Acyl ghrelin improves cardiac function in heart failure and increases fractional shortening in cardiomyocytes without calcium mobilization

Abstract

Background and Aims

Ghrelin is an endogenous appetite-stimulating peptide hormone with potential cardiovascular benefits. Effects of acylated (activated) ghrelin were assessed in patients with heart failure and reduced ejection fraction (HFrEF) and in ex vivo mouse cardiomyocytes.

Methods and results

In a randomized placebo-controlled double-blind trial, 31 patients with chronic HFrEF were randomized to synthetic human acyl ghrelin (0.1 µg/kg/min) or placebo intravenously over 120 min. The primary outcome was change in cardiac output (CO). Isolated mouse cardiomyocytes were treated with acyl ghrelin and fractional shortening and calcium transients were assessed. Acyl ghrelin but not placebo increased cardiac output (acyl ghrelin: 4.08 ± 1.15 to 5.23 ± 1.98 L/min; placebo: 4.26 ± 1.23 to 4.11 ± 1.99 L/min, P < 0.001). Acyl ghrelin caused a significant increase in stroke volume and nominal increases in left ventricular ejection fraction and segmental longitudinal strain and tricuspid annular plane systolic excursion. There were no effects on blood pressure, arrhythmias, or ischaemia. Heart rate decreased nominally (acyl ghrelin: 71 ± 11 to 67 ± 11 b.p.m.; placebo 69 ± 8 to 68 ± 10 b.p.m.). In cardiomyocytes, acyl ghrelin increased fractional shortening, did not affect cellular Ca²⁺ transients, and reduced troponin I phosphorylation. The increase in fractional shortening and reduction in troponin I phosphorylation was blocked by the acyl ghrelin antagonist D-Lys 3.

Conclusion

In patients with HFrEF, acyl ghrelin increased cardiac output without causing hypotension, tachycardia, arrhythmia, or ischaemia. In isolated cardiomyocytes, acyl ghrelin increased contractility independently of preload and afterload and without Ca²⁺ mobilization, which may explain the lack of clinical side effects. Ghrelin treatment should be explored in additional randomized trials.

Clinical Trial Registration

ClinicalTrials.gov Identifier: NCT05277415

Structured Graphical Abstract

Structured Graphical Abstract.

Study protocol and main findings. LVEF = Left ventricular ejection fraction; NYHA = New York Heart Association; Acyl = Acylated (Activated); FS = Fractional Shortening; F/F0 (Fluo-3 Ca2+) = Amplitude of Ca2+ transients was measured as change in the fluo-3 fluorescence signal (F) divided by the fluorescence immediately before a stimulation pulse given under control conditions (F0); Phospho cTnI = Phosphorylated cardiac troponin I; Ser 23-24 = serine 23-24 (Phosphorylation site); TnI = troponin I.

See the editorial comment for this article ‘Acyl-ghrelin therapy for heart failure: already a novel inotrope or even more?’, by E.A. Jankowska and P. Ponikowski, https://doi.org/10.1093/eurheartj/ehad220.

Introduction

Chronic heart failure (HF) with reduced ejection fraction (HFrEF) is characterized in part by compensatory but maladaptive neurohormonal activation and cardiac remodelling. Drugs that antagonize or modulate neurohormonal activation reduce the risk of death from any cause and hospitalization for HF and are cornerstones of guideline-based HFrEF therapy.¹ However, none of these drugs directly target cardiac contractility.²

Despite effective therapy, chronic HFrEF progresses to advanced HF, with worsening quality of life and functional capacity, frequent hospitalizations for worsening HF, intolerance to neurohormonal modulator drugs, and high risk of death.³ Drugs that increase contractility and cardiac output (CO) (inotropes) are sometimes necessary in these patients. However, conventional inotropes increase cytosolic Ca²⁺ concentrations which increases contractility but also causes adverse effects and may increase mortality.^4-14 Thus conventional inotropes are not recommended for long-term use and may be considered only for short-term use in select patients with acute and/or advanced HF.¹ Myosin activators increase systolic ejection time but improved clinical outcomes only modestly in GALACTIC-HF.¹⁵ Thus, drugs that increase contractility and improve outcomes are a major unmet need in HFrEF.^1,2

Ghrelin is an endogenous 28 amino acid anabolic peptide hormone with a molecular weight of 3371 g/mol. Ghrelin is released from the stomach in response to fasting and weight loss^16-18 and is activated by acylation of amino-acid 3.^19,20 Degradation takes place and plasma half time is 24–30 min. Acyl ghrelin binds the growth hormone (GH) secretagogue receptor (GHSR) and stimulates GH release¹⁶ and acts as a centrally acting appetite stimulant.²¹ Acyl ghrelin receptors (i.e. GHSRs) are widely distributed in cardiac and skeletal muscle and endothelium.²² In rat HF models, ghrelin increased CO and fractional shortening²³; in rat myocardial infarction models, ghrelin reduced cardiac sympathetic activity and left ventricular (LV) remodelling^24,25 and apoptosis.²⁶ Small studies in human HF suggested ghrelin may improve CO,²⁷ LV ejection fraction (LVEF), and exercise capacity and reduce muscle wasting.²⁸ These data refer to variable forms of acyl ghrelin and predominantly the non-acylated inactive form.^29-36 The safety, clinical efficacy, and mechanisms of action of acyl ghrelin in HF are unknown.

In a single centre randomized double-blind placebo-controlled clinical trial, the safety and efficacy of intravenous acyl ghrelin was assessed in patients with HFrEF. To generate hypotheses for potential underlying mechanisms for any observed clinical effect, contractility, cellular Ca²⁺ transients, and phosphorylation of cardiac troponin I (cTnI), which regulates Ca²⁺ sensitivity,^37,38 in response to acyl ghrelin in isolated mouse cardiomyocytes from sham and HF mice, was also assessed.

Methods human trial

Patients

Patients were ambulatory and had symptoms and signs of moderate–severe chronic HF, New York Heart Association (NYHA) class III or ambulatory class IV, and a LVEF ≤40%. Detailed inclusion and exclusion criteria are listed in Supplementary data online, Table S1.

Intervention: acyl ghrelin

Synthetic acylated human ghrelin (ghrelin (human) acetate, product number 4071265; Clinalfa, Bachem, Bubendorf, Switzerland) hereafter referred to as acyl ghrelin, was purchased from Bachem under license from Daiichi Sankyo (Tokyo, Japan). For each patient, a stock solution was prepared from multiple vials of powder acyl ghrelin (100 μg acyl ghrelin/vial in phosphate buffer). The content of each vial dissolved in 1 mL sterile water for infusion (B. Braun, Melsungen, Germany) and visually inspected to be a clear and colourless solution, prior to adding 0.001 g (0.02 mL of 50 g/L stock) human albumin (for concentration 0.001 g/mL = 0.1%) to each vial and further mixed with normal saline (NaCl 9 mg/mL, B. Braun). The proportions of stock solution and saline were according to patient body weight, resulting in a total volume of 100 mL (to ensure the same volume infusion for all patients). The final intravenous infusion rate was 0.50 mL/min volume (total volume 60 mL over 120 min), equivalent to 0.1 µg (30 pmol)/kg/min. The dose was chosen based on target engagement in previous human physiologic studies.^27,39 The infusion was given for 120 min and continued until all measurements after the 120 min time point had been completed (thus delivering some extra acyl ghrelin beyond 120 min). As a result, the total infusion time was 171 (161–172) min for acyl ghrelin and 171 (163–175) min for placebo. The total amount of acyl ghrelin delivered was 17.1 (16.1–17.2; min 11.1; max 17.5) μg/kg.

Control: placebo

A solution of physiological saline (NaCl 9 mg/mL, B. Braun) was infused at the same volume and rate (0.50 mL/min) and total duration as the acyl ghrelin infusion.

Procedures and data collection

A CONSORT diagram is shown in Supplementary data online, Figure S1. The detailed study protocol and procedures are shown schematically in Figure 1. Patients arrived to the laboratory in the fasting state, signed informed consent, and had eligibility criteria confirmed (see Supplementary data online, Table S1). Two peripheral venous catheters were inserted in the right arm (one in the antecubital vein for infusion and one on the dorsal hand for blood sampling; the left arm was not accessible due to repeat echocardiography and other measurements on the left side). Blood samples were drawn and analysed, including estimated glomerular filtration rate (eGFR) and plasma glucose (as part of inclusion/exclusion criteria). Thereafter, patients underwent additional examinations, and then a standardized breakfast of 500 kcal was consumed ad lib without coffee or tea. Patients were randomized in parallel by block randomization in groups of four using paper envelopes to acyl ghrelin or placebo and the infusions were given over 120 min. A dedicated unblinded study nurse prepared acyl ghrelin and placebo for infusion and patients, and all other investigators and study personnel were blinded. Immediately prior to, during, and after the 120 min infusion, and also 30 min after stopping the infusion, with patients remaining in the semi-recumbent position, CO and other parameters were measured, symptoms were assessed, and blood sampling and echocardiography were performed (Figure 1). After the assessment 30 min post-infusion, patients were offered a meal and observed in the clinical study unit for 60 min or longer; underwent symptom assessment, physical examination, and ECG; and were discharged to home in the late afternoon. Patients returned 2–5 days later for repeat measurements. Patients were followed prospectively for 30 days after treatment for morbidity and mortality outcomes.

Figure 1.

Human HFrEF RCT: study procedures.

Pre-specified efficacy outcomes

The primary outcome was difference between acyl ghrelin and placebo in change in CO from start of infusion (time 0) to end of infusion (time 120 min). Numerous secondary outcome measures from inert gas rebreathing, beat-to-beat haemodynamics, echocardiography, and plasma biomarkers in response to the 120 min infusion of acyl ghrelin or placebo, 30 min after stopping the infusion, and 2–5 days after stopping the infusion were also assessed.

Pre-specified safety outcomes

Safety outcomes included changes in heart rate and systolic blood pressure, hypotension and symptomatic hypotension, ischaemia, and arrhythmia during or after the infusion and clinical outcomes. There were no formal stopping rules. All interpretation and decisions during treatment were based on clinical judgement by the blinded investigator.

Hypotension was assessed in two ways: the difference in change in blood pressure between the groups and number of patients experiencing hypotension defined as a drop of systolic blood pressure of >20 mmHg or to <80 mmHg.

Conventional inotropes increase heart rate, and acyl ghrelin has been reported to reduce heart rate in healthy human subjects by reducing sympathetic and stimulating parasympathetic drive.^24,40,41 Conventional inotropes also cause ventricular tachyarrhythmias. Therefore, heart rate and arrhythmia were monitored continuously during and for 30 min after infusion using beat-to-beat plethysmography. Heart rate was reported in the overall population and also separately in patients with neither atrial nor ventricular pacing.

Ischaemia was assessed in two ways: The difference in change in high-sensitivity troponin T (hsTnT) between the groups was quantified, and ischaemia in any single patient was defined as symptoms, ECG changes, and hsTnT values consistent with ischaemia. QTc was assessed manually on ECGs as they were recorded.

Patients were followed prospectively from the 2–5 day visit until 30 days after treatment. Morbidity and mortality outcomes were all-cause death, HF hospitalization, heart transplantation, and LV assist device.

Data collection methods and definitions

Cardiac output

Non-invasive resting CO was assessed in duplicate at each measurement time point using the Innocor® device (Innovision, Odense, Denmark). Each measurement was immediately assessed for quality according to instructions and criteria from the manufacturer. Measurement that did not meet quality criteria were discarded and replaced by a new measurement. The Innocor® is a non-invasive device that measures pulmonary blood flow using an inert gas-rebreathing technique and measures oxygen uptake (VO₂) directly. It has been validated for CO against the Fick and thermodilution methods.^42,43 The coefficient of variation is low (5%–7%),⁴³ similar to the gold-standard Fick method⁴⁴ and superior to other non-invasive methods.⁴⁴ Pulmonary blood flow measured by inert gas rebreathing using the Innocor® and corrected for the intrapulmonary shunt has been shown to provide a reliable estimate of CO.⁴³

Continuous haemodynamic monitoring

Immediately before (time 0), during, and up to 30 min after the infusion, a plethysmography-based (finger-cuff) approach was used to continuously measure beat-to-beat blood pressure, heart rate, and rhythm (Nexfin®, BMEYE B.V., Amsterdam, the Netherlands). The data were monitored continuously, and data reported were averaged over the 15 min following the inert gas-rebreathing measurements.

ECG

Electrocardiogram records were collected and assessed upon arrival to the laboratory, at 60 and 120 min of infusion, prior to discharge, and at the 2–5 day follow-up visit. The purpose of ECGs was to assess eligibility and to provide safety monitoring during infusion and to clear the patient for discharge. Ischaemia and QT changes were manually assessed by a blinded investigator assessing ECGs as they were recorded.

Echocardiography

Echocardiography was performed by a technician blinded to treatment allocation, and all echocardiography images were analysed and interpreted by one independent investigator blinded to treatment allocation and clinical history of the patients. LVEF was measured using the Teichholz method. Tricuspid annular plane systolic excursion (TAPSE) was assessed with M-mode echocardiography. Speckle tracking was used for LV longitudinal strain. Changes in strain over time was calculated using the average of the regional strain from septal and inferior segments. Stroke volume (measured by echocardiography) was derived from the LV outflow tract (LVOT) area and the LVOT velocity time integral (VTI) (LVOT area × LVOT VTI). The primary outcome measure, CO, was measured by inert gas rebreathing as described above, but as an additional exploratory outcome measure, CO was obtained also from the echocardiography-derived stroke volume × heart rate.

Symptoms and signs

The following symptoms were assessed as yes/no prior to infusion and after 30, 60, and 120 min of infusion, 30 min after infusion was stopped, and at the 2–5 days follow-up; headache, dizziness, dyspnoea, central chest pain, flushing, sleepiness, gastrointestinal symptoms, hunger, thirst, and other symptoms. The following signs from physical examination were recorded upon reporting to the laboratory, prior to discharge and at 2–5 days: rales, peripheral oedema, jugular venous distension, hepatomegaly, and S3 gallop.

Blood samples

Blood samples were collected at time points shown in Figure 1. Blood was collected in ethylenediaminetetraacetic acid (EDTA) and serum tubes and immediately centrifuged, and plasma and serum were aliquoted and stored at −70°C until analysis. Separate samples were dedicated for acyl ghrelin pharmacokinetics and were prepared with a protease inhibitor cocktail consisting of 5.5 μL 10 mM KR-62436 (dipeptidyl peptidase 4 [DPP4] inhibitor) in dimethyl sulfoxide (DMSO) and a SIGMAFAST® protease inhibitor tablet (both from Sigma-Aldrich Corp., St. Louis, MO, USA) dissolved in 2100 µL of distilled water (50× stock). Blood samples were drawn using 6 mL EDTA plasma tubes and immediately put on ice, and 160 µL of the 50× protease inhibitor cocktail was added. Tubes were vortexed for 10 s and centrifuged at 4°C, 10 min, 2500 relative centrifugal force (RCF, or g). The resulting supernatant (plasma) was then pipetted into Eppendorf tubes (450 µL each) and immediately frozen at −70°C until analysed.

Serum and plasma biomarkers

Kidney function was measured by serum creatinine levels at Karolinska University Hospital central laboratory. eGFR was calculated according to the Chronic Kidney Disease Epidemiology Collaboration formula (eGFR = 142×min (Scr/κ, 1) α×max (Scr/κ, 1) −1.200 × 0.9938 age × 1.012 [if female). N-terminal pro B-type natriuretic peptide (NT-proBNP) was analysed by proBNPII (Roche Diagnostics, Bromma, Sweden). Plasma glucose measurements were from EDTA-containing whole venous blood tubes and analysed with a photometric point-of-care technique, the HemoCue® Glucose 201 RT (Ängelholm, Sweden). High-sensitivity troponin T and additional biomarkers were analysed in the Karolinska University Hospital central laboratory.

Acyl ghrelin and pharmacokinetics

Plasma concentrations of acyl ghrelin were assayed by a custom ELISA using electrochemiluminescence detection. Plasma samples were thawed and vortexed. They were then analysed in duplicate on 96-well multi-spot plates (Meso Scale Diagnostics, Rockville, MD, USA) coated with capture antibodies against acyl ghrelin according to manufacturer instructions. The Meso Scale Diagnostics Sector Imager 2400 was used to read the plates. The coefficients of variation (CV%) calculated from plasma concentrations after curve fitting were 9.3/3.5 (intra-/inter-assay CV%). The lower limit of detection was 3.2 ng/L and upper limit was set at 10 000 ng/L.

Statistics

Continuous variables were tested for normal distribution using D’Agostino & Pearson and Shapiro–Wilk normality tests. A majority of haemodynamics and echocardiography variables were normally distributed, and therefore for consistency, all are reported as mean ± standard deviation. A majority of laboratory variables were non-normally distributed, and therefore, for consistency, all are reported as median (interquartile range). Categorical variables are reported as number (%).

In order to compare the effect of acyl ghrelin vs. placebo on continuous variables over time, a two-way repeated measures analysis of variance (2W-RM-ANOVA) was used to evaluate treatment effect within treatment groups. When appropriate, sphericity was not assumed and a Geisser–Greenhouse correction was applied. The 2W-RM-ANOVA has two factors: treatment (two levels: acyl ghrelin or placebo) and time (three levels: time 0, after 60 min infusion, and after 120 min infusion). The main analysis to be considered in this statistical model is the interaction (treatment × time) where we test the significance of the null-hypothesis that the differences between treatments are the same at all time points. Post-hoc testing with Tukey adjusted P-values for repeated measures were used for changes over time within each treatment group.

Sample size

The power calculation was based on difference in change of CO and conducted as described in Supplementary data online, Table S2. With a power of 80%, two-sided alpha of 0.05 and an assumed 10% minimal treatment difference, the sample size required 10 patients in each group. Accounting for additional margins, sample size was set to 15 patients in each group.

Ethics

Acyl ghrelin has been used previously in studies in healthy humans and in patients with HF and other conditions. The trial was conducted according to International Conference on Harmonization and Good Clinical Practice guidelines and the Declaration of Helsinki. The trial was approved by the local (Stockholm) ethics committee (number 2008/1:12 and 2008/1695-31). In this physiological study of an endogenous peptide, the ethics committee waived the need for medical products agency (MPA) approval, based on a previous waiver for the same treatment in another study of gastrointestinal effects of acyl ghrelin (MPA waiver number 159:2007/16373; ethics Stockholm number 2007/119-31/1). All patients provided written informed consent.

Methods ex vivo mouse cardiomyocyte studies

Mouse cardiomyocyte isolation from sham and heart failure mice

Twelve- to 16-week-old C57BL6 mice with and without HF were studied. For the HF mouse model, myocardial infarction (MI) was induced by permanent ligation of the left coronary artery, as previously described.⁴⁵ Briefly, mice underwent thoracotomy and subsequent left coronary ligation; ∼80% survived and developed HF during the follow-up period (4–6 weeks). Sham mice underwent a sham procedure with thoracotomy. In HF mice, HF was verified using echocardiography and after euthanasia by measuring the weight of the heart and the lungs, and comparisons were made to sham mice. For HF and sham mice, after euthanasia, single cardiomyocytes were isolated from the left and right ventricles (mouse hearts are dominated by left ventricles), and atria were excluded, following the protocols developed by the Alliance for Cellular Signalling (AfCS Procedure Protocol ID PP00000 125) as previously described.⁴⁶

Cytosolic [Ca²⁺] and cell fractional shortening in response to acyl ghrelin, vehicle, and acyl ghrelin plus D-Lys 3

Mouse cardiomyocytes were incubated with the fluorescent indicator Fluo-3 AM followed by washing for >5 min as previously described.⁴⁷ Cardiomyocytes were plated on laminin-coated glass bottom dishes following the protocols developed by the Alliance for Cellular Signalling (AfCS Procedure Protocol ID PP00000 125) as previously described, which enables cardiomyocyte contraction in a load-independent fashion.^48,49 The dishes were placed in a custom built perfusion/stimulation chamber and continuously perfused with O₂/CO₂ (95/5%) bubbled Tyrode solution with the following composition (in mM): NaCl 121, KCl 5.0, CaCl₂ 1.8, MgCl₂ 0.5, NaH2PO₄ 0.4, NaHCO₃ 24, EDTA 0.1, and glucose 5.5. Cardiomyocytes were stimulated to contract using an electrical field between two platinum electrodes attached to the perfusion/stimulation chamber. Fluorescence was measured using a confocal microscope Bio-Rad MRC 1024 unit attached to a Nikon Diaphot inverted microscope (40–60× oil immersion lenses). Confocal images were analysed offline using ImageJ (National Institutes of Health; https://imagej.net/ij). Line scan confocal images were obtained from paced cardiomyocytes with the line scan running along the long axis of the cell. Changes in the emitted fluorescence, representing changes in free cytoplasmic Ca²⁺, were quantified. The amplitude of Ca²⁺ transients was measured as the change in the fluo-3 fluorescence signal (F) divided by the fluorescence immediately before a stimulation pulse given under control conditions (F0). Fractional cell shortening was calculated from the line scan images as the fractional change in the cell length from rest to maximal contraction.

For acyl ghrelin experiments, cardiomyocytes were superfused with physiological buffer (Tyrode solution) or Tyrode + acyl ghrelin (Bachem, Bubendorf, Switzerland) (100 nM) or Tyrode + acyl ghrelin + D-Lys 3 (3 μM). D-Lys 3 is a specific growth hormone secretagogue (i.e. acyl ghrelin) receptor 1a (GHSR1a) antagonist. The 100 nM concentration was chosen with the aim of saturating the receptor. This was based on previous studies with Chinese hamster ovary and human embryonic kidney cells transfected with the acyl ghrelin receptor, which established the EC₅₀ at 2.5 nM and suggested a plateau of the EC curve around 100 nM.^16,50 All cells were perfused for 15 min with the respective solution before measurements, except in some experiments where D-Lys 3 was introduced in perfusion system 10 min before adding acyl ghrelin. All experiments were performed at room temperature (∼24^°C).

Cardiac troponin I immunoblotting and phosphorylation

Post-translational modifications of myofibrillar proteins, e.g. troponin, can regulate the myofibril Ca²⁺ sensitivity. For instance, phosphorylation of troponin I serine 22–23 (human) and 23–24 (mouse) has been linked to reduced myofibrillar Ca²⁺ sensitivity.^37,38 Antibodies targeting phosphorylated serine 23–24 residues on mouse troponin I were used to test if acyl ghrelin changed the protein phosphorylation. Cardiomyocytes isolated from sham mice (24 aliquots of cells derived from 4 animals) and from HF mice (6 aliquots of cells derived from 2 animals) were treated with vehicle, acyl ghrelin, or acyl ghrelin + D-Lys as above and were pelleted and homogenated. Protein lysates were separated by electrophoresis and transferred onto membranes. Membranes were incubated with primary antibody: rabbit phospho-troponin I (cardiac) (Ser 23–24) antibody (Cell Signaling #4004, BioNordika, Solna, Sweden), rabbit troponin I (Cell Signaling #4002). Then infrared-labelled secondary antibodies (IRDye 680 and IRDye 800, 1:5000, Licor, Bad Homburg, Germany) were used. Immunoreactive bands were analysed using the Odyssey Infrared Imaging System. Band densities were quantified with Image J, normalized to total cTnI and final data expressed as fold increase compared with the vehicle group.

Statistics

Statistical comparison between two groups was performed using Student’s t-test (unpaired). For comparison between >2 groups, ANOVA was used. A P < 0.05 was used as definition for statistical significance. Average data was presented as mean ± standard error of the mean (SEM).

Ethics

All animal experiments were performed under the ethics approvals Stockholm N19/15 and N273/15.

Results human trial

Patients

Thirty-four patients consented and were screened. Three did not meet eligibility criteria, and 31 patients were randomized to intravenous acyl ghrelin vs. placebo (see Supplementary data online, Figure S1). One patient (placebo) experienced repeated urinary urgency and dizziness early after start of infusion and placebo infusion was interrupted and the patient was excluded and replaced, yielding 15 patients in each group. Baseline characteristics are shown in Table 1 and were generally similar between the groups. Median age was 70 and 71 years, respectively, and 13% were women in both groups. Baseline CO was 4.08 ± 1.15 L/min (acyl ghrelin group) and 4.26 ± 1.23 L/min (placebo group) and baseline LVEF was 27.4 ± 7.4% and 26.7 ± 12.2%, respectively.

Table 1.

Selected baseline clinical characteristics

Variable	Acyl ghrelin n = 15 (50%)	Placebo n = 15 (50%)
Demographics
Age (years)	70 (62–76)	71 (55–75)
Male sex	13 (87%)	13 (87%)
Medical history
Duration of HF (years)	8 (2–16)	9 (5–13)
Ischaemic heart disease	7 (47%)	11 (73%)
Prior MI	6 (40%)	11 (73%)
Prior CABG	3 (27%)	7 (63%)
Prior PCI	5 (45%)	4 (36%)
Prior stroke/TIA	2 (18%)	2 (18%)
History of atrial fibrillation or flutter	12 (80%)	10 (67%)
Hypertension	11(73%)	8 (53%)
Hyperlipidaemia	10 (67%)	1 (73%)
Pulmonary disease	4 (27%)	3 (20%)
Diabetes mellitus	7(47%)	6 (40%)
Chronic kidney disease	3 (20%)	4 (27%)
Treatment
ACEi/ARB	15 (100%)	15 (100%)
Beta blocker	15 (100%)	15 (100%)
MRA	11 (73%)	14 (93%)
Loop diuretic	13 (87%)	14 (93%)
Digoxin	1 (7%)	3 (20%)
Amiodarone	2 (13%)	3 (20%)
Statin	9 (60%)	10 (66%)
Insulin	2 (13%)	6 (40%)
Oral glucose lowering treatment	3 (20%)	6 (40%)
CRT	8 (53%)	10 (67%)
ICD	14 (93%)	11 (73%)
Secondary prevention	5 (33%)	2 (13%)
Primary prevention	9 (60%)	10 (66%)
Physical examination
BMI (kg/m2)	29 (25–30)	30 (28–32)
Systolic blood pressure, mm Hg	100 (85–110)	105 (92–127)
Heart rate, b.p.m., overall population	70 (60–80)	70 (70–74)
Heart rate, b.p.m., non-paced	69 (65–77)	70 (60–70)
NYHA class III	14 (93%)	14 (93%)
NYHA class IV	1 (7%)	1 (7%)
LVEF, %	27.4 ± 7.4	26.7 ± 12.2
Cardiac output, L/min	4.08 ± 1.15	4.26 ± 1.23
Laboratory
eGFR, mL/min/1.73m2	61 (46–66)	61 (47–71)
Haemoglobin, g/dL	132 (124–147)	127 (116–145)
NT-proBNP, ng/L	2160 (1060–3890)	2180 (780–4630)
ECG
Sinus rhythm	5 (33%)	7 (47%)
Atrial fibrillation/flutter	5 (33%)	7 (47%)
Atrial pace	5 (33%)	1 (7%)
Ventricular pace	7 (47%)	10 (67%)

Blood pressure was obtained with manual cuff measurement on the exam table prior to infusion.

Heart rate was obtained through pulse palpation for 1 min on the exam table prior to infusion.

Heart rate overall is based on all 15 patients in each group.

Heart rate non-paced is based on six patients in the acyl ghrelin group and four patients in the placebo group. Many patients had CRT; therefore, heart rate is reported in the overall population and also separately in patients with non-paced atria and ventricles. Many patients were switching between native and paced atrial and ventricular depolarization; the ECG data shown are the dominant rhythm on three ECGs as judged by the investigator.

LVEF and cardiac output are immediately before infusion and are shown also in Figure 2.

eGFR was calculated by the CKD-EPI method; CKD is defined as eGFR <60 mL/min/1.73m².

Laboratory and ECG data are at time of presentation to the laboratory the morning of the study and used for screening.

BMI, body mass index; b.p.m., beats per minute; NYHA class, New York Heart Association class; MI, myocardial infarction; CABG, coronary artery bypass grafting; PCI, percutaneous coronary intervention; TIA, transient ischaemic attack; eGFR, estimated glomerular filtration rate; NT-proBNP, N-terminal pro B-type natriuretic peptide; CRT, cardiac resynchronization therapy; ICD, implantable cardioverter defibrillator; MRA, mineralocorticoid receptor antagonist; ACEi, angiotensin-converting-enzyme inhibitor; ARB, angiotensin receptor blocker.

Primary outcome measure

Figure 2A shows mean (± standard deviation) CO from baseline and during 120 min infusion of acyl ghrelin vs. placebo, and 30 min after discontinuation of the infusion. There were no missing values at any of the measurement points. In the acyl ghrelin group, CO increased from 4.08 ± 1.15 L/min at baseline to 5.23 ± 1.98 L/min at 120 min (28% increase, within group P < 0.001). In the placebo group, CO changed from 4.26 ± 1.23 at baseline to 4.11 ± 1.99 L/min at 120 min (within group P = 0.89). The P-value for the interaction of time * treatment assignment was <0.001.

Figure 2.

Human HFrEF RCT: primary and secondary efficacy outcome measures. The P-interaction is for the treatment * time interaction and compares the effect in patients with HFrEF of acyl ghrelin vs. placebo on continuous variables over time using a two-way repeated measures ANOVA (2W-RM-ANOVA). Values represent means and error bars represent standard deviations. A Primary outcome: cardiac output by inert gas rebreathing immediately prior to, during, and after acyl ghrelin/placebo infusion. In the acyl ghrelin group, CO increased with infusion and fell after stop of infusion. In the placebo group, there was no significant change in CO, P-interaction time * treatment <0.001. B Secondary outcomes. Panel A: left ventricular ejection fraction (LVEF) by echocardiography. There was a nominal but not statistically significant increase in LVEF in acyl ghrelin vs. placebo. Panel B: Tricuspid annular plane systolic excursion (TAPSE) by echocardiography. There was a nominal but not statistically significant increase in TAPSE in acyl ghrelin vs. placebo. Panel C: Stroke volume (SV) by echocardiography. There was a significant difference in change of SV in favour of acyl ghrelin, P = 0.021. Panel D: Segmental strain by echocardiography. Speckle tracking was used for assessment of left ventricular peak longitudinal strain. Changes in strain over time were calculated using the average of the regional strain from septal and inferior segments. There was a nominal but not statistically significant increase in longitudinal strain in acyl ghrelin vs. placebo.

Secondary outcome measures

Figure 2B panels A–D show secondary outcome measures by echocardiography. There was a nominal increase in LVEF (Figure 2B panel A) and TAPSE (representing right ventricular function) (Figure 2B panel B), a significant increase in LV stroke volume (Figure 2B panel C), and a nominal increase in LV global strain (Figure 2B panel D).

Table 2 shows additional secondary outcome measures obtained by inert gas rebreathing. Pulmonary blood flow increased with acyl ghrelin (3.61 ± 0.70 to 4.08 ± 0.90 L/min) but not with placebo (3.82 ± 1.06 to 3.66 ± 1.12; P-interaction 0.008). The VO₂ remained constant, but the SpO₂ declined slightly during acyl ghrelin infusion, leading to an increase in the calculated intrapulmonary shunt (Table 2). Thus, with acyl ghrelin treatment, the measured pulmonary blood flow increased less than the CO because with increased CO, more blood was shunted through nonventilated space and was not captured by the inert gas-rebreathing method, which measures only blood flow through ventilated areas of the lungs. This also explains why SpO₂ declined slightly with acyl ghrelin treatment.

Table 2.

Selected secondary haemodynamic outcome measures during 2 h infusion and 30 min after stopping infusion

Variable	Acyl ghrelin				Placebo				P-value interactiona
	Start of infusion 0 min	After 60 min infusion	End of infusion 120 min	30 min after stop infusion	Start of infusion 0 min	After 60 min infusion	End of infusion 120 min	30 min after stop infusion
PBF, L/min	3.61 ± 0.70	3.76 ± 0.76	4.08 ± 0.90b	4.14 ± 0.97	3.82 ± 1.06	3.52 ± 1.08	3.66 ± 1.12	3.30 ± 1.02	0.008
A-V O2 diff, %	27.6 ± 14.3	23.6 ± 12.9b	20.9 ± 8.5b	21.6 ± 12.1	26.4 ± 9.0	28.5 ± 11.3	26.6 ± 10.7	29.0 ± 15.1	0.0049
SpO2, %	95.2 ± 3.4	93.5 ± 3.5b	93.4 ± 3.6b	94.7 ± 3.9	95.7 ± 1.9	95.2 ± 2.0	95.7 ± 1.8	95.9 ± 2.3	0.0054
Shunt, %	9.9 ± 10.6	16.8 ± 11.8b	18.0 ± 12.3b	12.6 ± 14.3	9.5 ± 6.7	11.7 ± 10.1	10.5 ± 8.6	10.3 ± 10.7	0.0083
SvO2, %	67.9 ± 14.4	70.0 ± 13.8	71.7 ± 10.1	73.3 ± 12.8	69.3 ± 8.6	66.8 ± 10.5	69.1 ± 9.9	67.1 ± 14.6	0.094
SV, ml	59.10 ± 17.82	69.64 ± 22.75b	79.83 ± 30.30b	74.31 ± 26.09	62.0 ± 15.7	58.9 ± 17.7	60.7 ± 15.1	55.1 ± 15.9	<0.001
VO2, L/min	0.20 ± 0.07	0.19 ± 0.07	0.19 ± 0.08	0.19 ± 0.08	0.20 ± 0.08	0.20 ± 0.09	0.19 ± 0.08	0.18 ± 0.08	0.88

The primary outcome measure (cardiac output from inert gas rebreathing) and secondary outcome measures from echocardiography are shown in Fig. 2 and not in this Table.

Heart rate and blood pressure are considered safety measures and are shown in Table 3.

All parameters in this Table 2 were obtained from inert gas rebreathing.

The P-value in the right column is the treatment * time interaction and compares the effect of acyl ghrelin vs. placebo on continuous variables over time using a two-way repeated measures ANOVA (2W-RM-ANOVA).

P < 0.05 vs. baseline within the same group (group or placebo group).

Measured data: VO₂, SpO₂, PBF.

Calculated data: SV, A-V O₂ diff, Shunt, SvO₂.

PBF = pulmonary blood flow; A-VO₂ diff = atrial-venous oxygen difference; SpO₂ = oxygen saturation; SvO₂ = mixed venous oxygen saturation; SV = stroke volume; VO₂ = oxygen uptake.

Safety measures

Safety measures during intravenous acyl ghrelin and placebo are shown in Table 3. Systolic blood pressure declined minimally in both groups, consistent with patients resting in the semi-recumbent position (acyl ghrelin: 104.7 ± 18.4 to 99.0 ± 21.1 mmHg; placebo: 120.2 ± 24.4 to 115.5 ± 22.3 mmHg, P-interaction 0.83). Heart rate did not increase in either group. Instead, heart rate decreased slightly in the acyl ghrelin group overall (70.7 ± 11.3 to 67.1 ± 11.1 in acyl ghrelin vs. 68.7 ± 7.6 to 67.6 ± 9.5 in placebo, P-interaction 0.15) and more so in the acyl ghrelin group with non-paced QRS complexes (76.6 ± 13.1 to 69.3 ± 12.9 in acyl ghrelin vs. 68.0 ± 6.1 to 66.3 ± 11.4 in placebo, P-interaction 0.34). Among symptoms assessed as yes or no, flushing at 60 min was more frequent in the acyl ghrelin group compared with placebo (n = 7 vs. n = 0). During and after infusion, there were no changes in eGFR, potassium, liver function biomarkers, or hsTnT.

Table 3.

Safety and pharmacokinetics outcome measures during 2 h infusion and 30 min after stopping infusion

Variable	Acyl ghrelin				Placebo				P-value a interaction
	Start infusion 0 min	After 60 min infusion	End of infusion 120 min	30 min after stop infusion	Start infusion 0 min	After 60 min infusion	End of infusion 120 min	30 min after stop infusion
Vital parameters
dBP, mm Hg	58.6 ± 8.4	56.5 ± 8.7	56.1 ± 10.3	58.9 ± 7.5	63.5 ± 10.1	66.5 ± 8.0	63.4 ± 8.2	68.4 ± 10.1	0.11
sBP, mm Hg	104.7 ± 18.4	99.0 ± 17.6	99.0 ± 21.1	105.7 ± 20.5	120.2 ± 24.4	118.2 ± 21.3	115.5 ± 22.3	123.6 ± 25.2b	0.83
mBP, mm Hg	74.9 ± 11.3	71.4 ± 10.2	70.9 ± 13.1	75.5 ± 12.0	82.2 ± 14.1	84.1 ± 11.9	81.1 ± 12.1	87.0 ± 14.5b	0.23
HR, bpm	70.7 ± 11.3	68.5 ± 11.2	67.1 ± 11.1	68.2 ± 11.4	68.7 ± 7.6	68.6 ± 8.1	67.6 ± 9.5	68.1 ± 8.6	0.15
HR non-paced, bpm	76.6 ± 13.1	71.3 ± 13.5	69.3 ± 12.9	71.8 ± 13.6	68.0 ± 6.1	69.0 ± 5.0	66.3 ± 11.4	68.3 ± 10.8	0.34
Hypotension	0	0	0	1 (7%)	0	1 (7%)	0	0	ND
Symptomatic hypotension	0	0	0	0	0	0	0	0	ND
Symptoms
Headache	0	1 (7%)	0	0	2 (13%)	0	1 (7%)	1 (7%)	ND
Dizziness	1 (7%)	0	0	0	1 (7%)	0	0	0	ND
Dyspnoea	2 (13%)	2 (13%)	2 (13%)	0	3 (20%)	1 (7%)	1 (7%)	2 (13%)	ND
Chest pain	1 (7%)	0	0	0	0	0	0	0	ND
Flushing	1 (7%)	7 (47%)	6 (40%)	1 (7%)	0	0	1 (7%)	1 (7%)	ND
Sleepiness	8 (43%)	6 (40%)	10 (67%)	6 (40%)	9 (60%)	6 (40%)	9 (60%)	3 (20%)	ND
Gastrointestinal symptoms	1 (7%)	2 (7%)	2 (7%)	1 (7%)	3 (20%)	4 (27%)	6 (40%)	4 (27%)	ND
Hunger (1–10)	3 (1–5)	5 (2–7)	5 (2–8)	5 (4–9)	2 (1–5)	3 (1–5)	4 (2–5)	6 (3–7)	ND
Thirst (1–10)	4 (2–5)	4 (2–6)	5 (4–8)	5 (3–8)	5 (2–5)	5 (2–6)	5 (1–7)	4 (3–7)	ND
Other symptoms	2 (13%)	2 (13%)	2 (13%)	1 (7%)	3 (20%)	3 (20%)	3 (20%)	3 (20%)	ND
Biomarkers
Acyl ghrelin, ng/L	76 (42–131)	6841 (4755–>10 000)	6536 (4835–>10 000)	503 (372–833)	60 (40–78)	64 (32–85)	59 (37–73)	58 (37–89)	<0.001
NT-proBNP, ng/L	2080 (1080–3980)	2340b (1100–4580)	2630b (1220–4840)	2820 (1280–5170)	2380 (787–4060)	2490b (918–4770)	2750b (860–4900)	2820 (968–4690)	0.63
hsTnT, ng/L	18 (15–30)	19 (16–31)	17 (15–30)	17 (15–29)	19 (10–30)	20 (11–29)	20 (11–29)	21 (12–31)	0.67
eGFR, mL/min/1.73 m2	66 (48–70)	62 (48–72)	58 (49–70)	60 (49–68)	62 (49–78)	62 (49–76)	62 (53–74)	63 (53–71)	0.31
ASAT (ukat/L)	0.41 (0.33–0.47)	0.44 (0.36–0.47)	0.41 (0.33–0.46)	0.43 (0.36–0.49)	0.37 (0.27–0.45)	0.40 (0.30–0.44)	0.37 (0.24–0.45)	0.34 (0.30–0.50)	0.97
ALAT (ukat/L)	0.31 (0.24–0.38)	0.31 (0.31–0.40)	0.31 (0.21–0.39)	0.32 (0.23–0.41)	0.35 (0.25–0.48)	0.35 (0.23–0.49)	0.34 (0.23–0.48)	0.33 (0.22–0.47)	0.58
GT (ukat/L)	0.77 (0.65–1.40)	0.79 (0.62–1.40)	0.80 (48) (0.62–1.30)	0.82 (49) (0.62–1.40)	0.97 (0.47–1.50)	0.94 (0.46–1.50)	0.95 (0.48–1.50)	0.98 (0.48–1.50)	0.51

Blood pressure and heart rate were obtained from beat-to-beat plethysmography and are reported as median values of 15 min of beat-to-beat sampling.

Heart rate overall is based on all 15 patients in each group.

Heart rate non-paced is based on six patients in the acyl ghrelin group and four patients in the placebo group.

The P-value in the right column is the treatment * time interaction and compares the effect of acyl ghrelin vs. placebo on continuous variables over time using a two-way repeated measures ANOVA (2W-RM-ANOVA).

P < 0.05 vs. baseline within the same group (acyl ghrelin group or placebo group).

b.p.m. = beats per minute; HR = heart rate; dBP = diastolic blood pressure; sBP = systolic blood pressure; mBP = mean blood pressure; ND = not done; NT-proBNP = N-terminal pro B-type natriuretic peptide; hsTnT = high-sensitive Troponin T; eGFR = estimated glomerular filtration rate; ASAT = aspartate amino transferase; ALAT = alanine amino transferase; GT = gamma-glutamyl transferase.

ECG assessment of ischaemia did not reveal any findings suggestive of ischaemia. ECG assessment of QTc and potential changes in QTc was impaired by lack of an ECG at start of infusion (time 0), artefacts, irregular rhythms, and QRS complexes that were paced and/or prolonged and/or shifted between native and paced. There were variable changes in QTc over time, but none of these changes were judged to be related to treatment.

During and immediately after the 120 min infusion, NT-proBNP increased in both groups, from 2080 (1080–3980) to 2820 (1280–5170) ng/L in the acyl ghrelin group and from 2380 (787–4060) to 2820 (968–4690) ng/L in the placebo group (P = 0.63) (Table 3 and Supplementary data online, Figure S2). hsTnT did not change in either group, from 18 (15–30) to 17 (15–29) ng/L in the acyl ghrelin group and from 19 (10–30) to 21 (12–31) ng/L in the placebo group (P = 0.67) (Table 3 and Supplementary data online, Figure S3).

Outcome and safety measures at 2–5 day follow-up visit and long term

To assess any potential remaining treatment effect and any potential emerging safety concerns, patients returned to the clinic 2–5 days after treatment for follow-up measurements (see Supplementary data online, Table S3). In patients treated with acyl ghrelin, CO had not fully returned to baseline. There were no effects or differences between groups in heart rate or blood pressure. Between baseline and the 2–5 day follow-up, NT-proBNP increased somewhat in the acyl ghrelin group, from 2080 (1080–3980) to 2450 (942–4920) ng/L, but not in the placebo group, from 2380 (787–4060) to 2020 (762–2870) ng/L (P = 0.01) (see Supplementary data online, Table S3). This was a general effect, with no patient experiencing any dramatic change (see Supplementary data online, Figure S2). hsTnT did not change on a group level, from 18 (15–30) to 20 (16–29) ng/L in the acyl ghrelin group and from 19 (10–30) to 22 (11–29) ng/L in the placebo group, P = 0.36 (see Supplementary data online, Table S3). However, there was one outlier value in the acyl ghrelin group, where hsTnT increased approximately four-fold, from 38 to 164 ng/L between baseline and follow-up (which occurred on day 4 for this patient) (see Supplementary data online, Figure S3). This patient had severe non-revascularizable coronary disease, but did not report any symptoms or have any ECG changes at the follow-up on day 4 and was alive without any events at 30 day follow-up. There were no changes or difference between groups in eGFR, high-sensitivity C-reactive protein, liver enzymes, triglycerides, haemoglobin, or white blood cell count. At vital status follow-up at 30 days after treatment, there were no patients in either group who had died, been hospitalized for heart failure, or received heart transplantation or a left ventricular assist device.

Results ex vivo mouse cardiomyocyte studies

In the human HFrEF randomized trial above, acyl ghrelin increased CO without tachycardia, arrhythmia, ischaemia, or hypotension, suggesting acyl ghrelin may increase cardiac contractility through a mechanism different from conventional inotropes. But whether the increased CO was load dependent (a result of reduced afterload) and whether it was a result of increased cytosolic Ca²⁺ could not be studied. To assess mechanisms for increased contractility, fractional shortening, Ca²⁺ transients, and post-translational modifications to troponin I were investigated in isolated mouse cardiomyocytes from HF and sham mice incubated with acyl ghrelin, acyl ghrelin + D-Lys (a selective acyl ghrelin receptor antagonist), or vehicle.

Cardiomyocyte fractional shortening and Ca²⁺ transients

In mice with LAD ligation, development of HF was verified at 4–6 weeks. Supplementary data online, Figure S4A shows that HF mice compared with sham mice had reduced fractional shortening and increased heart rate, end-diastolic and end-systolic dimensions, and interventricular septum and posterior wall thickness. Supplementary data online, Figure S4B shows that after euthanasia, hearts and lungs from HF mice compared with sham mice had greater weights both in absolute terms and as a proportion of total body weight. Figure 3A shows fractional shortening during contraction in sham cardiomyocytes isolated from six mice, after 15 min incubation with vehicle (average of n = 54 cardiomyocytes), acyl ghrelin (average of n = 67 cardiomyocytes), and acyl ghrelin + D-Lys, a selective acyl ghrelin receptor antagonist (average of n = 18 cardiomyocytes). Figure 3B shows the same for HF cardiomyocytes (isolated from n = 3 mice; incubated with vehicle [average of n = 43 cardiomyocytes], acyl ghrelin [average of n = 63 cardiomyocytes], and acyl ghrelin + D-Lys [average of n = 16 cardiomyocytes]). Fractional shortening was greater in acyl ghrelin than in vehicle-treated cells, with a similar difference in sham vs. in HF cardiomyocytes. D-Lys blocked the effect of acyl ghrelin. Subsequently, whether the increased fractional shortening was due to changes in Ca²⁺ transients was investigated. Figure 3C shows Ca²⁺ transients measured during three contractions in a representative sham cardiomyocyte. Figure 3D shows the same in a representative HF cardiomyocyte. The rate of increase and decay in cytosolic Ca²⁺ concentration were similar in representative cells incubated with vehicle, acyl ghrelin, and acyl ghrelin + D-Lys, in both sham and HF cardiomyocytes. Figure 3E shows the same Ca²⁺ transients averaged in the 54, 67, and 18 sham cardiomyocytes, respectively. Figure 3F shows the same averaged in the 43, 63, and 16 HF cardiomyocytes, respectively. Acyl ghrelin did not affect Ca²⁺ transients, suggesting acyl ghrelin does not increase cytosolic Ca²⁺ concentrations but instead achieves increased contractility potentially through increased myofilament response to Ca²⁺ (i.e. Ca²⁺ sensitization). There was a tendency toward reduced Ca²⁺ transients with D-Lys.

Figure 3.

Ex vivo studies of sham and heart failure mouse cardiomyocyte fractional shortening during contraction (A and B), CA2+ transients in representative cells (C and D), CA2+ transients quantified (E and F), and cardiac troponin I phosphorylation (G and H). Cardiomyocytes were isolated from sham and HF mice were incubated for 15 min with acyl ghrelin, vehicle, or acyl ghrelin + D-Lys (a specific acyl ghrelin receptor antagonist). Cells were electrically stimulated to contract at 60 b.p.m. The number of cardiomyocytes analysed were for sham cells: n = 54 treated with vehicle, n = 67 treated with acyl ghrelin, and n = 18 treated with acyl ghrelin + D-Lys; and for HF cells: n = 43, n = 63, and n = 16, respectively. A and B Cardiomyocyte fractional shortening. Contractility is assessed by percent fractional shortening (FS). In both sham and HF cardiomyocytes, acyl ghrelin increased factional shortening in an acyl ghrelin receptor specific fashion. Since this is ex vivo, these effects are independent of loading conditions (left ventricular preload or afterload). Bars show average % fractional shortening (FS) in each group. Error bars show standard deviation. * P < 0.001. C and D Ca²⁺ transients obtained from three single representative cells each from sham and HF mice (one for vehicle, one for acyl ghrelin, one for acyl ghrelin + D-Lys) during three beats. The Ca²⁺-transient amplitudes were not different after acyl ghrelin treatment in neither sham nor HF cells. Left: Representative recordings using confocal line scan imaging of three separate cardiomyocytes loaded with the fluorescent Ca2+ indicator fluo-3 (Fluo-3 AM), electrically paced at 1 Hz and treated with vehicle, acyl ghrelin, or acyl ghrelin + D-Lys. The images show each cardiomyocyte contracting three times with an increase in fluorescence intensity, in early systole, representing an increase in cytosolic Ca²⁺ concentrations. The arrows point to an indentation indicating the change in length of the cardiomyocyte during systolic contraction (fractional shortening, which is shown quantified in Figure 2). The indentation (fractional shortening) is greater in the representative cardiomyocyte treated with acyl ghrelin. Right: Plots of quantified fluorescence representing cytosolic Ca²⁺ fluorescent indicator as a function of time, during three electrically paced action potentials in the same three representative cardiomyocytes shown on the left. F/F0 (Fluo-3 Ca²⁺) on Y-axis represents amplitude change of Ca²⁺ fluorescent indicator, i.e. fold increase peak fluorescence upon electrical stimulation over baseline fluorescence, during three action potentials during 3 s (x-axis). E and F Average Ca2+ transients from all the cardiomyocytes studied. There was no difference between vehicle vs. acyl ghrelin vs. acyl ghrelin + D-Lys in neither sham nor HF cells. Data are presented as mean ± standard deviation. The number of cardiomyocytes analysed is indicated in the figure and corresponds to the same cells analysed for fractional shortening measurements (Figure 3A). G and H. Representative immunoblots of phosphorylated and total cTnI from sham and HF cardiomyocytes (Original Western blots are shown in Supplementary data online, Figures S5A–S5D). Cardiomyocytes were isolated from sham and HF mice and were aliquoted in different tubes and incubated for 15 min with acyl ghrelin (eight aliquots for sham and two aliquots for HF), vehicle (eight aliquots for sham and two aliquots for HF), or acyl ghrelin + D-Lys (eight aliquots for sham and two aliquots for HF). In the presence of acyl ghrelin, the phosphorylation signal was lower compared with vehicle. Pretreatment of cardiomyocytes with the ghrelin receptor antagonist D-Lys 3 prevented the reduction in cTnI phosphorylation. This indicates that the inotropic effect of ghrelin could potentially involve post-translational modifications of cTnI.

Cardiac troponin I phosphorylation

Figure 3G shows representative immunoblots of phosphorylated and total cTnI from sham and HF cardiomyocytes. Figure 3H shows quantification of immunoblot band intensities. In the presence of acyl ghrelin, the phosphorylation signal was lower compared with vehicle. Pretreatment of cardiomyocytes with the ghrelin receptor antagonist D-Lys 3 prevented the reduction in cTnI phosphorylation. This indicates that the inotropic effect of ghrelin could potentially involve post-translational modifications of cTnI.

Discussion

In this randomized double-blind placebo-controlled trial in patients with HFrEF, a 120 min infusion of acyl ghrelin increased CO by 28% without adverse effects such as hypotension, tachycardia, arrhythmia, or ischaemia. This was due to a significant increase in LV stroke volume since heart rate was unchanged or even slightly reduced. At the 2–5 day follow-up, one patient in the acyl ghrelin group had an increase in hsTnT, and there was an increase in the median NT-proBNP in the acyl ghrelin but not placebo group. Experimental ex vivo mouse cardiomyocyte studies showed that acyl ghrelin increased cardiomyocyte fractional shortening (contractility) in a load-independent fashion and without Ca²⁺ mobilization and potentially through reducing phosphorylation of cTnI, which is a known regulator of Ca²⁺ sensitivity (Structured Graphical Abstract). This potentially novel mechanism is different from that of conventional inotropes (which increase cytosolic Ca²⁺ leading to increased contractility but also harmful adverse effects) and myotropes (which increase systolic ejection time). The absence of Ca²⁺ mobilization and the reduction in cTnI phosphorylation observed in cardiomyocytes may explain the absence of adverse clinical effects with acyl ghrelin observed in the clinical trial.

Catecholamines (e.g. dobutamine) and phosphodiesterase-3 inhibitors (e.g. milrinone) increase cytosolic Ca²⁺ concentrations, causing tachycardia, arrythmias, and ischaemia. Levosimendan is a Ca²⁺-sensitizing agent but appears to have also phosphodiesterase inhibitory and thus Ca²⁺-mobilizing effects.² These agents also cause hypotension and are neutral on or worsen clinical outcomes.^4-14 Omecamtiv mecarbil is an orally available myosin activator which increases systolic ejection time and thus systolic function but had only a modest effect on reducing HF outcomes in the phase III GALACTIC-HF trial.¹⁵ Thus, drugs that increase contractility and improve outcomes are a major unmet need in HFrEF.^1,2 In our human HFrEF trial, CO was increased without tachycardia, arrhythmia, or hypotension, which, if confirmed, has important implications for a potential future ‘safe’ contractile agent (inotrope or myotrope). During the 2 h infusion, there was a slight increase in NT-proBNP in both groups, which is consistent with the recumbent state and increased venous return. At follow-up 2–5 days after the 2 h infusion, the elevated NT-proBNP persisted more in the acyl ghrelin than in the placebo group. The reasons are unknown. The short half-time of acyl ghrelin and the different time course of acyl ghrelin treatment and NT-proBNP effects suggest that these findings may be chance. Acyl ghrelin caused pulmonary and systemic vasodilation and the relatively elevated NT-proBNP levels at follow-up may be a rebound phenomenon. Ghrelin is expressed in cardiomyocytes and in fact colocalizes with proBNP in the same cardiomyocyte intracellular compartments.⁵¹ After coronary artery bypass grafting, both ghrelin and NT-proBNP may increase despite improved haemodynamics.⁵² Several studies have reported associations between ghrelin and growth hormonal-related peptides and natriuretic peptides but with poorly understood regulatory mechanisms and possibly different patterns in health and disease.^53,54

Concentrations of and appetite and anabolic effects of ghrelin have been studied in normal human subjects^39,55-68 and patients with cancer cachexia,^69,70 chronic obstructive pulmonary disease cachexia,⁷¹ growth hormone deficiency,^72,73 obesity,⁷⁴ the metabolic syndrome,⁷⁵ and Cushing’s syndrome.⁷⁶ Effects were variable and inconsistent. Small human studies in HF suggested ghrelin may improve CO²⁷ and LVEF,²⁸ but these do not specify whether the acyl ghrelin compound used was acylated or not. It is also unclear which form of ghrelin that is responsible for the observed actions in these studies, since acyl and des-acyl have different binding sites and effects in cardiomyocytes.⁷⁷

The mechanisms of action and mechanism(s) responsible for the clinical benefit of acyl ghrelin in the present study are unknown and probably multifactorial. The growth hormone secretagogue (acyl ghrelin) receptor (GHSR) is a 7-transmembrane G-protein coupled receptor with numerous downstream signalling pathways. It heterodimerizes with other G-protein coupled receptors, which may result in Gs, Gi, or Gq effects. There are four (and possibly five) different ghrelin receptors in the heart and vasculature with poorly understood function.^78,79 The Ca²⁺-handling machinery is complex.^80,81 Thus, Ca²⁺ handling in response to acyl ghrelin and the mechanisms by which acyl ghrelin increases cardiomyocyte contractility without mobilizing Ca²⁺ need to be further studied. The reduction in cTnI phosphorylation, if confirmed, could be one mechanism for increased Ca²⁺ sensitivity, but the downstream signalling from the acyl ghrelin receptor(s) responsible for this potential post-translational modification are unknown and need to be elucidated.

If the observed effects of acyl ghrelin on cardiac contractility and CO are confirmed, they provide a rationale for further clinical development. Chronic acyl ghrelin treatment could be associated with adverse effects related to GH, glucose intolerance, and lipid metabolism. However, acyl ghrelin may also have potentially additional beneficial metabolic and cardioprotective effects, such as reduced inflammation⁸² and apoptosis,⁸³ improved endothelial function through improved nitric oxide bioavailability,⁷⁵ and increased lean body mass.²⁸ Furthermore, acyl ghrelin has also been observed to have vasodilatory effects. This is as major limitation of existing inotropic drugs, where it leads to hypotension. In the present study, a major potential advantage was that acyl ghrelin did not reduce blood pressure, but since acyl ghrelin increased CO, by definition there was also a reduction in systemic vascular resistance. The ex vivo contractility studies show that acyl ghrelin increased contractility in a load-independent fashion, but it cannot be ruled out that the increase in CO in the human trial was due to improved contractility but also due to reduced systemic vascular resistance and afterload.

Limitations

The trial was conducted prior to widespread use of angiotensin receptor–neprilysin inhibitors and sodium–glucose co-transporter 2 inhibitors. The clinical trial was a physiological investigation with an endogenous peptide hormone rather than a novel molecule. Therefore, medical products agency approval was exempt, and there were no formal pharmacokinetic, pharmacodynamic, or thorough QT/QTc (QTQ) studies, and there was no formal reporting of adverse events. Nevertheless, the extensive safety monitoring during and after the 120 min infusion, 2–5 days after treatment, and vital status at 30 days, ensured both protection of the participating research subjects and assessment of potential safety issues with acyl ghrelin treatment for design of future trials.

Clinical or ex vivo dose or concentration response studies were not performed, and thus the optimal dose remains unknown. In the clinical study the single dose of 0.1 µg (30 pmol)/kg/min was based on target engagement in previous human physiologic studies.^27,39 In the ex vivo study the concentration was based on previous cellular studies.^16,50 Acyl ghrelin concentrations were measured during treatment, but many measurements reached the upper limit of the assay despite multiple dilutions, and their values may not be reliable. Cardiac output was not measured invasively, but the non-invasive inert gas-rebreathing method and the Innocor® device have been validated against both Fick and thermodilution^42,43 and allowed consistent and rapid collection of data during the intensive intravenous infusion protocol, with minimal risk and discomfort to patients. However, this precluded measurement of some other haemodynamic parameters, such as filling pressure or pulmonary vascular resistance. Echocardiography images were not interpreted by a core lab. Ejection fraction is reported by Teichholz. The protocol was intensive and focused on CO and did not allow time for lengthy echocardiography examinations. Therefore, ejection fraction by other methods was not consistently available. Randomization did not employ any techniques such as minimization to reduce differences between groups; thus, some baseline characteristics differed somewhat between groups. However, age, sex distribution, and ejection fraction were similar, and the difference in baseline mean CO (ghrelin 4.08 L/min vs. placebo 4.26 L/min was much smaller than the difference in response: (acyl ghrelin 4.08 to 5.23 L/min vs. placebo: 4.26 to 4.11 L/min). Cardiac output was the pre-specified primary outcome. There were several secondary outcomes, and there was no adjustment for multiplicity. Ex vivo cardiomyocyte studies included measurement of contractility (fractional shortening) and Ca²⁺ transients. The exploratory ex vivo data do not allow any conclusions regarding the mechanism of increased contractility. In cardiomyocytes, there was a possible tendency toward reduced Ca²⁺ transients with D-Lys, which may reflect chance or that D-Lys may have some non-specific effects in the experimental setting. Electromechanical properties, sarcoendoplasmic reticulum Ca²⁺ contents using caffeine sparks, or experimental measures of Ca²⁺ sensitivity were not assessed. However, the absence of changes in cytoplasmic Ca²⁺ transients following acyl ghrelin treatment suggested that the L-type Ca²⁺ currents, sarcoplasmic reticulum Ca²⁺ release, and total SR [Ca²⁺] would not be affected.

Conclusion

In patients with HFrEF, 120 min of intravenous acyl ghrelin compared with placebo improved CO without causing adverse hypotension, arrhythmias, tachycardia, or ischaemia. Ex vivo murine cardiomyocyte studies suggested that acyl ghrelin increased cardiomyocyte contractility in a load-independent fashion and without Ca²⁺ mobilization. This proof-of-concept clinical trial with exploratory ex vivo data should be further explored in further clinical and mechanistic studies.

Author contributions

Lars H Lund, Camilla Hage, Gianluigi Pironti, Tonje Thorvaldsen, Ulrika Ljung-Faxén, Stanislava Zabarovskaja, Kambiz Shahgaldi, Dominic-Luc Webb, Per M. Hellström, Daniel C. Andersson, and Marcus Ståhlberg

Supplementary data

Supplementary data is available at European Heart Journal online.

Data availability

Scientific and medical researchers can request access to data after publication of the primary manuscript in a peer-reviewed journal. Researchers may contact lars.lund@ki.se with inquiries.

Funding

Funded by grants to LHL from The Swedish Research Council (grant 523-2014-2336), the Swedish Heart-Lung foundation (grants 20150557 and 20190310), Karolinska Institutet (grant 2-70/2014), and Stockholm County Council (grant 20140220). CH was supported by the Swedish Research Council (grant 20180899). DCA was supported by the Swedish Heart-Lung Foundation, Swedish Society for Medical Research (SSMF).