ABSTRACT
-
Background/Aims
Metabolic dysfunction-associated steatotic liver disease (MASLD) is a growing cause of cirrhosis and its complications. Given its close association with type 2 diabetes mellitus (T2DM), evaluating whether sodium-glucose cotransporter-2 inhibitors (SGLT2is) can mitigate the progression of liver fibrosis is clinically important. We examined the association between SGLT2i use and liver fibrosis progression in patients diagnosed with MASLD and T2DM.
-
Methods
We conducted a target trial emulation study using a retrospective, active comparator new-user design among adults with MASLD, T2DM, and low-to-intermediate Fibrosis-4 (FIB-4≤2.67) scores who initiated treatment with either SGLT2is or dipeptidyl peptidase-4 inhibitors (DPP-4is) at Mass General Brigham or Asan Medical Center from 2013 to 2023. The primary outcome was the progression to advanced fibrosis (FIB-4>2.67), confirmed on ≥2 occasions within 1 year. The secondary outcome was the development of major adverse liver outcomes (MALO), including incident cirrhosis, decompensation events, hepatocellular carcinoma, or liver transplantation.
-
Results
Among 16,901 eligible patients, 2,571 propensity score-matched pairs were identified with balanced baseline characteristics. During follow-up (median, 3.7 years), fibrosis progression occurred at a rate of 3.46/100 person-years in SGLT2i users and 4.44 in DPP4i users. SGLT2i use was associated with a lower risk of fibrosis progression (HR 0.78, 95% CI 0.67–0.89; P<0.001). No significant difference in MALO incidence was observed. Subgroup analyses showed a consistent association among users of metformin, statins, and aspirin.
-
Conclusions
SGLT2i use was associated with reduced risk of fibrotic progression compared to DPP4i use in adults with MASLD and T2DM.
-
Keywords: Fatty liver; Fibrosis; Liver fibrosis; MASLD
Study Highlights
• This large retrospective cohort study involving 5,142 patients with type 2 diabetes and MASLD demonstrated that sodium-glucose cotransporter-2 inhibitor use was associated with a 22% lower risk of liver fibrosis progression compared to DPP-4i use.
• The protective association remained consistent across multiple analytic approaches and clinically relevant subgroups, including patients with low baseline fibrosis risk and those with BMI<30 kg/m2.
• Mediation analyses revealed that the hepatoprotective effects extended beyond metabolic improvements alone, suggesting direct modulation of hepatic inflammatory and fibrotic pathways.
• These findings support the potential role of SGLT2is in preventing liver disease progression in this high-risk population.
Graphical Abstract
INTRODUCTION
Metabolic dysfunction-associated steatotic liver disease (MASLD) has emerged as a leading cause of chronic liver disease and its complications, with estimates suggesting it will affect over 100 million people in the United States by 2030 [
1]. The rising prevalence of MASLD parallels the increasing rates of cardiometabolic risk factors, including obesity, insulin resistance, and type 2 diabetes mellitus (T2DM) [
2]. Currently, there are no pharmacologic agents approved by the U.S. Food and Drug Administration for the treatment of MASLD in the absence of steatohepatitis and clinically significant fibrosis. However, the strong association between MASLD and T2DM presents an opportunity to evaluate whether glycemic control agents approved for the treatment of T2DM may also have dual efficacy in mitigating the progression of liver fibrosis.
Glucagon-like peptide-1 receptor agonists [
3-
5] (GLP-1RAs) and thiazolidinediones [
6,
7], which are used in the treatment of T2DM, have been shown to be effective in promoting the resolution of MASH and in reducing fibrotic progression. Sodium-glucose cotransporter-2 inhibitor (SGLT2i), an increasingly utilized class of oral antihyperglycemic agents approved for the treatment of T2DM, chronic kidney disease (CKD), heart failure, and atherosclerotic cardiovascular disease, also have significant chemopreventive potential. SGLT2is block renal glucose reabsorption and have pleiotropic effects that may contribute to the mitigation of fibrotic progression, including blood glucose reduction, modest weight loss, and cardiorenal protection [
8,
9]. In animal models, SGLT2is have been shown to reduce lipogenesis and oxidative stress, properties that may be protective in MASLD [
10-
12].
Clinical studies investigating the association between SGLT2is and the progression of fibrotic liver disease in patients with MASLD are limited and often involve small sample sizes [
13-
16]. A systematic review of randomized controlled trials evaluating SGLT2is in MASLD reported improvements in liver transaminase levels, body mass index (BMI), lipid profiles, and glycemic control [
9]. However, the impact of SGLT2is on fibrotic progression, particularly in patients with low-to-moderate risk of clinically significant fibrosis, remains poorly understood. In this study, we investigated the association between SGLT2i use and the risk of both fibrotic progression and major adverse liver outcomes (MALOs) in patients with both MASLD and T2DM treated at two large medical centers.
MATERIALS AND METHODS
The study protocol was reviewed and approved by the Institutional Review Board of Massachusetts General Hospital (No. 2024P001205) and Asan Medical Center (2023-1481). The requirement for obtaining informed consent was waived considering the retrospective nature of the study. The study adhered to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.
Study design and data source
We designed this retrospective cohort study to emulate a target trial evaluating the effect of SGLT2is compared with dipeptidyl peptidase-4 inhibitors (DPP-4is) on the progression of liver fibrosis in patients with MASLD and T2DM. DPP-4i was selected as the active comparator because it represents a similarly positioned second- to third-line antidiabetic therapy with a neutral effect on hepatic outcomes. The key components and specifications of the emulated target trial are detailed in
Supplementary Table 1.
We used two large-scale electronic health record databases: the Research Patient Data Registry (RPDR) from the Mass General Brigham (MGB) network in Boston, MA, and the Asan BiomedicaL research Environment (ABLE) from Asan Medical Center (AMC) in Seoul, Republic of Korea. The RPDR captures longitudinal clinical data from over 4 million patients, while the ABLE system encompasses data from more than 5.5 million patients who have received care at AMC.
Study population
The study included adult patients (aged ≥18 years) diagnosed with both T2DM and MASLD who initiated treatment with either an SGLT2i or a DPP-4i between January 1, 2013, and December 31, 2023, at either MGB or AMC (
Fig. 1). The index date was defined as the date of the first prescription for the respective antidiabetic agent. To restrict the cohort to individuals with early-stage liver disease, we included only those with a baseline Fibrosis-4 (FIB-4) score in the low (<1.3) or intermediate (1.3–2.67) risk range. MASLD was defined using a hierarchical approach as follows: (1) International Classification of Diseases, 9th and 10th Revisions (ICD-9/10) diagnosis codes or (2) Hepatic Steatosis Index (HSI)>36 in the absence of ICD codes. HSI was selected because its components (BMI, aspartate aminotransferase [AST] levels, alanine aminotransferase [ALT] levels, and sex) were routinely available, and it has been extensively validated for MASLD detection in T2DM populations (areas under the receiver operating characteristic curves: 0.79–0.98) [
17]. We further validated the use of an HSI>36 for MASLD identification through manual chart review of 200 randomly selected patients. Those identified as having MASLD based on an HSI>36 demonstrated a positive predictive value of 91.0%, consistent with ICD code-based validation results. Exclusion criteria included a prior diagnosis of hepatitis B or hepatitis C virus infection, cirrhosis, hepatic decompensation, hepatocellular carcinoma (HCC), liver transplant, bariatric surgery, or a baseline FIB-4 score>2.67. Patients with less than six months of follow-up or prior exposure to either drug class were also excluded.
To minimize the impact of confounding variables, we employed 1:1 propensity score (PS) matching with a caliper of 0.2 between SGLT2i and DPP-4i initiators. PSs were calculated based on demographic, clinical, laboratory, and treatment-related variables assessed at baseline. Patients were followed from the index date until the occurrence of study outcomes or the end of the study period (October 31, 2024), whichever occurred first.
Exposures and covariates
The exposure was the initiation of an SGLT2i, with the comparator being the new use of a DPP-4i. The index date was the first prescription of either agent, and prior exposure to either drug class was excluded. Concomitant medications were also considered. Comorbidities such as hypertension, dyslipidemia, and coronary artery disease (CAD), CKD, cerebrovascular accident (CVA), peripheral vascular disease (PVD), and liver-related conditions were identified based on at least two ICD-9 or ICD-10 codes prior to the index date. Laboratory values, including liver enzymes, platelet count, serum creatinine, lipid levels, and hemoglobin A1c (HbA1c), were extracted from the closest measurements taken within one year prior to the index date. Medication exposure was defined based on prescriptions recorded during the one-year period prior to cohort entry. Full definitions of variables, diagnostic codes, and ascertainment windows are provided in
Supplementary Tables 2–
4.
The primary outcome was the progression of liver fibrosis, defined as a transition from a low or intermediate FIB-4 risk category to a high-risk category (FIB-4>2.67). To ensure robustness, fibrosis progression was considered confirmed only when two separate FIB-4 scores exceeding the high-risk threshold were observed within a one-year interval. The date of the second qualifying measurement was used as the event time. The secondary outcome was the development of MASLD-associated MALO, a composite endpoint that included incident cirrhosis, hepatic decompensation, HCC, or liver transplantation. Each component of this outcome was defined based on ICD-9 and ICD-10 codes according to the operational definitions described in
Supplementary Tables 2,
3. For each patient, the event date was defined as the earliest occurrence of any of the included components.
To adjust for confounding variables, we estimated PSs using logistic regression to model the probability of initiating an SGLT2i vs. a DPP-4i. The PS model included baseline covariates measured before or on the index date, including demographic characteristics (age, sex, race), metabolic factors (BMI, weight), comorbidities (hypertension, dyslipidemia, CAD, CKD, PVD, CVA, heart failure, chronic obstructive pulmonary disease (COPD), hyperthyroidism, and hypothyroidism), and co-medications (metformin, aspirin, statins, niacin, fibrates, bile acid sequestrants, insulin, angiotensin-converting enzyme inhibitors, and angiotensin receptor blockers). Laboratory parameters such as AST, ALT, total bilirubin, platelet count, total cholesterol, high-density lipoprotein (HDL), low-density lipoprotein (LDL), triglycerides, and HbA1c were also included. Additionally, liver fibrosis markers (FIB-4 score, AST to platelet ratio index [APRI] score) and calendar time (index year and month) were considered. Patients initiating SGLT2i were matched at a 1:1 ratio with those initiating DPP-4i using nearest neighbor matching without replacement, with a caliper width of 0.2 standard deviations of the logit of the PS. We achieved excellent covariate balance after matching, with absolute standardized mean differences (SMDs) below 0.1 for all included variables (
Supplementary Fig. 1).
The primary analysis was based on an intention-to-treat (ITT) approach, in which follow-up commenced at the index date and continued until one of the following events occurred: study outcome, treatment discontinuation, death, or administrative censoring (October 31, 2024). A 90-day extension after the final prescription was allowed to account for delayed discontinuation. A per-protocol analysis was also performed, in which follow-up ended at the earliest of the ITT endpoints or at the time of crossover or addon between treatment groups (i.e., DPP-4i use among SGLT2i users or vice versa).
We calculated incidence rates and 95% confidence intervals (CIs) for both the primary and secondary outcomes. Time-to-event analyses were conducted using Cox proportional hazards models within the matched cohort to estimate hazard ratios (HRs) and 95% CIs for fibrosis progression. For the composite secondary outcome (MALO), we also constructed cumulative incidence curves and conducted cause-specific hazard modeling.
Subgroup analyses were conducted to explore potential effect modification across clinically relevant strata. These subgroups included healthcare system (MGB vs. AMC), baseline fibrosis risk (low vs. intermediate FIB-4 score), BMI category (<30 vs. ≥30 kg/m2), glycemic control (HbA1c ≤8% vs. >8%), MASLD diagnostic source (ICD-based vs. HSI/imaging-based), and the use of concomitant medications such as metformin, aspirin, and statins. PS-matched Cox models were constructed within each subgroup to assess the consistency of the treatment effect. Sensitivity analyses included re-estimation of the PS and repetition of matching procedures, as well as landmark analyses restricted to patients with a minimum of 1 year of follow-up. To evaluate whether changes in BMI, body weight, HbA1c, or total cholesterol levels mediated the association between SGLT2i use and the risk of fibrosis progression, we conducted causal mediation analyses using Aalen’s additive hazards models. These analyses treated each laboratory variable as a time-varying covariate and estimated the total effect of treatment, the direct effect, and the indirect (mediation-like) effect through the mediator.
All statistical analyses were performed using R version 4.2.1 (R Foundation for Statistical Computing, Vienna, Austria). Two-sided P-values<0.05 were considered statistically significant.
RESULTS
Baseline characteristics
Of the 16,901 identified patients who initiated treatment with SGLT2i or DPP-4i, a total of 2,571 new SGLT2i users were compared with 2,571 DPP-4i users after PS matching. The groups were well balanced across all baseline characteristics, with all SMDs being less than 0.1, except for data source (SMD=0.119) (
Table 1,
Supplementary Table 5). The mean age was 63.1 years in the SGLT2i group and 62.8 years in the DPP-4i group, and 55.0% were male in both groups. Most patients were from the MGB network, and the racial/ethnic distribution was similar between the two groups. The rates of comorbidities (e.g., hypertension, dyslipidemia, CKD) and the use of co-medications (e.g., metformin and statins) were comparable. Liver-related laboratory values, including the FIB-4 score, were also similar, with approximately 80% of patients classified as low fibrosis risk at baseline (FIB-4<1.3).
Among patients with MASLD and T2DM, the use of SGLT2i was associated with a significantly lower risk of fibrosis progression compared with DPP-4i. In the ITT analysis, the incidence rate of fibrosis progression was 3.46 per 100 person-years (PYs) in the SGLT2i group and 4.44 per 100 PYs in the DPP-4i group, resulting in an HR of 0.78 (95% CI 0.67–0.89) (
Table 2). The cumulative incidence curves for fibrosis progression showed a clear separation between the groups over time (
Fig. 2A). In the per-protocol analysis, which censored patients at the time of treatment discontinuation, the protective association remained consistent (HR 0.78, 95% CI 0.67–0.90,
Supplementary Fig. 2), with a similar divergence observed in the cumulative incidence curves.
In the ITT analysis, the incidence of MALO was low and comparable between the two groups, with rates of 0.49 and 0.51 events per 100 PYs in the DPP-4i and SGLT2i groups, respectively. The corresponding HR was 1.06 (95% CI 0.73–1.52), indicating no statistically significant difference (
Table 2,
Fig. 2B). Similar findings were observed in the per-protocol analysis.
Prespecified subgroup analyses were performed to assess the consistency of the primary outcome across clinically relevant strata within the PS-matched cohort (
Fig. 3). SGLT2i use was associated with a reduced risk of fibrosis progression compared with DPP-4i use across various patient subgroups. The protective association remained significant among patients with a low baseline FIB-4 score (HR 0.79, 95% CI 0.66–0.96), those with an HbA1c level greater than 8% (HR 0.76, 95% CI 0.62–0.93), statin users (HR 0.79, 95% CI 0.67–0.93), and patients aged <65 years (HR 0.77, 95% CI 0.60–0.98). The effect size appeared consistent among metformin users (HR 0.81, 95% CI 0.68–0.97) and aspirin users (HR 0.84, 95% CI 0.71–1.01), although the latter did not reach statistical significance. Moreover, metformin use was associated with a lower incidence of fibrosis progression in both treatment groups, with combined SGLT2i and metformin treatment showing the most favorable outcome (
Supplementary Table 6,
Supplementary Fig. 3). In a sensitivity analysis stratified by MASLD diagnostic source, the protective association remained consistent across the ICD-based MASLD cohort (HR 0.84, 95% CI 0.71–0.98) and the HSI or imaging-based MASLD cohort (HR 0.73, 95% CI 0.52–1.05). A per-protocol version of the subgroup analysis is presented in
Supplementary Figure 4, showing a generally similar trend across the strata.
We conducted a landmark analysis limited to patients with at least one year of follow-up, followed by re-estimation of the PS and re-matching. In this cohort, SGLT2i use remained significantly associated with a reduced risk of fibrosis progression, consistent with the main analysis (
Supplementary Table 7,
Supplementary Figs. 5,
6). The association with MALO did not significantly differ between the two groups. Additional negative control outcome analyses demonstrated no significant difference between SGLT2i and DPP-4i users in the risk of incident fractures or lung cancer (
Supplementary Fig. 7).
Changes in BMI, body weight, HbA1c, and total cholesterol were evaluated as potential mediators of the association between SGLT2i use and fibrosis progression. The mediation effects were modest across all variables, accounting for 2.4% for BMI, 6.2% for body weight, 1.0% for HbA1c, and 0.3% for total cholesterol. The majority of the observed association was attributable to the direct effect of SGLT2i use rather than through these intermediate metabolic changes (
Supplementary Table 8).
DISCUSSION
In this large, retrospective cohort study emulating a target trial, we demonstrated that among patients with T2DM and MASLD, the use of SGLT2i was associated with a significantly lower risk of liver fibrosis progression compared with the use of DPP-4i. This association was consistent across various analytic strategies, including ITT, per-protocol, and landmark analyses. Furthermore, the protective effect remained robust across clinically relevant subgroups, including patients with a low baseline fibrosis burden, those with a BMI<30 kg/m2, and those receiving concomitant statins or metformin.
Mechanistically, the observed benefits of SGLT2is are biologically plausible and supported by experimental evidence suggesting that these agents have multifactorial hepatoprotective effects. Preclinical studies have demonstrated that SGLT2i reduces hepatic steatosis by suppressing de novo lipogenesis and enhancing lipid oxidation, thereby decreasing liver fat accumulation—an upstream driver of fibrosis progression [
18]. In a study involving histological examination, treatment with canagliflozin for 24 weeks showed significant histological improvements in scores for steatosis, lobular inflammation, ballooning, and fibrosis stage among patients with MAFLD and T2DM [
19]. Additionally, treatment with empagliflozin in patients with MAFLD and T2DM showed histological improvements in steatosis, ballooning, and fibrosis compared to those treated with a placebo [
20]. In patients with T2DM, SGLT2i use was associated with weight reductions and decreased adiposity due to the elimination of glucose in the urine, thereby aiding in the management of obesity, T2DM, and MASLD [
21]. However, beyond weight loss, SGLT2i use leads to a reduction in the insulin/glucagon ratio, resulting in hepatic responses including reduced glucose uptake, decreased glycogen and de novo lipid synthesis, and stimulation of β-oxidation and ketogenesis. Finally, SGLT2i has been shown to improve insulin sensitivity and reduce hepatic inflammation, leading to significant histological improvements in patients with MASLD [
22,
23]. Our mediation analyses further support this notion, demonstrating that reductions in body weight and improvements in glucose control only partially account for the observed association between SGLT2i use and a lower risk of fibrosis progression. This suggests that the hepatoprotective effects of SGLT2i extend beyond metabolic improvements alone and may involve direct modulation of hepatic inflammatory and fibrotic pathways.
Our findings align with and expand upon an emerging body of evidence supporting the potential hepatoprotective effects of SGLT2i in patients with T2DM and MASLD. Most notably, the recent DEAN trial evaluated dapagliflozin in patients with biopsy-confirmed MASH, many of whom also had concomitant T2DM [
24]. In this DEAN trial, 48 weeks of dapagliflozin treatment resulted in significantly higher rates of MASH improvement without worsening fibrosis compared to placebo, as well as fibrosis improvement without worsening MASH [
24]. Additionally, the antifibrotic benefits appeared to be consistent regardless of diabetes status or baseline metabolic risk factors. These results directly corroborate our study, which demonstrated a reduced risk of fibrosis progression based on serial FIB-4 assessments in a large real-world cohort. Additionally, a 72-week prospective trial conducted by Takahashi et al. [
25] also demonstrated that ipragliflozin significantly improved liver histology compared with standard antidiabetic therapy. Moreover, meta-analyses and systematic reviews have consistently found that SGLT2is improve non-invasive markers such as liver fat, stiffness, and liver enzymes [
9,
26,
27], although data on hard fibrosis endpoints remain limited.
Our subgroup analyses revealed consistent benefits among patients with low baseline fibrosis risk, supporting the notion that early intervention may yield greater advantages before advanced fibrosis develops. In contrast, the protective association was attenuated among patients with intermediate FIB-4 scores; however, this may be partly due to the smaller sample size of this subgroup and reduced statistical power. These findings align with the hypothesis that SGLT2i may be more effective in slowing the initiation and progression of fibrogenic processes in early-stage disease. Interestingly, the benefits of SGLT2i were also evident in patients with lower BMI, further suggesting that mechanisms beyond weight reduction may contribute to the attenuation of fibrosis.
A notable strength of our study lies in the use of two large, hospital-based cohorts that incorporate extensive laboratory and clinical data. Unlike previous claims-based studies that primarily relied on administrative data without serial laboratory measurements, our study leveraged comprehensive longitudinal data, including repeated FIB-4 scores, which enabled a detailed evaluation of the risk of fibrosis progression over time. Importantly, many prior studies assessing the impact of SGLT2i use in patients with MASLD have focused on severe liver outcomes such as HCC, hepatic decompensation, or liver-related mortality [
28-
31]. While these endpoints are clinically relevant and are often captured in large administrative database studies, they typically reflect advanced stages of liver disease and may only manifest after prolonged observation periods. In contrast, our approach, which emulates a target trial with PS matching and serial non-invasive fibrosis assessments, allowed us to examine earlier trajectories and detect differences in fibrosis progression before the onset of overt hepatic complications. Importantly, repeated assessments of fibrosis progression in real-world practice offer unique advantages over claims-based studies that rely solely on diagnosis codes. While histological confirmation through liver biopsy remains the reference standard for fibrosis staging, it is not feasible to perform serial biopsies in large populations for research purposes. Therefore, our ability to longitudinally capture fibrosis progression using validated noninvasive measurements in a well-characterized cohort represents a significant methodological strength.
Several limitations warrant consideration. First, as with all observational studies, the possibility of residual confounding cannot be entirely eliminated, despite our use of PS matching and target trial emulation methodology. The potential for misclassification bias because of incomplete records of alcohol consumption and other alternative causes of liver disease in electronic medical record-based studies represents an inherent limitation of our observational design. Moreover, imaging studies performed outside our healthcare networks were not available, and baseline elastography data were substantially limited, precluding the inclusion of liver stiffness measurements in the baseline fibrosis assessment. Although we accounted for a comprehensive set of potential confounders—including demographics, comorbidities, laboratory parameters, concomitant medications, and other clinical variables—unmeasured factors such as lifestyle behaviors (e.g., diet and exercise) could still influence treatment selection and the risk of fibrosis progression. Additionally, we were unable to include diabetic microvascular complications and urine albumin levels as covariates because of database coding discrepancies and substantial missing data, which may have limited our ability to adjust for disease severity. Second, while the FIB-4 score is a validated non-invasive marker for fibrosis progression, it remains a surrogate measure, and histological confirmation via liver biopsy was not available in this study. Additionally, concomitant medications affecting FIB-4 components may not have been fully ascertained, although our target trial emulation design with PS matching has minimized such bias. Third, our analysis grouped all SLGT2i into a single exposure category, which precludes the assessment of potential differences among individual agents. Given emerging evidence suggesting variability in metabolic and cardiovascular benefits across the SGLT2i class, further studies should consider evaluating agent-specific effects on liver outcomes. Additionally, future studies comparing SGLT2i users to a broader reference population of patients with DM and MASLD not using these specific medication classes will be important to establish absolute treatment benefits and provide reference incidence rates for the broader population. Moreover, although the data source distribution remained slightly imbalanced after PS matching, additional adjustment for data source did not materially change the results (HR 0.77 vs. 0.78), and stratified analyses showed consistent effect directions but with statistically marginal associations due to reduced sample sizes. Finally, our study population predominantly consisted of patients with early-stage MASLD, which limits our ability to draw conclusions regarding long-term effects.
In conclusion, among patients with T2DM and early-stage MASLD, the use of SGLT2is was associated with a significantly lower risk of liver fibrosis progression compared to the use of DPP-4is. These findings highlight the potential role of SGLT2i as part of a strategy to prevent the progression of liver disease in this high-risk population. Further prospective studies and randomized trials are warranted to confirm these observations and explore their implications for clinical practice.
FOOTNOTES
-
Access to Data and Data Analysis
J Choi and RT Chung had full access to the data used in this study and take responsibility for its integrity and for the accuracy of the analyses.
-
Data Availability
Data from the hospital cohorts are available on request from the corresponding author. The data are not publicly available due to the privacy policy.
-
Authors’ contributions
J Choi and RT Chung are the guarantors of the article. All authors had full access to the data used in this study and take responsibility for its integrity and for the accuracy of the analyses. J Choi and RT Chung were responsible for the conception and design of the study; the acquisition, analysis, and interpretation of data; and drafting of the manuscript. J Choi and D Fulop performed the statistical analyses. All authors were responsible for the data acquisition, critical revision of the manuscript, and approved the final version of the manuscript.
-
Acknowledgements
This study was supported by NIH R01 CA255621, U01 CA288375.
This funding source had no role in the design of this study and will not have any role during its execution, analyses, interpretation of the data, or decision to submit results.
-
Conflicts of Interest
Jonggi Choi received a research grant from Gilead Sciences. Other authors have nothing to declare.
SUPPLEMENTARY MATERIAL
Supplementary material is available at Clinical and Molecular Hepatology website (
http://www.e-cmh.org).
Supplementary Figure 1.
Propensity score distribution before and after matching. Kernel density plots of propensity scores by treatment group before and after 1:1 propensity score matching. DPP-4, dipeptidyl peptidase-4; SGLT2, sodium-glucose cotransporter-2.
Supplementary Figure 2.
Cumulative Incidence of (A) fibrosis progression and (B) major adverse liver-related outcome (MALO) by perprotocol analysis. DPP-4, dipeptidyl peptidase-4; SGLT2, sodium-glucose cotransporter-2.
Supplementary Figure 3.
Cumulative Incidence of fibrosis progression according to metformin use. DPP-4, dipeptidyl peptidase-4; SGLT2, sodium-glucose cotransporter-2.
Supplementary Figure 4.
Subgroup analysis of fibrosis progression by treatment group using per-protocol analysis. DPP-4i, dipeptidyl peptidase-4 inhibitor; SGLT2i, sodium-glucose cotransporter 2 inhibitor; HR, hazard ratio; CI, confidence intervals; PS, propensity score; MGB, Mass General Brigham; AMC, Asan Medical Center; FIB-4, Fibrosis-4 index; BMI, body mass index; HbA1c, hemoglobin A1c; ICD, International Classification of Diseases; HSI, Hepatic Steatosis Index.
Supplementary Figure 5.
Cumulative incidence of fibrosis progression in the landmark analysis among patients with ≥1 year of follow-up (intention-to-treat analysis). DPP-4, dipeptidyl peptidase-4; SGLT2, sodium-glucose cotransporter-2.
Supplementary Figure 6.
Cumulative incidence of fibrosis progression in the landmark analysis among patients with ≥1 year of follow-up (per-protocol analysis). DPP-4, dipeptidyl peptidase-4; SGLT2, sodium-glucose cotransporter-2.
Supplementary Figure 7.
Cumulative incidence of negative control outcomes: (A) incident fracture and (B) lung cancer. DPP-4, dipeptidyl peptidase-4; SGLT2, sodium-glucose cotransporter-2.
Figure 1.Study flow and cohort definition. T2DM, type 2 diabetes; SGLT2, sodium glucose cotransporter 2; DPP-4, dipeptidyl peptidase 4; MASLD, metabolic dysfunction-associated steatotic liver disease; HBV, hepatitis B; HCV, hepatitis C; CKD, chronic kidney disease; FIB-4, Fibrosis-4.
Figure 2.Cumulative incidence of (A) fibrosis progression and (B) major adverse liver-related outcome in SGLT2 inhibitor users vs. DPP-4 inhibitor users. Kaplan–Meier curves showing cumulative incidence of (A) fibrosis progression and (B) major adverse liver-related outcomes (MALO: cirrhosis, HCC, decompensation, or transplant). SGLT2, sodium-glucose cotransporter 2; DPP-4, dipeptidyl peptidase-4; MALO, major adverse liver-related outcome; HCC, hepatocellular carcinoma.
Figure 3.Subgroup analysis of fibrosis progression by treatment group. Forest plot showing adjusted HRs for progression to high-risk fibrosis (FIB-4≥2.67) across clinical subgroups comparing SGLT2i use vs. DPP-4i use. HRs, hazard ratios; FIB-4, Fibrosis-4; SGLT2i, sodium-glucose cotransporter 2 inhibitor; DPP-4i, dipeptidyl peptidase-4 inhibitor; CI, confidence intervals; PS, propensity score; MGB, Mass General Brigham; AMC, Asan Medical Center; BMI, body mass index; HbA1c, hemoglobin A1c; ICD, International Classification of Diseases; HSI, Hepatic Steatosis Index.
Table 1.Baseline characteristics of the study population after propensity score matching
Table 1.
|
Characteristic |
SGLT2i user (n=2,571) |
DPP-4i user (n=2,571) |
SMD |
|
Data source |
|
|
|
|
MGB network |
1,883 (73.2) |
1,744 (67.8) |
0.119 |
|
Asan Medical Center |
688 (26.8) |
827 (32.2) |
|
|
Demographics |
|
|
|
|
Age (yr) |
63.1±11.9 |
62.8±13 |
0.024 |
|
Sex, male |
1,415 (55.0) |
1,407 (54.7) |
0.006 |
|
Race and ethnicity*
|
|
|
0.036 |
|
Asian |
823 (32.0) |
863 (33.6) |
|
|
Black |
229 (8.9) |
220 (8.6) |
|
|
White |
1,273 (49.5) |
1,238 (48.2) |
|
|
Other |
246 (9.6) |
250 (9.7) |
|
|
Body mass index (kg/m2) |
|
|
0.022 |
|
<25 |
178 (6.9) |
187 (7.3) |
|
|
25–29.9 |
1,487 (57.8) |
1,460 (56.8) |
|
|
≥30 |
906 (35.2) |
924 (35.9) |
|
|
Comorbidity†
|
|
|
|
|
Hypertension |
2,105 (81.9) |
2,081 (80.9) |
0.024 |
|
Dyslipidemia |
1,819 (70.8) |
1,777 (69.1) |
0.036 |
|
Chronic kidney disease |
544 (21.2) |
527 (20.5) |
0.016 |
|
Coronary artery disease |
1,009 (39.2) |
1,005 (39.1) |
0.003 |
|
Peripheral vascular disease |
384 (14.9) |
385 (15.0) |
0.001 |
|
Cerebrovascular accident |
519 (20.2) |
503 (19.6) |
0.016 |
|
Heart failure |
561 (21.8) |
546 (21.2) |
0.014 |
|
Chronic obstructive pulmonary disease |
487 (18.9) |
468 (18.2) |
0.019 |
|
Hypothyroidism |
127 (4.9) |
118 (4.6) |
0.016 |
|
Hyperthyroidism |
428 (16.6) |
398 (15.5) |
0.032 |
|
Medications |
|
|
|
|
Metformin use |
1,788 (69.5) |
1,781 (69.3) |
0.006 |
|
Aspirin use |
1,258 (48.9) |
1,267 (49.3) |
0.007 |
|
Statin use |
1,892 (73.6) |
1,890 (73.5) |
0.002 |
|
Fibrate use |
135 (5.3) |
121 (4.7) |
0.025 |
|
Nicotinic acid use |
15 (0.6) |
12 (0.5) |
0.016 |
|
Bile acid sequestrant use |
25 (1.0) |
23 (0.9) |
0.008 |
|
Thiazolidinedione use |
61 (2.4) |
54 (2.1) |
0.018 |
|
GLP-1RA use |
299 (11.6) |
267 (10.4) |
0.040 |
|
Sulfonylurea use |
974 (37.9) |
976 (38.0) |
0.002 |
|
Insulin use |
1,421 (55.3) |
1,397 (54.3) |
0.019 |
|
ACE inhibitor use |
920 (35.8) |
901 (35.0) |
0.015 |
|
Angiotensin receptor blocker use |
769 (29.9) |
786 (30.6) |
0.014 |
|
Baseline laboratory tests |
|
|
|
|
Platelet count |
252±73 |
253±73 |
0.007 |
|
Aspartate aminotransferase |
24±11 |
24±13 |
0.022 |
|
Alanine aminotransferase |
28±19 |
28±21 |
0.023 |
|
Total cholesterol |
159±44 |
158±43 |
0.014 |
|
Serum creatinine |
1.01±0.5 |
1.01±0.5 |
0.005 |
|
Estimated GFR (MDRD) |
76.6±24.9 |
77.0±27.3 |
0.063 |
|
Estimated GFR (CKD-EPI) |
77.2±23.0 |
78.0±24.6 |
0.033 |
|
Total bilirubin |
0.5±0.3 |
0.5±0.3 |
0.014 |
|
High-density lipoprotein |
45±13 |
45±12 |
0.002 |
|
Low-density lipoprotein |
84±35 |
84±33 |
0.007 |
|
Triglyceride |
185±156 |
182±220 |
0.018 |
|
Hemoglobin A1c |
8.1±1.5 |
8.1±1.5 |
0.001 |
|
FIB-4 score‡
|
1.25±0.51 |
1.25±0.52 |
0.009 |
|
FIB-4 score: low risk |
2,037 (79.2) |
2,026 (78.8) |
0.011 |
|
FIB-4 score: intermediate risk |
534 (20.8) |
545 (21.2) |
|
Table 2.Association of SGLT2 inhibitor use with clinical outcomes
Table 2.
|
Exposure |
No. of events |
Person-years |
Incidence per 100 person-years (95% CI) |
Hazard ratio (95% CI) |
|
Intention-to-treat |
|
|
|
|
|
Primary outcome: fibrosis progression |
|
|
|
|
|
SGLT2i |
358 |
10,332 |
3.46 (3.12–3.84) |
0.78 (0.67–0.89) |
|
DPP-4i |
432 |
9,729 |
4.44 (4.03–4.88) |
Reference |
|
Secondary outcome: major adverse liver-related outcomes |
|
|
|
SGLT2i |
60 |
11,681 |
0.51 (0.39–0.66) |
1.06 (0.73–1.52) |
|
DPP-4i |
55 |
11,250 |
0.49 (0.37–0.64) |
Reference |
|
Per-protocol |
|
|
|
|
|
Primary outcome: fibrosis progression |
|
|
|
|
|
SGLT2i |
317 |
10,228 |
3.10 (2.77–3.46) |
0.78 (0.67–0.90) |
|
DPP-4i |
382 |
9,605 |
3.98 (3.59–4.40) |
Reference |
|
Secondary outcome: major adverse liver-related outcomes |
|
|
|
SGLT2i |
52 |
10,559 |
0.49 (0.37–0.65) |
1.11 (0.73–1.70) |
|
DPP-4i |
38 |
8,621 |
0.44 (0.31–0.61) |
Reference |
Abbreviations
Asan BiomedicaL research Environment
AST to platelet ratio index
aspartate aminotransferase
dipeptidyl peptidase-4 inhibitor
glucagon-like peptide-1 receptor agonist
International Classification of Diseases
major adverse liver outcome
metabolic dysfunction-associated steatotic liver disease
peripheral vascular disease
Research Patient Data Registry
sodium-glucose cotransporter-2 inhibitor
standardized mean difference
REFERENCES
- 1. Estes C, Razavi H, Loomba R, Younossi Z, Sanyal AJ. Modeling the epidemic of nonalcoholic fatty liver disease demonstrates an exponential increase in burden of disease. Hepatology 2018;67:123-133.
- 2. Kabbany MN, Conjeevaram Selvakumar PK, Watt K, Lopez R, Akras Z, Zein N, et al. Prevalence of nonalcoholic steatohepatitis-associated cirrhosis in the United States: An analysis of National Health and Nutrition Examination Survey data. Am J Gastroenterol 2017;112:581-587.
- 3. Armstrong MJ, Gaunt P, Aithal GP, Barton D, Hull D, Parker R, et al. Liraglutide safety and efficacy in patients with nonalcoholic steatohepatitis (LEAN): a multicentre, double-blind, randomised, placebo-controlled phase 2 study. Lancet 2016;387:679-690.
- 4. Newsome PN, Buchholtz K, Cusi K, Linder M, Okanoue T, Ratziu V, et al.; NN9931-4296 Investigators. A placebo-controlled trial of subcutaneous semaglutide in nonalcoholic steatohepatitis. N Engl J Med 2021;384:1113-1124.
- 5. Loomba R, Hartman ML, Lawitz EJ, Vuppalanchi R, Boursier J, Bugianesi E, et al.; SYNERGY-NASH Investigators. Tirzepatide for metabolic dysfunction-associated steatohepatitis with liver fibrosis. N Engl J Med 2024;391:299-310.
- 6. Musso G, Cassader M, Paschetta E, Gambino R. Thiazolidinediones and advanced liver fibrosis in nonalcoholic steatohepatitis: a meta-analysis. JAMA Intern Med 2017;177:633-640.
- 7. Lian J, Fu J. Pioglitazone for NAFLD patients with prediabetes or type 2 diabetes mellitus: a meta-analysis. Front Endocrinol (Lausanne) 2021;12:615409.
- 8. Zinman B, Wanner C, Lachin JM, Fitchett D, Bluhmki E, Hantel S, et al.; EMPA-REG OUTCOME Investigators. Empagliflozin, cardiovascular outcomes, and mortality in type 2 diabetes. N Engl J Med 2015;373:2117-2128.
- 9. Li H, Hou Y, Xin W, Ding L, Yang Y, Zhang Y, et al. The efficacy of sodium-glucose transporter 2 inhibitors in patients with nonalcoholic fatty liver disease: a systematic review and meta-analysis. Pharmacol Res 2025;213:107647.
- 10. Nasiri-Ansari N, Nikolopoulou C, Papoutsi K, Kyrou I, Mantzoros CS, Kyriakopoulos G, et al. Empagliflozin attenuates nonalcoholic fatty liver disease (NAFLD) in high fat diet fed ApoE(-/-) mice by activating autophagy and reducing ER stress and apoptosis. Int J Mol Sci 2021;22:818.
- 11. Anstee QM, Goldin RD. Mouse models in non-alcoholic fatty liver disease and steatohepatitis research. Int J Exp Pathol 2006;87:1-16.
- 12. Perakakis N, Chrysafi P, Feigh M, Veidal SS, Mantzoros CS. Empagliflozin improves metabolic and hepatic outcomes in a non-diabetic obese biopsy-proven mouse model of advanced NASH. Int J Mol Sci 2021;22:6332.
- 13. Latva-Rasku A, Honka MJ, Kullberg J, Mononen N, Lehtimäki T, Saltevo J, et al. The SGLT2 inhibitor dapagliflozin reduces liver fat but does not affect tissue insulin sensitivity: a randomized, double-blind, placebo-controlled study with 8-week treatment in type 2 diabetes patients. Diabetes Care 2019;42:931-937.
- 14. Kahl S, Gancheva S, Straßburger K, Herder C, Machann J, Katsuyama H, et al. Empagliflozin effectively lowers liver fat content in well-controlled type 2 diabetes: a randomized, double-blind, phase 4, placebo-controlled trial. Diabetes Care 2020;43:298-305.
- 15. Eriksson JW, Lundkvist P, Jansson PA, Johansson L, Kvarnström M, Moris L, et al. Effects of dapagliflozin and n-3 carboxylic acids on non-alcoholic fatty liver disease in people with type 2 diabetes: a double-blind randomised placebo-controlled study. Diabetologia 2018;61:1923-1934.
- 16. Cusi K, Bril F, Barb D, Polidori D, Sha S, Ghosh A, et al. Effect of canagliflozin treatment on hepatic triglyceride content and glucose metabolism in patients with type 2 diabetes. Diabetes Obes Metab 2019;21:812-821.
- 17. Vilar-Gomez E, Chalasani N. Non-invasive assessment of non-alcoholic fatty liver disease: clinical prediction rules and blood-based biomarkers. J Hepatol 2018;68:305-315.
- 18. Moon JS, Hong JH, Jung YJ, Ferrannini E, Nauck MA, Lim S. SGLT-2 inhibitors and GLP-1 receptor agonists in metabolic dysfunction-associated fatty liver disease. Trends Endocrinol Metab 2022;33:424-442.
- 19. Akuta N, Kawamura Y, Watanabe C, Nishimura A, Okubo M, Mori Y, et al. Impact of sodium glucose cotransporter 2 inhibitor on histological features and glucose metabolism of nonalcoholic fatty liver disease complicated by diabetes mellitus. Hepatol Res 2019;49:531-539.
- 20. Lai LL, Vethakkan SR, Nik Mustapha NR, Mahadeva S, Chan WK. Empagliflozin for the treatment of nonalcoholic steatohepatitis in patients with type 2 diabetes mellitus. Dig Dis Sci 2020;65:623-631.
- 21. Wilding JP. The role of the kidneys in glucose homeostasis in type 2 diabetes: clinical implications and therapeutic significance through sodium glucose co-transporter 2 inhibitors. Metabolism 2014;63:1228-1237.
- 22. Alatrach M, Laichuthai N, Martinez R, Agyin C, Ali AM, Al-Jobori H, et al. Evidence against an important role of plasma insulin and glucagon concentrations in the increase in EGP caused by SGLT2 inhibitors. Diabetes 2020;69:681-688.
- 23. Ferrannini E, Muscelli E, Frascerra S, Baldi S, Mari A, Heise T, et al. Metabolic response to sodium-glucose cotransporter 2 inhibition in type 2 diabetic patients. J Clin Invest 2014;124:499-508.
- 24. Lin J, Huang Y, Xu B, Gu X, Huang J, Sun J, et al. Effect of dapagliflozin on metabolic dysfunction-associated steatohepatitis: multicentre, double blind, randomised, placebo controlled trial. BMJ 2025;389:e083735.
- 25. Takahashi H, Kessoku T, Kawanaka M, Nonaka M, Hyogo H, Fujii H, et al. Ipragliflozin improves the hepatic outcomes of patients with diabetes with NAFLD. Hepatol Commun 2022;6:120-132.
- 26. Androutsakos T, Nasiri-Ansari N, Bakasis AD, Kyrou I, Efstathopoulos E, Randeva HS, et al. SGLT-2 inhibitors in NAFLD: expanding their role beyond diabetes and cardioprotection. Int J Mol Sci 2022;23:3107.
- 27. Mo M, Huang Z, Liang Y, Liao Y, Xia N. The safety and efficacy evaluation of sodium-glucose co-transporter 2 inhibitors for patients with non-alcoholic fatty liver disease: An updated meta-analysis. Dig Liver Dis 2022;54:461-468.
- 28. Krishnan A, Schneider CV, Mukherjee D, Woreta TA, Alqahtani SA. Adverse liver and renal outcomes after initiating SGLT-2i and GLP-1RA therapy among patients with diabetes and MASLD. J Diabetes 2025;17:e70069.
- 29. Yen FS, Wei JC, Wang C, Hou MC, Yu TS, Huang Y, et al. Sodium-glucose cotransporter2 inhibitors and associated liver-related outcomes in diabetes patients. Diabetes Res Clin Pract 2025;223:112174.
- 30. Mao X, Zhang X, Kam L, Chien N, Lai R, Cheung KS, et al. Synergistic association of sodium-glucose cotransporter-2 inhibitor and metformin on liver and non-liver complications in patients with type 2 diabetes mellitus and metabolic dysfunction-associated steatotic liver disease. Gut 2024;73:2054-2061.
- 31. Bea S, Ko HY, Bae JH, Cho YM, Chang Y, Ryu S, et al. Risk of hepatic events associated with use of sodium-glucose cotransporter-2 inhibitors versus glucagon-like peptide-1 receptor agonists, and thiazolidinediones among patients with metabolic dysfunction-associated steatotic liver disease. Gut 2025;74:284-294.
, Daniel Fulop3,*, Vy H. Nguyen4, Eric Przybyszewski2,3, Jiunn Song2,3, Allison Carroll3, Megan Michta2, Erik Almazan5, Tracey G. Simon2,3, Raymond T. Chung2,3
