SGLT-2 Inhibitors and Hepatic Decompensation Risk in MASLD
A nationwide Korean claims analysis has placed SGLT-2 inhibitors in direct comparison with two major drug classes used in patients with metabolic disease and fatty liver. In adults with metabolic dysfunction-associated steatotic liver disease (MASLD), SGLT-2 inhibitor use was associated with a lower risk of hepatic decompensation than thiazolidinediones, while the risk looked similar to that seen with glucagon-like peptide-1 receptor agonists. The study is clinically relevant because hepatic decompensation marks the transition from compensated liver disease to a far more dangerous phase, where ascites, variceal bleeding, liver failure, and transplant enter the picture.
The comparison matters because MASLD rarely exists in isolation. Patients often carry diabetes, obesity, hypertension, and other metabolic burdens that shape both drug selection and liver outcomes. A therapy that improves glycemia yet also lowers the probability of a major hepatic event would carry a different kind of value than one that affects glucose alone. In that setting, head-to-head data are more useful than indirect comparisons drawn from separate trials with different populations and endpoints.
Head-to-Head Findings Across Two Comparator Classes
Investigators used a nationwide healthcare claims database in Korea spanning 2014 through 2022 and identified adults aged 40 years or older with MASLD who initiated an SGLT-2 inhibitor or a comparator drug. After 1:1 propensity score matching, the analysis included 22,550 patients in the SGLT-2 inhibitor versus GLP-1RA comparison and 191,628 patients in the SGLT-2 inhibitor versus TZD comparison. The median age in both matched cohorts was 57 years. Men made up 60% of the SGLT-2 inhibitor and GLP-1RA cohort and 72% of the SGLT-2 inhibitor and TZD cohort.
The primary endpoint was a composite of hepatic decompensation events: ascites, oesophageal varices with bleeding, hepatic failure, or liver transplant. In the comparison with GLP-1RA, SGLT-2 inhibitor use produced a hazard ratio of 0.93, with a 95% confidence interval of 0.76 to 1.14. In practical terms, that estimate did not show a meaningful difference between the two drug classes for hepatic decompensation risk. The separation appeared only in the comparison with TZDs, where SGLT-2 inhibitor use carried a hazard ratio of 0.77, with a 95% confidence interval of 0.72 to 0.82.
The data point is sharp. Against TZDs, SGLT-2 inhibitors were associated with a materially lower risk of hepatic decompensation.
Why the Contrast with TZDs May Have Emerged
The result fits a biologically plausible pattern. SGLT-2 inhibitors improve glycemic control while promoting modest weight loss, natriuresis, and better insulin sensitivity. Those effects may reduce the metabolic stress that drives steatosis, inflammation, and eventual fibrotic progression. In a liver already burdened by dysfunctional lipid handling and systemic insulin resistance, even incremental metabolic improvement can matter over time.
Thiazolidinediones occupy a different therapeutic niche. They improve insulin sensitivity and have long been discussed in metabolic liver disease, but their broader safety profile and fluid-retention effects can complicate clinical use, especially in patients with advanced cardiometabolic disease. In a claims-based population where hepatic decompensation was the endpoint, the balance of real-world treatment effects may therefore favor SGLT-2 inhibitors over TZDs. The data do not prove a mechanistic pathway, but they do align with the clinical expectation that metabolic therapies do not all translate into equal downstream liver protection.
Sex-Stratified Results Point to a Stronger Signal in Women
Secondary analyses added another layer. When compared with TZDs, SGLT-2 inhibitors were associated with lower hepatic decompensation risk in both sexes, but the effect estimate was stronger in women than in men. The hazard ratio in men was 0.87, with a 95% confidence interval of 0.80 to 0.94. In women, the hazard ratio was 0.62, with a 95% confidence interval of 0.55 to 0.69.
That differential should be read carefully. Sex-stratified findings in observational work can reflect biology, prescribing patterns, baseline risk, or residual confounding, and claims data cannot disentangle those forces completely. Even so, the size of the separation in women invites attention. It may reflect differences in metabolic phenotype, progression patterns, or treatment response, but the study design cannot identify the cause. What it can do is flag a subgroup in which the apparent association was stronger and the confidence bounds stayed well below unity.
Clinical Meaning for MASLD Management
For clinicians treating MASLD, the comparison lands in a familiar but still important place: not every glucose-lowering drug appears to offer the same hepatic benefit. SGLT-2 inhibitors were not superior to GLP-1RA for the composite hepatic endpoint in this cohort, yet they performed better than TZDs. For patients who need diabetes therapy and also carry liver risk, that difference may influence prescribing decisions, especially when the goal is to avoid progression toward decompensation rather than simply to improve laboratory values.
The endpoint selection also matters at the bedside. Ascites, variceal bleeding, hepatic failure, and transplant represent hard clinical outcomes, not surrogate markers. By using those events as the primary composite, the analysis moved beyond fatty liver imaging or enzyme changes and asked a more consequential question: which treatment class tracked with fewer major hepatic failures over time? The answer favored SGLT-2 inhibitors over TZDs and found no clear difference versus GLP-1RA.
Limits of Claims Data and the Need for Clinical Confirmation
Observational data can reveal real-world associations across large populations, but they cannot replace randomized evidence. Claims databases do not contain the full clinical texture of MASLD severity, fibrosis stage, lifestyle factors, alcohol exposure, body composition, or adherence. Propensity score matching improves balance, yet unmeasured confounding can persist, especially in patients selected for one drug class over another because of renal function, cardiovascular disease, obesity, or prior treatment history.
The cohort also reflects people aged 40 years and older in Korea, so generalizability to younger patients or other health systems is not guaranteed. The findings are still clinically useful because they compare active therapies used in everyday practice, but they should be interpreted as comparative effectiveness signals rather than definitive proof of causation. Future randomized trials or carefully phenotyped prospective studies will be needed to determine whether the lower event rate with SGLT-2 inhibitors versus TZDs reflects a true hepatic protection effect, a selection pattern in prescribing, or both.
Even with those constraints, the central result is difficult to ignore: in MASLD, SGLT-2 inhibitors tracked with fewer hepatic decompensation events than TZDs and performed similarly to GLP-1RA on the primary endpoint. For a disease defined by silent progression until advanced liver failure emerges, that is the kind of endpoint difference clinicians notice.