ApoA-1 is a major protein component of HDL-C. Both ApoA-1 and HDL-C suppress the progression of arteriosclerosis by binding to excess cholesterol in the periphery and transporting it to the liver [16]. However, it should be noted that high HDL-C levels do not necessarily reduce the risk of atherosclerosis. As for HDL, it is important to be able to extract excess cholesterol from the walls of the blood vessels, which is called the cholesterol efflux capacity [17]. A cholesteryl ester transfer protein inhibitor, Evacetrapib, which substantially increases the levels of HDL-C and lower LDL-C, reportedly did not reduce cardiovascular risk and mortality and hence is not being marketed [18]. Further, impaired glucose tolerance has been reported to reduce the ability of HDL to extract cholesterol [19]. HDL-C has two subspecies, namely HDL-C 2 and HDL-C 3. Large, cholesterol-rich HDL-C 2 is inversely associated with plasma TG and insulin resistance, while small, cholesterol-poor HDL-C 3 is not [20]. HDL-C subfractions or measurement of cholesterol efflux capacity was not performed in this study. Increased levels of ApoA-1 are more strongly associated with a lower risk of coronary artery disease, compared with increased levels of HDL-C [21]. Furthermore, ApoA-1 is associated with NAFLD. A Korean study reported that patients with lower ApoA-1 levels showed a higher prevalence of NAFLD, even if they were not obese [22]. In another study on mice, the incidence of NAFLD was higher in ApoA-1 deficient mice [23]. These reports suggest that lower levels of ApoA-1 are associated with a higher incidence of arteriosclerotic diseases and NAFLD.

ApoB is a protein component of LDL-C particles, and the ApoB level represents the number of LDL-C particles [24]. A higher ApoB level is associated with a higher risk of arteriosclerotic diseases and NAFLD [25, 26]. Similarly, RLP-C is a metabolite produced during the degradation of lipoproteins, such as very low-density lipoproteins (VLDLs) or chylomicrons. Higher RLP-C levels were associated with an increased incidence of arteriosclerotic diseases and NAFLD [27, 28] and a higher TG/HDL-C ratio with a higher risk of arteriosclerotic diseases and NAFLD [29, 30].

Our findings showed significant reductions in HbA1c and fasting plasma glucose levels, indicating a potential beneficial effect of luseogliflozin on diabetes control. Similar to other reports on luseogliflozin [9, 31], this study showed reductions in BMI, body weight, and waist circumference. Although the visceral fat mass was not measured accurately, waist circumference has been reported to correlate with the visceral fat mass area [32]. No reduction in skeletal muscle mass was observed in this study. Similarly in Sasaki’s study [9], there was only a small decrease in the skeletal muscle mass index, while BMI and the total fat mass decreased significantly. We routinely explained to the patients to continue the original diet and exercise therapy but did not ask them to follow a special diet or exercise intensively after starting luseogliflozin. Luseogliflozin is reported to regulate essential amino acids, short-chain fatty acids, and intestinal bacteria which leads to an increase in skeletal muscle mass [33]. Therefore, we believe that a significant reduction in skeletal muscle mass might not occur, however, it is important to consider individual differences and carefully evaluate the patients’ body compositions.

As for lipids, HDL-C and ApoA-1 levels significantly increased, indicating a reduced risk of arteriosclerotic diseases and NAFLD. Although we did not measure the cholesterol efflux capacity of HDL, it is reported that impaired glucose tolerance reduces cholesterol efflux capacity [19]. As such, an improvement in diabetes control, likely improves the cholesterol efflux capacity of HDL, leading to a reduced risk of arteriosclerotic diseases. Additionally, TG and RLP-C levels, as well as the TG/HDL-C ratio, significantly decreased, indicating a reduced risk of arteriosclerotic diseases and NAFLD. Therefore, changes in lipid levels can potentially reduce the risk and suppress the progression of diabetic complications. No significant changes were observed in LDL-C and ApoB levels. In a study by Ejiri et al. [34] using luseogliflozin in patients with diabetes and heart failure with baseline lipid profiles at normal range, no significant changes were observed in small dense LDL-C, TG, or LDL-C levels. In a retrospective analysis by Kubota et al. [35], 63 patients with T2DM on usual doses of SGLT2-i (Empagliflozin, Canagliflozin, Dapagliflozin, or Tofogliflozin), showed a significant decrease in total cholesterol and non-HDL-C levels, but no changes in TG, HDL, or LDL-C levels. However, this study was limited owing to its observational nature and use of multiple SGLT2-i. In another study of dapagliflozin, the LDL-C level showed no significant change. However, subsequent research revealed that the amount of small dense LDL-C, the main risk factor for arteriosclerosis among LDL particles, was sufficiently suppressed [36]. In our study, changes in the levels of small dense LDL-C could not be determined, as this assessment was not covered by insurance and therefore not measured. However, several surrogate markers, such as LDL-C/ApoB [37] and TG/ApoB [38] ratios, could indicate and identify the changes in small dense LDL-C. Hence, we evaluated these markers and compared them with those of the dapagliflozin study which showed a significant decrease in small dense LDL-C [36]. Surprisingly, the changes in these markers were similar to those in the dapagliflozin study. In our study, the LDL-C/ApoB ratio showed a change from 1.1 to 1.1 (p = 0.373), similar to that in the dapagliflozin study (change from 1.2 to 1.2, p value is not available). The TG/ApoB ratio showed a change from 2.4 to 1.9 (p = 0.007), similar to that in the dapagliflozin study (change from 1.5 to 1.3, p value is not available) (Additional file 3). Since the dapagliflozin study reported a significant reduction of small dense LDL-C, we recognized that the LDL-C/ApoB ratio is not accurate enough. On the other hand, the dapagliflozin study reported a significant reduction of the TG/ApoB ratio which was similar to our study that also showed a significant reduction of TG/ApoB, which likely meant reduction in the small dense LDL-C.

For assessing the liver function, FLI was measured to determine the risk of fatty liver, and the FIB-4 index and type IV collagen level were measured to screen for fibrosis in the liver. In our study, FLI was reduced significantly from 68.0 to 53.4 after 24 weeks (p < 0.01), whereas no significant changes were observed in fibrosis markers, which was similar to the findings of Sumida et al. at 24 weeks [31], however other studies with 52 weeks of follow-up showed a significant decrease [39, 40]. These differences could be due to the difference in the sample size and observation period. Changes in the FIB-4 index or type IV collagen level may be observed with even longer follow-ups or in patients with more severe liver dysfunction with fibrosis. Although the presence of fatty liver was not evaluated using images in this study, fatty liver was likely suppressed based on the FLI changes; indeed, a previous study reported a reduction in intrahepatic fat content following luseogliflozin administration in patients with T2DM and NAFLD [31]. Additionally, the ALT and γ-GTP levels were significantly decreased. A meta-analysis that aimed to evaluate the effect of SGLT2-i on NAFLD based on the changes in liver enzyme levels and liver fat volume concluded that SGLT2-i can significantly decrease ALT levels and liver fat, accompanied by weight loss [41]. Another study on Ipragliflozin in patients with diabetes and NAFLD also reported a significant decrease in the ALT and γ-GTP levels, along with the pathological improvement of NAFLD [42]; thus, luseogliflozin might reduce the risk of and improve NAFLD. Improvement in liver enzyme levels was observed, even though the study participants did not show severe liver dysfunction. Therefore, we deduced that earlier administration of SGLT2-i can suppress NAFLD progression over a period of 24 weeks.

While assessing the renal function, red blood cell count and hematocrit values increased significantly, which might not be due to dehydration, but due to the fibroblast-like cells regaining their ability to produce erythropoietin, because of the reduction in excessive glucose reabsorption in the renal tubules [43]. A decrease in the UACR was observed in this study, though statistically insignificant. Kubota et al. reported that UACR levels were significantly reduced only in the microalbuminuria and overt albuminuria groups, especially in patients with decreased body weight, blood pressure, and blood glucose [35]. As for luseogliflozin, a study reported a significant decrease in ACR [9], but two other studies did not, similar to our study [44, 45]. These differences could be probably due to variations in the sample size of the participants and the renal stages. Overall, luseogliflozin might have protective effects on the kidney and renal tubules.

Improvements in lipid profile and liver function may be attributed to many possible mechanisms. First, hyperglycemia and hyperinsulinemia cause fat accumulation in the liver, enlargement of adipose cells, and fat deposition in muscles, which could lead to insulin resistance [46]. When hyperglycemia and hyperinsulinemia are alleviated by SGLT2-i administration, these changes are alleviated, which eventually leads to the improvement of obesity and fatty liver. Our study also reports that indicators related to NAFLD, including FLI and ALT, are likely to decrease in patients with higher baseline serum C-peptide levels or body fat percentage, which means that patients with higher insulin resistance can benefit from SGLT2-i. Second, β-oxidation in the liver is also activated by SGLT2-i [4], leading to reduced production of VLDLs or RLP-C. Third, peroxisome proliferator-activated receptor alpha (PPAR-α), activated when SGLT2-i is administered [47], increases the production of ApoA-1, which eventually leads to the suppression and improvement of NAFLD [48].

Multivariate analysis in our study also revealed that the changes in the TG level were most closely correlated to the FLI change. Previous studies have shown that hypertriglyceridemia is one of the risk factors for NAFLD [49] and that the prevalence and degree of hypertriglyceridemia significantly correlate with the severity of NAFLD [50, 51]. It is known that the accumulation of triglycerides in hepatocytes increases fatty acid β-oxidation. When this occurs in the presence of mitochondrial abnormalities, it leads to a free radical formation with consequent cell injury, inflammation, and fibrosis [52]. Our study findings indicate that reducing TG levels is the key to improve liver function.

This study had several limitations. First, is the small sample size of 25 patients, and a shorter study period of 24 weeks. A larger and longer study is required to further investigate the durability of the changes. Second, the severity of liver dysfunction in the participants was mild, which should be taken into consideration; more changes might occur with a larger and longer-term evaluation, especially in patients with higher levels of liver enzymes. Changes in the measures for liver fibrosis, such as the FIB-4 index or type IV collagen, might need a longer follow-up of approximately 1 year. Third, levels of HDL-C, HDL-C subspecies (HDL-C 2 and HDL-C 3), or cholesterol efflux capacity were not measured in this study, which is required to precisely assess the anti-arteriosclerotic function of HDL. Fourth, we evaluated only the markers of fatty liver, but not the fatty liver itself. Even if the markers of fatty liver are improved, it is not clear whether the fatty liver is actually improved, as abdominal ultrasonography was not performed in this study. Furthermore, the most robust approach for diagnosing NAFLD needs the evaluation of TG accumulation via magnetic resonance imaging or vibration‐controlled transient elastography. Fifth, we did not perform a detailed lipoprotein analysis to clarify whether TG decline is mainly due to chylomicrons or VLDLs. TG-rich lipoproteins generally include VLDL and chylomicrons, but chylomicrons are less likely to cause arteriosclerosis unless they form remnants. A complete lipoprotein evaluation should be performed for further information. Sixth, the study did not measure the levels of small dense LDL-C as it was not covered by insurance; therefore, potential changes caused due to LDL-C remained undetected. Seventh, serum insulin levels were not measured. The measurement of serum insulin levels in patients who were not administered insulin might provide more information regarding the changes in insulin resistance. Lastly, although the dosage of drugs for diabetes or dyslipidemia did not change a month prior to or during the study, the potential effects caused by traditional treatments should be considered.