Primary and secondary endpoints

The primary endpoint was a change in LVEF assessed by Simpson’s biplane method at rest and during dobutamine stress echocardiography. Secondary endpoints were changes in exercise capacity and changes in other echocardiographic measures of systolic function including: global longitudinal strain (GLS) at rest, global longitudinal strain rate (GLSR) at rest, and wall motion score index (WMSI) at rest and during dobutamine stress echocardiography.

Study drug and dosage

Subjects had a minimum 2 week washout period for their glucose lowering therapy prior to the first baseline visit. The study drugs liraglutide/placebo subcutaneous injections and metformin tablets were titrated in an identical manner in both periods: 0.6 mg liraglutide/placebo od + 500 mg metformin bid was increased after 14 days to 1.2 mg od + (1000 mg + 500 mg) daily and to 1.8 mg od + 1000 mg bid after 28 days [11]. Efforts were made to give subjects the same dosage of study drug in both periods.

Ethics and dissemination

This study was approved by the Regional Committee on Biomedical Research Ethics of the Capital Region of Denmark and the Danish Medicines Agency. The study has been carried out in accordance with the ICH-GCP (International Conference on Harmonization-Good Clinical Practice) standards and was monitored by the GCP-unit for eastern Denmark. Written informed consent was obtained from each participant.

Dobutamine stress echocardiography

The details of echocardiography protocol have been described previously [11]. In short, two-dimensional echocardiography was performed at rest and during dobutamine infusion using a M5S transducer (Vivid E9, GE Vingmed Ultrasound, Horten, Norway) and analyzed off-line (GE EchoPAC V. 112). Dobutamine was administered an incremental regimen to reach target HR for each stress level [12]. LVEF was calculated using the Simpson biplane method [13] from contrast-enhanced images. Baseline images were used as reference for each stress level and for the examinations in the following visits in an attempt to get comparable visualization of the LV. The echocardiography examinations were performed by four investigators (AS, OWN, OK, and PK). Global longitudinal strain (GLS) and strain rate (GLSR) was assessed using 2D speckle tracking and calculated as the average of the peak systolic values for the apical 4-chamber, 2-chamber, and long-axis views. Wall motion score (WMS) was assessed using a 16-segment model of the left ventricle and graded by the following score: 1 = normal or hyperkinetic; 2 = hypokinetic; 3 = akinetic; 4 = dyskinetic or aneurysmal [13]. Wall motion score index (WMSI) was calculated as the sum of the scores divided by the number of segments visualized. An abnormal stress response was defined as stress induced regional wall motion abnormalities (RWMA), either because of increased WMS at peak stress compared to rest (ischemic response) or because of decreased WMS at low stress (biphasic) or peak stress (viable). LV mass index (2D method) and relative wall thickness (RWT) was calculated as recommended [13]. All echocardiography analyses were performed by one observer (PK). Consensus for LVEF was achieved by an average of the LVEF measurements between two observers in any questionable cases. WMS assessment was also reviewed by one senior cardiologist (AS or OWN). If the discrepancy between the observer (PK) and the senior cardiologist involved more than one segment classified as abnormal or more than two points in total WMS score, the images were also assessed by a second senior cardiologist and consensus was obtained.

Cycle ergometer exercise tolerance test

A standard cycle ergometer exercise tolerance test was performed with a workload appropriate for each subject: a starting work load of 25 W with an increasing work load of 25 or 50 W every 2 min. Subjects were encouraged to exercise until maximal exhaustion. The maximal exercise capacity was expressed as total exercise duration and as estimated metabolic equivalents (METs) [14]: METs = [12 × workload(watt)/weight(kg) + 3.5]/(3.5 ml/kg/min).

Statistical analyses

Power-calculation has been described previously [11]. A post hoc power calculation analysis was added based on the actual data. The SD for the difference between two values for the measurements in placebo period was used. At rest a SD of 5.0 % was observed, and with n = 30 patients a paired analysis provided 80 % power to detect a minimum detectable difference of 2.7 %. For low stress, peak stress and recovery a SD of 6.5, 7.3 and 9 % was observed and with n = 29, n = 24, and n = 29 this provided 80 % power to detect difference of 3.5, 4.4 and 4.9 %, respectively. A dropout rate of approximately 20 % was estimated. Continuous variables were summarized as the mean ± SD or medians with interquartile ranges, and categorical variables were summarized as percentages. The intention-to-treat (ITT) population was defined as subjects who completed minimum one measurement series in one period. The per-protocol population was defined as subjects who completed both measurement series in both intervention periods. A linear-mixed model with random effects for subjects and fixed effects for period and treatment was used in the analysis of the treatment effect on the primary endpoint in the ITT-population. The paired t test was used to compare the treatment effects of liraglutide and placebo for the per-protocol population. For non-normally distributed data Wilcoxon signed-rank test was used. Subgroup analyses were performed according to baseline DSE response. A two-sided p value of less than 0.05 was considered statistically significant. Reproducibility for LVEF was assessed in four randomly selected patients with four levels of stress (baseline, low stress, peak stress and recovery), totaling 16 measurements per observer. Interobserver variability for three observers (PK, OWN, AS) and intraobserver variability for one observer (PK) were calculated by obtaining variance estimates using two-way and one-way analysis of variance (ANOVA) and were expressed as a coefficient of variation (CoV) and a coefficient of repeatability (COR) [15, 16]. All statistical analyses were performed with Stata 13.1 (StataCorp, TX, USA).

Effect of liraglutide on systolic function

Liraglutide did not affect any significant changes in LV end-diastolic and end-systolic volume (Additional file 1: Table S2). Nine subjects in the per-protocol population had abnormal stress response at baseline and sub-group analyses did not reveal any significant changes in LVEF at any stress level (Additional file 1: Table S3).

Effect of liraglutide on exercise test performance

Exercise tolerance test results were available for 21 subjects for all 4 visits. No significant differences in the maximal achieved METS was observed between the liraglutide and placebo treatment period (METS −0.13 ± 0.6 vs. −0.42 ± 0.87; 95 % CI −0.22 to 0.80; p = 0.244). The total exercise time was slightly reduced at the end of both treatment periods, but with no difference in the treatment effect (−25 vs. −24 s; p = 0.980, respectively).

Effect of liraglutide on metabolic and hemodynamic parameters

Systolic and diastolic blood pressure measurements at study visits or during dobutamine stress did not change significantly; however, a significant increase in the resting heart rate was observed (6.2 beats per minute (bpm); 95 % CI 0.8–11.5; p = 0.001). Heart rate was also increased after liraglutide treatment at low stress (10.3 bpm; 95 % CI 0.2–20.4; p = 0.046) and peak stress levels (5.3 bpm; 95 % CI 1.2–9.5; p = 0.014) (Additional file 1: Table S4).

Reproducibility

For LVEF measurement the interobserver and intraobserver variability were, SD 3.4 and 2.2 %; CoV 4.4 and 2.9 %; COR 6.6 and 6.1 %, respectively.

Safety and compliance

Safety and adverse events

In general the study drugs were well-tolerated as only three subjects discontinued the study due to intolerance to medication. Two patients experienced MI during the study and were subsequently withdrawn from the study. Both of these events were deemed by the investigators to be unrelated to treatment. One patient experienced an allergic reaction during the DSE at baseline visit. This patient subsequently underwent ergometer stress echocardiography without contrast during all four visits.

Strengths and limitations

Although comparable DSE scans were aimed for at all four levels of stress and between visits by using reference images from the baseline visit, we cannot exclude that some differences in the scanning positions may have contributed to the overall variation. However, the use of contrast-enhanced 2D echo has shown to increase the accuracy and reproducibility echocardiographic examinations [41]. The homogeneity of our study populations limits the application of our results to other patient groups such as patients with heart failure or patients with longer duration of diabetes and poor glycemic control.

Perspectives

The presence of subclinical LV dysfunction and increased risk of heart failure in subjects with T2D and CAD necessities exploration of glucose lowering medication that may improve and prevent further deterioration of cardiac function. In this context, an understanding of how GLP-1 RAs affect the myocardial function during conditions of myocardial stress is of great importance. Any improvement in systolic function during stress may prevent the long-term deterioration of LV performance and progression to heart failure. Although, no significant improvement in systolic function was observed in our study, we did not observe any deterioration in cardiac function. Thus, liraglutide may be a safe treatment option in patients with cardiac risk factors and preserved LVEF. The recently published LEADER trial showed improved effect on CV mortality after liraglutide treatment [42]. Notably, no increased risk of hospitalization for heart failure was observed. We believe that our study adds to the further understanding of the results from the various long-term GLP-1 RA trials.