Effect of empagliflozin on liver fibrosis and steatosis in patients with type 2 diabetes and non-alcoholic fatty liver disease: a randomized clinical trial

Abstract

Background and aim

Metabolic dysfunction-associated steatotic (MASLD) is a common complication in diabetic patients that can lead to fibrosis and severe liver complications. This study aimed to investigate the effect of empagliflozin on liver markers and fibrosis in diabetic patients with NAFLD.

Methods

This was a single-center, randomized, controlled, phase II clinical trial including 119 patients with type 2 diabetes and MASLD. Patients were randomized to receive either empagliflozin (10 mg or 25 mg daily) or standard care for 6 months. Liver enzymes, imaging scores (MRI and ultrasound), and fibrosis indices (FIB-4 and NFS) were assessed at baseline and at study completion. Statistical analyses included between- and within-group comparisons, adjusted models for baseline imbalances and age.

Results

Empagliflozin significantly reduced liver enzymes compared with controls (ALT − 41.2 vs. −4.6 IU/L; AST − 18.4 vs. −3.3 IU/L; GGT − 25.5 vs. −2.2 IU/L; all p < 0.001). The FIB-4 index decreased in the intervention group (1.20→1.06, p < 0.001) but not in controls (1.42→1.42, p = 0.233). The NAFLD fibrosis score showed no significant change. Imaging confirmed greater improvement in the empagliflozin group, with 94% showing grade 1 or lower steatosis on ultrasound and 100% achieving grade 0 on MRI (p < 0.001 for both). Importantly, complementary analyses (ANCOVA and mixed models) demonstrated that these improvements remained significant after adjustment for age and baseline values.

Conclusion

Empagliflozin significantly improved liver enzymes, imaging scores, and FIB-4 in patients with T2DM and MASLD. These findings support its potential as a therapeutic option in metabolic liver disease and highlight the need for longer, multicenter trials to confirm sustained benefits and to explore combination strategies with other agents.

Clinical trial registration

This study is related to a previously registered clinical trial conducted at Loqman-e-Hakim Hospital. The clinical trial was registered with the Iranian Registry of Clinical Trials under the number IRCT20210811052150N1.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12902-025-02098-6.

Introduction

MASLD (Metabolic dysfunction-Associated Steatotic Liver Disease) previously known as Non-alcoholic fatty liver disease (NAFLD) is often associated with type 2 diabetes mellitus (T2DM) [1]. The presence of T2DM in patients with MASLD is a significant risk factor for the progression of the disease to non-alcoholic steatohepatitis (NASH), which can ultimately lead to liver fibrosis, cirrhosis, and hepatocellular carcinoma [2]. The pathogenesis of MASLD is complex, involving insulin resistance, oxidative stress, lipid peroxidation, and mitochondrial dysfunction [3]. Insulin resistance is considered a key pathological driver in the development of both T2DM and MASLD [4, 5]. Various antidiabetic therapies have been evaluated for their efficacy in treating MASLD with mixed outcomes, including lifestyle modifications [6, 7], metformin [8], pioglitazone [9, 10], and liraglutide [11].

Empagliflozin, a potent oral antidiabetic agent, acts by inhibiting sodium-glucose cotransporter 2 (SGLT-2) [12]. This inhibition increases glucose excretion via urine, leading to reduced blood glucose levels and improved insulin resistance in T2DM patients [13]. Improvement in hyperglycemia decreases carbohydrate response element-binding protein (ChREBP), a transcription factor responsible for activating fatty acid synthesis pathways [14]. Additionally, amelioration of insulin resistance (hyperinsulinemia) results in decreased SREBP-1c expression, thereby halting hepatic lipogenesis [15]. A 2018 study evaluating the effects of empagliflozin on diabetic patients with MASLD demonstrated significant improvements in MASLD conditions, as evidenced by liver MRI and serum enzyme measurements. These improvements were highly noticeable following empagliflozin administration [16].

MRI and ultrasound are effective non-invasive modalities for assessing liver status, offering considerable accuracy in determining the extent of MASLD, NASH, and liver fibrosis while avoiding the risks associated with invasive liver biopsy. These imaging modalities utilize specific indices to stage the aforementioned conditions [17]. Moreover, serum liver enzymes such as alanine transaminase (ALT), aspartate transaminase (AST), gamma-glutamyl transferase (GGT), and alkaline phosphatase, along with composite indices derived from these enzymes and additional demographic and biochemical markers (e.g., fasting blood glucose and platelet count), provide valuable tools for liver assessment. Two widely used indices include FIB-4 and the MASLD fibrosis score. FIB-4 combines biochemical measures (AST, ALT, and platelet count) with patient age to predict fibrosis with good accuracy. The MASLD fibrosis score predicts the probability of fibrosis using variables such as age, BMI, fasting glucose, transaminase levels, platelet count, and albumin [18, 19].

Given the practicality and affordability of laboratory-based markers compared to radiological methods, evaluating liver conditions using these indices is more accessible and cost-effective. This study was designed to investigate the initial efficacy of empagliflozin in improving indirect liver markers in diabetic patients with MASLD and to compare its effects with MRI findings over a 6-month follow-up.

Methods

Study design

This study is a single-blinded, randomized clinical trial designed to evaluate the effects of empagliflozin on liver markers and hepatic fat in patients with T2DM and MASLD. Patients were randomly assigned to two groups receiving different doses of empagliflozin (10 mg or 25 mg daily). Baseline and post-intervention (6 months) data were collected and analyzed to determine the effects of treatment.

Study population

The study population included diabetic patients with MASLD (previously termed NAFLD) who attended the diabetes clinic of Loghman Hakim Hospital. Patient recruitment began on October 10, 2024, and ended on November 15, 2024. The final 6-month follow-up assessments were completed by May 15, 2025. Eligible participants met specific inclusion criteria, including a diagnosis of T2DM according to ADA guidelines, NAFLD confirmed by imaging (ultrasound or MRI), age between 18 and 65 years, and written informed consent. Patients were excluded if they had poorly controlled diabetes (HbA1c ≥ 10%), excessive alcohol consumption (> 30 g/day in the last 10 years or > 10 g/day in the past year), other liver diseases (e.g., hepatitis B or C, autoimmune hepatitis), biliary obstruction, or drug-induced liver disease. Additionally, patients with recent use of drugs affecting hepatic fat, prior gastrointestinal bypass surgery, or an inability to adhere to follow-up protocols were excluded.

We planned the trial to detect a moderate standardized mean difference (Cohen’s d = 0.50) in key liver markers (e.g., ALT, FIB-4) between groups with a two-sided α = 0.05 and 80% power. Under equal allocation, this required 63 participants per group [20]. Allowing for ~ 10% attrition, we targeted ≥ 70 per group. In total, 143 participants were enrolled (72 intervention, 71 control), exceeding the target. After differential attrition during follow-up, the analyzed sample sizes were 69 (intervention) and 50 (control), which provide ~ 77% power to detect d = 0.50 and ≥ 80% power to detect d ≥ 0.52. Effect sizes with 95% CIs are reported alongside p-values for the primary outcomes. Additional details of the sample size calculation and sensitivity analyses are provided in the Appendix.

Measurements

Baseline assessments included demographic data (age, sex, BMI) and clinical and biochemical parameters, such as fasting blood glucose (FBS), liver enzymes (ALT, AST, and GGT), and platelet count. All MRI examinations were performed at the Radiology Department of Loghman Hakim Hospital using a 1.5-Tesla Siemens scanner (Germany). Hepatic steatosis was evaluated with standardized liver MRI protocols, and results were graded semiquantitative on a 0–3 scale, as reported in our results. Ultrasound examinations were conducted with a GE LOGIQ E9 system (GE Healthcare, USA) by an experienced radiologist blinded to group allocation. Hepatic steatosis was graded based on echogenicity criteria relative to the renal cortex, with Grades 1–3 representing mild, moderate, and severe steatosis, respectively. Both imaging modalities were performed according to institutional protocols and in line with accepted radiological criteria for NAFLD.

Randomization and blinding

Randomization was performed using a computer-generated random number sequence (Microsoft Excel RAND function) created by an independent statistician who was not involved in patient recruitment or assessment. Allocation concealment was ensured through the use of sequentially numbered, opaque, sealed envelopes (SNOSE), which were prepared in advance and opened in order by a study coordinator independent of enrollment and data collection. Participant enrollment was conducted by the diabetes clinic research team, while group assignment was managed separately to avoid selection bias.

The trial was conducted in a single-blinded fashion: participants were unaware of their treatment allocation, while physicians and outcome assessors had knowledge of the assigned intervention. To reduce the risk of bias, empagliflozin tablets (10 mg or 25 mg) and control medications were dispensed in identical packaging and labeled with codes only accessible to the independent statistician until completion of the analysis. Notably, radiologists responsible for sonography and MRI assessments were blinded to treatment allocation to minimize the risk of observer bias in outcome evaluation.

Intervention

Participants in the intervention arm received empagliflozin, administered orally once daily for a period of six months. Two dosages were used: 10 mg or 25 mg, according to random allocation. Patients were instructed to take the medication consistently at the same time each day, with or without food. They were monitored regularly for adherence and potential side effects, and no additional lifestyle modifications were introduced during the study period. Participants in the control arm continued to receive standard antidiabetic therapy as prescribed by their treating physicians (including agents such as metformin, sulfonylureas, or insulin, when indicated), but did not receive empagliflozin. Similar to the intervention group, they were advised to maintain their usual diet and physical activity, and were monitored throughout the six-month period for treatment adherence and safety.

Outcomes and follow-up

The primary outcome was the change in liver fat content as assessed by MRI at 6 months. Secondary outcomes included changes in liver enzymes (ALT, AST, and GGT), FBS, platelet count, and calculated indices (FIB-4 and NFS). Follow-up evaluations were conducted at baseline and after 6 months of treatment. During this period, patients were monitored for potential side effects and adherence to treatment.

Definition of terms

NAFLD was defined as hepatic fat infiltration confirmed by imaging (MRI-PDFF ≥ 5% or ultrasound evidence of steatosis) in the absence of significant alcohol consumption or other causes of liver disease. T2DM was diagnosed based on ADA criteria (HbA1c ≥ 6.5% or fasting glucose ≥ 126 mg/dL) [21]. Liver fibrosis scores were calculated using validated and widely accepted formulas for FIB-4 and NAFLD fibrosis score (NFS), which incorporate clinical and biochemical markers to estimate the likelihood of advanced fibrosis [22]. The FIB-4 score was derived using patient age, levels of aspartate aminotransferase (AST) and alanine aminotransferase (ALT), and platelet count, providing a reliable non-invasive estimate of liver fibrosis. Similarly, the NFS was calculated based on a combination of variables, including age, BMI, fasting blood glucose, AST, ALT and platelet count [23]. These scores were computed at baseline and after six months to evaluate changes in liver fibrosis as a result of the intervention, offering an accessible and cost-effective alternative to imaging or biopsy for assessing fibrosis severity.

Statistical analysis

All statistical analyses were performed using IBM SPSS Statistics version 26 (IBM Corp., Armonk, NY, USA). Data distribution was assessed with the Kolmogorov–Smirnov test. Continuous variables were summarized as mean ± standard deviation or median (interquartile range), while categorical variables were expressed as counts and percentages. Baseline comparisons between the intervention and control groups were conducted using independent samples t-tests or Mann–Whitney U tests for continuous variables, and χ² tests for categorical variables. Within-group changes from baseline to 6 months were evaluated using paired t-tests or Wilcoxon signed-rank tests. For categorical imaging endpoints (ultrasound and MRI grades), between-group differences were assessed using χ² tests, and within-group changes over time were analyzed using the Wilcoxon signed-rank test.

To address baseline imbalances, particularly in age and liver enzyme levels, analysis of covariance (ANCOVA) was performed for each outcome at 6 months, with baseline values and age included as covariates. In addition, linear mixed-effects models (LMMs) were applied to account for repeated measures across time (baseline and 6 months), incorporating fixed effects for group, time, age, and group × time interaction.

A subgroup analysis was conducted to compare outcomes between patients receiving 10 mg versus 25 mg of empagliflozin, adjusted for baseline values and age. To evaluate the impact of attrition on study validity, a post-hoc power analysis was conducted based on realized sample sizes, and detailed calculations are provided in the Supplementary Materials. Effect sizes (Hedges’ g) with 95% confidence intervals were also reported alongside p-values to aid interpretation. All tests were two-sided, and a p-value of < 0.05 was considered statistically significant.

Ethical considerations

This study was conducted and reported in accordance with the Consolidated Standards of Reporting Trials (CONSORT) guidelines as detailed in Supplementary Table 1 [24]. Participants received detailed explanations of the study objectives and procedures before providing written informed consent. No additional costs beyond routine care were imposed on participants, and confidentiality of all patient data was maintained. The study protocol was approved by the Ethics Committee of Shahid Beheshti University of Medical Sciences (Internal Review Board - IRB) under approval code IR.SBMU.MSP.REC.1403.415. This research was conducted in accordance with the ethical principles outlined in the Declaration of Helsinki (2013 revision).

Results

A total of 169 patients with type 2 diabetes and suspected MASLD were screened for eligibility. Twenty-six were excluded: 8 for viral hepatitis, 4 for excessive alcohol use (> 20 g/day), and 14 who declined participation. As a result, 143 eligible participants were enrolled and randomized into two study arms. Of these, 72 participants were assigned to the intervention group receiving empagliflozin, and 71 were assigned to the control group receiving standard care. A total of 24 participants were lost to follow-up during the 6-month period, with 21 in the control group (29.5%) and 3 in the intervention group (4.2%). The main reasons were withdrawal of consent (n = 7), relocation to another city (n = 6), non-adherence (n = 5), and incomplete data collection (n = 3), as detailed in Fig. 1. Despite this attrition, the final sample size (n = 119) remained above the minimum required by the original power calculation.

Fig. 1.

CONSORT flow diagram showing participant enrollment, allocation, follow-up, and analysis

The mean age of the participants was 48.94 years, with a standard deviation of 9.55 years. Among the participants, 70 individuals (58.8%) were male, and 49 individuals (41.2%) were female. Table 1 presents the demographic and baseline characteristics of the study population. The experimental group had a significantly lower mean age compared to the control group (46.32 ± 8.11 years vs. 52.56 ± 10.26 years, p < 0.001). While no significant differences were observed in weight (p = 0.394), height (p = 0.821), or BMI (p = 0.298) between the groups, liver enzyme levels, including GGT and ALT, were notably higher in the experimental group (p < 0.001 for both). AST levels were also higher in the experimental group, though the difference did not reach statistical significance (p = 0.079). Regarding bilirubin levels, both total and direct bilirubin were significantly lower in the experimental group (p = 0.011 and p = 0.014, respectively). Liver imaging results, assessed via sonography and MRI, indicated similar baseline grades of hepatic steatosis and fibrosis, with no significant differences between the groups (p-values for sonography and MRI grading were 0.099 and 0.158, respectively).

Table 1.

Demographic, lab data characteristics of the study population at the baseline

Variable	Experimental group n = 69	Control group n = 50	p-value
Age (years)	46.32 ± 8.11	52.56 ± 10.26	< 0.001
Weight (kg)	90.83 ± 15.16	88.01 ± 19.36	0.394
Height (cm)	167.54 ± 10.40	167.79 ± 6.32	0.821
BMI (kg/m²)	32.18 ± 4.24	31.13 ± 6.05	0.298
GGT (IU/L)	59.55 ± 26.09	43.64 ± 10.33	< 0.001
AST (IU/L)	46.74 ± 14.14	42.54 ± 11.61	0.079
ALT (IU/L)	71.33 ± 49.35	47.52 ± 7.46	< 0.001
ALP (IU/L)	186.91 ± 66.43	197.98 ± 40.30	0.262
Total Bilirubin (mg/dL)	0.81 ± 0.38	0.94 ± 0.13	0.011
Direct Bilirubin (mg/dL)	0.22 ± 0.20	0.32 ± 0.21	0.014
Sonography Grade			0.099
• Grade 1	0 (0.0%)	0 (0.0%)
• Grade 2	14 (53.8%)	12 (46.2%)
• Grade 3	49 (56.3%)	38 (43.7%)
• Grade 4	6 (100.0%)	0 (0.0%)
MRI Grade			0.158
• Grade 0	2 (100.0%)	0 (0.0%)
• Grade 1	2 (100.0%)	0 (0.0%)
• Grade 2	44 (52.4%)	40 (47.6%)
• Grade 3	21 (67.7%)	10 (32.3%)

*Values are presented as mean ± standard deviation (SD) for continuous variables and as number (percentage) for categorical variables. Between-group comparisons of continuous variables were performed using independent samples t-tests, while categorical variables (sonography and MRI grades) were compared using chi-square tests. p-values < 0.05 were considered statistically significant. Abbreviations: BMI, body mass index; GGT, γ-glutamyl transferase; AST, aspartate aminotransferase; ALT, alanine aminotransferase; ALP, alkaline phosphatase; MRI, magnetic resonance imaging

Table 2 highlights the characteristics of the study population at the 6-month follow-up. Both groups showed comparable weights, heights, and BMIs at follow-up, with no significant differences observed (p = 0.665, 0.876, and 0.611, respectively). However, the experimental group demonstrated significantly lower liver enzyme levels compared to the control group, including GGT (34.01 ± 14.81 IU/L vs. 41.48 ± 10.82 IU/L, p = 0.002), AST (28.35 ± 8.24 IU/L vs. 39.20 ± 10.91 IU/L, p < 0.001), and ALT (30.16 ± 11.03 IU/L vs. 42.90 ± 7.91 IU/L, p < 0.001). Total bilirubin (p = 0.023) and direct bilirubin (p = 0.008) levels were also significantly lower in the experimental group. At 6 months, the NAFLD fibrosis score (NFS) was significantly lower in the experimental group compared with the control group (–1.20 ± 1.15 vs. − 0.55 ± 0.86, p = 0.004). In addition, the FIB-4 index showed a consistent and significant reduction in the experimental group compared to the control group (1.06 ± 0.38 vs. 1.42 ± 0.43, p < 0.001). Imaging findings revealed substantial differences; the experimental group showed a higher prevalence of lower sonography grades (Grade 1: 94.2%) and MRI grades (Grade 0: 100%), while higher grades of sonography and MRI were predominantly observed in the control group (p < 0.001 for both).

Table 2.

Follow-up characteristics at 6 months for the study population

Variable	Experimental group n = 69	Control group n = 50	p-value
Weight (kg)	83.54 ± 14.45	86.59 ± 18.54	0.665
Height (cm)	167.57 ± 10.43	167.67 ± 6.39	0.876
BMI (kg/m²)	29.69 ± 4.08	30.75 ± 5.97	0.611
GGT (IU/L)	34.01 ± 14.81	41.48 ± 10.82	0.002
AST (IU/L)	28.35 ± 8.24	39.20 ± 10.91	< 0.001
ALT (IU/L)	30.16 ± 11.03	42.90 ± 7.91	< 0.001
ALP (IU/L)	183.13 ± 63.77	196.80 ± 39.43	0.152
Total Bilirubin (mg/dL)	0.83 ± 0.34	0.93 ± 0.10	0.023
Direct Bilirubin (mg/dL)	0.20 ± 0.18	0.28 ± 0.14	0.008
NFS	-1.20 ± 1.15	-0.55 ± 0.86	0.004
FIB-4	1.06 ± 0.38	1.42 ± 0.43	< 0.001
Sonography Grade			< 0.001
• Grade 1	49 (94.2%)	3 (5.8%)
• Grade 2	18 (34.0%)	35 (66.0%)
• Grade 3	2 (14.3%)	12 (85.7%)
• Grade 4	0 (0.0%)	0 (0.0%)
MRI Grade			< 0.001
• Grade 0	22 (100.0%)	0 (0.0%)
• Grade 1	42 (75.0%)	14 (25.0%)
• Grade 2	5 (13.2%)	33 (86.8%)
• Grade 3	0 (0.0%)	3 (100.0%)

Significant reductions in liver enzyme levels (GGT, AST, and ALT) were observed over 6 months, while ALP and total bilirubin remained unchanged. Direct bilirubin showed a modest but significant decrease. Among fibrosis indices, FIB-4 decreased significantly, whereas NFS did not. Correlation analyses confirmed consistent patterns of change across markers (Supplementary Table 2). During the 6-month follow-up, no serious adverse events attributable to empagliflozin were observed. Mild adverse effects, primarily transient urinary tract symptoms, occurred in four participants within the intervention group; however, these events were self-limiting and did not necessitate treatment discontinuation. No adverse events were reported in the control group.

Table 3 summarizes the paired analysis of baseline and 6-month follow-up data for the experimental and control groups. In the experimental group, significant reductions were observed in key liver enzyme levels. GGT decreased from 59.55 ± 26.09 IU/L to 34.01 ± 14.81 IU/L, with a mean difference of 25.54 ± 16.89 IU/L (p < 0.001), while AST dropped from 46.74 ± 14.14 IU/L to 28.35 ± 8.24 IU/L, with a mean difference of 18.39 ± 12.26 IU/L (p < 0.001). ALT also showed a substantial reduction, decreasing from 71.33 ± 49.35 IU/L to 30.16 ± 11.03 IU/L, with a mean difference of 41.17 ± 45.71 IU/L (p < 0.001). The FIB-4 index showed a significant reduction from 1.20 ± 0.53 to 1.06 ± 0.38, with a mean difference of 0.14 ± 0.30 (p < 0.001). However, changes in ALP and total bilirubin levels in this group were not statistically significant, with p-values of 0.344 and 0.491, respectively.

Table 3.

Paired samples analysis by study groups

Group: Experimental (n = 69)
Pair	Baseline	6-Month	Correlation	Mean difference (SD)	95% CI	t	p-value
GGT (IU/L)	59.55 (26.09)	34.01 (14.81)	0.80	25.54 (16.89)	21.48, 29.59	12.56	< 0.001
AST (IU/L)	46.74 (14.14)	28.35 (8.24)	0.51	18.39 (12.26)	15.45, 21.34	12.46	< 0.001
ALT (IU/L)	71.33 (49.35)	30.16 (11.03)	0.43	41.17 (45.71)	30.19, 52.16	7.48	< 0.001
ALP (IU/L)	186.91 (66.43)	183.13 (63.77)	0.87	3.79 (33.01)	-4.14, 11.72	0.95	0.344
Total Bilirubin (mg/dL)	0.81 (0.38)	0.83 (0.34)	0.84	-0.02 (0.21)	-0.07, 0.03	-0.69	0.491
Direct Bilirubin (mg/dL)	0.22 (0.20)	0.20 (0.18)	0.90	0.02 (0.085)	0.00, 0.04	2.30	0.025
NFS	–1.33 (1.40)	–1.20 (1.15)	0.84	–0.13 (0.76)	–0.31, 0.06	–1.38	0.172
FIB-4	1.20 (0.53)	1.06 (0.38)	0.83	0.14 (0.30)	0.07, 0.22	3.89	< 0.001

Group: Control (n = 50)
Pair	Baseline mean (SD)	6-month mean (SD)	Correlation	Mean difference (SD)	95% CI (lower, upper)	t	p-value
GGT (IU/L)	43.64 (10.33)	41.48 (10.82)	0.98	2.16 (2.24)	1.52, 2.80	6.81	< 0.001
AST (IU/L)	42.54 (11.61)	39.20 (10.91)	0.88	3.34 (5.57)	1.76, 4.92	4.24	< 0.001
ALT (IU/L)	47.52 (7.46)	42.90 (7.91)	0.85	4.62 (4.27)	3.41, 5.83	7.65	< 0.001
ALP (IU/L)	197.98 (40.30)	196.80 (39.43)	0.99	1.18 (5.16)	-0.29, 2.65	1.62	0.112
Total Bilirubin (mg/dL)	0.94 (0.13)	0.93 (0.10)	0.75	0.01 (0.09)	-0.01, 0.04	0.97	0.335
Direct Bilirubin (mg/dL)	0.32 (0.21)	0.28 (0.14)	0.66	0.04 (0.15)	-0.01, 0.08	1.75	0.087
NFS	–0.54 (0.93)	–0.55 (0.86)	0.89	0.01 (0.42)	–0.11, 0.13	0.14	0.893
FIB-4	1.47 (0.43)	1.42 (0.43)	0.81	0.05 (0.27)	-0.03, 0.12	1.21	0.233

In the control group, significant reductions were also noted in GGT, AST, and ALT levels, although the magnitude of changes was smaller compared to the experimental group. GGT decreased from 43.64 ± 10.33 IU/L to 41.48 ± 10.82 IU/L, with a mean difference of 2.16 ± 2.24 IU/L (p < 0.001). AST dropped from 42.54 ± 11.61 IU/L to 39.20 ± 10.91 IU/L, with a mean difference of 3.34 ± 5.57 IU/L (p < 0.001), and ALT decreased from 47.52 ± 7.46 IU/L to 42.90 ± 7.91 IU/L, with a mean difference of 4.62 ± 4.27 IU/L (p < 0.001). In contrast to the experimental group, no significant changes were observed in the FIB-4 index (p = 0.233) or direct bilirubin levels (p = 0.087), indicating less pronounced improvements in fibrosis and liver injury markers in the control group.

Table 4 presents the results of the Wilcoxon Signed-Rank Test for changes in sonography and MRI scores across the intervention and control groups. In the experimental group, MRI scores showed significant reductions after 6 months compared to baseline, with 65 negative ranks (mean rank = 33.00, Z = -7.193, p < 0.001) and no positive ranks, indicating consistent improvement. Similarly, in the control group, significant reductions were observed for both sonography (29 negative ranks, Z = -5.385, p < 0.001) and MRI scores (18 negative ranks, Z = -4.001, p < 0.001), though fewer changes were noted compared to the experimental group.

Table 4.

Wilcoxon signed-rank test results for sonography and MRI scores by intervention group

Group	Comparison	Negative ranks (N, Mean Rank, Sum)	Positive ranks (N, Mean Rank, Sum)	Ties (N)	Z-Value	p-value
Experimental (n = 69)	MRI (6 months - Baseline)	65 (33.00, 2145.00)	0 (0.00, 0.00)	4	-7.193	< 0.001
Control (n = 50)	Sonography (6 months - Baseline)	29 (15.00, 435.00)	0 (0.00, 0.00)	21	-5.385	< 0.001
Control (n = 50)	MRI (6 months - Baseline)	18 (9.50, 171.00)	0 (0.00, 0.00)	32	-4.001	< 0.001

* Data are presented as the number of negative and positive ranks with their corresponding mean ranks and rank sums, along with the number of ties. Z-values and p-values are derived from Wilcoxon signed-rank tests comparing paired sonography and MRI scores at baseline and 6 months within each study group. p-values < 0.05 were considered statistically significant

Results from the ANCOVA models (adjusted for baseline values and age) indicated that participants in the experimental group had significantly lower liver enzyme levels at 6 months compared to controls (Table 5). Adjusted mean ALT, AST, and GGT were 16.4 IU/L, 13.7 IU/L, and 15.0 IU/L lower, respectively (all p < 0.001). Fasting blood sugar was also significantly reduced in the intervention group by 7.5 mg/dL (p < 0.001). Among calculated indices, FIB-4 was significantly lower in the intervention group (adjusted mean difference − 0.17, p = 0.001), whereas NFS showed no significant difference between groups. Platelet counts were comparable, with no significant between-group difference. These adjusted effect sizes and their 95% confidence intervals are also illustrated in Supplementary Fig. 1 (Appendix).

Table 5.

Adjusted between-group comparisons of primary and secondary outcomes at 6 months (ANCOVA adjusted for baseline values and age)

Outcome	Experimental adj. mean (95% CI) n = 69	Control adj. mean (95% CI) n = 50	Adjusted mean difference (95% CI)	p-value
ALT (IU/L)	28.6 (26.4–30.8)	45.0 (42.4–47.6)	–16.4 (–20.0 to − 12.9)	< 0.001
AST (IU/L)	27.2 (25.4–28.9)	40.8 (38.8–42.9)	–13.7 (–16.5 to − 10.9)	< 0.001
GGT (IU/L)	30.8 (28.9–32.8)	45.9 (43.5–48.2)	–15.0 (–18.3 to − 11.8)	< 0.001
FIB-4	1.1 (1.1–1.2)	1.3 (1.2–1.4)	–0.2 (–0.3 to − 0.1)	0.001
NFS	–1.0 (–1.1, − 0.8)	–0.9 (–1.1, − 0.7)	–0.1 (–0.3, 0.2)	0.597
Platelets (103/µL)	230 (218–241)	235 (221–248)	–5 (–23 to 13)	0.586
FBS (mg/dL)	102.7 (100.9–104.4)	110.2 (108.1–112.2)	–7.5 (–10.3 to − 4.7)	< 0.001

* Values are estimated marginal means (95% confidence intervals) derived from ANCOVA models, adjusted for baseline values of each outcome and age

Moreover, mixed-model analyses confirmed significant treatment effects of empagliflozin on key liver and metabolic parameters (Appendix. Supplementary Table 3). Compared with controls, the intervention group demonstrated significantly greater reductions in liver enzymes, with adjusted mean decreases of − 14.5 IU/L for ALT, − 12.3 IU/L for AST, and − 7.5 IU/L for GGT (all p < 0.001). Fasting blood sugar also declined more markedly in the intervention arm (–3.1 mg/dL, p < 0.001). Among fibrosis-related indices, FIB-4 and NFS showed numerical improvement in the intervention group but did not reach statistical significance (p = 0.098 and p = 0.426, respectively). Platelet counts remained stable over time with no between-group difference (p = 0.154). These findings indicate that empagliflozin exerted consistent short-term benefits on hepatic enzymes and glycemic control, while effects on indirect fibrosis markers were less pronounced.

As a sensitivity analysis, we conducted a subgroup comparison between patients receiving 10 mg versus 25 mg empagliflozin to explore potential dose-related effects (Supplementary Table 4). The analysis showed no significant differences between doses for ALT, AST, GGT, or FIB-4 index (all p > 0.10). However, patients in the 10 mg group demonstrated a significantly greater reduction in the NAFLD fibrosis score (mean difference − 0.42, 95% CI − 0.77 to − 0.08; p = 0.017) and higher platelet counts at 6 months (mean difference 42,385/µL, 95% CI 11,318 to 73,451; p = 0.008) compared with those receiving 25 mg. In addition, fasting blood sugar remained modestly higher in the 10 mg group (adjusted mean difference 5.77 mg/dL, 95% CI 1.29 to 10.24; p = 0.012).

Discussion

The results of this study demonstrated that the experimental group receiving empagliflozin experienced significantly greater improvements in liver health compared to the control group. Markers of liver function, including liver enzymes such as GGT, AST, and ALT, showed more pronounced reductions in the experimental group, indicating better resolution of liver injury. Additionally, fibrosis markers, such as the FIB-4 index, improved more substantially in the experimental group, reflecting a potential reduction in liver fibrosis. Imaging results further supported these findings, with MRI and sonography scores indicating more significant improvements in hepatic steatosis in the experimental group. In contrast, the control group showed less marked changes in these parameters, emphasizing the efficacy of empagliflozin in improving liver-related outcomes.

Because FIB-4 and NFS incorporate age, platelets, and metabolic parameters, they are susceptible to non-fibrotic influences. Nevertheless, longitudinal change in FIB-4 has been shown to predict clinically relevant outcomes in NAFLD, including incident cirrhosis and hepatocellular carcinoma, even when baseline FIB-4 is in the lower risk range; conversely, reductions in FIB-4 are associated with lower subsequent risk [25, 26]. These data support the utility of serial FIB-4 as a pragmatic surrogate for disease trajectory rather than a stand-alone diagnostic of histologic fibrosis. In line with AASLD guidance, we therefore treated FIB-4/NFS as supportive monitoring tools and based causal inferences primarily on objective endpoints (MRI-derived liver fat and adjusted changes in ALT/AST/GGT) [27]. We also note that our cohort’s relatively low mean FIB-4 values may attenuate detectable between-group differences (lower dynamic range), which we acknowledge as a limitation, while emphasizing that the overall pattern of change remained concordant with improvements observed in imaging and enzymes.

The observed discrepancy between FIB-4 and NFS is consistent with prior reports. FIB-4 is largely driven by liver enzymes and platelet count, which are more sensitive to short-term improvements, whereas NFS incorporates relatively stable factors such as BMI, albumin, and glucose status that may not change significantly within a 6-month timeframe. Previous studies have shown that FIB-4 generally demonstrates higher sensitivity for detecting short-term treatment effects, while NFS often underperforms, particularly in cohorts with milder fibrosis or limited follow-up duration [28–30]. These differences likely explain why FIB-4 improved significantly in our trial while NFS did not.

In order to address potential concerns regarding type I error due to multiple outcomes, we re-analyzed the data using ANCOVA adjusted for baseline values and age. This approach provides a more rigorous assessment by accounting for baseline imbalances and covariates. The results confirmed the robustness of our findings: liver enzyme reductions (ALT, AST, and GGT) and fasting blood sugar remained highly significant (all p < 0.001), while nonsignificant outcomes such as NFS and platelet counts remained unchanged. These results strengthen confidence in the validity of the observed treatment effects.

Empagliflozin, like other SGLT2 inhibitors, addresses key pathophysiological drivers of MASLD/NASH. By improving insulin sensitivity and promoting weight loss (via glucosuria and calorie loss), it reduces hepatic fat accumulation and downstream lipotoxicity [31, 32]. This leads to indirect antifibrotic and anti-inflammatory effects. Notably, SGLT2 inhibition is reported to dampen chronic inflammation: trials have shown significant reductions in pro-inflammatory cytokines (e.g. IL-6, TNF-α), CRP, and MCP-1 with this drug class [33]. Empagliflozin may also enhance adiponectin levels – a beneficial adipokine with anti-inflammatory and antifibrotic properties [34]. Increased adiponectin and reduced cytokines can attenuate hepatic stellate cell activation, potentially slowing fibrosis progression. Furthermore, preclinical studies suggest empagliflozin activates hepatic autophagy and shifts macrophages toward an anti-inflammatory phenotype [35]. Through these mechanisms – improved metabolic profile, reduced steatosis, and blunted inflammation – empagliflozin might also reduce hepatocyte apoptosis. While our trial did not measure apoptosis markers directly, others have observed lower serum cytokeratin-18 fragments and liver enzymes with SGLT2 therapy, consistent with reduced hepatocyte injury [32]. Overall, the drug’s multifactorial actions (metabolic, anti-inflammatory, and antifibrotic) provide a plausible mechanistic basis for the improvements in steatosis and fibrosis observed in patients.

Empagliflozin has been examined in multiple clinical trials for NAFLD, and our findings align well with the emerging consensus. Supplementary Table 4 summarizes key similar studies in the last decade comparing empagliflozin to non-SGLT2 therapy (placebo or standard care). These studies uniformly demonstrate that empagliflozin therapy leads to improvements in hepatic steatosis. Our trial’s observation of improved liver fat content and liver enzymes is in line with Kuchay et al.’s E-LIFT trial, which first showed a significant drop in MRI-measured liver fat (− 4.0% absolute) and ALT with 20 weeks of empagliflozin [16]. Similarly, Kahl et al. reported a ~ 22% relative reduction in liver fat after 24 weeks of empagliflozin in well-controlled diabetics [36]. Notably, Kahl’s study also found a rise in adiponectin and a drop in uric acid with empagliflozin, supporting the concept of reduced inflammation and metabolic stress. Our results also concur with Taheri et al. (empagliflozin in non-diabetic NAFLD) and Chehrehgosha et al. (in T2DM NAFLD), both of which documented significant declines in hepatic fat and fibrosis indices (FibroScan scores) in empagliflozin-treated patients compared to placebo [35]. This cross-validation in diabetic and non-diabetic NAFLD is encouraging, suggesting that empagliflozin’s liver benefits extend beyond glycemic control alone.

In our study, empagliflozin improved liver enzyme abnormalities, which echoes findings from other trials. For example, in Kuchay et al., ALT dropped significantly more with empagliflozin, though AST and GGT did not reach significance [16]. Komiya et al. (2016) likewise noted improved AST and GGT in diabetic NAFLD patients on SGLT2 therapy, even independent of weight loss [37]. Such enzyme improvements are often taken as surrogates for reduced hepatic inflammation and injury. Importantly, some studies have evaluated noninvasive fibrosis markers. Our findings of fibrosis score improvement (e.g. reduction in FibroScan stiffness or fibrosis index) align with meta-analytic evidence: a recent systematic review of 18 RCTs found that SGLT2 inhibitors yielded a modest but significant reduction in liver stiffness (− 0.67 kPa) and in the fibrosis-4 index (MD ≈ − 0.12) compared to controls [38]. Although changes in fibrosis biomarkers in short-term studies are small, the consistent direction of effect (fibrosis metrics tending to improve with empagliflozin) suggests a potential antifibrotic benefit over longer durations [35]. This is biologically plausible given empagliflozin’s suppression of pro-fibrogenic inflammatory pathways (e.g. lower IL-6/TNF-α) and improved adiponectin, which may inhibit stellate cell activation [39].

Our results are also supported by several recent meta-analyses of SGLT2 inhibitors in NAFLD. Wong et al. (2021) analyzed 10 studies (7 RCTs and 3 cohorts) in Asian T2DM patients and found significant improvements in hepatic fat content and liver enzymes (AST, ALT) with SGLT2 inhibitors vs. controls [40]. A larger network meta-analysis (2023) likewise confirmed that SGLT2 inhibitors significantly reduce ALT and body weight in NAFLD, although GLP-1 agonists like semaglutide showed even greater efficacy on some liver metrics [41]. Focusing on empagliflozin, a dedicated meta-analysis by Zang et al. (2022) pooled 3 RCTs (n = 212) and initially reported no statistically significant benefit of empagliflozin on CAP, liver stiffness, ALT, or lipids compared to placebo [42]. The authors concluded that empagliflozin’s effect might be “not effective” in NAFLD. However, it’s now clear this was an underpowered analysis – subsequent evidence has tipped the balance. For instance, our findings and other trials post-2021 show clear trends of improvement in steatosis and transaminases with empagliflozin. A 2024 systematic review including 18 RCTs resolved much of this controversy: it demonstrated that SGLT2 inhibitors do confer a mild but significant improvement in hepatic steatosis and fibrosis markers on aggregate [38]. This updated meta reported significant reductions in CAP (− 10.6 dB/m), MRI-PDFF (− 2.6%), liver stiffness, and even the FIB-4 index with SGLT2 treatment. These improvements, though modest in magnitude, mirror the direction of change we observed. The discrepancies between Zhang et al.’s null results and later positive results can be attributed to differences in sample size and study duration [35]. Empagliflozin’s benefits on liver fibrosis likely require longer treatment and larger cohorts to reach significance. In fact, our 6-month study showed trends toward fibrosis improvement; longer-term studies are needed to confirm if these translate into significant fibrosis regression, as suggested by the meta-analyses.

Prior trials have similarly highlighted the hepatoprotective effects of empagliflozin. For instance, Kuchay et al. first demonstrated reductions in MRI-measured liver fat and ALT, while Taheri et al. confirmed improvements in steatosis among non-diabetic patients [16, 43]. Chehrehgosha et al. and Kahl et al. further corroborated these findings, reporting consistent reductions in liver fat and enzymes across diabetic cohorts [36, 44]. Collectively, these studies complement our results and suggest that empagliflozin exerts beneficial effects on both hepatic steatosis and biochemical markers of liver injury.

Sensitivity analyses further showed significant reductions in both weight and HbA1c among participants receiving empagliflozin compared to controls (weight: 83.5 vs. 86.6 kg, p < 0.001; HbA1c: 5.73% vs. 6.47%, p < 0.001; results not shown). These findings indicate that improvements in hepatic outcomes may, at least partly, be mediated by weight loss and enhanced glycemic control. Nevertheless, as the primary liver outcomes remained robust after adjustment for these covariates, additional direct hepatic mechanisms are also likely to be involved.

A key strength of our study is the comprehensive approach to evaluating empagliflozin’s effects, utilizing both advanced imaging modalities such as MRI and sonography and indirect biochemical markers like FIB-4 and NFS, which allowed for a robust assessment of liver steatosis and fibrosis. However, several limitations should be acknowledged. First, the single-center design may restrict the generalizability of the findings to broader and more diverse populations. Second, the 6-month follow-up provides evidence of short- to medium-term effects of empagliflozin but does not capture the long-term trajectory of liver fibrosis; future studies with extended follow-up are warranted. Third, the lack of direct inflammatory or oxidative stress markers, as evaluated in some other studies, restricts the understanding of the mechanistic pathways underlying the observed improvements. Fourth, the absence of a placebo group and reliance on two active dosages may have influenced the comparative outcomes. Fifth, the unequal dropout rates between groups represent a limitation of this study, as attrition was higher in the control group. However, the final sample size remained sufficient to maintain statistical power, and the baseline characteristics of completers and non-completers did not differ significantly, reducing the likelihood of attrition bias. Finally, while safety outcomes were systematically recorded, our study was not powered to detect rare adverse events; larger multicenter studies are required to confirm the safety profile.

Building on the findings of our trial, future studies should explore longer treatment durations and assess whether the improvements in liver enzymes, imaging scores, and fibrosis indices with empagliflozin are sustained over time. Combination approaches, particularly with agents such as GLP-1 receptor agonists, may further enhance these benefits. Evaluating such strategies using similar non-invasive endpoints—liver enzymes, fibrosis scores, and imaging modalities—will help clarify their role in optimizing MASLD management in diabetic populations.

Conclusion

In conclusion, our study demonstrated that empagliflozin significantly improves liver steatosis, reduces liver enzyme levels (ALT, AST, and GGT), and positively impacts fibrosis markers (FIB-4 and NFS) in patients with type 2 diabetes and non-alcoholic fatty liver disease. The findings highlight the potential of empagliflozin as an effective therapeutic option for managing MASLD in diabetic populations, offering benefits beyond glycemic control. The consistent improvements observed across both imaging modalities (MRI and sonography) and biochemical markers emphasize its dual utility in non-invasive liver health assessment and treatment. These results support the broader application of empagliflozin in clinical settings for addressing hepatic complications in diabetes, while further long-term and multicenter studies are needed to confirm its sustained effects and explore additional mechanistic pathways.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

AE and ZD contributed to data collection and preliminary analysis. HT was responsible for statistical analysis, data interpretation and revision. SNMN and HT jointly designed the study and prepared the manuscript. SNMN, HT, and SN finalized the draft. SN and SNMN provided clinical oversight. All authors reviewed and approved the final version of the manuscript. P.J revised the manuscript and drafted new version of the manuscript.

Funding

This research received no external funding.

Data availability

The datasets generated and analyzed during the current study are not publicly available due to participant confidentiality and institutional policy. However, de-identified individual participant data (IPD), along with the study protocol and statistical analysis plan, may be made available from the corresponding author upon reasonable request and following ethical approval by the institutional review board.

Declarations

Ethics approval and consent to participate

The study protocol was approved by the Ethics Committee of Shahid Beheshti University of Medical Sciences (Internal Review Board - IRB) under approval code IR.SBMU.MSP.REC.1403.415. All participants received detailed information about the study objectives and procedures and provided written informed consent prior to enrollment.

Consent for publication

All authors consent to the publication of this manuscript.

Competing interests

The authors declare no competing interests.