Empire HF trial  was designed, conducted and reported in accordance with a protocol in compliance with Good Clinical Practice and in accordance with the Declaration of Helsinki. All participants signed informed consent prior to inclusion. The Empire HF study protocol was approved by the relevant institutional review board (Danish National Committee on Health Research Ethics, number H-17010756).
This present Empire HF Biomarker study was an exploratory study of the Empire HF trial, which has been described, and published [26, 27]. The trial was an investigator-initiated, multicentre, randomized, double-blind, placebo-controlled trial, in which HFrEF patients were randomly assigned to empagliflozin or matching placebo (1:1) for 12 weeks between June 29, 2017, through September 10, 2019.
A full list of inclusion and exclusion criteria is provided in appendix p3. Briefly, stable HFrEF patients aged ≥ 18 years, with New York Heart Association (NYHA) functional class I–III symptoms and left ventricular ejection fraction (LVEF) of 40% or less were eligible. Patients with known type 2 diabetes were treated in accordance with European and national guidelines and were required to have a glycated haemoglobin (HbA1C) level in the range of 48–83 mmol/mol (6.5–10.0%) and stable doses of anti-glycaemic treatment within the last 30 days. Exclusion criteria included symptomatic hypotension with a systolic blood pressure < 95 mmHg, estimated glomerular filtration rate ≤ 30 mL/min/1.73 m2, admission for HF within 30 days, or admission for hypoglycaemia in the past 12 months .
All study related procedures including blood samples were performed at two sites (Odense University Hospital, and Herlev and Gentofte University Hospital). Eligible patients were randomly assigned in a double-blind fashion using a fixed-randomization schedule, according to blocks of 10. Patients underwent an examination program at randomization, including clinical examination, blood tests, and were evaluated in the outpatient clinic after 45 ± 10 days for examination of volume status, and recording of adverse events. A blood test was performed to assess renal function and electrolyte levels. At follow-up, the investigation program was repeated after 12 weeks.
Blood tests were performed at baseline and repeated at 12 weeks follow-up. Plasma concentrations of GDF-15, and hsTNT were analysed on fasting blood samples at baseline and follow-up, which were immediately centrifuged upon collection and stored at − 80 °C. A batch analysis was performed at a central laboratory (GDF-15 Roche Elecsys Assay; Roche Diagnostics GmbH, Mannheim, Germany) . Plasma concentrations of hsCRP were analysed on fasting blood samples at baseline and follow-up on the Atellica essay, and Cobas8000 platform with a coefficient variation < 10%.
Echocardiographic protocol has been previously described in detail . In brief, transthoracic echocardiography was performed on a Vivid e9 ultrasound system (General Electric, Horten, Norway), and stored digitally for offline analysis. For two-dimensional and Doppler images, three consecutive beats for patients in sinus rhythm and multiple beats for patients with atrial fibrillation were measured and averaged. Left ventricular (LV) end systolic volume (LVESV), LV end-diastolic volume (LVEDV), and left atrial volume index (LAVi) were assessed using the biplane method of disks. LV dimensions were measured from frozen end-diastolic and end-systolic 2D images in the parasternal long axis to assess LV mass (LVM).
Plasma volume was estimated as: (1 − hematocrit) × (α + [β × weight in kg]), in which α = 1530 and β = 41 for men, whereas α = 864 and β = 47.9 for women [29, 30].
Primary efficacy measure was the between-group difference in the change of plasma GDF-15 from baseline to 12 weeks follow-up. Secondary exploratory measures included the between-group differences in the changes of hsCRP and hsTNT.
No specific sample size estimation was performed for this post-hoc analysis, and study population size was the sample size of the main Empire HF trial . The primary statistical analysis was based on all included patients with available plasma GDF-15 measurement based on the per protocol principle. We replicated the analyses with the full study cohort, and findings remained consistent with the per protocol analysis in a sensitivity analysis (Additional file 1: Online Appendix p4).
Plasma GDF-15 was log-transformed due to the skewed distribution. The between group difference in GDF-15 was analysed using a linear mixed effect model with a random intercept to account for repeated measurements from the same individual and reported as a ratio of change value with 95% confidence intervals (95% CIs) for the between-group changes.
For normally distributed measurements, the aforementioned statistical analysis was used but reported as a coefficient value with 95% CIs. To reduce the risk for false discoveries in a post-hoc study, we adjusted for age, sex, body mass index (BMI), estimated glomerular filtration (eGFR), and diabetes.
Pearson’s correlation coefficients, the corresponding two-sided 95%CIs, and p-values were calculated to examine the association between the change in the log-transformed plasma GDF-15 concentrations from baseline to 12 weeks and the following changes in the cardiac, renal, and metabolic measurements; LVESV, LVEDV, LAVI, LVM, systolic blood pressure, the metabolic parameters; weight, BMI, HbA1C, and the renal parameters; haematocrit, estimated glomerular filtration rate (eGFR), and plasma volume.
A two-sided significance level of 0.05 was considered significant. Data frequencies are expressed as mean standard deviation (SD) for normally distributed variables, number (%) for categorical variables, and median with interquartile range (IQR) for non-normally distributed variables. Statistical analyses were conducted using Stata Statistical Software, version 16 (Stata Corp, College. Station, Texas, USA). The Empire HF is publicly registered on ClinicalTrials.gov, NCT03198585, and EudraCT, 2017-001341-27.