Effects of Canagliflozin on Fatty Liver Indexes in Patients with Type 2 Diabetes : A Meta-analysis of Randomized Controlled Trials

PURPOSE: Non-alcoholic fatty liver disease (NAFLD) affects about 75% of patients with type 2 diabetes mellitus (T2DM). We conducted a meta-analysis to determine the effect of canagliflozin on fatty liver indexes in T2DM patients. METHODS: A literature search of PubMed, Embase and Cochrane was conducted up to March 30, 2017. The liver function test and lipid profile were extracted from randomized controlled trials (RCTs) to evaluate the effect of canagliflozin on fatty liver. Weighted mean differences (WMDs) or relative risks and 95% confidence intervals (CIs) were computed by using either fixed or randomeffects models. Sensitivity analysis and publication bias were evaluated. RESULTS: Our results showed that canagliflozin decreased serum concentrations of alanine amino transferase (WMD: -11.68 [95% CI: -18.95, -10.95]; P<0.001), aspartate amino transferase (WMD: -7.50 [95% CI: -10.61, -4.38]; P<0.001), gammaglutamyl transferase (WMD: -15.17 [95% CI: -17.73, -12.61]; P<0.001), triglycerides (WMD: -0.10 [95% CI: -0.15, -0.05]; P<0.001) but increased low-density lipoprotein cholesterol (WMD: 0.1 [95% CI: 0.06, 0.13]; P<0.001), high-density lipoprotein cholesterol (WMD: 0.06 [95% CI: 0.05, 0.07]; P<0.001) at week 26 or 52. CONCLUSIONS: Our results indicated that canagliflozin may have a protective effect on fatty liver in T2DM patients. The limitation was that the liver biopsy was hard to obtain in published studies. More RCTs specified on NAFLD are needed to get further information. This article is open to POST-PUBLICATION REVIEW. Registered readers (see “For Readers”) may comment by clicking on ABSTRACT on the issue’s contents page. ____________________________________________________________________


INTRODUCTION
Type 2 diabetes mellitus (T2DM) is a chronic and progressive metabolic disease that is associated with comorbidities, including non-alcoholic fatty liver disease (NAFLD) (1).NAFLD shared some pathogenetic requisites with T2DM, such as obesity and insulin resistance (2), and affects about 75% of patients with T2DM (3).The prognosis for patients with concomitant NAFLD and T2DM is worsened due to increased risk for life-threatening sequela such as cardiovascular disease and hepatocellular carcinoma (4).Therefore, antidiabetic drugs which have effect on improving NAFLD would be beneficial and suitable for T2DM patients with NAFLD.
Canagliflozin is a sodium glucose co-transporter 2 (SGLT2) inhibitor developed for the treatment of adults with T2DM (5).Canagliflozin promotes urinary glucose excretion, resulting in decreased plasma glucose, a mild osmotic diuresis and a net caloric loss (6,7).Canagliflozin provides improvements in glycosylated hemoglobin, body weight and systolic blood pressure, and is generally well tolerated.Studies which assessed the effects of SGLT2 inhibitors on hepatic steatosis suggested the potential application of this class for the treatment of NAFLD (8,9).
Several previous research suggeted that canagliflozin might benefit NAFLD.Shiba K et al. found that canagliflozin attenuated the development of hepatocellular carcinoma in a mouse model of human non-alcoholic steatohepatitis (10).Seko Y et al. conducted a retrospective study and found SGLT2 inhibitors significantly decreased the transaminase activities in Japanese patients with NAFLD and T2DM (11).Takase T et al. conducted an observational study in Japanese patients with T2DM, and found that ipragliflozin significantly decreased body mass index, waist circumference, gamma-glutamyl transferase and triglycerides (12)._____________________________________________ ABBREVIATIONS: NAFLD= non-alcoholic fatty liver disease; T2DM= type 2 diabetes mellitus; WMD= weighted mean difference; CI=95% confidence interval; ALT= alanine amino transferase; AST=aspartate amino transferase; GGT=gamma-glutamyl transferase; ALP= alkaline phosphatase; TG= triglycerides; HDL=high-density lipoprotein; LDL= lowdensity lipoprotein; SGLT2=sodium glucose cotransporter 2; RCT= randomized controlled trials; RR = relative risk However, these studies are not randomized controlled trials and mainly Japanese population.Currect meta-analysis associated with SGLT2 inhibitors only mentioned lipid or alanine amino transferase (ALT) and didn't take them as main outcomes (13)(14)(15).So we conducted a meta-analysis to evaluate the effect of canagliflozin on NAFLD through liver function and lipid profile.

Search strategy and selection of articles
We searched PubMed, EMBASE, Cochrane Central Register of Controlled Trials (CENTRAL) and ClinicalTrials.govdatabases up to March 30, 2017 to identify eligible randomized controlled trials (RCTs) using keyword combinations of ("Sodium glucose co-transporter" OR SGLT2 OR SGLT-2 OR "SGLT 2" OR Tofogliflozin OR Apleway OR Deberza OR CSG452 OR Empagliflozin OR Jardiance OR dapagliflozin OR Farxiga OR Forxiga OR Canagliflozin OR Invokana OR Sotagliflozin OR LX4211 OR luseogliflozin OR Lusefi OR Ipragliflozin OR Suglat OR remogliflozin OR BHV091009 OR sergliflozin OR GW869682X OR ertugliflozin OR MK-8835 OR PF-04971729) AND (RCT OR random).Only human studies were included.Two reviewers (Li B, Wang Y) independently screened titles and/or abstracts for relevance followed by full-text article assessments for inclusion.Studies were included if: (1) The participants were non-pregnant adults (aged over 18 years) with T2DM.(2) The treatment intervention was canagliflozin monotherapy or combination therapy with any approved agent or not.(3) The study design was randomized, double-blind, placebo-controlled, or active-controlled, parallelgroup study.Articles were excluded if they were letters, editorials, conference abstracts, reviews, and commentaries.For multiple publications in the same RCT, only the article with the most comprehensive data was included.Searching results are depicted in Figure 1.

Quality assessment of the trials
The quality of RCTs was assessed with the Cochrane risk of bias tool, which is the recommended approach for assessing the risk of bias in studies included in Cochrane reviews.This tool assesses the risk of bias in 2 parts, addressing the following specific domains: sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective outcome reporting, and other issues (16).

Data extraction
Two reviewers (Li B, Wang Y) independently extracted relevant information for the meta-analysis.The extracted data included the characteristics of each study (author, year, study design, treatment, mean age, race, mean glycosylated hemoglobin and follow-up time), and clinical outcomes (change percentage of alanine amino transferase (ALT), aspartate amino transferase (AST), gamma-glutamyl transferase (GGT), alkaline phosphatase (ALP), triglycerides (TG), high-density lipoprotein (HDL), low-density lipoprotein (LDL), LDL/HDL, non-HDL cholesterol) of the control and canagliflozin groups in each study.Two authors separately performed data extraction.The differences were discussed and resolved.

Data synthesis and statistical analysis
To synthesize the efficacy outcomes, Review Manager 5 (The Cochrane Collaboration, Oxford, UK) was used to calculate the estimates and 95% confidence intervals (CIs) of the weighted mean differences (WMDs) between the intervention group (canagliflozin 100 or 300mg daily) and the control group for quantitative variables and relative risks (RRs) for categorical variables, using either fixed or random effects models with an inverse variance method.P values less than 0.05 were considered statistically significant.Heterogeneity among the trials was assessed using the χ 2 test defined as a P value less than 0.10 and was further quantified through the I 2 statistics.In order to evaluate the stability of results without estimation bias from individual study, sensitivity analysis was performed by exclusion of each study one by one.This process of excluding one study at a time allowed for identification of any single article that might have a large influence on the final results.Publication bias was evaluated using the funnel plot method.

Literature search and characteristics of the included studies
The search strategy initially identified 4434 articles.After selection, 11 randomized, double-blind, placebo-controlled or active-controlled, parallelgroup trials met the selection criteria, with a total enrollment of 6745 patients with T2DM.The whole literature search process was summarized in Figure 1 .61]for the total (P<0.001)(Fig.2c).

DISCUSSION
The main findings of this meta-analysis included two parts.First, canagliflozin significanltly decreased serum concentrations of ALT, AST and GGT at week 26 and 52, indicating it might have a protective effect on liver.Second, canagliflozin reduced TG but increased LDL-C and HDL-C levels at week 26 and 52 ， which was consistant with previous meta-analysis, but a little confusing.There is no doubt that fasting plasma TG was pravently investigated and tightly related with NAFLD (17,18).While no benefits or harm of Statins were observed on liver disease although they are confidently used to reduce LDL-cholesterol and prevent cardiovascular risk (19).So the reduced TG level of our meta-analysis indicated canagliflozin might be helpful to NAFLD.How did canagliflozin affect LDL and HDL cholesterol level?In a study of hamsters with dietinduced dyslipidemia, Briand F et al. found empagliflozin moderately increased ketone production and LDL cholesterol levels by switching energy metabolism from carbohydrate to lipid utilization.The catabolism of (3)H-cholesteryl oleate-labeled LDL cholesterol injected intravenously was significantly reduced by 20%, indicating that empagliflozin reduced intestinal cholesterol absorption (20).Canagliflozin may raise LDL cholesterol levels through the same mechanism with empagliflozin, which are the reduced catabolism and reduced intestinal absorption.
What kind of LDL and HDL cholesterol subspecies did SGLT-2 inhibitors affect?Hayashi T et al. conducted a single center, open-label, randomized, prospective study in human to determine how SGLT-2 inhibitors affect LDL and HDL cholesterol subspecies.They found that dapagliflozin suppresses potent atherogenic small dense LDL cholesterol and increased HDL2 cholesterol, a favorable cardiometabolic marker.In their opinion, the elevated level of LDL cholesterol levels after treatment with dapagliflozin was due to increased concentrations of the less atherogenic large buoyant LDL cholesterol (21).
The results of our meta-analysis were consistent with the previous three meta-analysis, two of which only analyzed lipid change after canagliflozin treatment (13,14) and the third one analyzed lipid and ALT only (15).Compared to these studies, the advantage of our study was that: 1) We focused on the effect of canagliflozin on the fatty liver indexes; 2) We made an analysis on the reason why LDL and HDL cholesterol were elevated and the meaning of this change to cardiovascular risk.It is well known that meta-analysis has certain unavoidable limitations.Although we had limited this analysis to well designed RCTs and performed quality assessment to reduce the possible selective bias, the present meta-analysis still had several potential limitations.First, this meta-analysis compared canagliflozin with placebo and other active antidiabetic drugs because the active controlled trials were so few to conduct a metaanalysis.Second, the data of liver biopsy, which is essential for the diagnosis of non-alcoholic fatty liver disease/non-alcoholic steatohepatitis (22), is insufficient.It is hard to obtain in current pulished studies.

CONCLUSION
We conduct a meta-analysis on canagliflozin effect on fatty liver indexes in T2DM patients.Our results showed canagliflozin decreased serum concentrations of ALT, AST, GGT, TG but increased LDL and HDL cholesterol levels at week 26 or 52.Our results indicated that canagliflozin may have a protective effect on fatty liver.The limitation was that liver biopsy was hard to obtain.More RCTs specified on NAFLD are expected to make further conclusion.

Figure 1 .
Figure 1.Summary of study identification, inclusion, and exclusion.

Figure 2a .
Figure 2a.Forest plot depicting the ALT level with canagliflozin versus control group.

Figure 2b .
Figure 2b.Forest plot depicting the AST level with canagliflozin versus control group.

Figure 2c .
Figure 2c.Forest plot depicting the GGT level with canagliflozin versus control group.

Figure 3a .
Figure 3a.Forest plot depicting the TG level with canagliflozin versus control group.

Figure 3b .
Figure 3b.Forest plot depicting LDL-C with canagliflozin versus control group.

Figure 3c .
Figure 3c.Forest plot depicting HDL-C with canagliflozin versus control group.

Table 2 .
Baseline liver function and lipid levels of included studies