Introduction

Nonalcoholic fatty liver disease (NAFLD), which is the most common chronic liver disease worldwide, is classified into nonalcoholic fatty liver (NAFL) and nonalcoholic steatohepatitis (NASH) [1, 2]. The prevalence of NAFLD has increased with the rise in obesity and metabolic syndrome and is associated with lifestyle changes. In patients with obesity and type 2 diabetes mellitus, the prevalence of NAFLD is as high as 57–98% and 62–69%, respectively [3,4,5]. Insulin resistance and obesity are closely related to the development and progression of NAFLD [6]. The most common complications of NAFLD are type 2 diabetes mellitus and impaired glucose tolerance [6,7,8,9].

A positive correlation between the visceral fat mass and the hepatocyte fat content in NAFLD patients has been reported [10, 11]. The mechanisms through which insulin resistance and obesity contribute to the onset and progression of NAFLD/NASH appear to be those described in the two-hit theory by Day et al. [12, 13] and the multiple parallel hits hypothesis by Tilg et al. [14]. The two-hit theory suggests that the first “hit”, steatosis, increases the sensitivity of the liver to the second “hits” that mediate liver injury [13]. In the multiple parallel hits hypothesis, endoplasmic reticulum stress and related signaling networks, adipocytokines, and insulin resistance are suggested as the central pathways responsible for NASH [14]. Cusi et al. recently reported on the beneficial effects of pioglitazone on NASH in patients with prediabetes or type 2 diabetes mellitus [15]. However, the only established therapies for NAFLD are dietary and exercise therapies [1, 2].

Sodium glucose co-transporter 2 (SGLT2) inhibitors, which are used to treat type 2 diabetes mellitus, lower blood glucose levels by inhibiting the reabsorption of glucose in the kidney and promoting the urinary glucose excretion [16]. In addition to a sustained antihyperglycemic effect, SGLT2 inhibitors have also been reported to have a body weight-lowering effect that is accompanied by a reduction in abdominal fat [17]. SGLT2 inhibitors were found to reduce the fat content in the liver, improve inflammation, and prevent progression of fibrosis in NAFLD model mice [18]. The results of phase 3 studies conducted in Japan and overseas suggested that in patients with type 2 diabetes mellitus, liver function parameters might be improved by the administration of the SGLT2 inhibitor canagliflozin [16, 19,20,21]. However, the effects on liver function and blood glucose levels and the safety of canagliflozin in patients with abnormal hepatic function have not been clarified.

Alanine aminotransferase (ALT), which indicates liver inflammation or injury, is a candidate marker for NAFLD. In recent years, it has been reported that ALT is a useful indicator of the progression of liver fibrosis in NAFLD patients [22, 23]. An ALT level >30 U/L is the criterion established in the guidelines of the Ministry of Health, Labour and Welfare in Japan for the routine health checkup program. Therefore, we conducted a pooled and subgroup analysis of three Japanese clinical studies by baseline ALT [16, 19, 24] to investigate the effects on liver function and blood glucose levels and the safety of canagliflozin in a high ALT subgroup (baseline ALT >30 U/L).

Methods

We carried out a post hoc analysis of three previous studies in Japan: a 12-week monotherapy study [24], a 24-week monotherapy study [16], and a 52-week monotherapy or combination therapy study [19]. This post hoc analysis was divided into two parts: Study 1 (a pooled analysis of the 12- and 24-week studies) and Study 2 (the 52-week study). The aim of Study 1 was to compare the efficacy on liver function and blood glucose levels and the safety of canagliflozin 100 mg with those of placebo at 12 weeks in a high ALT subgroup (baseline ALT >30 U/L). The aims of Study 2, which included a larger number of patients than Study 1, were (1) to evaluate the long-term efficacy and safety of canagliflozin 100 mg at 52 weeks, and (2) to analyze the modulating factors influencing the change of ALT in a high ALT subgroup. The improving effects on liver function and blood glucose levels and safety achieved with the different drugs (placebo and canagliflozin 100 mg) and for the different ALT categories (ALT >30 U/L, ALT ≤30 U/L) were comparatively evaluated. These studies were conducted in compliance with Good Clinical Practice guidelines and the Pharmaceutical Affairs Law in Japan, according to the ethical principles of the Declaration of Helsinki of 1964, as revised in 2008, and were approved by the institutional review boards at all of the participating institutions.

Study 1

Study 1 consists of a pooled analysis of two placebo-controlled studies (the 12-week study and 24-week study). The data from each placebo and canagliflozin 100 mg (the approved dose in Japan) group up to 12 weeks were pooled. Pooling confirmed that the patient backgrounds and the ALT profiles through 12 weeks were nearly identical in these two studies. In the high ALT subgroup of the 12-week study, the mean changes in ALT (±SD) from baseline to 12 weeks were −9.0 ± 10.7 and −2.0 ± 17.0 U/L in the canagliflozin 100 mg and placebo groups, respectively. In the 24-week study, the mean changes were −11.1 ± 12.5 and −3.6 ± 19.8 U/L, respectively.

Study design

The 12-week study [24] was a multicenter, randomized, placebo-controlled, double-blind, parallel-group, phase 2 dose-finding study in Japanese type 2 diabetes mellitus patients. Subjects received monotherapy with canagliflozin (50, 100, 200, or 300 mg) or placebo once a day for 12 weeks. The 24-week study [16] was a multicenter, randomized, placebo-controlled, double-blind, parallel-group phase 3 confirmatory study in Japanese type 2 diabetes mellitus patients. Subjects received monotherapy with canagliflozin (100 or 200 mg) or placebo once a day for 24 weeks. Patients with serious liver disease, defined as those who required hospitalization or surgery for treatment, and those with a medical history of hepatitis B or hepatitis C were excluded, as were patients with ALT or aspartate aminotransferase (AST) > 2.5 × ULN (40 U/L) on the day of screening. The exclusion criteria of ALT and AST levels were set in consideration of the patients’ safety and ethics. The amount of alcohol intake was not regulated in these two studies.

Study 2

Study design

The 52-week study [19] was a multicenter, randomized, open-label, long-term, phase 3 study in Japanese patients with type 2 diabetes mellitus. Subjects received canagliflozin (100 or 200 mg) once a day for 52 weeks, either as monotherapy or in combination with another oral antihyperglycemic drug (sulfonylurea, glinide, α-glucosidase inhibitor, biguanide, thiazolidinedione, or dipeptidyl peptidase-4 inhibitor). In combination therapy groups, the antihyperglycemic drugs were given from ≥83 days before the first day of canagliflozin treatment. Patients with serious liver disease, hepatitis B, or hepatitis C were excluded, as were patients with ALT or AST > 2.5 × ULN (40 U/L) on the day of screening. The amount of alcohol intake was not regulated in this study.

Outcomes

The primary endpoint is the change in ALT level from baseline. The secondary endpoints are the changes in the following parameters: liver function-related parameters (AST and γ-glutamyl transpeptidase [γ-GTP]) and efficacy-related parameters (glycosylated hemoglobin [HbA1c], fasting blood glucose, insulin, body weight, body mass index [BMI], and waist circumference). Adverse events (AEs), serious AEs (SAEs), and laboratory tests (low-density lipoprotein cholesterol [LDL-c], high-density lipoprotein cholesterol [HDL-c], triglyceride [TG], free fatty acid [FFA], uric acid, and total ketone bodies) were also evaluated.

AEs were classified according to system organ class and preferred term using MedDRA/J version 15.1 (Japanese Maintenance Organization, Tokyo, Japan).

Statistical analyses

The full analysis set was used for the analysis of liver function-related parameters (e.g., ALT, AST), efficacy-related parameters (e.g., HbA1c), and laboratory test values. The safety analysis set was used for the analysis of AEs.

In Study 1, the change in ALT after 12 weeks of treatment with canagliflozin 100 mg was compared with that after 12 weeks of treatment with placebo in the high ALT subgroup. The comparisons between treatment groups were performed using analyses of covariance with the baseline value of parameters as a covariate. For reference, similar analyses were also performed for the low ALT subgroup. The last observation carried forward (LOCF) method was used to impute missing values.

In Study 2, the effects of treatment for 52 weeks with canagliflozin 100 mg on liver function-related parameters, efficacy-related parameters, and laboratory test values in each subgroup were assessed by the paired t test. Missing values were imputed by the LOCF method. To identify factors affecting a decrease in ALT in the high ALT subgroup, single correlation and multiple regression analyses (stepwise variable selection method; both entry and retention criteria were set at P = 0.05) were performed with change in ALT as a dependent variable and related factors as explanatory variables. These were performed in the canagliflozin 100 mg group in Study 2, in which more subjects were observed for a longer period.

Continuous data are summarized based on the number of patients (n), mean, and standard deviation (SD), and discrete data are summarized based on the n and percentage values for each category. Tests were two-sided with a 5% significance level, and 95% confidence intervals (CIs) were calculated. The statistical analysis was performed by Takumi Information Technology Inc. (Tokyo, Japan), using SAS 9.4 (SAS Institute Inc., Cary, NC, USA).

Results

Study 1

Baseline characteristics

The mean ± SD ALT at baseline was 45.6 ± 13.9 U/L in the canagliflozin 100 mg group and 48.4 ± 14.2 U/L in the placebo group in the high ALT subgroup. The HbA1c levels were 8.07 ± 0.70% in the canagliflozin 100 mg group and 8.10 ± 0.72% in the placebo group in the high ALT subgroup (Table 1). Additionally, the mean HbA1c at baseline did not markedly differ between the high ALT subgroup and the low ALT subgroup, either in the canagliflozin 100 mg group or the placebo group. Conversely, the mean body weight, BMI, waist circumference, and homeostatic model assessment-insulin resistance (HOMA-IR) were all higher in the high ALT subgroup than in the low ALT subgroup, both in the canagliflozin 100 mg group and in the placebo group.

Table 1 Baseline characteristics of patients (Study 1)
Full size table

Efficacy

At 12 weeks, the ALT level was significantly lower in the canagliflozin 100 mg group than in the placebo group, only in the high ALT subgroup (Fig. 1a). A significant reduction in the AST level was also observed in the canagliflozin 100 mg group, but only in the high ALT subgroup (Fig. 1b). In both ALT subgroups, HbA1c and body weight were significantly lower in the canagliflozin 100 mg group than in the placebo group (Fig. 1c, d). Waist circumference was significantly lower in the canagliflozin 100 mg group than in the placebo group, but only in the high ALT subgroup; in the canagliflozin and placebo group, the change from baseline was −1.90 ± 2.78 and −0.51 ± 2.65 cm, respectively, in the high ALT subgroup (P = 0.0132) and −1.39 ± 2.93 and −0.70 ± 2.67 cm, respectively, in the low ALT subgroup (no significant difference). Additionally, in both ALT subgroups, fasting insulin level was significantly lower in the canagliflozin 100 mg group than in the placebo group; the change from baseline was −4.056 ± 4.792 and −1.671 ± 5.324 μIU/mL, respectively, in the high ALT subgroup (P = 0.0163) and −1.599 ± 3.858 and −0.372 ± 3.228 μIU/mL, respectively, in the low ALT subgroup (P = 0.0067) (Suppl. 1).

Fig. 1
figure 1

Change in parameters after administration of canagliflozin (Study 1). ALT (a), AST (b), HbA1c (c), and body weight (d) at baseline and 12 weeks. Data are presented as mean ± SD. Low ALT subgroup (unfilled circle ALT ≤ 30 U/L; n = 109) and high ALT subgroup (unfilled square, ALT >30 U/L; n = 59) in the placebo group. Low ALT subgroup (filled circle, n = 117) and high ALT subgroup (filled square, n = 47) in the canagliflozin 100 mg group. An analysis of covariance by baseline value was performed for the change from before to after treatment for the comparison of the placebo and canagliflozin 100 mg groups. *P < 0.05; **P < 0.0001. ALT alanine aminotransferase; AST aspartate aminotransferase; CAN canagliflozin 100 mg; HbA1c glycosylated hemoglobin; NS not significant; PBO placebo

Full size image

Safety

The overall incidence of AEs in the high ALT subgroup was similar across treatment groups [canagliflozin 100 mg: 48.9% (95% CI 34.1–63.9), placebo: 50.8% (95% CI 37.5–64.1)]. The incidences of AEs related to the study drug in the high ALT subgroup were 19.1% (95% CI 9.1–33.3) and 11.9% (95% CI 4.9–22.9) with canagliflozin 100 mg and placebo, respectively. In the canagliflozin 100 mg group, no clear difference in the overall incidence of AEs related to the study drug was found between ALT subgroups [19.1% (95% CI 9.1–33.3) and 23.9% (95% CI 16.5–32.7) in the high and low ALT subgroups, respectively]. There was no clear difference in the overall incidence of SAEs in the high ALT subgroup across treatment groups. There were no SAEs related to the study drug in either treatment group in the high ALT subgroup (Table 2).

Table 2 Adverse events (Study 1)
Full size table

In the high ALT subgroup, the incidences of symptomatic hypoglycemia (6.4 vs. 0.0%), female genital infections (6.3 vs. 0.0%), and osmotic diuresis (6.4 vs. 1.7%) were higher in the canagliflozin 100 mg group than in the placebo group. Among these AEs, the incidences of symptomatic hypoglycemia (6.4 vs. 0.9%) and osmotic diuresis (6.4 vs. 2.6%) in the canagliflozin 100 mg group were higher in the high ALT subgroup than in the low ALT subgroup. There were no hepatic AEs in the canagliflozin 100 mg group in both subgroups. Total ketone bodies were significantly increased in the canagliflozin 100 mg group in both subgroups (Suppl. 1).

Study 2

Baseline characteristics

In the canagliflozin 100 mg group, the mean ± SD ALT at baseline was 49.1 ± 18.6 U/L in the high ALT subgroup and 19.3 ± 5.5 U/L in the low ALT subgroup. The AST and γ-GTP levels were both higher in the high ALT subgroup than in the low ALT subgroup. No clear difference in the platelet count was found between the subgroups. The mean ± SD HbA1c at baseline was 8.21 ± 0.91% in the high ALT subgroup and 7.97 ± 0.86% in the low ALT subgroup. The mean age was lower in the high ALT subgroup than in the low ALT subgroup (53.4 ± 10.5 vs. 59.6 ± 10.2 years). The proportion of men, the HOMA-IR, body weight, BMI, waist circumference, and the proportion of subjects with complications of hypertension or hyperlipidemia were higher in the high ALT subgroup (Table 3). The proportion of patients using thiazolidinedione in the low ALT subgroup was higher than that of the high ALT subgroup. Although the relationship between combination of thiazolidinedione and baseline ALT level was not clear, the beneficial effects of pioglitazone were reported in patients with NASH and type 2 diabetes [15].

Table 3 Baseline characteristics (Study 2)
Full size table

Efficacy

In the high ALT subgroup, the administration of canagliflozin 100 mg resulted in improvement in liver function parameters (ALT, AST) starting from the early stages of treatment, and this improvement persisted through 52 weeks (Fig. 2a, b). The mean change in ALT level at 12 weeks was similar to that in the canagliflozin treatment group in the high ALT subgroup in Study 1.

Fig. 2
figure 2

Change in parameters through 52 weeks (Study 2). Change in ALT (a), AST (b), HbA1c (c), and body weight (d) over 52 weeks. Data are presented as mean ± SD. Low ALT subgroup (unfilled square, ALT ≤ 30 U/L; n = 352–389) and high ALT subgroup (filled square ALT >30 U/L; n = 179–195) in the canagliflozin 100 mg group. ALT alanine aminotransferase, AST aspartate aminotransferase, CAN canagliflozin 100 mg, HbA1c glycosylated hemoglobin

Full size image

At 52 weeks in the high ALT subgroup, the mean absolute and percent change ± SD in ALT from baseline was −16.0 ± 18.8 U/L (−30.0 ± 27.9%). The mean changes in AST and γ-GTP levels were −9.0 ± 14.1 and −28.2 ± 73.9 U/L, respectively (Suppl. 2). These changes were appreciably greater than those in the low ALT subgroup (ALT, AST, γ-GTP: 0.4 ± 27.8, 0.2 ± 7.4, and −3.6 ± 38.6 U/L, respectively). Neither the presence nor absence of the concomitant use of another antihyperglycemic drug with canagliflozin nor the type of concomitant drug used resulted in any clear differences in the ALT lowering effect of canagliflozin 100 mg in the high ALT subgroup (data not shown).

A reduction in ALT after the use of canagliflozin 100 mg was found in 88.7% of the subjects in the high ALT subgroup (Suppl. 3). In Study 2, a marked increase in ALT was observed at 28 weeks in one subject who discontinued canagliflozin due to jaundice and bile duct carcinoma. It was concluded by the investigator that this patient had been afflicted with bile duct carcinoma prior to enrollment in the study.

The HbA1c and body weight decreased in both subgroups, which persisted through 52 weeks (Fig. 2c, d). The HbA1c, fasting blood glucose, insulin, body weight, and waist circumference were all found to be significantly decreased in both ALT subgroups at 52 weeks (Suppl. 2). At 52 weeks, there were no clear changes in LDL-c levels in either subgroup. The HDL-c level was significantly increased in both subgroups (mean change ± SD from baseline at 52 weeks: 2.9 ± 7.3 and 5.0 ± 8.0 mg/dL in the high and low ALT subgroups, respectively; both P < 0.0001). The FFA level was slightly but significantly increased in both subgroups (mean change ± SD from baseline at 52 weeks: 0.075 ± 0.263 and 0.088 ± 0.231 mEq/L, respectively; both P < 0.0001). The TG level was not found to be significantly decreased in the high ALT subgroup (Suppl. 2).

Safety

The overall incidence of AEs in the canagliflozin 100 mg group was 80.0% (95% CI 73.7–85.4) and 82.3% (95% CI 78.1–85.9) in the high ALT and low ALT subgroups, respectively (Table 4). The incidences of AEs related to the study drug were 28.7% (95% CI 22.5–35.6) in the high ALT subgroup and 33.4% (95% CI 28.7–38.3) in the low ALT subgroup. There were no clear differences in the overall incidence of SAEs or SAEs related to the study drug between the two subgroups [SAEs: 4.6% (95% CI 2.1–8.6) in the high ALT subgroup, 5.4% (95% CI 3.4–8.1) in the low ALT subgroup; SAEs related to the study drug: 1.0% (95% CI 0.1–3.7) in the high ALT subgroup, 0.3% (95% CI 0.0–1.4) in the low ALT subgroup]. Although the incidences of AEs associated with symptomatic hypoglycemia, asymptomatic hypoglycemia, female genital infection, and osmotic diuresis were >5%, the incidences in the high ALT subgroup were similar to those in the low ALT subgroup. There was one hepatic AE related to the study drug in the low ALT subgroup only: hepatic function abnormal. Total ketone bodies were significantly increased (97.5 ± 265.2 and 158.9 ± 728.9 μmol/L in the high and low ALT subgroups, respectively; both P < 0.0001).

Table 4 Adverse events (Study 2)
Full size table

Analysis of modulating factors of the change in ALT

To investigate the subject characteristics and factors associated with the change in ALT levels induced by canagliflozin in the high ALT subgroup, correlation and multiple regression analyses were performed for the changes in ALT at 12 and 52 weeks. The results of the correlation analysis in the high ALT subgroup showed a negative correlation between the change in ALT and baseline ALT at both 12 weeks (r = − 0.605, P < 0.0001) and 52 weeks (r = − 0.638, P < 0.0001). The other parameters had low or no correlation with the change in ALT (Table 5). The results of the multiple regression analysis performed with change in ALT at 12 weeks and 52 as the dependent variable and the baseline values of each of the parameters as explanatory variables are shown in Suppl. 4. At 12 weeks, the baseline ALT and HOMA-IR were selected as explanatory variables (R 2 = 38.3%). At 52 weeks, the baseline ALT, baseline HOMA-IR, and baseline LDL-c were selected as explanatory variables (R 2 = 44.1%). From the multiple regression analysis with the change in related parameters containing baseline ALT as explanatory variables, baseline ALT, change in body weight, and change in LDL-c were identified as explanatory variables (R 2 = 40.9%) at 12 weeks, and baseline ALT, change in HbA1c, change in body weight, change in LDL-c, and change in TG were selected as explanatory variables (R 2 = 47.7%) at 52 weeks. In all regression analyses, the baseline ALT accounted for most of the effect, and the contribution of the other variables was small (partial R 2 = 1.3–2.4%, Suppl. 4).

Table 5 Relationship between the change in ALT after the administration of canagliflozin and various parameters in the high ALT subgroup (Study 2)
Full size table

Discussion

The present post hoc analysis was performed using the data from Japanese phase 2 and 3 studies of canagliflozin, and the results revealed that canagliflozin improved hepatic function tests in type 2 diabetes mellitus patients with high ALT levels. This effect persisted through 52 weeks. The incidence of AEs or SAEs in the high ALT subgroup in the canagliflozin treatment group was comparable to that of the placebo group and the low ALT subgroup. The results of correlation analysis and multiple regression analysis showed that the baseline ALT was the main influencing factor of the change in ALT.

It was confirmed that in patients with high ALT levels, canagliflozin not only lowers blood glucose levels, body weight, and insulin levels, but also lowers the levels of liver function parameters such as ALT, AST, and γ-GTP. To inhibit the progression of NAFLD, it is important to control body weight, HbA1c, and ALT levels [23, 25]. Seko et al. reported that a reduction of 30% or more in the baseline ALT level was associated with amelioration of the NAFLD activity score and progression of liver fibrosis in NAFLD patients [22]. In Study 2 of the present post hoc analysis, the ALT level decreased by approximately 30% by canagliflozin in the high ALT subgroup at 52 weeks. Furthermore, SGLT2 inhibitors have been reported to inhibit the progression of fibrosis or decrease serum ALT levels in NASH and diabetic mouse models, as well as type 2 diabetes mellitus patients with NAFLD [18, 26,27,28,29]. Collectively, the results of the present analysis suggest that treatment with SGLT2 inhibitors may provide a clinical benefit to type 2 diabetes mellitus patients with NAFLD.

Obesity, insulin resistance, and enhanced expression of inflammatory adipocytokines, such as IL-6 and TNFα, in adipose tissue are among the factors associated with NAFLD as suggested by the two-hit theory [12, 13] and the multiple parallel hits hypothesis [14]. SGLT2 inhibitors have been reported to reduce abdominal visceral fat [17], improve insulin sensitivity [30], and suppress the genetic expression of inflammatory markers in the liver of mice fed a high-fat diet [27]. In our study, body weight, waist circumference, and fasting insulin level were significantly reduced in the canagliflozin 100 mg group, which suggests abdominal visceral fat reduction and improvement of insulin sensitivity. Hence, the body weight lowering effect and the subsequent improvement of insulin resistance and expression of inflammatory adipocytokines by canagliflozin treatment may result in the improvement of liver function.

Leiter et al. reported that the liver function improving effects of canagliflozin were related to decreases in blood glucose levels and body weight [21]. In contrast, Komiya et al. reported that the decrease in ALT by 24-weeks administration of ipragliflozin in type 2 diabetes patients with impaired hepatic function did not depend on the decrease in body weight [27]. The results of our multiple regression analysis between the subject characteristics or the change in related factors and the change in ALT level at 12 and 52 weeks by canagliflozin suggested that the baseline ALT value was the main explanatory factor for the ALT lowering effect of canagliflozin. These results indicate that canagliflozin further lowered ALT in patients with higher baseline ALT values. The change in body weight was identified as one of the explanatory factors for the change in ALT at 12 and 52 weeks. These results support the liver function improving mechanisms of SGLT2 inhibitors through the body weight lowering effect, which were hypothesized above and reported by Leiter et al. However, there were some parameters for which the underlying mechanisms explaining the relationship with the change in ALT remain unclear. Except for the baseline ALT value, the contributions of the parameters identified as the explanatory factors for the change in ALT were small in the multiple regression analysis. These results may derive from other unknown mechanisms.

In Studies 1 and 2, the incidence of AEs or SAEs in the high ALT subgroup in the canagliflozin treatment group was similar to that of the placebo group and the low ALT subgroup. Consistent with previous reports [31, 32], the incidences of genital infection and osmotic diuresis were higher in the canagliflozin 100 mg group. The increase of free fatty acids and ketone bodies were observed in the canagliflozin 100 mg group, probably due to enhance lipolysis in adipose tissue and hepatic ketogenesis that is attributed to energy loss by urinary glucose excretion [33].

The present study has some limitations. In this analysis, impaired liver function was defined as ALT >30 U/L. Patients with ALT or AST >2.5 × ULN, serious liver disease, or a history of hepatitis B or hepatitis C infection were excluded from this analysis. In addition, patients did not always have a confirmed NAFLD diagnosis [2] because histological diagnosis by liver biopsy or ultrasonography was not always performed for all patients in the high ALT subgroup. Furthermore, the amount of alcohol intake was not regulated in patients in this study. In the high ALT subgroup, the mean platelet count value, a marker of liver fibrosis, was in the normal range at baseline. In the future, it will be necessary to conduct a more detailed investigation in patients with confirmed diagnoses of NAFLD.

In conclusion, the results of this post hoc analysis suggest that in type 2 diabetes mellitus patients with impaired liver function, such as those with NAFLD, canagliflozin may improve liver function, reduce both HbA1c and body weight, and also be well tolerated.