The Effect of Liraglutide on Left Ventricular Function in Chronic Heart Failure Patients With and Without Type 2 Diabetes Mellitus

Background The number of patients with diabetes in Denmark has doubled in the preceding 10-years period and has now increased to 271.000 individuals (5% of the population). T2D is a major risk factor of CHF and CHF per se is associated with insulin resistance, thus T2D and CHF often co-exists, and is associated with markedly impaired prognosis. Glycemic control in patients with the combination of T2D and CHF is complicated and the currently available treatments have proven to be inadequate in clinical trials. Therefore, new treatment modalities of hyperglycaemia in CHF are warranted.

Glucagon-like peptide 1 (GLP-1) is a naturally existing hormone, which is secreted from the incretine system. A beneficial effect of GLP-1 on cardiac function has recently been suggested in small clinical studies, demonstrating improved left ventricular ejection fraction (LVEF) in both diabetic and non-diabetic patients. Liraglutide (Victoza®) is a GLP-1-analogue developed for the treatment of T2D. However, the impact of Liraglutide on cardiac function is currently under investigation.

Objectives Primary objective: To investigate the effect of Liraglutide 1.8 mg once daily compared to placebo on LVEF in CHF patients with and without T2D after 24 weeks of treatment.

Secondary objectives: To investigate the effect of Liraglutide 1.8 mg once daily compared to placebo on left ventricular diastolic function, on plasma levels of NT-proBNP, on symptoms of heart failure and quality of life over 24 weeks of treatment.

Design Multicenter, randomized, double blind study evaluating the impact of 24 weeks treatment of Liraglutide versus placebo on cardiac function in CHF patients with and without T2D.

Population A total of 240 patients with documented systolic CHF (50% with T2DM) will be randomised. The effect of Liraglutide on left ventricular systolic and diastolic function will be evaluated by advanced echocardiography using 2D, 3D, and tissue Doppler imaging.

Primary outcome parameter is change in LVEF from visit 1 to week 24

Perspectives Irrespective of the outcome, the study will contribute with essential information regarding treatment of CHF. Potentially Liraglutide can improve heart function substantially, thus changing the prognosis of CHF-patients worldwide.

Detailed statistical analyses plan

Outcomes Primary objective To investigate the effect of liraglutide uptitrated to a maximum of 1.8 mg once daily compared with placebo for 24 weeks on the difference in left ventricular ejection fraction (LVEF) from baseline to follow-up in chronic heart failure patients (CHF) patients with and without type 2 diabetes (T2D).

Secondary objectives To investigate the effect of liraglutide compared with placebo in terms of change from baseline to follow-up of

1. Left ventricular contractile function measured by I. Global longitudinal strain by speckle tracking II. Mitral plane systolic annular max velocity by TDI (s', average of six points)
2. Left ventricular end-systolic volume (LVESV) and end-diastolic volume (LVEDV)
3. Grade of diastolic function
4. Functional capacity measured by a 6-minute walk test (6MWT)
5. Plasma NT-proBNP levels
6. Blood pressure
7. Quality of life evaluated by the MLHF questionnaire
8. Hospitalisations due to cardiovascular disease
9. All-cause mortality (will be confirmed by the Central Personal Register)

The investigators who were blinded to liraglutide versus placebo assessed all the outcomes.

Investigators will analyse data as treatment group A and B, and only when all primary analyses have been performed and approved then data can be unblinded. The managing committee has contributed to the preparation of the analysis plan and did also approve it.

Sample size Primary objective The primary outcome parameter is change in LVEF from visit 1 to week 24. A total of 241 patients were randomised to allow for approximately 20% of the randomised patients with major protocol deviations or withdrawal. With 90% power and with a two-sided significance level of 0.05, the sample size analysis estimated the study to detect a difference of 2.5% in LVEF in absolute values. The power analysis was based on a standard deviation of 6% in delta EF (absolute units). An absolute difference of 2.5% in change from baseline between the two treatment groups was based on clinical relevance as determined from studies of the impact of ACE inhibitors and betablockers on LVEF. These key outcome ACE-inhibitor and key betablocker trials showed a relative risk reduction in outcome of approximately 25 - 30% during treatment.

Stratification The randomisations were performed using two stratification variables: Gender (female/male) and diabetes (yes/no).

Data management Data were collected real time in a custom-made web based electronic case record form (eCRF) hosted by an external data management unit at Aarhus University. In compliance with GCP, a study manual on collection of endpoints and procedures were developed, and quality assurance of relevant instruments collected in sponsors trial master file.

Data on other adverse events assessed by the investigator as clinically significant, including abnormal laboratory values, were collected and recorded on standardised forms at each contact. These data were reported to the relevant authorities in accordance with current legislation and ICH-GCP guidelines. An unblinded data monitoring committee, consisting of independent cardiologists and endocrinologists, were established to evaluate events during study conduct.

Statistical analysis Data will be analysed according to the intention-to-treat principle. Sites were encouraged to obtain a final measure of the primary endpoint as soon as possible after a patient terminated study medication prematurely. Multiple imputation methods will be applied to handle missing data on the primary endpoint.

A per protocol analysis will also be performed. The per protocol population will consist of all patients who completed the study with a documented valid baseline, 24 week assessment of the primary objective and no major protocol deviations.

Data will be analysed as the difference from baseline to follow-up. Normally distributed variables will be presented as mean ± standard deviation (SD), whereas non-normally distributed variables will be presented as median [range]. Comparisons between groups will be performed by an unpaired Student's t-test or ANOVA when data are normally distributed. Mann-Whitney U test or Kruskal-Wallis one-way analysis will be used for non-normally distributed data. Subgroup analyses will be made with respect to: presence of diabetes, cause of heart failure (ischemic/non-ischemic), body mass index, NYHA-class, LVEF above/below mean, atrial fibrillation. Linear regression will be used to study correlation between outcome parameters and baseline patient characteristics. Categorical variables are presented as number and percentage; Chi-square-test is used to compare groups of categorical variables. P-values <0.05 are considered statistically significant. All calculations will be performed using a commercially available program (SPSS for Windows and Mac, version 22.0).

Population and handling of missing data Most statistical methods fail to handle any case that has a missing value, which may introduce bias or affect the representativeness of the results. Imputation preserves all cases by replacing missing data with a probable value based on other available information in the data set. Once all missing values have been imputed, the data set can then be analysed using standard techniques for complete data. Imputation methods will be applied to handle missing data on the primary endpoint.

Characteristics of patients at baseline A description of baseline characteristics by intervention group will be presented. Discrete variables will be summarised by frequencies and percentages. Percentages will be calculated according to the total number of patients in the particular group.

Continuous variables will be summarised using standard measures of central tendency and dispersion using either mean ± SD for data with normal distribution or median and interquartile range for non-normally distributed data.

Outline of figures and tables The first figure will be a consort flow chart of respectively GLP-1 versus placebo. The first tables will describe the baseline characteristics. Follow-up data will be described in tables.