Incretin‐Based Therapies for the Management of Nonalcoholic Fatty Liver Disease in Patients With Type 2 Diabetes

See Article on Page 2414. Potential conflict of interest: K.C. has received research support as principal investigator from the National Institutes of Health (NIH), Cirius, Echosens, Inventiva, Novartis, Novo Nordisk, Poxel, and Zydus. K.C. is a consultant for Allergan, AstraZeneca, BMS, Boehringer Ingelheim, Coherus, Deuterex, Eli Lilly, Genentech, Gilead, Janssen, Pfizer, Poxel, Novo Nordisk, and Sanofi‐Aventis. The prevalence of nonalcoholic fatty liver disease (NAFLD) in type 2 diabetes mellitus (T2DM) is believed to be ~70% and that of advanced fibrosis ~20%.1 Given the magnitude of the problem, there is a real sense of urgency to identify the best pharmacological approach to treat both diabetes and nonalcoholic steatohepatitis (NASH) in this population. Vitamin E is recommended for patients with biopsy‐proven NASH without diabetes,2 but did not have similar efficacy in a recent proof‐of‐concept randomized placebo‐controlled trial (RCT) in patients with T2DM.3 Regarding antidiabetic agents, there is a significant amount of data for some agents, but much more limited information for others.1 Among insulin sensitizers, whereas metformin has not shown significant benefit, pioglitazone has been incorporated into current treatment guidelines for patients with or without T2DM and biopsy‐proven NASH.2 Sodium‐glucose co‐transporter‐2 inhibitors (SGLT2i) promote glycosuria and weight loss, being considered today second‐line therapy for patients with diabetes at risk of cardiovascular disease, heart failure, or chronic kidney disease.6 Promising results have been observed for the treatment of NAFLD in animal models and in open‐label clinical trials.4 Recently, a placebo‐controlled RCT in patients with T2DM reported that canagliflozin promoted weight loss and improved hepatic insulin sensitivity and insulin secretion.7 In those with NAFLD, cangliflozin decreased intrahepatic triglycerides (IHTG) versus baseline, but this was of borderline significance when compared to placebo (−38% vs. −20%; P = 0.09). Of note, all patients received dietary advice at the onset of the trial, which likely caused the positive change in the placebo group. Reduction in IHTG did not appear to be treatment specific, but rather proportional to degree of weight loss, which was more pronounced with canagliflozin (−5.5% vs. −2.1%, respectively; P = 0.001). Patients who did not lose weight had no reduction in IHTG, but more patients on canagliflozin compared to placebo lost ≥5% of body weight and had a ≥30% reduction in IHTG (38% vs. 7%, P = 0.009). This study highlights the importance of lifestyle changes during studies in NAFLD (often even unintended by investigators) and the need for a placebo arm when investigating antidiabetic agents in NAFLD, given that past in vivo animal studies and open‐label clinical trials had suggested unique properties of SGLT2i, beyond weight loss.1 The above serves as an introduction to the role of incretin‐based therapies, the most studied group of all. In animal models of NAFLD, both glucagon‐like peptide‐1 receptor agonists (GLP‐1Ras)8 and dipeptidyl peptidase‐4 inhibitor (DPP‐IVi)10 improve many pathways linked to steatohepatitis, although their efficacy at the clinical level has been mixed (see Tables 1 and 2). In open‐label prospective trials, sitagliptin has been found to improve plasma aminotransferases,12 IHTG,14 and liver histology.15 But negative results have also been reported with sitagliptin in open‐label studies16 and in all three placebo‐controlled trials.18 Negative studies have been also reported with liraglutide,20 but most have been positive with an improvement in hepatic steatosis (HS)22 or liver histology.29 This is usually proportional to the magnitude of weight loss, as reported for other short‐31 and long‐acting32 GLP‐1RAs. However, given the many effects of GLP‐1RAs in animal models of NAFLD, one cannot exclude mechanisms other than weight loss at play. Unfortunately, placebo‐controlled studies with liver biopsies have been rare,20 and the field still awaits the results of large RCTs with GLP‐1RAs (or DPP‐IV inhibitors) using liver histology as the primary outcome, rather than surrogate biochemical or imaging endpoints. Table 1 - Effect of Sitagliptin in NAFLD Author N Duration (Weeks) Comparator Main Study Results Weight ALT Liver Fat (IHTG*) Open‐label studies Iwasaki et al. (2011) 30 16 None Not reported ↓ n/a Ohki et al. (2012) 82 74 Liraglutide, pioglitazone Unchanged ↓ n/a Fukuhara et al. (2014) 44 52 None Not reported Unchanged n/a Asakawa et al. (2015) 62 57 None Not reported Unchanged n/a Kato et al. (2015) 20 24 Glimepiride ↓ Not reported ↓ Alam et al. (2018) 40 52 Lifestyle Unchanged ↓ ↓† Sayari et al. (2018) 138 16 Sitagliptin + synbiotic ↓ ↓ n/a RCTs Smits et al. (2016) 18 12 Liraglutide or placebo Unchanged Unchanged No change vs. placebo Cui et al. (2016) 50 24 Placebo Unchanged Unchanged No change vs. placebo Joy et al. (2017) 12 24 Placebo Unchanged Unchanged No change vs. placebo‡ Statistically significant changes vs. comparison indicated by arrows.*Liver fat measured with MRI‐based imaging.†Improvement on histology (NAFLD activity score) greater with sitagliptin on paired liver biopsies.‡No significant improvement in liver histology on paired liver biopsies.Abbreviations: ALT, alanine aminotransferase; n/a, not applicable. Table 2 - Effect of Liraglutide in NAFLD Author N Duration (Weeks) Comparator Main Study Results Weight ALT Liver Fat Open‐label studies Ohki et al. (2012) 82 74 Sitagliptin, pioglitazone ↓ ↓ n/a Eguchi (2015) 19 24 Lifestyle ↓ ↓ ↓† Tang et al. (2015) 35 12 Insulin ↓ Unchanged Unchanged Feng et al. (2017) 87 24 Gliclazide, metformin ↓ ↓ ↓‡ Bouchi et al. (2017)* 17 24 Insulin alone ↓ ↓ ↓ Petit et al. (2017) 68 24 Insulin alone ↓ ↓ ↓ Matikainen et al. (2018) 22 16 Lifestyle ↓ Not reported ↓ RCTs Smits et al. (2016) 18 12 Sitagliptin or placebo Unchanged Unchanged Unchanged Armstrong et al. (2016) 52 48 Placebo ↓ ↓ ↓§ Vanderheiden et al. (2016)* 71 24 Insulin alone ↓ ↓ ↓ Frossing et al. (2018) 72 26 Placebo ↓ ↓ ↓ Statistically significant changes vs. comparison(s) indicated by arrows.*Liraglutide plus insulin vs. insulin alone.†Ten of 19 had a repeat liver biopsy; NAFLD activity score improved in 6.‡Reduced more vs. gliclazide (but not metformin).§Improvement on histology (NAFLD activity score) greater with liraglutide on paired liver biopsies.Abbreviations: ALT, alanine aminotransferase; n/a, not applicable. In the current issue of Hepatology, Yan et al.33 ask the important question as to the best add‐on to metformin (liraglutide, sitagliptin, or insulin) for the management of NAFLD in patients with uncontrolled diabetes. To this end, they conducted an open‐label, multicenter study across eight tertiary university centers in China. They randomized 75 patients with uncontrolled diabetes and NAFLD to either liraglutide, sitagliptin, or long‐acting basal insulin glargine for 26 weeks. The primary endpoint was the change in IHTG measured by magnetic resonance imaging (MRI) proton‐density fat fraction. Key secondary endpoints included glycated hemoglobin, body weight, and abdominal adiposity (subcutaneous adipose tissue [SAT] and visceral adipose tissue [VAT]), among others. All treatments led to a similar good glycemic control (mean <7.0% by 26 weeks). There was a significant 3.6‐kg weight loss with liraglutide, compared to 1.7 kg with sitagliptin and 1.2 kg with insulin (~5% vs. ~2% and ~1% of total body weight, respectively; P = 0.042 vs. insulin, but not significant [NS] versus sitagliptin). Both SAT and VAT decreased with liraglutide, as reported in earlier studies.23 From a baseline of ~15.5%, IHTG decreased with both incretin‐based therapies: 4.0% with liraglutide and 3.8% with sitagliptin for a similar relative liver fat reduction of ~20% (both P < 0.001 vs. baseline and insulin; NS between liraglutide vs. sitagliptin). The investigators concluded that liraglutide and sitagliptin were comparable add‐ons to metformin to treat NAFLD in patients with diabetes. Many past studies have examined the efficacy of sitagliptin (Table 1) and liraglutide (Table 2) in patients with NAFLD. Yet, this study is valuable by testing them head to head in patients with T2DM and NAFLD from China, where information is rather limited in this setting. The reduction of body weight and of IHTG with liraglutide were consistent with most previous studies (Table 2), where, to a significant extent, benefit was proportional to degree of weight change. The lack of a placebo arm to fully assess the effect of diabetes education during the trial, or of unintentional investigator treatment bias, is the major limitation of the study by Yan et al.33 It is possible that lifestyle changes accounted for the significant weight loss from baseline with sitagliptin because this DPP‐IVi and others are considered weight neutral.6 Consistent with its lack of effect on body weight, most studies have not observed a reduction of IHTG with sitagliptin (Table 1). This was particularly true for all three placebo‐controlled trials,18 where the average decrease of IHTG in the placebo arms (~10%‐20%) was similar to the relative decrease observed with sitagliptin by Yan et al.33 This reduction is also in agreement with the relative ~10%‐20% reduction of IHTG observed in placebo arms of other controlled trials in patients with steatohepatitis (NASH).7 Relative short duration of diabetes (~4 years) may have also promoted a better response than in previous studies, although the absolute reduction in MRI‐measured liver‐fat content was relatively small (mean reduction of ~4%) and none reached resolution of HS. Furthermore, Ohki et al.22 reported that in Japanese patients with T2DM and NAFLD, sitagliptin did not modify either body weight or the aspartate to aminotransferase to platelet ratio index, which did decrease with pioglitazone or liraglutide (i.e., the GLP‐1RA induced a weight loss of 3.8 kg).17 The unintended role of lifestyle intervention in the study by Yan et al.33 was also apparent in the trend toward weight loss observed with insulin therapy (–1.2 kg; P = 0.28), which is also discordant with the typical weight gain observed with insulin therapy. In a meta‐analysis of 14,250 patients treated with basal insulin for a mean of 27.7 weeks, weight gain occurred in 43 of 46 studies.36 In a recent multicenter study by Pan et al.37 involving 833 patients, where the majority were from China, insulin therapy with either basal insulin degludec or insulin glargine also led to weight gain (~2 kg). Despite inducing weight gain, insulin glargine may reduce HS, particularly in insulin‐naїve patients.38 This is in contrast to the increase observed with basal insulin peglispro, a hepatopreferential insulin that had minimal capacity to inhibit peripheral adipose tissue lipolysis (in contrast to all currently available insulins). This unique feature led to its discontinuation because it promoted hepatocyte triglyceride accumulation and an elevation in plasma aminotransferases.39 This cross‐talk between adipose tissue and the liver has significant relevance for future drug development and highlights how the pathophysiology of diabetes and NAFLD are closely interconnected. In summary, the study by Yan et al.33 adds to our efforts of identifying the optimal incretin‐based approach to treat NAFLD in T2DM, an increasing dilemma in this population. More studies are needed to test how different incretin‐based approaches may best work in combination with other available antidiabetic agents (i.e., SGLT2i or pioglitazone) and novel drugs under development. Inclusion of a placebo arm in future proof‐of‐concept studies will be mandatory to fully understand their therapeutic potential. Much work lies ahead before we may find the optimal approach for the management of NAFLD in T2DM.