features of insulin resistance in a heart failure population. This raises the possibility that increased insulin resistance could contribute to the adverse clinical outcome associated with aldosterone escape in heart failure. This hypothesis justifies further investigation in a focused prospective clinical study using gold standard measurements of insulin sensitivity.