Safety and efficacy of liraglutide in patients with type 2 diabetes and elevated liver enzymes: individual patient data meta-analysis of the LEAD program

Authors

  • M. J. Armstrong,

    Corresponding author
    1. Liver Unit, Queen Elizabeth University Hospital Birmingham, Birmingham, UK
    • Centre for Liver Research and NIHR Liver Biomedical Research Unit, University of Birmingham, Birmingham, UK
    Search for more papers by this author
  • D. D. Houlihan,

    1. Centre for Liver Research and NIHR Liver Biomedical Research Unit, University of Birmingham, Birmingham, UK
    2. Liver Unit, Queen Elizabeth University Hospital Birmingham, Birmingham, UK
    Search for more papers by this author
  • I. A. Rowe,

    1. Centre for Liver Research and NIHR Liver Biomedical Research Unit, University of Birmingham, Birmingham, UK
    2. Liver Unit, Queen Elizabeth University Hospital Birmingham, Birmingham, UK
    Search for more papers by this author
  • W. H. O. Clausen,

    1. Statistics, Larix A/S, Ballerup, Denmark
    Search for more papers by this author
  • B. Elbrønd,

    1. Global Medical Affairs, Novo Nordisk A/S, Soeborg, Denmark
    Search for more papers by this author
  • S. C. L. Gough,

    1. Oxford Centre for Diabetes, Endocrinology and Metabolism, and NIHR Oxford Biomedical Research Centre, Churchill Hospital, Oxford, UK
    Search for more papers by this author
    • Joint senior authors.
  • J. W. Tomlinson,

    1. Centre for Endocrinology, Diabetes and Metabolism, University of Birmingham, Birmingham, UK
    Search for more papers by this author
    • Joint senior authors.
  • P. N. Newsome

    Corresponding author
    1. Liver Unit, Queen Elizabeth University Hospital Birmingham, Birmingham, UK
    • Centre for Liver Research and NIHR Liver Biomedical Research Unit, University of Birmingham, Birmingham, UK
    Search for more papers by this author
    • Joint senior authors.

Correspondence to:

Dr M. J. Armstrong, Centre for Liver Research and NIHR Liver Biomedical Research Unit, 5th Floor IBR, University of Birmingham, Birmingham B15 2TT, UK.

E-mail: mattyarm@doctors.org.uk

Summary

Background

Non-alcoholic fatty liver disease has reached epidemic proportions in type 2 diabetes (T2D). Glucagon-like peptide-1 analogues are licensed in T2D, yet little data exist on efficacy and safety in liver injury.

Aim

To assess the safety and efficacy of 26-week liraglutide on liver parameters in comparison with active-placebo.

Methods

Individual patient data meta-analysis was performed using patient-level data combined from six 26-week, phase-III, randomised controlled T2D trials, which comprise the ‘Liraglutide Effect and Action in Diabetes’ (LEAD) program. The LEAD-2 sub-study was analysed to assess the effect on CT-measured hepatic steatosis.

Results

Of 4442 patients analysed, 2241 (50.8%) patients had an abnormal ALT at baseline [mean ALT 33.8(14.9) IU/L in females; 47.3(18.3) IU/L in males]. Liraglutide 1.8 mg reduced ALT in these patients vs. placebo (−8.20 vs. −5.01 IU/L; P = 0.003), and was dose-dependent (no significant differences vs. placebo with liraglutide 0.6 or 1.2 mg). This effect was lost after adjusting for liraglutide's reduction in weight (mean ALT difference vs. placebo −1.41 IU/L, = 0.21) and HbA1c (+0.57 IU/L, = 0.63). Adverse effects with 1.8 mg liraglutide were similar between patients with and without baseline abnormal ALT. In LEAD-2 sub-study, liraglutide 1.8 mg showed a trend towards improving hepatic steatosis vs. placebo (liver-to-spleen attenuation ratio +0.10 vs. 0.00; = 0.07). This difference was reduced when correcting for changes in weight (+0.06, = 0.25) and HbA1c (0.00, = 0.93).

Conclusions

Twenty-six weeks' liraglutide 1.8 mg is safe, well tolerated and improves liver enzymes in patients with type 2 diabetes. This effect appears to be mediated by its action on weight loss and glycaemic control.

Introduction

Non-alcoholic Fatty liver disease (NAFLD) and steatohepatitis are common complications in type 2 diabetes, and leading causes of liver disease worldwide. As a result of the alarming growth of type 2 diabetes and central obesity, NAFLD is expected to become a major cause of liver-related mortality and liver transplantation over the next 5 years. Despite this, there are currently no approved therapies of proven benefit for NAFLD in patients with type 2 diabetes.[1]

Glucagon-like peptide-1 (GLP-1) is an incretin hormone with a potent blood glucose-lowering action mediated via its ability to induce insulin secretion and reduce glucagon secretion in a glucose-dependent manner. Furthermore, GLP-1 slows gastrointestinal motility and increases satiety with reduced food intake.[2] Human GLP-1 is rapidly degraded by the enzyme dipeptidyl peptidase-4 and other endopeptidases, resulting in a short half-life of 1.5–2.0 min.[3] To overcome this, GLP-1 receptor agonists based on exendin-4 (exenatide) or human analogues (liraglutide), resistant to dipeptidyl peptidase-4, have recently been developed. Liraglutide has been produced using human recombinant DNA technology and shares 97% amino acid sequence homology with native human GLP-1.[4] The resultant once-daily subcutaneous administration has recently been licensed in America and Europe for use in type 2 diabetes. Twenty-six weeks' liraglutide therapy has been shown to reduce HbA1c by 1.0–1.5%, systolic blood pressure by 2–7 mmHg and weight by 2–3 kg in over 4000 patients with type 2 diabetes studied in the ‘Liraglutide Efficacy and Action in Diabetes’ (LEAD) programme.[5-10] In addition, the GLP-1 receptor agonists, exendin-4 and liraglutide, have been shown to improve liver enzymes, oxidative stress and hepatic steatosis in murine models.[11-13] In vitro data suggest that GLP-1 receptor agonists can act directly on human hepatocytes via a G-protein-coupled receptor[14, 15] and protect hepatocytes from fatty acid-related death.[16] These actions suggest that by direct or indirect metabolic mechanisms, liraglutide may be a promising option for the treatment of NAFLD.

To date, human studies investigating the effect of GLP-1 receptor agonists on the human liver have been limited to case reports,[17, 18] solitary case series[19] and uncontrolled open-label retrospective studies.[20] In light of this limited experience in patients with hepatic injury, the US Food and Drug Administration (FDA) and the European Medicines Agency (EMA) caution against the use of GLP-1 analogues in patients with mild, moderate and sever liver injury. The efficacy and safety of this group of drugs in liver disease, therefore, remain unproven. Our meta-analysis of individual patient-level data combined from the six phase III, multinational, randomised-controlled trials (RCT) that comprise the LEAD program, was performed to assess the safety and efficacy of liraglutide on liver parameters in comparison with an active-placebo control. In addition, a sub-study of the LEAD-2 trial was analysed to assess the effect of liraglutide on hepatic steatosis.

Methods

Study population and design

All subjects from the LEAD clinical development programme randomised to 0.6 mg, 1.2 mg, 1.8 mg/day liraglutide, oral antidiabetic drugs (OADs) or active/placebo, were included in this meta-analysis. The study design of the six phase III RCTs that comprise the LEAD programme are summarised in Supplementary Table S1.[5-10] In total, 4456 patients from 40 countries with type 2 diabetes who were unable to maintain glycaemic control (HbA1c ≥ 7%) with diet and exercise alone, or with oral antidiabetic treatment, were recruited to the LEAD program. This was designed primarily to compare liraglutide alone (or in combination with various OAD) to antihyperglycaemic therapies commonly used in type 2 diabetes. The original primary outcome of all the LEAD studies was change in HbA1c from baseline, with secondary outcomes including changes in fasting plasma glucose and weight. Exclusion criteria for the 6 individual LEAD trials included treatment with systemic corticosteroids, liver-specific symptoms and alanine transaminase (ALT) ≥2.5 times upper limit of normal (ULN) range of the standard laboratory, renal dysfunction (defined as ≥135 μmol/L in males, ≥110 μmol/L in females), cancer (except basal or squamous cell skin cancer), and seropositivity for hepatitis B (HBV) or hepatitis C virus (HCV). A detailed alcohol history was not taken, although patients with a prior diagnosis of alcohol-related liver disease were excluded. Liraglutide (or active-placebo) was injected once daily with a prefilled pen injection device for 26 weeks' (52 weeks in LEAD-3) duration. In all studies, the starting dose was liraglutide 0.6 mg/day, titrated after 7 days to 1.2 mg. In studies in which liraglutide 1.8 mg was evaluated, subjects were titrated to this dose after further 7 days at 1.2 mg.

LEAD-2 was a 26-week, multicentre, randomised, double-blinded, active-placebo control, phase III.[8] Subjects were randomised to 0.6, 1.2 or 1.8 mg/day liraglutide, 4 mg/day glimepiride or placebo, all in combination with metformin. LEAD-2 incorporated an optional sub-study to assess the efficacy of 26 weeks' treatment with liraglutide on hepatic steatosis and body fat composition. Inclusion criteria and treatment regimen for the sub-study were identical to the main trial.

Data collection and definitions

Baseline demographics and clinical/laboratory measures were recorded at randomisation. The metabolic syndrome was defined as ≥3 metabolic components.[21] For purposes of this retrospective analysis, serum ALT was used as a surrogate marker to estimate the proportion of LEAD participants with liver injury. The new ALT cut-offs as recommended by Prati et al. (>30 IU/L in males, >19 IU/L females) were used to define abnormality.[22] The NAFLD Fibrosis Score (NFS),[23] a well-validated scoring system, was retrospectively calculated in all subjects at baseline in an attempt to estimate the presence or absence of advanced liver fibrosis in the LEAD program.[23] The low cut-off score (<−1.455) has a negative predictive value of 88–93% and the high cut-off score (>+0.676) has a positive predictive value of 79–90% for the presence of advanced fibrosis in NAFLD in secondary care populations.[23, 24]

In the LEAD-2 sub-study,[25] hepatic steatosis was measured at randomisation and 26 weeks using single-slice, noncontrast-enhanced abdominal computerised tomography (CT). Directly comparing the CT attenuation of the liver to the spleen (internal control) to provide a liver-to-spleen attenuation ratio (LSAR) has previously been shown to be a valid tool for measuring the presence and severity of hepatic steatosis.[26, 27] In this study, the presence of hepatic steatosis was defined as an LSAR <1.0,[26] with increases in the ratio indicating reductions in steatosis.[27]

Safety profile

Individual patient-level data of trial withdrawal, treatment-emergent adverse events (AE) and serious adverse events (SAE) were combined from the LEAD programme to enable a descriptive comparison of the safety profile of 26 weeks' treatment with liraglutide (1.2 mg/1.8 mg) between patients with normal and abnormal baseline ALT. A treatment-emergent AE was defined as an event occurring between the first dose and the last dose (plus 7 days) or starting before first dose with increasing severity during treatment. All treatment emergent AEs with an incidence of 10% or more in any pooled treatment group (active-placebo, liraglutide 1.2 mg, liraglutide 1.8 mg) organised by system organ class and preferred term are reported.

Statistical analysis

The six phase III RCTs were combined to facilitate an individual patient-level data meta-analysis of liraglutide vs. active-placebo control. The analysis was based on an intention-to-treat (ITT) population (4442 out of 4456 recruited), defined as subjects from each of the individual trials who were randomised and exposed to at least one dose of study treatment (0.6 mg, 1.2 mg, 1.8 mg liraglutide, active-placebo control or other antidiabetic treatments). Descriptive statistics were applied to characterise the whole ITT study cohort and the individual treatment groups. All continuous clinical and laboratory variables are reported as means (s.d.) and categorical variables as numbers (percentages), unless stated. The significance level was set at < 0.05.

Changes in liver enzymes – meta-analysis

Change in ALT after 26 weeks' treatment was analyzed using an analysis of covariance (ancova) model with the trial, country, randomised treatment, previous antidiabetic medication, normality of ALT at baseline, and the interaction effect between randomised treatment and the normality of ALT at baseline as fixed effects. This analysis, therefore, allowed for adjustment of potentially confounding factors, including concomitant oral antidiabetic medications. Postbaseline data were imputed using the last-observation-carried-forward (LOCF) method in case of missing observations at week 26. The ancova model was then repeated for subjects with abnormal baseline ALT, where changes in weight and/or changes in HbA1c were included as covariates to investigate if the effect of 26 weeks' liraglutide (1.8 mg/day) treatment (vs. active-placebo) on ALT was independent of its effects on weight and/or glycaemic control.

Changes in hepatic steatosis – LEAD-2 sub-study only

Change in CT-measured hepatic steatosis (i.e. LSAR) after 26 weeks' treatment in the LEAD-2 sub-study was analyzed using a repeated-measures model with previous antidiabetic treatment, country and randomised treatment as fixed effects, and baseline values as covariates. Furthermore, interaction terms of previous antidiabetic treatment by visit, country by visit, randomised treatment by visit and baseline values by visit were included, and an unstructured covariance matrix for parameter of interest (LSAR) within the same subject was employed. The analysis was based on the ITT population in LEAD-2 to estimate change from baseline and comparison between liraglutide and active-placebo. The repeated-measures model with actual values (rather than ancova with LOCF) was selected to reduce bias in treatment effect and residual standard deviation (s.d.) estimates due to the relatively small numbers in each treatment arm. The repeated-measures model was repeated with changes in weight and/or changes in HbA1c set as covariates to investigate if the effect of 26 weeks' treatment with liraglutide 1.8 mg (vs. active-placebo) on LSAR was independent of its effects on weight and/or glycaemic control.

Results

Baseline demographics – meta-analysis

A total of 4442 patients were included in the ITT individual patient-level data meta-analysis of the LEAD programme, of which 2734 (61.5%) received liraglutide and 524 (11.8%) received active-placebo to enable comparison. The remaining 1184 (26.7%) were randomised to other antidiabetic medications (analysed, but data not presented). The baseline demographics and clinical characteristics were similar amongst the patients who received either liraglutide (n = 2734) or active-placebo (n = 524) injections (Table 1). The mean age was 55.9 (s.d.: 10.1) years with predominance towards Caucasian race (78.6%); 62% of the cohort had the metabolic syndrome at baseline with a mean BMI of 31.5 (s.d.: 5.4) kg/m2 and poor glycaemic control [HbA1c mean 8.3% (s.d.: 1.0)].

Table 1. Baseline demographics and clinical parameters of LEAD 1–6 trials: intention-to-treat (ITT) cohort
  LEAD program Total Liraglutide vs. Placebo
TotalPlacebo Lira 0.6 mg Lira 1.2 mg Lira 1.8 mg
  1. Values are mean (s.d.) unless stated otherwise. Percentages include missing values.

  2. a

    ≥3 components of the metabolic syndrome [NCEP-ATP III criteria (21)].

  3. b

    Normal reference range for ALT is ≤19 IU/L for females and ≤30 IU/L for males (22).

ITT, N (% of randomised)4442 (99.7)3258 (99.7)524 (99.2)475 (100)896 (99.8)1363 (99.8)
Male gender, N (%)2378 (53.5)1732 (53.2)288 (55.0)277 (58.3)449 (50.1)718 (52.7)
Age (years)55.9 (10.1)55.7 (10.1)55.7 (9.8)55.9 (10.2)55.9 (10.0)55.6 (10.2)
Ethnicity, N (%)
White3493 (78.6)2556 (78.5)419 (80.0)356 (74.9)697 (77.8)1084 (79.5)
Asian/Hawaiian/ Pacific Is.565 (12.7)417 (12.8)61 (11.6)103 (21.7)104 (11.6)149 (10.9)
Black/African American257 (5.8)198 (6.1)27 (5.2)11 (2.3)76 (8.5)84 (6.2)
American Indian/Alaskan5 (0.1)4 (0.1)2 (0.4)0 (0.0)1 (0.1)1 (0.1)
Other94 (2.1)67 (2.1)10 (1.9)5 (1.1)18 (2.0)34 (2.5)
Previous OAD, N (%)
Combination2703 (60.9)2025 (62.2)408 (77.9)323 (68.0)455 (50.8)839 (61.6)
Monotherapy1467 (33.0)1055 (32.4)116 (22.1)152 (32.0)350 (39.1)437 (32.1)
Diet only272 (6.1)178 (5.5)0 (0.0)0 (0.0)91 (10.2)87 (6.4)
Metabolic parameters
Metabolic syndromea, N (%)2754 (62.0)2011 (61.7)324 (61.8)275 (57.9)570 (63.6)842 (61.8)
Systolic BP (mmHg)131.3 (15.2)130.9 (15.0)131.5 (15.0)131.1 (14.8)130.4 (14.8)130.9 (15.2)
Weight (kg)88.6 (18.9)88.7 (19.0)90.3 (18.5)85.2 (17.6)88.7 (19.2)89.4 (19.4)
BMI (kg/m2)31.5 (5.4)31.5 (5.4)31.9 (5.2)30.3 (4.9)31.7 (5.4)31.7 (5.5)
Waist circumference (cm)
Male106.9 (13.4)107.1 (13.5)108.0 (13.2)104.9 (11.9)107.8 (13.7)107.0 (14.1)
Female102.4 (13.6)102.4 (13.5)104.0 (12.9)98.1 (12.8)102.3 (13.6)103.2 (13.7)
Total cholesterol (mmol/L)4.9 (1.2)4.9 (1.2)5.0 (1.2)4.9 (1.1)5.0 (1.2)4.9 (1.2)
Triglycerides (mmol/L)2.3 (1.9)2.3 (1.8)2.5 (2.4)2.2 (1.4)2.2 (1.6)2.2 (1.8)
FPG (mmol/L)9.7 (2.4)9.8 (2.4)9.8 (2.3)10.1 (2.4)9.8 (2.5)9.7 (2.4)
HbA1c (%)8.3 (1.0)8.4 (1.0)8.4 (1.0)8.4 (1.0)8.4 (1.1)8.3 (1.0)
Duration of diabetes (years)7.7 (5.7)7.7 (5.6)8.6 (5.9)7.3 (4.9)7.1 (5.5)7.9 (5.7)
Liver enzymes (IU/L)
Total ALT
N 441532355174738891356
Mean (s.d. )29.4 (16.6)28.9 (16.1)28.2 (15.7)29.0 (15.8)28.5 (15.5)29.5 (16.7)
Normal ALTb
N (% ITT)2174 (49.2)1627 (50.3)269 (52.0)252 (53.3)449 (50.5)657 (48.5)
Mean (s.d.)19.1 (5.6)18.9 (5.6)18.4 (5.3)19.4 (5.6)19.2 (5.8)18.7 (5.6)
Abnormal ALT
N (% ITT)2241 (50.8)1608 (49.7)248 (48.0)221 (46.7)440 (49.5)669 (51.5)
Mean (s.d.)39.4 (17.7)39.1 (16.8)38.8 (16.3)40.0 (16.5)38.0 (16.5)39.6 (17.3)

In all, 50.8% (2241/4415; 27 missing data) patients had an abnormal ALT at baseline, with mean values of 33.8 (s.d.: 14.9) IU/L in females and 47.3 (18.3) IU/L in males. A high NFS (>+0.676) was found in 6.3% (266/4238; mean score 1.13) of patients suggesting the presence of advanced liver fibrosis (Stages F3/F4 on Kleiner classification[28]). The presence of advanced liver fibrosis, however, could not be confidently excluded in 61.7% (2613/4238) of the diabetic patients who scored an indeterminate value with the NFS (−1.455 to +0.676). Advanced fibrosis was predicted to be absent in 32.1% (1359/4238) subjects with a low NFS (<−1.455).

Safety profile – meta-analysis

The frequency of adverse events or subsequent withdrawal rates from liraglutide (1.2/1.8 mg) was similar between patients with or without abnormal ALT at baseline (Table 2). The incidence of gastrointestinal and hepatobiliary serious adverse events with liraglutide 1.2 mg or 1.8 mg was comparable in patients with abnormal baseline ALT (1.2 mg, 1.1%; 1.8 mg, 0.6%) and with normal ALT at baseline (1.2 mg, 1.1%; 1.8 mg, 0.9%).

Table 2. Safety profile of 1.2 and 1.8 mg liraglutide in patients with normal and abnormal liver enzymes (ALT) in LEAD 1–6 trials
  Patients with normal ALT at baseline, N (%) Patients with abnormal ALT at baseline, N (%)
PlaceboLira 1.2 mgLira 1.8 mgPlaceboLira 1.2 mgLira 1.8 mg
  1. a

     General/other includes: congenital, familial and genetic disorders; eye disorders; general disorders and administration site conditions; injury (i.e. fracture), poisoning and procedural complications (i.e. surgery); psychiatric disorders; reproductive system; skin disorders; and vascular disorders.

Safety population269 (100)449 (100)659 (100)248 (100)440 (100)699 (100)
Overall withdrawal rate74 (27.5)96 (21.4)133 (20.2)73 (29.4)89 (20.2)111 (15.9)
Participants with any SAE18 (6.7)35 (7.8)43 (6.5)11 (4.4)29 (6.6)32 (4.6)
Participants with AE178 (66.2)358 (79.7)495 (75.1)163 (65.7)353 (80.2)547 (78.3)
Withdrawal due to AE7 (2.6)35 (7.8)76 (11.5)6 (2.4)42 (9.5)50 (7.2)
Participants with SAE, by system organ class
Gastrointestinal disorders0 (0.0)4 (0.9)5 (0.8)1 (0.4)4 (0.9)3 (0.4)
Hepatobiliary disorders0 (0.0)1 (0.2)1 (0.2)1 (0.4)1 (0.2)2 (0.3)
Cardiac disorders3 (1.1)7 (1.6)9 (1.4)2 (0.8)6 (1.4)5 (0.7)
Infections and infestations5 (1.9)6 (1.3)6 (0.9)1 (0.4)5 (1.1)5 (0.7)
Metabolism and nutrition disorders1 (0.4)1 (0.2)2 (0.3)1 (0.4)0 (0.0)1 (0.1)
Neoplasms (benign, malignant and unspecified)3 (1.1)3 (0.7)6 (0.9)0 (0.0)5 (1.1)9 (1.3)
Renal and urinary disorders1 (0.4)1 (0.2)1 (0.2)0 (0.0)0 (0.0)0 (0.0)
Respiratory, thoracic and mediastinal disorders1 (0.4)0 (0.0)1 (0.2)0 (0.0)2 (0.5)1 (0.1)
Nervous system disorders1 (0.4)4 (0.9)6 (0.9)1 (0.4)3 (0.7)1 (0.1)
Musculoskeletal and connective tissue disorders1 (0.4)3 (0.7)7 (1.1)1 (0.4)2 (0.5)1 (0.1)
Investigations0 (0.0)0 (0.0)0 (0.0)1 (0.4)0 (0.0)0 (0.0)
Endocrine disorders0 (0.0)0 (0.0)0 (0.0)0 (0.0)5 (1.1)3 (0.4)
General/othera6 (2.2)7 (1.6)6 (0.9)3 (1.2)6 (1.4)6 (0.9)
AE (any severity) with an incidence of 10% or more in any treatment group, by system organ class and preferred term
Gastrointestinal disorders47 (17.5)207 (46.1)286 (43.4)45 (18.1)199 (45.2)325 (46.5)
Diarrhoea11 (4.1)52 (11.6)70 (10.6)12 (4.8)55 (12.5)118 (16.9)
Nausea12 (4.5)103 (22.9)141 (21.4)14 (5.6)86 (19.5)164 (23.5)
General disorders and administration site conditions26 (9.7)68 (15.1)77 (11.7)20 (8.1)65 (14.8)85 (12.2)
Infections and infestations92 (34.2)159 (35.4)215 (32.6)83 (33.5)194 (44.1)263 (37.6)
Upper respiratory tract infection22 (8.2)38 (8.5)40 (6.1)11 (4.4)45 (10.2)53 (7.6)
Investigations25 (9.3)45 (10.0)44 (6.7)15 (6.0)29 (6.6)56 (8.0)
Metabolism and nutrition disorders17 (6.3)78 (17.4)94 (14.3)22 (8.9)66 (15.0)86 (12.3)
Musculoskeletal and connective disorders43 (16.0)80 (17.8)104 (15.8)27 (10.9)88 (20.0)123 (17.6)
Nervous system disorders33 (12.3)87 (19.4)120 (18.2)33 (13.3)102 (23.2)138 (19.7)
Headache16 (5.9)36 (8.0)59 (9.0)23 (9.3)61 (13.9)76 (10.9)

Change in ALT – meta-analysis

Twenty-six weeks' treatment with liraglutide 1.8 mg significantly reduced ALT in patients with abnormal baseline readings in comparison with active-placebo (−8.20 vs. −5.01 IU/L; P = 0.003) (Figure 1). This effect was dose-dependent with greater reductions in ALT seen with the 1.8 mg dose than with the 1.2 mg dose (1.8 vs. 1.2 mg difference, −1.49 IU/L, P = 0.09) and 0.6 mg daily (1.8 vs. 0.6 mg, −2.61 IU/L; P = 0.02). The improvements in ALT with liraglutide 1.8 mg vs. active-placebo were eliminated on correcting for change in weight (corrected mean difference vs. placebo, −1.41 IU/L; P = 0.21) and in HbA1c (corrected mean difference vs. placebo, 0.57 IU/L; P = 0.63). No significant differences in ALT were seen between placebo and the lower doses of liraglutide (0.6 mg/1.2 mg).

Figure 1.

Changes in ALT with 26 weeks' treatment of liraglutide vs. placebo in T2D patients with abnormal (left) and normal (right) ALT at baseline. Meta-analysis of LEAD-1 to LEAD-6.

Change in hepatic steatosis – LEAD-2 sub-study

In LEAD-2, the presence of hepatic steatosis was confirmed on CT imaging in 64.4% (96/149) of individuals at baseline, of which 58.3% (56/96) had at least 30% hepatic steatosis on CT (LSAR <0.8[29]). In keeping with the ALT data (above), the effect of liraglutide on LSAR appeared to be dose-dependent. At 26 weeks, there was a trend towards an improvement in LSAR with liraglutide 1.8 mg (n = 23) compared with the 11 patients on active-placebo [mean difference +0.10 (95% CI: −0.01 to +0.20); P = 0.07]. This difference at 26 weeks was reduced when correcting for changes in weight [mean difference +0.06 (95% CI: −0.04 to +0.15); P = 0.25] and HbA1c [mean difference 0.00 (95% CI: −0.11 to +0.10); P = 0.93]. No significant differences in LSAR were seen between placebo and the lower doses of liraglutide 0.6 mg [mean difference 0.00 (95% CI: −0.11 to +0.10); P = 0.90] and 1.2 mg [mean difference 0.02 (95% CI: −0.09 to +0.12); P = 0.73].

Discussion

This individual patient-level data meta-analysis of the LEAD programme demonstrates that 26 weeks' treatment with liraglutide 1.8 mg/day is well tolerated, safe to use and results in significant improvements in liver enzymes in patients with type 2 diabetes and asymptomatic liver injury. The efficacy of 26-week liraglutide on liver enzymes is dependent on drug dosage and appears to be mediated by its effect on weight change and glycaemic control. Gastrointestinal symptoms, namely nausea and diarrhoea, are the most common adverse events associated with liraglutide, but are mainly transient in nature (<2 weeks) and occur no more frequently in patients with abnormal liver enzymes.

Half of the patients in the meta-analysis had abnormal liver transaminases at baseline in keeping with previous diabetes trials with comparable patient characteristics.[20] Furthermore, almost two-thirds (64.4%) of subjects in the LEAD-2 sub-study had hepatic steatosis confirmed on CT. Although magnetic resonance spectroscopy (MRS) is the recognised non-invasive ‘gold standard’ for quantifying hepatic steatosis,[30] our figures for steatosis are in agreement with rates (56.9–69.5%) previously reported in large cross-sectional studies that utilised ultrasound and/or MRS.[31, 32] Entry criteria to the LEAD programme included negative HBV/HCV serology and no history of steroid use and/or alcoholic liver disease, leading us to conclude that the majority of these cases are likely due to NAFLD. This view is further supported by the compelling data that exist linking NAFLD to type 2 diabetes, obesity and the metabolic syndrome,[32, 33] all of which were prevalent in our study population. However, due to the lack of a detailed alcohol consumption history and serology for rarer liver conditions (such as autoimmune disease and haemochromatosis), other causes of liver damage in our study population cannot be categorically excluded.

Clinical trials in NAFLD demonstrate that reductions in ALT correlate with histological improvements in liver inflammation.[34-37] Here, we report significant improvements in the liver injury biomarker ALT with 26 weeks of liraglutide 1.8 mg (−8.20 IU/L from a baseline mean of 39.6 IU/L) in comparison with an active-placebo control. Previous studies with GLP-1 agonists, namely exenatide, have demonstrated a similar magnitude of reduction in ALT from baseline, but lack comparison with a placebo control.[20] Although significant, the effects of 26-week liraglutide 1.8 mg on ALT seen in our study may have been diluted by the placebo effect. Placebo effects have previously been reported to be as high as 19–30% in prospective NAFLD trials,[34, 38] which is of little surprise given the established efficacy of lifestyle modification in such metabolic conditions.[39]

Our study highlights a trend towards improvements in hepatic steatosis with liraglutide 1.8 mg in comparison with placebo controls over 26 weeks (P = 0.07). This, together with the significant baseline changes in hepatic steatosis with liraglutide 1.8 mg, reinforces findings from a previous case-report[18] and a case-series investigating the histological effects of exenatide in eight patients with type 2 diabetes.[19] The latter group reported decreased non-alcoholic steatohepatitis (NASH) activity (defined as reduced hepatocyte ballooning and inflammation and/or steatosis) in 3 of the 8 type 2 diabetic patients receiving 28-week exenatide.[19]

Our data suggest that the effect of 26-week liraglutide on weight loss and to a similar extent its effect on glycaemic control are the main factors in significantly reducing ALT in comparison with active-placebo controls. Although these effects appear to be correlated with liraglutide's benefit on weight and HbA1c, in the absence of a prospective study powered specifically for liver end-points, we are not able to rule out the possibility of a direct effect of liraglutide on liver injury. There are increasing murine and in vitro data to support a direct mechanism of GLP-1 receptor agonists on the liver, over and above its role as an incretin hormone.[11, 13-15] Not only has the GLP-1 receptor been identified on both murine and human hepatocytes,[12, 14, 15] but GLP-1 receptor agonist treatment in cell culture decreases triglyceride and free fatty acid stores in the absence of insulin.[13, 15] Furthermore, recent in vitro evidence would suggest that GLP-1 receptor agonists markedly improve the ability of the hepatocyte to handle excess free fatty acids and lipid production by modulating lipid transport, beta-oxidation and de novo lipogenesis,[11, 13, 15] all of which have been implicated in the pathogenesis of NAFLD.

Our study has a number of strengths. First, this is the first individual patient-level data meta-analysis (six large double-blinded RCTs) to focus on the effects of a human GLP-1 analogue on liver parameters in patients with type 2 diabetes, and the first to report comparisons with an active-placebo control whilst controlling for several confounding factors, including concomitant OAD treatment (i.e. TZDs, metformin). Second, the meta-analysis provides a descriptive overview of the safety profile of liraglutide in type 2 diabetic patients with and without abnormal blood liver enzymes prior to treatment. Even though the long-term adverse events remain unknown, this study provides valuable reassurance in the safe short-term use of liraglutide in the presence of mild-to-moderate liver injury (ALT >ULN to <2.5 times ULN) and potential steatosis.

The main limitation of this study is that the six RCTs combined in this meta-analysis were powered on glycaemic control and not changes in liver parameters. It is important to note that differences in patient populations between the six trials, in terms of previous exposure to antidiabetic therapy and baseline abnormality of liver enzymes, were included as fixed effects in this analysis. Nevertheless, there may be a degree of heterogeneity (despite similar eligibility criteria) between the six trials included in this meta-analysis. Finally, the lack of liver biopsy precludes the ability to accurately validate the severity of underlying liver injury with regard to the NFS predictions and most importantly, to validate the accuracy of ALT as a serial marker of liver inflammation in our cohort.

In conclusion, our large-scale study highlights that 26 weeks' treatment with liraglutide 1.8 mg has an acceptable safety profile and significantly improves liver enzymes vs. placebo in patients with type 2 diabetes and asymptomatic liver injury. These effects appear to be mediated by the effect of liraglutide on weight loss and glycaemic control. Our data support the rationale to prospectively investigate GLP-1 analogues in liver injury associated with type 2 diabetes and the metabolic syndrome.

Authorship

Guarantor of the article: MJA is the guarantor and takes full responsibility for the integrity of the data from inception to the published article.

Author contributions: MJA, BE, SCLG, JWT and PNN contributed to the concept and design of the study. WHOC and MJA performed the statistical analysis. MJA wrote the first draft of the manuscript. MJA, DDH, IAC, BE, SCLG, JWT and PNN contributed to the redrafting of the manuscript and the final submitted version. All authors approved the final version of the manuscript.

Acknowledgements

Declaration of personal interests: The authors thank Ali Falahati (Novo Nordisk A/S, Denmark) for statistical support; and Professor Wolfgang E. Schmidt (St Josef-Hospital, Germany) who contributed to the study's initial advisory meetings. SCLG has served on advisory boards for Novo Nordisk, Eli Lilly, Sanofi-aventis and Takeda, and has received honoraria for lectures given on behalf of Novo Nordisk, Eli Lilly, Sanofi-aventis, Takeda and GSK, BE and WHOC are employees of Novo Nordisk and own shares in Novo Nordisk as part of an employment incentive programme. PNN has received funding from Novo Nordisk for a clinical trial of liraglutide in NAFLD. MJA, DDH and JWT have no conflict of interests to declare.

Declaration of funding interests: MJA is in receipt of a Wellcome Trust Clinical Research Fellowship. DDH is in receipt of an MRC Clinical Research Fellowship. Novo Nordisk funded the phase III trials analysed in this manuscript.

Ancillary