Circulating 3‐hydroxy butyrate predicts mortality in patients with chronic heart failure with reduced ejection fraction

Abstract

Aims

In patients with chronic heart failure with reduced ejection fraction (HFrEF), myocardial ketone metabolism is increased and short‐term treatment with the ketone body 3‐hydroxy butyrate (3‐OHB) has beneficial haemodynamic effects. In patients with HFrEF, we investigated whether the level of circulating 3‐OHB predicted all‐cause mortality and sought to identify correlations between patient characteristics and circulating 3‐OHB levels.

Methods and results

We conducted a cohort study in 218 patients with HFrEF. Plasma 3‐OHB levels were measured using high‐performance liquid chromatography tandem mass spectrometry. Data on all‐cause mortality were obtained by reviewing the patients' medical records, which are linked to the national ‘Central Person Registry’ that registers the timing of all deaths in the country. Mean left ventricular ejection fraction was 35 ± 8.6%, mean age was 67 ± 10 years, 54% were New York Heart Association II, and 27% had type 2 diabetes mellitus. Median follow‐up time was 7.3 (interquartile range 6.3–8.4) years. We observed large variations in 3‐OHB levels between patients (median 59 μM, range: 14–694 μM). Patients with 3‐OHB levels above the median displayed a markedly increased risk of death compared with those with low levels {hazard ratio [HR]: 2.1 [95% confidence interval (CI): 1.3–3.5], P = 0.003}. In a multivariate analysis, 3‐OHB predicted mortality independently of known chronic heart failure risk factors [HR: 1.004 (95% CI: 1.001–1.007), P = 0.02] and with a similar statistical strength as N‐terminal pro‐brain natriuretic peptide (NT‐proBNP) [HR: 1.0005 (95% CI: 1.000–1.001), P = 0.02]. For every 100 μmol increase in plasma 3‐OHB, the hazard of death increased by 49%. The following factors significantly predicted 3‐OHB levels in the univariate analysis: free fatty acids (FFAs) [β: 238 (95% CI: 185–292), P < 0.0001], age [β: 2.43 (95% CI: 1.14–3.72), P < 0.0001], plasma insulin {β: −0.28 [95% CI: −0.54–(−0.02)], P = 0.036}, body mass index {β: −3.15 [95% CI: −5.26–(−0.05)], P = 0.046}, diabetes [β: 44.49 (95% CI: 14.84–74.14), P = 0.003], glycosylated haemoglobin [β: 1.92 (95% CI: 0.24–3.59), P = 0.025], New York Heart Association class [β: 26.86 (95% CI: 5.99–47.72), P = 0.012], and NT‐proBNP [β: 0.03 (95% CI: 0.01–0.04), P = 0.001]. In a multivariate analysis, only FFAs predicted 3‐OHB levels [β: 216 (95% CI: 165–268), P > 0.001].

Conclusions

In patients with HFrEF, circulating 3‐OHB was a strong predictor of all‐cause mortality independently of NT‐proBNP. Circulating FFAs were the best predictor of 3‐OHB levels.

Introduction

Ketone bodies are produced in the liver and are of vital importance for energy generation in the heart during fasting, exercise, and severe illness. ¹ Patients with severe heart failure [heart failure with reduced ejection fraction (HFrEF)] have elevated circulating levels of ketone bodies ² and augmented myocardial utilization of the ketone body 3‐hydroxy butyrate (3‐OHB). ³ , ⁴ It has been hypothesized that ketone bodies may act as a ‘super fuel’ for the failing heart. ⁵ In support of this, our group recently discovered that 3‐OHB infusion increases cardiac output with 2 L/min and left ventricular ejection fraction (LVEF) by 8% in patients with HFrEF. ⁶ Circulating ketone bodies are increased in patients with ST‐elevation myocardial infarction (STEMI), ⁷ and so is exhaled breath acetone in acute and chronic HFrEF. ⁸ , ⁹ In a study of 76 patients with HFrEF, circulating ketone did not independently predict outcome after 30 days. ¹⁰ However, it remains unknown whether circulating levels of 3‐OHB in patients with chronic HFrEF are associated with their long‐term prognosis.

Furthermore, we lack knowledge about the determinants of circulating 3‐OHB in a large cohort of chronic HFrEF patients with reduced ejection fraction, as only one minor study in patients with chronic HFrEF has reported a correlation between 3‐OHB and LVEF, free fatty acids (FFAs), and glucose levels. ²

In the present study, we investigated the prognostic impact of circulating 3‐OHB levels with respect to all‐cause mortality during long‐term follow‐up in patients with chronic HFrEF. Moreover, we investigated the association between circulating 3‐OHB levels and validated chronic heart failure risk factors and metabolic parameters to identify the clinical, paraclinical, and metabolic determinants of circulating 3‐OHB levels.

Methods

Design

In a historical cohort of 218 chronic HFrEF patients, we analysed circulating 3‐OHB, insulin, N‐terminal pro‐brain natriuretic peptide (NT‐proBNP), and circulating metabolic substrates and performed a long‐term follow‐up of all‐cause mortality. The 218 patients with HFrEF in the present study had previously been included in a double‐blind, randomized, placebo‐controlled clinical trial of 24 weeks' treatment with liraglutide vs. placebo. ¹¹ Patients were included from 2012 to 2015. The main inclusion criteria were (i) stable chronic HFrEF on maximally tolerated heart failure treatment (i.e. maximally tolerated doses of beta‐blocker, angiotensin‐converting enzyme inhibitor/angiotensin 2 receptor blocker, and aldosterone antagonist) and (ii) New York Heart Association (NYHA) classes I–III. Among the 241 patients included in the original study, 23 had insufficient plasma for 3‐OHB analysis. The original study was approved by the Central Denmark Regional Committees on Health Research Ethics.

Laboratory measurements

Blood samples were drawn at the randomization visit of the trial after an overnight fast. After collection, samples were stored at −80°C. Plasma 3‐OHB was measured using hydrophilic interaction liquid chromatography tandem mass spectrometry as previously described. ¹² The lower limit of quantification of 3‐OHB with this method is around 3 μmol/L with a relative standard deviation of reproducibility below 10%. NT‐proBNP was measured by chemiluminescent micro‐particle immunoassay on an Abbott Architect i2000SR (Abbott, Germany) as described by the manufacturer. FFAs were determined by enzymatic calorimetric assay (Trichem, Denmark). Plasma insulin was determined by enzyme‐linked immunosorbent assay (Mercodia, Sweden). ¹³ Homeostatic model assessment (HOMA) index for insulin resistance was calculated as glucose (mmol/L) × insulin (mUi/L) divided by 22.5.

Echocardiographic measurements

Examinations were performed using a Vivid 9 scanner (Version BT12; GE Vingmed Ultrasound, Horten, Norway). Two‐ and three‐dimensional echocardiography was performed according to international scientific recommendations. ¹⁴ Left ventricular opacification was enhanced using the commercially available ultrasound contrast agent SonoVue (Bracco, Initios Medical AB, Copenhagen, Denmark). All images were stored digitally for subsequent analysis. To minimize variability, all analyses were performed by one experienced echo technician who was blinded to treatment.

Clinical outcome assessment

Data on all‐cause mortality were obtained by reviewing the patients' medical records. In Denmark, these records are linked to the national ‘Central Person Registry’, which registers the timing of all deaths in the country.

Statistical analysis

Normal distribution was evaluated using QQ plots and transformed as appropriate. Continuous variables are presented as mean ± standard deviation (SD) if normally distributed. Non‐normally distributed data are presented as median [interquartile range (IQR)]. Discrete variables are presented as frequencies and percentages. To compare groups, the Mann–Whitney test was used for skewed continuous variables. The associations of risk factors and biomarkers of cardiovascular disease and 3‐OHB levels were analysed using simple and multiple linear regression models with robust standard errors to account for heteroscedasticity in variance. Beta values are presented with 95% confidence intervals (CIs). Predicted value and residuals were used to validate the models. Time‐to‐event data were evaluated using Kaplan–Meier estimates and compared using the log‐rank test. Hazard ratios (HRs) were calculated, using Cox regression model, and a Cox proportional hazards model was computed to adjust for confounding. Only variables reaching statistical significance in the univariate analysis were included in the multivariate analysis. Assumption of proportional hazards was checked on the basis of Schoenfeld residuals after fitting the model. A two‐tailed P‐value < 0.05 was considered statistically significant. Statistical analysis was done using STATA Version 15 (StataCorp, USA), and graphs were produced using GraphPad Prism Version 8.

Results

Baseline characteristics

Baseline characteristics are shown in Table 1 . The median age was 67 years, 90% were male, 61% had ischaemic heart disease, 27% had type 2 diabetes, and NYHA class ranged from I to III with 54% as NYHA II. Mean LVEF was 35% and median NT‐proBNP was 396 (IQR: 158–795 ng/L). Median 3‐OHB was 59 μM with a large variation between patients (IQR: 37–117 μM) (Figure  1 ). FFA levels were 0.54 (IQR: 0.42–0.67) mmol/L in the cohort. HOMA‐IR index for the cohort was 2.3 (IQR: 1.5–4.2), and patients with and without diabetes differed significantly [3.9 (IQR: 2.1–7.6) vs. 2.0 (IQR: 1.3–3.6), P < 0.0001].

Table 1.

Baseline characteristics (N = 218)

General
Age (years)	67 [59–73]
Sex (male/female)	197/21
Type 2 diabetes, N (%)	59 (27)
Smoking, N (%)	45 (21)
Ischaemic heart disease, N (%)	133 (61)
Atrial fibrillation, N (%)	62 (28)
Cause of HFrEF, N (%)
Ischaemic heart disease	140 (64)
Dilated cardiomyopathy	41 (19)
Previous valve disease	9 (4)
Hypertension	6 (3)
Other	15 (7)
Unknown	7 (3)
Heart failure severity, N (%)
Mild (LVEF > 40%)	67 (30.5)
Moderate (LVEF 30–40%)	84 (39)
Severe (LVEF < 30%)	67 (30.5)
Clinical and laboratory measurements
Systolic blood pressure (mmHg)	127 ± 19
Diastolic blood pressure (mmHg)	76 ± 11
Heart rate (beats/min)	66 [60–71]
BMI (kg/m2)	28.6 [25.7–31.8]
6 min walk test (m)	462 ± 91
Minnesota questionnaire	22 [12–40]
NYHA class, N (%)
I	70 (32)
II	118 (54)
III	27 (12)
HbA1c (mmol/L)
Diabetes	49.8 ± 9.9
No diabetes	38.4 ± 3.9
HOMA index	2.3 [1.5–4.2]
Diabetes	3.9 [2.1–7.6]
No diabetes	2.0 [1.3–3.6]
eGFR (mL/min/1.73 m2)	80 ± 20
Plasma NT‐proBNP (ng/L)	369 [158–795]
Plasma 3‐hydroxy butyrate (μmol/L)	59 [37–117]
Serum free fatty acids (mmol/L)	0.54 [0.42–0.67]
Fasting plasma insulin (pmol/L)	48 [34–86]
Echocardiographic measurements
LVEF (%)	35 ± 8.6
E‐velocity (m/s)	0.67 [0.52–0.84]
A‐velocity (m/s)	0.57 [0.44–0.76]
E/e′	11 [8.1–14.3]
E‐deceleration time (ms)	153 [131–178]
Left atrial volume index (mL/m2)	37 [30–46]
Global longitudinal strain (%)	11.3 ± 3.1
Medication, N (%)
ACE inhibitors	142 (65.1)
Angiotensin receptor blockers	73 (33.5)
Beta‐blockers	198 (90.8)
Diuretics	160 (73.4)
Aldosterone receptor antagonists	99 (45.4)
Metformin	43 (19.7)
Sulfonylureas	8 (3.7)
Insulin	10 (4.6)
Statins	168 (77.1)
Aspirin	150 (68.8)

ACE, angiotensin‐converting enzyme; BMI, body mass index; eGFR, estimated glomerular filtration rate; HbA1c, glycosylated haemoglobin; HFrEF, heart failure with reduced ejection fraction; HOMA, homeostatic model assessment; LVEF, left ventricular ejection fraction; NT‐proBNP, N‐terminal pro‐brain natriuretic peptide; NYHA, New York Heart Association.

Data are shown as mean ± SD and median [interquartile range], unless specified as N (%).

Figure 1.

Frequency distribution of fasting plasma 3‐hydroxy butyrate (3‐OHB) levels in patients with HF (N = 218). Patients are categorized according to plasma 3‐OHB levels with 20 μmol/L increments on the x‐axis.

Circulating ketone bodies predict all‐cause mortality

Data on vital status were available in all patients. A total of 70 deaths (32%) were recorded during a median follow‐up time of 7.3 years (IQR: 6.3–8.4). Kaplan–Meier survival curves of patients with above‐ and below‐median 3‐OHB levels are shown in Figure 2 . HR in patients with 3‐OHB levels above vs. below the median was 2.1 (95% CI: 1.3–3.5, P = 0.003).

Figure 2.

Kaplan–Meier survival estimates for patients with plasma 3‐hydroxy butyrate (3‐OHB) levels above (red line) and below (blue line) the median. Patients above the median had increased risk of all‐cause mortality compared with the patients below the median level of 3‐OHB. CI, confidence interval; HR, hazard ratio.

In a univariate analysis of survival data, 3‐OHB, NYHA class, NT‐proBNP, age, LVEF, FFA, and estimated glomerular filtration rate were all predictors of all‐cause mortality (Table  2 ). In a multivariate analysis, only LVEF, age, 3‐OHB, NT‐proBNP, and NYHA class remained independent predictors of mortality. FFA did not predict mortality in the multivariate analysis. The HR for 3‐OHB was 1.004 (95% CI: 1.001–1.007, P = 0.01); that is, for a 100 μmol increase in 3‐OHB, the hazard for all‐cause mortality was increased by 1.004¹⁰⁰ = 49%. There was no difference between subsequent treatment allocation groups (i.e. 24 weeks of liraglutide or placebo).

Table 2.

Hazard ratios for all‐cause mortality

Risk factor	HR (95% CI)	P‐value
Univariate analysis (N = 216)
Plasma 3‐OHB (μmol/L) a	1.004 (1.002–1.006)	<0.0001
NYHA class	1.27 (1.12–1.44)	<0.0001
NT‐proBNP (ng/L) b	1.0004 (1.0001–1.001)	<0.0001
Age (years)	1.07 (1.04–1.10)	<0.0001
LVEF (%)	0.95 (0.93–0.98)	0.001
Free fatty acids (mmol/L) c	3.27 (1.41–7.60)	0.006
eGFR (mL/min/1.73 m2)	0.98 (0.98–0.99)	0.04
Body mass index (kg/m2)	0.95 (0.90–1.01)	0.10
Mean arterial blood pressure (mmHg)	0.99 (0.97–1.00)	0.12
Diabetes	1.36 (0.82–2.26)	0.23
Male sex	1.50 (0.60–3.73)	0.38
Smoking history	1.19 (0.68–2.09)	0.53
Multivariate analysis (N = 216), χ 2 (model): 0.0001
LVEF (%)	0.96 (0.93–0.99)	0.008
Age (years)	1.05 (1.01–1.08)	0.009
Plasma 3‐OHB (μmol/L) a	1.004 (1.001–1.007)	0.02
NT‐proBNP (ng/L) b	1.0005 (1.000–1.001)	0.02
NYHA class	1.19 (1.02–1.38)	0.02
Free fatty acids (mmol/L) c	2.49 (0.71–8.70)	0.15
eGFR (mL/min/1.73 m2)	0.99 (0.98–1.01)	0.49

3‐OHB, 3‐hydroxy butyrate; CI, confidence interval; eGFR, estimated glomerular filtration rate; HR, hazard ratio; LVEF, left ventricular ejection fraction; NT‐proBNP, N‐terminal pro‐brain natriuretic peptide; NYHA, New York Heart Association.

P‐values < 0.05 in bold.

^a Per 1 μmol/L increase.

^b Per 1 ng/L increase.

^c Per 1 mmol/L increase.

Patient and metabolic determinants of circulating 3‐hydroxy butyrate

Determinants of 3‐OHB are presented in Table 3 . In a univariate analysis, FFA [β: 238 (95% CI: 183–297), P > 0.0001], age [β: 2.43 (95% CI: 1.14–3.72), P < 0.0001], body mass index {β: −3.15 [95% CI: −5.26–(−0.05)], P = 0.046}, diabetes [β: 44.49 (95% CI: 14.84–74.14), P = 0.003], NT‐proBNP [β: 0.03 (95% CI: 0.01–0.04), P = 0.001], and glycosylated haemoglobin [β: 1.92 (95% CI: 0.24–3.59), P = 0.025] were all significantly correlated with 3‐OHB. These associations are presented in Table 3 and depicted in Figure 3 A–D . After adjustment for other significant covariates, only FFA remained significant [β: 216 (95% CI: 165–268), P > 0.0001]. There was no association between 3‐OHB levels and storage time of the samples [β: 1.01 (95% CI: 0.99–1.01), P = 0.1].

Table 3.

Predictors of circulating 3‐hydroxy butyrate

Risk factor	β‐coefficient (95% CI)	R 2	P‐value
Univariate analysis (N = 218)
Free fatty acids (mmol/L)	238 (185–292)	0.27	<0.0001
Age (years)	2.43 (1.14–3.72)	0.06	<0.0001
NT‐proBNP (ng/L)	0.03 (0.01–0.04)	0.05	0.001
Type 2 diabetes mellitus	44.49 (14.84–74.14)	0.04	0.003
NYHA class	26.86 (5.99–47.72)	0.03	0.012
Haemoglobin A1c (mmol/L)	1.92 (0.24–3.59)	0.02	0.025
Plasma insulin (pmol/L)	−0.28 (−0.54–(−0.02))	0.02	0.036
Body mass index (kg/m2)	−3.15 (−5.26–(−0.05))	0.02	0.046
Current smoker	29.29 (−5.47–64.04)	NS	0.10
HOMA index	−3.2 (−7.5–1.1)	NS	0.14
eGFR (mL/min/1.73 m2)	−0.41 (−1.09–0.27)	NS	0.24
Fasting blood glucose (mmol/L)	4.96 (−3.75–13.67)	NS	0.26
Male sex	1.50 (0.60–3.73)	NS	0.38
Mean arterial blood pressure (mmHg)	−0.17 (−1.23–0.89)	NS	0.75
LVEF (%)	−0.22 (−1.79–1.35)	NS	0.77
Multivariate analysis (N = 218), R 2  = 0.36
Free fatty acids (mmol/L)	216 (165–268)		>0.0001
Diabetes	33 (0–66)		0.06
NT‐proBNP (ng/L)	0.01 (0.00–0.03)		0.08
Age (years)	0.98 (−0.23–2.18)		0.11
Body mass index (kg/m2)	−2.51 (−5.74–0.71)		0.13
Plasma insulin (pmol/L)	−0.14 (−0.4–0.12)		0.28
Haemoglobin A1c (mmol/L)	0.84 (−1.02–2.69)		0.38
NYHA class	2.9 (−8.38–14.18)		0.61

CI, confidence interval; eGFR, estimated glomerular filtration rate; HOMA, homeostatic model assessment; LVEF, left ventricular ejection fraction; NS, not significant; NT‐proBNP, N‐terminal pro‐brain natriuretic peptide; NYHA, New York Heart Association.

P‐values < 0.05 in bold.

Figure 3.

Univariate linear regression analysis. The associations between plasma 3‐hydroxy butyrate (3‐OHB) and (A) free fatty acid (FFA), (B) age, (C) N‐terminal pro‐brain natriuretic peptide (NT‐proBNP), and (D) glycosylated haemoglobin (HbA1c). Plasma 3‐OHB, NT‐proBNP, and FFA depicted on a Log2‐scale.

Discussion

In a cohort of 218 patients with HFrEF, circulating 3‐OHB levels were an independent predictor of long‐term, all‐cause mortality even when adjusting for established heart failure risk factors. Circulating 3‐OHB levels displayed only weak associations between known biomarkers and risk factors in chronic HFrEF but were associated with circulating FFAs.

Circulating ketone levels and long‐term prognosis in patients with chronic heart failure with reduced ejection fraction

Increased 3‐OHB levels are associated with an increased risk of new‐onset HFrEF, especially in women. ¹⁵ Small cohort studies have suggested that total circulating ketone bodies and exhaled acetone may act as biomarkers to detect acute HFrEF. ¹⁶ Furthermore, it has been demonstrated that urinary ketones increase as NYHA class increases and LVEF decreases. ¹⁷ In 102 patients with non‐ischaemic HFrEF and a mean LVEF of 50%, fasting exhaled breath acetone levels were associated with pulmonary artery wedge pressure, NYHA class, and BNP. ⁸ Exhaled breath acetone has also demonstrated prognostic value in patients with chronic HFrEF ⁹ but displays a non‐linear relationship with 3‐OHB and may not reflect actual blood ketone levels, ¹⁸ in part because acetone only constitutes 25% of circulating ketone bodies.

In 369 patients with STEMI, a recent study reported an association between non‐fasting ketone levels 24 h after reperfusion and functional outcomes at a 4 month follow‐up. ⁷ This was not a study in patients with heart failure, illustrated by an LVEF of 54% at follow‐up. Intriguingly, patients with higher circulating ketone body levels had larger infarct sizes and lower LVEF at a 4 month follow‐up, which heralds a larger risk for future development of heart failure. The authors discussed that it remains unknown whether the increased circulating ketone bodies reflect a maladaptive response or the reverse. In patients with arrhythmogenic cardiomyopathy, elevated plasma‐3‐OHB was a predictor of a combined endpoint consisting of mainly arrhythmic events. ¹⁹

Our study demonstrated that plasma 3‐OHB measured after an overnight fast was strongly associated with all‐cause mortality in patients with chronic HFrEF. In our study, survival curves separated visually after 2 years of follow‐up, highlighting the potential of circulating 3‐OHB as a long‐term prognostic marker. The association had strength similar to that of established risk factors such as LVEF, NT‐proBNP, and NYHA functional class. Of note, 3‐OHB predicted mortality independently of and with a similar statistical strength as did NT‐proBNP. The wide range of 3‐OHB‐levels from 14 to 694 μM conveyed a considerable differentiation in patient risk; that is, for every 100 μM increase in plasma 3‐OHB, the risk of death increased by 40%. Therefore, our findings call for larger studies on the prognostic value of 3‐OHB in patients with chronic HFrEF.

In knock‐out mice with heart failure, which are unable to oxidize 3‐OHB, inability to utilize ketones is detrimental to the prognosis. ²⁰ This suggests that increased ketone utilization is a beneficial, adaptive response to the increased energetic demands of the heart during chronic heart failure. ²¹ This is supported by the finding that 3‐OHB extraction from the blood is increased even in patients with mild to moderately symptomatic HFrEF (NYHA II–III) and increases with decreasing LVEF. ²² The correlation between 3‐OHB and long‐term mortality in the present study obviously raises the question whether this reflects a causal relationship and, if so, whether increased 3‐OHB levels are adaptive or maladaptive. Recent studies have found beneficial haemodynamic effects of therapeutic ketone supplements in patients with HFrEF and healthy individuals. ⁶ , ²³ The cardiac effects of a long‐term increase in circulating ketone bodies, and thus myocardial uptake, ⁴ require further research.

Determinants of circulating 3‐hydroxy butyrate levels in patients with chronic heart failure with reduced ejection fraction

Circulating ketone levels increase when hepatic ketone production is increased or utilization or excretion is reduced. ¹ , ²⁴ Augmented production results from a rising supply of ketogenic substrates, mainly FFAs, and activation of the hepatic ketogenic mechanism. ²⁵ Insulin and glucagon are the key regulatory hormones of ketogenesis. ²⁶ Norepinephrine, cortisol, and growth hormone augment lipolysis and thus circulating FFAs and ketone production. ²⁷ In this context, ketone bodies may, similarly to lactate, be viewed as a stress indicator, and it seems plausible that the circulating concentrations of ketone bodies may, to some extent, reflect the levels of ketogenic stress hormones and indirectly the degree of cardiac and overall physiological stress. Basal serum levels of 3‐OHB in humans fall in the low micro‐molar range but rise to a few hundred μM after 12–16 h of fasting. ²⁸ Levels of 3‐OHB increase to 1–2 mM after 2 days of fasting ²⁹ or during acute, severe illness. ³⁰

To date, the largest published study of 3‐OHB levels in patients with heart failure included 45 hospitalized patients with an average LVEF of 39% and various heart failure aetiologies, that is, chronic HFrEF, cardiac valve disease, pulmonary hypertension, and pericardial constriction. ² This seminal paper showed that 3‐OHB correlates with plasma FFA, pro‐atrial natriuretic peptide, growth hormone, noradrenaline, interleukin‐6, LVEF, and pulmonary artery wedge pressure. In a multivariate analysis, levels of circulating ketone bodies only correlated independently with FFAs, plasma glucose, and LVEF. ² Smaller studies have reported 3‐OHB levels ranging from 80 to 100 μmol/L in chronic, stable HFrEF patients, similar to our findings. In patients with advanced HFrEF (NYHA III–IV), the average 3‐OHB level was 360 μmol/L ³¹ and in patients with STEMI, it was 520 μmol/L. ⁷ As ketone production and metabolism are affected by neuro‐hormonal activation, it is conceivable that fasting ketone levels measured in stable patients are more representative for the long‐term ketone levels in the individual patient and display less variability, than in the acutely sick or fasting individuals. In line with this, the range of circulating ketones was wider and the mean values were higher in the papers by Lommi et al. ² and de Koning et al. ⁷ than in the present study.

Our study of ketones included the largest cohort to date of chronic HFrEF patients with a contemporary HFrEF diagnosis, comprehensive echocardiographic assessment, and treated according to contemporary guidelines. Our findings document a large variation in fasting circulating 3‐OHB levels between patients. We confirm the association between 3‐OHB levels and FFA levels described in HFrEF patients and healthy subjects. ² , ²⁵ These findings suggest an increased lipolytic flux, even in patients receiving modern HFrEF treatment. However, FFA levels displayed an R ²‐value of only 0.27 in the univariate analysis, indicating that other factors determine circulating 3‐OHB levels. Circulating 3‐OHB has been linked to increased NT‐proBNP in patients without heart failure with predominantly ischaemic heart disease. ³² We observed only weak correlations, suggesting that the mechanism of increased 3‐OHB levels in patients with HFrEF is not strongly mediated by NT‐proBNP. Overall, we report that 3‐OHB levels in patients with chronic HFrEF display large inter‐individual differences, correlate with circulating FFAs, but also display a large residual variation that cannot be explained by other known chronic heart failure risk factors.

Strengths and limitations

The strengths of this study include the largest cohort to date of patients with a well‐defined diagnosis of chronic HFrEF, a long follow‐up time of 7.3 years, and data on vital status in all patients. In addition, circulating 3‐OHB was measured using a validated and precise method with a high reproducibility in stable patients in a well‐defined metabolic state after an overnight fast. We also included patients with diabetes, which increases the external validity of our findings. We included FFAs, insulin, and body mass index in our analysis, making the cohort well characterized from a metabolic standpoint. This was a multicentre study, which increases its generalizability. There are inherent design‐related limitations of a retrospective cohort study. We have adjusted for confounders, using multivariate analysis. However, residual confounding is always possible, and causal inferences should be made with caution. Ninety per cent of the cohort was male, which decreases external validity. As the majority of patients were in NYHA functional class II and the mean NT‐proBNP was 360 ng/L, the cohort does not represent patients with advanced, severe HFrEF.

In the present study, we utilized hydrophilic interaction liquid chromatography tandem mass spectrometry for the determination of circulating 3‐OHB. This method is costly due to equipment and utensil prices. Furthermore, it requires some technical expertise. However, cheaper enzyme‐linked immunosorbent assays are commercially available for the detection of circulating 3‐OHB.

Conclusions

In patients with chronic HFrEF, high ketone levels predicted all‐cause mortality independently of known heart failure risk factors. Blood ketone levels were strongly associated with FFA levels. Further studies are needed to elucidate the additional value of circulating 3‐OHB levels in prognostication and the possible beneficial effects of exogenous ketone supplementation.

Conflict of interest

The authors have nothing to declare.

Funding

This work was supported by the Novo Nordisk Foundation (Novo Nordisk Fonden) (grant no. NNF17OC0028230), the Danish Heart Foundation (Hjerteforeningen) (grant no. 19‐R135‐A9280‐22126), and the Independent Research Fund Denmark (Danmarks Frie Forskningsfond) (grant no. 8020‐00120A).

Christensen, K. H. , Nielsen, R. R. , Schou, M. , Gustafsson, I. , Jorsal, A. , Flyvbjerg, A. , Tarnow, L. , Bøtker, H. E. , Kistorp, C. , Johannsen, M. , Møller, N. , and Wiggers, H. (2024) Circulating 3‐hydroxy butyrate predicts mortality in patients with chronic heart failure with reduced ejection fraction. ESC Heart Failure, 11: 837–845. 10.1002/ehf2.14476.