Glucagon-like peptide-1 receptor agonists (GLP-1RAs), first approved by the Food and Drug Administration for their use in type 2 diabetes in 2005, have gained growing interest in the management of metabolic dysfunction–associated steatotic liver disease (MASLD). The attraction of GLP-1RAs is the benefit in the control of hyperglycemia alongside beneficial effects on weight. There is clear evidence that they reduce steatosis both histologically and on MRI.1 A number of meta-analyses have recently been published, the largest of which identified 31 studies, of which 25 were sufficiently robust for analysis showing benefits in noninvasive steatosis scores and transaminase levels.2 A meta-analysis of histological outcomes identified 4 studies of which 2 were at high risk of bias and only covered a total of 406 patients but did suggest GLP-1RAs improved histological features.3 One other cohort study failed to show the benefit of GLP-1RA in the prevention of HCC in 2526 patients with incident MASLD.4 There remains a paucity of data on longer-term outcomes in chronic liver disease and cirrhosis and therefore this study aims to address an important gap in the evidence base. The choice of an active-comparator new-user design, applied to a large multinational cohort, brings a robust design to address the issue of retrospective data collection. Using a DPP4i control to minimize crossover to GLP-1RA due to lack of benefits and more recent guidance against combination therapy was a particularly well-considered choice.5 This study represents the largest cohort studied so far looking at hepatic endpoints with 91,479 participants in the GLP-1RA group.6 The attention to detail in regression modeling and the accounting for a large number of confounders adds to the strength of the study, although there is a reliance on the size of the cohort to protect from other factors that are often present in data obtained from ICD coding, which can often be prone to errors and incomplete data. The one variable that was not reported is the ethnic diversity of the study population which does introduce a level of confounding and may also limit the applicability of results to practice outside of Scandinavia. Another concern is that reliance on coding means that we cannot be certain that cofactors to the etiology of liver disease were not missed. The decision to focus purely on endpoints of disease however does remove the concern over the misclassification of disease stage usually encountered in retrospective data. The primary outcomes under study were defined as acute and subacute hepatic failure, chronic hepatic failure, hepatic failure unspecified, other and unspecified cirrhosis of the liver, portal hypertension, hepatorenal syndrome, esophageal varices with or without bleeding, liver transplantation, and HCC, all compounded as a single measure of "serious liver events." The rationale for the use of this compound outcome measure was previously explored in an analysis of coding within the same data resources used in this study and proposed to have a >90% accuracy to recorded clinical events. On closer inspection of the original validation group of a cohort encompassing cryptogenic and alcohol-associated disease, this had missing crucial data points for 16% of the cohort. The validation group also identified that ascites correlated poorly in coding with chronic liver disease–associated ascites due to the lack of clarity in alternate etiology in coding, hence the absence of a common complication of advanced cirrhosis from the composite outcome measure and a source of potential confounding.7 The use of a composite outcome does come with the benefit of increased reliability by not wholly relying on one single disease outcome overcoming the issues of partial or missing coding data. One part of this composite that did not seem to fit with the others was the acute and subacute hepatic failure code. This is a rare and clearly defined disease entity and is not classically associated with MASLD, nor does it fit with the other codes associated with chronic disease outcomes. The reason for excluding data from the 18–34-year age group was not made clear. Notably, studies using the same Scandinavian public health data resources as those used in this study have quantified a significant disease burden in this younger age group that is associated with an increased long-term mortality risk.8 A recently published prospective cohort study of 35,860 Chinese patients also suggested that an earlier onset of MASLD was linked with a higher risk of liver cancer with an adjusted HR of 2.66 for those under the age of 45.9 The exclusion of this patient group from the analysis has the potential to affect both the primary and secondary outcome measures and the conclusions drawn therein. The conclusion that GLP-1RA use was associated with a reduced risk of serious liver events despite the large cohort size remains unproven with many of the CIs for the adjusted HRs spanning 1.0 in the individual subset analyses. 11.3% of the original GLP-1RA cohort was removed from analysis due to the addition of DPP4i, although as discussed this is unlikely to have had a great deal of effect on outcomes. Reintroducing them to the analysis pulled the adjusted HR to 0.93 from 0.85 with a 95% CI of 0.83–1.04. A similar effect was observed upon restricting outcomes to primary diagnoses providing an adjusted HR of 0.9 with a 95% CI of 0.78–1.04. The breakdown in the different GLP-1 RA drugs fell heavily in favor of liraglutide which accounted for 81.3% of the cohort. In terms of benefit, there is evidence to suggest that liraglutide may not be as effective as newer GLP-1 RAs as it has less associated weight loss and a smaller reduction in glycated hemoglobin levels (HbA1c). Several recent meta-analyses comparing liraglutide and semaglutide concluded there was a significant difference in efficacy, favoring semaglutide.10,11 The other factor to consider is that the dosing of GLP-1RA for diabetes is lower than that for weight loss and as such limiting analysis by indication of diabetes only may downplay the dose-dependent benefit seen in randomized controlled trials for liver histological regression of MASLD.3 This may account for the smaller-than-anticipated adjusted HR when compared with other previous studies. Overall, even this large observational cohort from a well-recognized, high-quality data resource was not able to draw strong conclusions in the ongoing debate over the efficacy of GLP-1RAs in reducing progression toward decompensated cirrhosis. Nonetheless, in keeping with previous studies, it shows a signal toward benefit in the MASLD population. Although the study shows promise for cirrhosis endpoints, GLP-1RA efficacy in cirrhosis remains a large gap in our evidence base, due to the majority of randomized controlled trials focusing on precirrhotic disease. Ultimately the findings from this study further emphasize the need for large prospective studies of GLP-1RA in patients with MASLD cirrhosis.
No takes yet. Share an insight, caveat, or question.
Yeh et al. (2024) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: