Empagliflozin Reduces Liver Fat in Individuals With and Without Diabetes

Abstract

OBJECTIVE

To examine the effect of empagliflozin on liver fat content in individuals with and without type 2 diabetes (T2D) and the relationship between the decrease in liver fat and other metabolic actions of empagliflozin.

RESEARCH DESIGN AND METHODS

Thirty individuals with T2D and 27 without were randomly assigned to receive in double-blind fashion empagliflozin or matching placebo (2:1 ratio) for 12 weeks. Participants underwent 75-g oral glucose tolerance testing and measurement of liver fat content with MRS before therapy and at study end. Hepatic glucose production before the start of therapy was measured with 3-³H-glucose.

RESULTS

Empagliflozin caused an absolute reduction of 2.39% ± 0.79% in liver fat content compared with an increase of 0.91% ± 0.64% in participants receiving placebo (P < 0.007 with ANOVA). The decrease in liver fat was comparable in both individuals with diabetes and those without (2.75% ± 0.81% and 1.93% ± 0.78%, respectively; P = NS). The decrease in hepatic fat content caused by empagliflozin was strongly correlated with baseline liver fat content (r = −0.62; P < 0.001), decrease in body weight (r = 0.53; P < 0.001), and improvement in insulin sensitivity (r = −0.51; P < 0.001) but was not related to the decrease in fasting plasma glucose or HbA_1c or the increase in hepatic glucose production.

CONCLUSIONS

Empagliflozin is effective in reducing liver fat content in individuals with and without T2D. The decrease in liver fat content is independent of the decrease in plasma glucose concentration and is strongly related to the decrease in body weight and improvement in insulin sensitivity.

Graphical Abstract

Introduction

The prevalence of metabolic dysfunction–associated steatotic liver disease (MASLD) has markedly increased in recent decades, and currently, MASLD is the most prevalent chronic liver disease (1). Although hepatic steatosis is largely reversible, 25% of patients with MASLD experience progression to metabolic eysfunction–associated steatohepatitis (MASH) and liver cirrhosis (1). Therefore, prevention of MASLD will lead to a reduction in the incidence of MASH and its subsequent complications. Obesity and type 2 diabetes (T2D) are major risk factors for MASLD and MASH (1–3). It is estimated that ∽80% of obese individuals and patients with T2D manifest MASLD (4–6). Furthermore, longitudinal studies have reported that obesity is an independent risk factor for MASLD, and it increases the risk of MASLD 3.5-fold in individuals without diabetes (4–6).

Currently, there is no approved therapy for MASLD/MASH. Caloric restriction with attendant weight loss has been shown to be beneficial in reducing hepatic steatosis (7). Several classes of glucose-lowering agents, including thiazolidinediones and glucagon-like peptide 1 receptor agonists, have also been shown to reduce liver fat content in patients with T2D (8,9).

Sodium–glucose cotransporter 2 (SGLT2) inhibitors are a novel class of glucose-lowering drugs (10). Clinical studies have demonstrated that SGLT2 inhibitors exert multiple systemic actions, including improvement in insulin sensitivity and insulin secretion, increased total-body fat oxidation (11), reduced hospitalization for heart failure, and slowed progression of diabetic kidney disease (12–15). A secondary analysis of data from cardiovascular outcome trials has demonstrated that members of this class reduce markers of MASLD in patients with T2D (16). Dapagliflozin, empagliflozin, and canagliflozin have also been shown to reduce liver fat content in patients with T2D with MASLD measured with MRS (17–24).

The aims of the current study were to 1) examine whether empagliflozin is effective in reducing liver fat content in individuals without diabetes and in those with T2D and 2) examine the relationship between the metabolic actions of SGLT2 inhibitors and the decrease in liver fat content in individuals with T2D and in those without diabetes.

Research Design and Methods

Participants

A total of 36 individuals with T2D and 36 without diabetes were recruited to the study. One individual without diabetes dropped out because of COVID-19 infection. Thirty and 27 participants with and without T2D, respectively, completed all study procedures, including the MRI measurement; five and eight participants, respectively, completed clinical and metabolic measurements but did not complete the MRI measurement because of claustrophobia or overweight status and therefore were not included in the present report.

Inclusion and exclusion criteria are provided in the Supplementary Material. Table 1 lists the characteristics of both groups.

Table 1.

Baseline patient characteristics

	Diabetes			No diabetes			P (T2D vs. no diabetes)
	Empagliflozin (n = 20)	Placebo (n = 10)	P	Empagliflozin (n = 18)	Placebo (n = 9)	P
Age, years	54 ± 2	57 ± 2	NS	44 ± 3	47 ± 3	NS	<0.05
Female sex, %	42	45	NS	44	55	NS	NS
BMI, kg/m2	32.4 ± 1.1	33.2 ± 1.2	NS	33.9 ± 2.0	35.1 ± 2.1	NS	NS
Diabetes duration, years	5.9 ± 1.5	6.0 ± 2.1	NS			NS	NS
FPG, mg/dL	149 ± 8	159 ± 14	NS	105 ± 2	104 ± 1	NS	<0.0001
2-h glucose, mg/dL	268 ± 15	276 ± 21	NS	150 ± 8	144 ± 10	NS	<0.0001
HbA1c, %	7.5 ± 0.4	7.8 ± 0.5	NS	5.5 ± 0.1	5.6 ± 0.1	NS	<0.0001
Baseline medications
Drug naive	45	55	NS
Metformin	40	33	NS
Metformin + SU	15	22	NS
eGFR, mL/min ⋅ 1.73 m2	100 ± 3	98 ± 4	NS	97 ± 4	98 ± 5	NS	NS
ALT, IU	36 ± 4	28 ± 3	NS	35 ± 4	31 ± 5	NS	NS
AST, IU	24 ± 2	19 ± 1	NS	22 ± 2	20 ± 2	NS	NS
ALT/AST ratio	0.70 ± 0.05	0.70 ± 0.06	NS	0.65 ± 0.05	0.68 ± 0.05	NS	NS
FIB-4	0.98 ± 0.12	0.80 ± 0.06	NS	0.70 ± 0.10	0.73 ± 0.07	NS	NS
APRI	0.33 ± 0.06	0.23 ± 0.02	NS	0.28 ± 0.04	0.25 ± 0.03	NS	NS
NFS	−0.42 ± 0.31	−0.32 ± 0.25	NS	−0.78 ± 0.31	−0.55 ± 0.23	NS	NS
Liver fat, %	13.9 ± 1.5	12.9 ± 1.6	NS	12.1 ± 1.6	11.1 ± 1.6	NS	NS
n with >5% liver fat	18	9	NS	17	6	NS	NS

eGFR, estimated glomerular filtration rate; SU, sulfonylurea.

The study was approved by the University of Texas Health Science Center at San Antonio Institutional Review Board, and informed written consent was obtained from all individuals before participation. The study protocol is available at ClinicalTrial.Gov (identifier NCT03193684).

Research Design

Eligible participants underwent 1) 75-g oral glucose tolerance testing (OGTT), 2) measurement of hepatic fat content with MRS, and 3) 7- to 8-h measurement of endogenous glucose production (EGP) with 3-³H-glucose infusion. Two hours after the start of 3-³H-glucose infusion (3 h in T2D), participants were randomly assigned (2:1 ratio) to receive in double-blind fashion (patients and investigators) 25 mg empagliflozin or matching placebo. Participants were randomly assigned to the empagliflozin or placebo group based on fasting plasma glucose (FPG) <140 or ≥140 mg/dL (for patients with T2D), estimated glomerular filtration rate <80 or ≥80 mL/min ⋅ 1.73 m², and BMI <32 or ≥32 kg/m². After hepatic glucose production (HGP) measurement, participants continued to receive the therapy to which they were assigned in double-blind fashion for 12 weeks, at which time the OGTT and measurement of hepatic fat content were repeated. Participants were instructed to take empagliflozin or placebo every morning before breakfast and continue their dietary habits unchanged during the study. Patients with T2D were asked to continue their background glucose-lowering medications unchanged. Pill counting was used to assess participant compliance.

OGTT

Before the start of therapy and at study end, participants underwent 75-g OGTT in the morning after a 10- to 12-h overnight fast. Participants were instructed to consume a high-carbohydrate diet for 3 days before the test. During OGTT, blood was drawn at −30, −15, 0, 30, 60, 90, and 120 min for measurement of plasma glucose, insulin, C-peptide, and free fatty acid (FFA) concentrations.

HGP Measurement

The basal rate of HGP was measured with 3-³H-glucose infusion as previously described (22,23). Briefly, a prime (40 μCi) continuous (0.4 μCi/min) infusion of 3-³H-glucose (Perkin Elmers, Waltham, MA) was started and continued until study end (8 h). Plasma glucose, insulin, FFAs, and ³H-glucose radioactivity were measured at baseline and every 30 min after drug ingestion. Three hours after the start of the study, participants ingested in double-blind fashion 25 mg empagliflozin or matching placebo.

At 6:00 a.m., participants voided, and the urine was discarded. Urine was collected from 6:00 to 9:00 a.m. (to 8:00 a.m. in individuals without diabetes) and from 9:00 a.m. to 2:00 p.m. (drug treatment period). Urinary volume and glucose concentration were measured to determine the urinary glucose excretion (UGE) rate. At 2:00 p.m., participants received a meal and returned home.

Hepatic Fat Measurement

Hepatic fat content was measured with MRS–proton density fat fraction using a 3-Tesla MRI Scanner (Siemen Healthcare Solutions, Malvern, PA) as previously described (25). All spectra were read blindly by two independent experienced observers (G.C. and M.B.) using iMRI software and an open-source Java-based graphical user interface that enables time domain analysis of MRS and spectroscopic imaging. If the variance was more than 10% between the two readers, the observers met to resolve the difference. Intrahepatic triglyceride content was calculated as fat fraction (area under the curve [AUC] fat peak/AUC fat peak + water peak). Measurements were corrected for T1 and T2 relaxation.

Liver Scores

AST, ALT, platelet count, and plasma albumin were measured at baseline and study end. The following fibrosis scores were calculated: AST/ALT ratio, fibrosis-4 (FIB-4) index (age, AST, ALT, and platelets), aspartate aminotransferase platelet ratio index (APRI), and nonalcoholic fatty liver disease fibrosis score (NFS; i.e., MASLD fibrosis score; age, BMI, impaired fasting glucose and diabetes, AST/ALT ratio, platelets, and albumin) (25).

Analytical Methods

Plasma glucose was measured using the glucose oxidase method (Analox Instruments, Stourbridge, U.K.; International Point of Care, Toronto, Ontario, Canada). Plasma insulin and C-peptide concentrations were measured by radioimmunoassay (IBL America, Minneapolis, MN). Plasma FFA concentration was measured spectrophotometrically (FUJIFILM Wako Chemicals). 3-³H-glucose radioactivity was determined on deproteinized barium/zinc plasma samples as previously described (23).

Data Analysis and Statistical Methods

Liver scores (AST/ALT ratio, FIB-4 index, APRI, and NFS) were calculated as previously described (25). OGTT-derived indices of insulin sensitivity and insulin secretion were calculated from the plasma glucose, insulin, and C-peptide concentrations during the OGTT as previously described (25). Insulin sensitivity was measured with the Matsuda index (26). The ratio between the incremental AUC of plasma C-peptide concentration to the incremental AUC of plasma insulin concentration during the OGTT (ΔC-pep/ΔI)_0–120 was used to provide a measure of insulin clearance. The incremental AUCs of plasma glucose, insulin, and C-peptide concentration during the OGTT were calculated using the trapezoid rule.

The basal rate of EGP was calculated as the rate of glucose appearance as previously described (23). The change in EGP during the last hour of (240–300 min) from baseline was considered the effect of the drug on EGP.

The primary comparison was the change in liver fat content from baseline to 12 weeks in participants receiving empagliflozin versus placebo in each treatment group. In a secondary analysis, we compared the effect size of empagliflozin on liver fat content in patients with T2D versus that in individuals without diabetes with two-way ANOVA.

To determine predictors of the change in liver fat content, a multivariate linear regression model was created for participants receiving empagliflozin, in which the change in liver fat content measured with MRI was the dependent variable and age, sex, ethnicity, BMI, HbA_1c, insulin sensitivity, FPG, EGP, and treatment were the independent factors. Possible interaction between significant predictors of change in liver fat content identified with the multivariate model was tested with mediation analysis. The Sobel test was used to test the significance of interaction.

Values are presented as mean ± SEM. A two-sided Student t test was used to compare mean differences between treatment arms. For comparison of the drug effect compared with placebo, two-way ANOVA was used with time and treatment as factors. To compare the effect size of empagliflozin between individuals with T2D and those without diabetes (decrease in liver fat and increase in HGP), the change from baseline in each parameter was computed, and two-way ANOVA was used with treatment (empagliflozin vs. placebo) and group (T2D vs. no diabetes) as factors.

A P value <0.05 was considered statistically significant.

Power Calculation

The primary aim of the study was to examine the metabolic changes that accompany the increase in HGP after empagliflozin administration in patients with T2D and in individuals without diabetes. The change in liver fat content caused by empagliflozin versus placebo was the main secondary outcome of the study. Therefore, the study was powered to detect a significant increase in HGP in participants treated with empagliflozin compared with placebo in each group. The sample size calculation based on our previous studies (23,27) determined that 32 completers in each group (individuals with T2D and those without diabetes) in a 2:1 randomization ratio (empagliflozin to placebo) provided >90% power to detect a significant increase in HGP after empagliflozin administration compared with placebo. To ensure 32 completers, we planned to recruit 36 individuals with T2D and 36 without.

Data and Resource Availability

Data sharing will be available upon proper request to the corresponding author.

Results

Table 1 lists the baseline characteristics of study participants. Patients with T2D were 10 years older than individuals without diabetes, but otherwise both groups were well matched. As anticipated, patients with T2D had higher fasting and 2-h plasma glucose concentrations and HbA_1c than individuals without diabetes. Twelve individuals without diabetes had normal glucose tolerance, and 15 had impaired glucose tolerance (2-h plasma glucose 140–199 mg/dL). Participants with and without diabetes receiving empagliflozin were well matched to their counterparts receiving placebo (Table 1).

Effect of Empagliflozin on Glucose Metabolism

Compared with placebo, empagliflozin caused a significant increase in UGE in both individuals with diabetes and those without (2.55 ± 0.41 and 1.59 ± 0.15 g/h, respectively; both P < 0.0001), with a tendency toward a greater increase in the group with diabetes (P = 0.10) (Table 2).

Table 2.

Placebo-subtracted change in metabolic parameters in individuals with and without diabetes receiving empagliflozin

	Patients with T2D	Individuals without diabetes	P
UGE, g/h	+2.55 ± 0.41	+1.59 ± 0.15	NS
Body weight, kg	−1.79 ± 0.25	−3.19 ± 0.48	NS
Percentage of body weight	−1.8	−3.4
FPG, mg/dL	−31 ± 6	−3 ± 2	<0.0001
HbA1c, %	−0.84 ± 0.2	−0.24 ± 0.10	<0.0001
EGP, mg/kg ⋅ min	+0.57 ± 0.12	+0.38 ± 0.10	NS
Liver fat content, %	−3.75 ± 0.79	−2.73 ± 0.73	NS
ALT, IU	−11 ± 3	−5 ± 4	NS
AST, IU	−5 ± 2	−2 ± 2	NS
Matsuda index	+2.4 ± 0.2	+2.3 ± 0.2	NS
Hematocrit, %	+1.9 ± 0.5	+1.3 ± 0.6	NS

Empagliflozin caused a significant decrease in HbA_1c in patients with T2D at 3 months (−0.84% ± 0.2%) compared with placebo (−0.24% ± 0.3%; P < 0.01), whereas HbA_1c did not change significantly in individuals without diabetes (−0.04% ± 0.05%; P = NS) (Table 2).

Fasting plasma glucose concentration at 12 weeks decreased by 24 ± 6 mg/dL in patients with T2D versus a small increase (+7 ± 10 mg/dL) in participants receiving placebo (P = 0.02). In individuals without diabetes, neither empagliflozin (−0.2 ± 2 mg/dL) nor placebo (−3 ± 2 mg/dL) significantly altered FPG concentration after 12 weeks. Insulin sensitivity, measured with the Matsuda index at baseline, was comparable in individuals with T2D and those without diabetes (2.39 ± 0.21 vs. 2.61 ± 0.22, respectively; P = NS) and significantly improved (P < 0.01) in both individuals with T2D and those without diabetes receiving empagliflozin by 48% and 35%, respectively, whereas it remained unaltered in participants receiving placebo (Table 2).

Consistent with previous results from our laboratory (22–24,27), empagliflozin ingestion caused a rapid increase in HGP in both individuals with diabetes and those without (Supplementary Fig. 1). The increase in HGP was slightly, but not significantly, greater in individuals with diabetes versus those without.

Participants with and without diabetes receiving empagliflozin experienced weight loss of 1.72 ± 0.25 and 2.97 ± 0.48 kg, respectively, compared with weight gain of 0.07 ± 0.05 and 0.22 ± 0.71 kg, respectively, in the placebo group (both P < 0.05) (Table 2).

Effect of Empagliflozin on Hepatic Fat Content

Twelve weeks of treatment with empagliflozin significantly reduced liver fat content compared with placebo. In the total study population (with and without diabetes), empagliflozin caused an absolute reduction in liver fat content of 2.39% ± 0.79%, compared with an increase of 0.91% ± 0.64% in participants receiving placebo (primary outcome), and the difference between the two was highly statistically significant (P < 0.007 with ANOVA). The effect of empagliflozin versus placebo remained significant when analyzed in each group (with and without diabetes) separately. The absolute decrease in liver fat content was −2.75% ± 0.81% (P = 0.03 vs. placebo) in patients with diabetes and −1.93% ± 0.78% (P = 0.02 vs. placebo) in individuals without diabetes (Fig. 1A). To compare the efficacy of empagliflozin with regard to liver fat reduction in individuals with versus those without diabetes, a post hoc analysis was performed to compare the placebo-subtracted change in liver fat in individuals with versus those without diabetes; a comparable reduction in liver fat was observed in both groups (P = NS).

Figure 1.

A: Absolute change in liver fat content (LFC; %) in patients with T2D and in individuals without diabetes receiving empagliflozin (EMPA) or placebo (PLC). B: Relationship between the decrease in hepatic fat content and baseline liver fat in patients with T2D and in individuals without diabetes. Open circles represent individuals without diabetes, and closed circles represent patients with T2D. C: Decrease in LFC in participants with baseline LFC >7% (in patients with T2D [n = 11] and those without diabetes [n = 11]) and in participants with baseline liver fat <7% (n = 9 and n = 7, respectively). Asterisks represent statistical significance as follows: *P < 0.05, **P < 0.01.

The change in liver fat content in individuals receiving empagliflozin was strongly associated with baseline liver fat content, regardless of glucose tolerance status (Fig. 1B). Participants with baseline liver fat content <7% experienced no significant change in liver fat content, whereas both individuals with diabetes and those without diabetes with liver fat content ≥7% experienced a robust decrease in liver fat content at 12 weeks of treatment. The absolute decrease in liver fat content in was −5.4% ± 0.91% and −4.17% ± 0.76%, respectively (both P < 0.01 vs. placebo) (Fig. 1C).

Baseline hepatic fat content was slightly but not significantly higher in those with T2D versus individuals without diabetes (13.6% ± 1.5% vs. 11.5% ± 1.6%; P = 0.10). Liver fat content tended to be higher in those with impaired (n = 15) versus normal glucose tolerance (n = 12; 12.54 ± 1.27 vs. 10.23 ± 1.19; P = 0.08). Baseline liver fat content was significantly and inversely correlated with the Matsuda index of insulin sensitivity (r = −0.39; P < 0.01), fasting plasma glucagon (r = 0.39; P < 0.01), adipocyte insulin resistance index (r = −0.28; P < 0.05), ALT (r = 0.39; P < 0.01), AST/ALT ratio (r = −0.25; P < 0.25), and APRI (r = 0.39; P < 0.01) but not with FIB-4 or NFS (Supplementary Table 1).

The decrease in liver fat content was positively correlated with the decrease in body weight (r = 0.53; P < 0.001) (Fig. 2A) and inversely correlated with the improvement in the Matsuda index of insulin sensitivity (r = −0.51; P < 0.001) (Fig. 2B). It was also significantly correlated with the increase in ALT (r = 40; P < 0.01) and APRI (r = 0.26; P < 0.05) and inversely correlated with the increase in plasma FFA concentration (r = −0.26; P < 0.05) and increase in HGP (r = −0.32; P < 0.01). The reduction in liver fat content was strongly and inversely associated with the change in insulin clearance rate (r = −0.48; P < 0.001) (Fig. 2C).

Figure 2.

A: Relationship between the absolute change from baseline to week 12 in hepatic fat content in participants receiving empagliflozin (n = 38; closed circles) and placebo (n = 19; open circles) and percentage decrease in body weight (r = 0.52; P < 0.001). B: Increase in the Matsuda index of insulin sensitivity (r = −0.51; P < 0.001). C: Change in the insulin clearance rate (r = −0.46; P < 0.01).

To examine the independent predictors of the change in liver fat content, we created a multivariate linear regression model with the change in liver fat content as the dependent variable and age, sex, BMI, diabetes duration, and change from baseline to 12 weeks in body weight, Matsuda index, UGE, FPG, HbA_1c, ALT, AST, plasma FFA concentration, HGP, APRI, AST/ALT ratio, and NFS as independent variables. Only the change in body weight and change in the Matsuda index of insulin sensitivity were significant independent predictors of the change in liver fat content, and the model including both variables explained 62% (r² = 0.62) of the change in liver fat content. Each 1 SD of weight loss (3.3 kg) was associated with 1.78% absolute reduction in liver fat, whereas each 1 SD improvement in the Matsuda index (0.54 units) was associated with 1.47% absolute reduction in liver fat content. Because weight loss and improvement in insulin sensitivity are closely associated, we used mediation analysis to examine whether the change in body weight contributed to the relationship between the change in liver fat and change in the Matsuda index and vice versa. The Sobel test was not significant for either (P = 0.28 for change in Matsuda index and P = 0.21 for change in body weight).

Adverse Events

Three participants (n = 2 with T2D and n = 1 without diabetes) receiving empagliflozin experienced a vaginal yeast infection during the trial. All three patients received therapy for yeast infection and completed the study. No other adverse events related to therapy were observed.

Conclusions

Approximately 80% of patients with T2D manifest increased hepatic fat content, and a similar percentage of obese individuals without diabetes also have MASLD (4–6). Previous studies have demonstrated that ∼20% of the relative reduction in liver fat content (17–21) is caused by all members of SGLT2 inhibitor class (empagliflozin, dapagliflozin, canagliflozin, and leusogliflozin) in patients with T2D with MASLD. The results of the current study extend those of previous studies and demonstrate that empagliflozin is as effective in lowering liver fat content in obese individuals without diabetes as it is in patients with T2D (Fig. 1). Empagliflozin caused absolute reductions of 2.75% and 1.93% in hepatic fat content in patients with T2D and in individuals without diabetes, respectively, which corresponded to relative reductions of 20% and 15% in liver fat, respectively. Not surprisingly, the decrease in liver fat content caused by empagliflozin was strongly and inversely related to baseline liver fat content in both patients with T2D and in individuals without diabetes (r = −62; P < 0.001) (Fig. 1B). Thus, in participants with baseline liver fat >7%, empagliflozin caused absolute reductions of 5.4% and 4.17% in hepatic fat content in those with T2D and those without, respectively, which corresponded to relative reductions of 25% and 22% in liver fat content, respectively, which is comparable to the reduction in liver fat caused by SGLT2 inhibitors in patients with T2D with MASLD. It remains to be seen whether such a decrease in liver fat content for a longer duration with empagliflozin will produce an improvement in liver histology in patients without diabetes with MASH. A previous study reported improvement in liver histology in patients with T2D with MASH treated with SGLT2 inhibitors (28). It is well established that therapies that decrease liver fat content (e.g., pioglitazone) (29) are associated with improvement in liver histology in patients with MASH with and without diabetes.

Although the cut point for MASLD diagnosis is 5.56%, the present results demonstrate that SGLT2 inhibitors are not equally effective across the entire range of liver fat content in individuals with MASLD, and future studies with a larger sample size will be required to determine the range of liver fat at which SGLT2 inhibitors are most effective.

It should be emphasized that participants in the current study were not enrolled based on liver fat content but rather on metabolic characteristics. Nonetheless, >80% of the patients with T2D and obese individuals without diabetes had baseline liver fat content >5%, which is the threshold for diagnosis of MASLD, emphasizing the high frequency of hepatic steatosis in this population.

The mechanisms responsible for the reduction in hepatic fat content caused by SGLT2 inhibitors are not fully understood (22–24). The primary action of empagliflozin is on the kidney to inhibit renal glucose uptake and cause glucosuria, which results in a reduction in plasma glucose concentration. Neither the amount of glucose excreted in the urine nor the decrease in FPG or the decrease in HbA_1c caused by empagliflozin in the current study was related to the decrease in hepatic fat content. Furthermore, despite comparable amounts of urinary glucose loss in individuals without diabetes and in patients with T2D, empagliflozin did not cause any significant reduction in the FPG concentration or HbA_1c in those without diabetes, consistent with the recently described renohepatic axis by our group (11,22). Nonetheless, the reduction in liver fat content was comparable in both groups, suggesting that the decrease in plasma glucose concentration caused by empagliflozin cannot explain the reduction in liver fat content.

We previously demonstrated that the reduction in plasma glucose concentration caused by SGLT2 inhibition is associated with a paradoxical increase in the basal rate of HGP in T2D (27). Consistent with previous studies, empagliflozin caused an increase in HGP of 0.46 mg/kg ⋅ min (18%) in patients with T2D. The decrease in hepatic fat content caused by empagliflozin was inversely correlated with the increase in HGP caused by the drug (r = −0.32; P < 0.01). However, this correlation was not significant in the multivariate regression analysis, suggesting that it reflects other metabolic actions of empagliflozin rather a direct mechanistic link between the increase in HGP and decrease in liver fat content. Similarly, although other metabolic factors were correlated with the decrease in liver fat content (e.g., Supplementary Table 1), none were significant predictors of the change in hepatic fat content in the multivariate model. Only weight loss and change in the Matsuda index of insulin sensitivity remained significantly and strongly correlated with the change in liver fat content in the multivariate model. Baseline liver fat content was weakly and nonsignificantly correlated with BMI, most likely because >90% of participants were overweight or obese. As expected, baseline hepatic fat content was inversely correlated with the Matsuda index of insulin sensitivity (r = −0.39; P < 0.01). Unlike baseline liver fat content, the change from baseline to week 12 in liver fat content was strongly correlated with weight loss and with the increase in the Matsuda index of insulin sensitivity, and both were independent predictors of the change in liver fat content in the multivariate model. Furthermore, in the multivariate regression model, weight loss plus Matsuda index explained approximately one-half of the variation in the change in hepatic fat content. Each 1 SD in the Matsuda index (0.5 units) and weight loss (3.3 kg) was associated with an absolute reduction in liver fat content of ∼1.5%. Although insulin sensitivity and body weight are strongly related, and weight loss associated with SGLT2 inhibition likely contributes to improvement in insulin sensitivity, both weight loss and the change in the Matsuda index caused by empagliflozin independently and directly contributed to the reduction in liver fat content. These observations emphasize the importance of obesity and insulin resistance in the etiology of MASLD.

Of note, although not statistically significant, the decrease in body weight was numerically greater in participants without T2D than in individuals with diabetes. A greater reduction in body weight in individuals without versus those with diabetes also has been reported with glucagon-like peptide 1 receptor agonist therapy and is explained by the decrease in plasma glucose concentration in those with T2D, which leads to a reduction in glucosuria. Furthermore, it has been shown that SGLT2 inhibition stimulates a secondary increase in food intake. It is possible that the secondary stimulation of food intake differs between those with and without diabetes.

We previously demonstrated that members of the SGLT2 inhibitor class increase plasma FFA concentration and total body fat oxidation (30). An increase in total body fat oxidation can improve insulin sensitivity independently of weight loss. The increase in plasma FFA concentration caused by empagliflozin was inversely and significantly correlated with the decrease in hepatic fat content. Because ∼60% of triglycerides, which accumulate in the liver, originate from FFAs released from adipocytes, one would expect an inverse relationship between the change in plasma FFA concentration and liver fat content. However, the correlation between plasma FFA concentration and hepatic fat content lost its statistical significance in the multivariate model. Therefore, it is likely that other metabolic actions of empagliflozin rather a direct effect on the liver are responsible for the decrease in hepatic fat content. Unfortunately, because of the risk of spreading COVID-19, the measurement of total body fat oxidation with indirect calorimetry, which was part of the original protocol, was discontinued between February 2020 and July 2022. Therefore, measurement of fat oxidation could not be performed in a majority of study participants during the COVID-19 epidemic, and we could not evaluate the relationship between the expected increase in total-body fat oxidation and the depletion of liver fat content. Lastly, we observed a strong inverse correlation between the decrease in liver fat content and increase in the metabolic clearance of insulin, consistent with previous observations from our group (31) and others (32). These results suggest that hepatic fat content is an important determinant of insulin degradation by the liver.

Liver indices (e.g., AST/ALT ratio, FIB-4, NFS, and APRI) are widely used as simple noninvasive tools for the diagnosis of MASLD and MASH and to monitor response to therapy. However, the change in these parameters correlated poorly or not at all with the decrease in liver fat content caused by empagliflozin in the current study. Only the change in APRI mildly correlated with the decrease in liver fat content. Simply monitoring the change in ALT after empagliflozin administration was a much stronger predictor of response to therapy than other liver indices (Supplementary Table 1).

The current study was performed in a single center and therefore included a relatively small number of participants. A larger multicenter study is warranted to confirm its results. Furthermore, the change in liver fat content was a secondary outcome of the study. Therefore, it was not powered to detect significant change in liver fat in individuals without diabetes or a difference in the SGLT2 impact in those with versus without diabetes. Nonetheless, we believe that the findings of the current study will foster larger multicenter studies with change in liver fat as the primary outcome in order to definitely establish SGLT2 inhibitors as therapeutic options in obese individuals without diabetes.

In summary, the results of the current study demonstrate that empagliflozin reduces liver fat content in obese individuals without diabetes, as it does in patients with T2D. Future larger studies will establish SGLT2 inhibitors as therapeutic options for liver fat in obese individuals without diabetes.

This article contains supplementary material online at https://doi.org/10.2337/figshare.24964044.