Reductions in liver enzymes are associated with anti‐hyperglycaemic and anti‐obesity effects of tofogliflozin in people with type 2 diabetes: Post‐hoc analyses

Abstract Aims How the pathology of type 2 diabetes (T2D), including hyperglycaemia and obesity, affects liver enzymes has not been clinically demonstrated. Thus, we compared time courses of gamma‐glutamyltransferase (GGT) and alanine aminotransferase (ALT) with those of fasting plasma glucose (FPG) and body weight (BW) during treatment with the SGLT2 inhibitor tofogliflozin for T2D. Materials and Methods We post‐hoc analysed preexisting data on 1046 people with T2D administered tofogliflozin or placebo for 24 weeks in four tofogliflozin studies. First, time courses of percent changes in variables during the intervention were analysed using a mixed effect model to explore the similarity of the time courses and to evaluate time‐treatment interactions. Second, clinical factors related to the percent changes in GGT and ALT were clarified using multivariate analyses. Results GGT levels and FPG values rapidly and significantly decreased via tofogliflozin as early as week 4, with decreases maintained until week 24. Conversely, BW and ALT decreased progressively until week 24. Time courses of FPG (p = .365, time‐treatment interaction) and GGT (p = .510) reductions were parallel between tofogliflozin and placebo from weeks 4 to 24, while BW and ALT reductions (p < .001, respectively) were not. Reductions in GGT at week 24 were associated with reductions in FPG and BW at week 24, whereas ALT reductions were only associated with reductions in BW. Conclusions Reductions in GGT and ALT were associated with the anti‐hyperglycaemic and anti‐obesity effects of tofogliflozin, respectively, in people with T2D. Therefore, GGT and ALT may be surrogate markers for hyperglycaemia and obesity in T2D.


| INTRODUC TI ON
3][4] Circulating levels of liver enzymes, including those for serum alanine aminotransferase (ALT) and gamma-glutamyltransferase (GGT), provide insights into the extent of hepatic fat accumulation and its related oxidative stress.ALT is the most specific marker of liver pathology and fat accumulation.Changes in ALT levels were correlated with hepatic fat levels measured by proton spectroscopy after weight loss. 5Furthermore, high ALT [6][7][8][9][10] and GGT 8,10 levels were identified as significant risk factors for T2D.
In a Japanese prospective cohort study, both ALT and GGT were identified as independent predictors of T2D after patients were stratified according to various risk factors, specifically fasting insulin, body mass index (BMI), waist-hip ratio, and high-sensitivity C-reactive protein values. 10These findings suggest that elevated liver enzymes values, particularly those for GGT and ALT, indicate the presence of metabolic abnormalities initiated by fat accumulation through weight gain and hyperglycaemia.Additionally, increased GGT and ALT levels were reported to be independent and additive risk factors for the development of T2D in people without fatty liver, suggesting that their increase may be individually reflected by the distinct pathology of T2D. 11However, the time-course of the elevation of each liver enzyme and its specific associations with hyperglycaemia and obesity have not been demonstrated clinically.To address this issue, careful observation of the time-course of liver enzymes values and a statistical evaluation of similarity of changes in GGT and ALT levels during interventions for hyperglycaemia and obesity may be useful.
A new class of antidiabetic agents, sodium/glucose cotransporter-2 (SGLT2) inhibitors, reduces both hyperglycaemia and body weight (BW), 12,13 thereby lowering liver enzyme levels. 14,15In the present study, we examined the time course of GGT and ALT levels during interventions for hyperglycaemia and obesity with the SGLT2 inhibitor, tofogliflozin.Further, we explored similarities of time courses of percent changes in variables and evaluated the associations of fasting plasma glucose (FPG) and BW with GGT and ALT in patients with T2D.

| Integrated studies and patients
The pooled post-hoc analysis included four phase 2 and phase 3 tofogliflozin trials (Tables S1 and S2) with durations of at least 24 weeks that included people with T2D and compared placebo to different doses of tofogliflozin or compared various doses of tofogliflozin either as monotherapy or as an add-on to other antidiabetic agents.The CSG003JP study (placebo or tofogliflozin [10, 20, and 40 mg] as monotherapy) was a 24-week randomized, double-blind, placebo-controlled, combined phase 2 and 3 trial. 14e CSG004JP study (tofogliflozin 20 and 40 mg as monotherapy) and the CSG005JP study (tofogliflozin 20 and 40 mg as add-ons to other oral antidiabetic agents) were both 52-week randomizedcontrolled, open-label, phase 3 trials. 16Details of these studies and their results, along with patient inclusion and exclusion criteria, were previously reported. 14,16The CSG006JP study (tofogliflozin 40 mg as monotherapy or an add-on to sulfonylurea or dipeptidyl peptidase-4 inhibitor) was a 24-week, multicentre, open-label trial. 17Individual-level data from the 24-or 52-week core treatment periods of each study were used in this pooled analysis.Details of each study with the main patient inclusion criteria are summarized in Table S1.All included studies were conducted in accordance with the Declaration of Helsinki and Good Clinical Practice.
The protocols had been reviewed and approved by the institutional review board of each participating centre (JapicCTI-101349, JapicCTI-101351, JapicCTI-101352 and JapicCTI-111572).All patients provided written informed consent prior to enrolment in the study.This consent included an agreement for further analysis and publication of the existing data acquired in those studies.
Therefore, the requirement for additional informed consent was waved due to the use of preexisting data.

| Measurements
Baseline laboratory values measured for the following variables:

| Overall statistical flow in this post hoc analyses
Each study involved in this post-hoc analysis included a plan for statistical analysis.However, since this is a post hoc analysis of several interventional studies, there was not a plan for the statistical analysis.However, we examined the data as follows.First, we

| Evaluation of clinical parameters at baseline
To assess the effects of obesity and glycemic status, respectively, on baseline parameters, participants were divided into two groups according to the presence of obesity (BMI ≥25 kg/m 2 ) and baseline median FPG levels (FPG ≥8.7 mmol/L [156 mg/dL]).
Analyses of combinations of correlations among baseline liver enzymes, FPG levels and BMI values were performed using Spearmanʼs rank order correlation coefficients.
To examine whether baseline FPG levels and BMI values might be associated with GGT and ALT baseline levels in the entire participant population, 12 variables at baseline (age, sex, duration of diabetes, HbA1c, FPG, HOMA-IR, HOMA-beta, BMI, eGFR, waist circumference, AST, ALT and GGT) were initially identified according to clinical considerations.

| Statistical evaluation of the time-course of variables during the entire intervention
To examine time courses of the following variables, percentage changes in FPG, BW, GGT, ALT, AST, ALP, BAP and NTx during the whole intervention were tested using a mixed effect model with treatment (tofogliflozin vs. placebo), time (FPG, BW, GGT, ALT, AST, and ALP; from week 4 to 24 and BAP, and NTx; week 12 and 24) and interaction between treatment and time as fixed effects, baseline value as a covariate, and subject as a random effect.

TA B L E 1 (Continued)
To examine whether the percent changes in FPG levels and BW might affect the percent changes in GGT and ALT levels at week 24 in participants receiving tofogliflozin, 13 variables at baseline (age, sex, duration of diabetes, HbA1c, FPG, HOMA-IR, HOMA-beta, BMI, eGFR, waist circumference, AST, GGT and ALT) and 2 variables at week 24 (percent changes in FPG levels and BW) were initially identified according to clinical considerations.In addition, one variable at week 24 (percent change in HOMA-IR) was added to the above used model for further consideration.Additionally, to examine whether the percent changes in FPG levels and BW might affect the those in GGT and ALT levels at week 4, the same procedure was performed using the percentage of changes in variables at week 4 instead of changes at week 24.Finally, to examine whether FPG and BW reductions might predict the percent changes in GGT and ALT levels at week 24, the percent changes in GGT and ALT levels at week 4 were added to the above baseline variables.These factors were identified using a multivariate regression model and stepwise variable selection (p < 0.05).
The participants' demographics were summarized using appropriate descriptive statistics (mean and standard deviation [SD]   for continuous variables, numeric values and percentages for categorical variables).Furthermore, differences in assessments across groups were analysed using analysis of variance (ANOVA) and Fisher's exact test.In the present study, all HbA1c values are presented in National Glycohemoglobin Standardization Program (NGSP) units.

| Association of basal GGT and ALT levels with clinical parameters
At baseline, both GGT and ALT levels were positively associated with FPG levels and BMI values (Table S3).Multivariate analyses suggested that GGT levels were positively associated with FPG levels not BMI values and that ALT levels were positively associated with BMI and HbA1c (Table S4).When participants were stratified according to BMI and FPG, basal levels of ALT and AST but not GGT and ALP were significantly higher in participants with a BMI ≥25 kg/m 2 (Table 1).Also, values for ALT, GGT and ALP but not AST were significantly higher in participants with FPG ≥8.7 mmol/L (156 mg/dL).

| Time-course of percent changes in GGT, ALT, FPG and BW during the intervention with tofogliflozin
FPG levels after the start of tofogliflozin therapy had rapidly and significantly decreased as early as week 4 (least squares mean: −14.4%, p < .001 vs. placebo).This decrease was maintained until week 24 (−16.7%,p < .001).Similarly, GGT levels had rapidly and significantly decreased at week 4 (−17.8%,p < .001),and the decrease was maintained until week 24 (−24.4%,p < .001).ALP levels were significantly decreased at week 4 (−5.3%, p < .01),which was maintained until week 24 (−6.9%,p < .001).On the other hand, BW decreased F I G U R E 1 Time-courses of percentage changes in levels of FPG (A), GGT (B), ALP (C), body weight (D), AST (E), and ALT (F).Observed least square mean (standard error).• placebo, • tofogliflozin.The time courses of the percent changes in FPG, GGT, ALP, body weight, AST and ALT during entire intervention were tested using a mixed effect model with treatment (tofogliflozin vs. placebo) and time (from week 4 to 24) and interaction between treatment and time as fixed effects with the baseline value as a covariate, and subject as a random effect.P value for time-treatment interaction was reported.ALP, alkaline phosphatase; ALT, alanine aminotransferase; AST, aspartate aminotransferase; FPG, fasting plasma glucose; GGT, gamma-glutamyltransferase.

| Clinical parameters that were associated with and predictive of GGT and ALT reductions at week 24 in participants receiving tofogliflozin
At week 24, reductions in both GGT and ALT were positively associated with reductions in FPG and BW (Table S5).
Because the time courses of these reductions were distinct according to the careful investigation and statistical evaluation, we analysed factors influencing these reductions at weeks 4 and 24 in participants receiving tofogliflozin using a multivariate stepwise method (Table 2, Table 3 and Table 4).The reduction at week 24 in GGT was independently associated with the reductions in FPG and BW at week 24 (Table 2), whereas the reduction at week 24 in ALT was associated with the reduction in only BW but not FPG (Table 2).In addition, we added the percent change in HOMA-IR at week 24 into the above multivariate model as a factor that might influence the reductions in GGT and ALT at week 24.The reductions of both liver enzymes were independently associated with the reduction in HOMA-IR at week 24; however, the influences of the reductions in FPG and BW for liver enzymes were maintained (Table S6).Similarly, at week 4, the reduction in GGT was associated with reductions in FPG and BW at week 4 (Table 3), whereas the overall reduction in ALT was not associated with either FPG or BW reductions (Table 3).Finally, the early reductions in FPG and BW at week 4 predicted the reduction in GGT at week 24, whereas a reduction in only BW at week 4 predicted the reduction in ALT (Table 4).

| DISCUSS ION
Although the ductal enzyme GGT and liver enzyme ALT are used clinically as severity indices of fatty liver disease, the mechanisms and time-course of their elevations remain unclear.The present study proved with the mixed-effects model for repeated measure approach that time courses of the decreases in GGT and ALT levels during the intervention to treat hyperglycaemia and obesity with the SGLT2 inhibitor, tofogliflozin, are distinct from each other.Also, the time courses of decreases in FPG levels and BW were distinct.
Multivariate analyses indicated associations of BW reductions with both GGT and ALT reductions.However, FPG reductions were associated with a greater reduction in GGT but not ALT.These findings indicate that GGT and ALT reflect distinct pathologies of T2D, that is, glycemic status and BW, respectively.GGT but not ALT reflects glycemic status whereas both enzymes reflect obesity in people with T2D.
GGT is a ubiquitous cell membrane and plasma-circulating enzyme involved in the cellular metabolism of glutathione (GSH).As GSH is a major antioxidant, GGT is considered a protective enzyme,

Regression coefficient p
Percent change in body weight at week 24 (>1%) contributing to the maintenance of the cellular redox status. 18,19[22] Hyperglycaemia leads to the generation of reactive oxygen species via NADPH oxidases. 23Based on these findings, we speculate that in the present study, GGT levels decreased during the intervention with tofogliflozin mainly via a reduction in hyperglycaemia-associated oxidative stress, which should be investigated in the future.
In the present study, baseline ALP levels were significantly higher in the participants with higher FPG levels but did not differ between those with and without obesity (Table 1).ALP reduction was significantly correlated with FPG reduction, but not BW reduction at week 24 (Table S4).Elevated serum levels of ALP were associated with endothelial dysfunction in hypertensive individuals. 24Also, it was experimentally suggested that the robust induction of vascular ALP activity by oxidative stress was related to vascular calcification. 25evated levels of ALP were shown to interact with an oxidative stress-mediated elevation in GGT and were related to mortality risk in patients with chronic kidney disease. 26These findings suggest an association between ALP and GGT through oxidative stress.On that basis, we speculate that tofogliflozin reduces ALP levels mainly via its anti-hyperglycaemic effects leading to the reduction in hyperglycaemia-mediated oxidative stress.Further investigations will be needed because there are several ALP isozymes to be considered.
Indeed, in the present study, the time-course of BAP change was similar to the ALP reduction.Adjusted for age, sex, duration of diabetes, HbA1c, FPG, HOMA-IR, HOMA-β, BMI, eGFR, waist circumference, AST, ALT, GGT, the percentage change in FPG at week 4 and the percent change in body weight at week 4.

TA B L E 4 Potential predictors influencing percentage of changes in GGT
and ALT levels at week 24 in participants receiving tofogliflozin.

Percent change in GGT levels at week 4
Stepwise variable selection was used for the analyses.Adjusted for age, sex, duration of diabetes, HbA1c, FPG, HOMA-IR, HOMA-β, BMI, eGFR, waist circumference, AST, ALT, GGT, the percentage change in FPG at week 4 and the percent change in body weight at week 4. Potential factors influencing percentage of changes in GGT and ALT levels at week 4 in participants receiving tofogliflozin.
TA B L E 3Note: Stepwise variable selection was used for the analyses.