Second-line antidiabetic drugs: friend or foe of the liver.

INTRODUCTION
Diabetes mellitus (DM) is a modifiable risk factor associated with advanced liver disease, including cirrhosis, hepatic decompensation, and liver-related death. Recently, with the growing interest in metabolic dysfunction-associated steatotic liver disease (MASLD), the importance of DM as a cardiometabolic factor has been increasingly emphasized.1 Insulin resistance, the primary pathophysiological mechanism of DM, promotes hepatic inflammation and fibrosis through an increase in mitochondrial oxidative stress. Furthermore, it is also implicated in hyperglycemia-induced oncogenic process, contributing to the development of hepatocellular carcinoma (HCC).2-4 Therefore, optimal antidiabetic treatment is crucial for preventing the progression of chronic liver disease to advanced stages in diabetic patients.
Metformin, the first-line antidiabetic agent, has been demonstrated to confer protective effects against HCC and overall mortality in patients with DM compared to non-users.5,6 Based on this knowledge, the recently updated guidelines from the American Association for the Study of Liver Disease suggest that metformin use for the prevention of HCC in patients with chronic liver disease and DM.7 However, due to inconsistencies across studies regarding its effect on the risk of hepatic decompensation, HCC, and liver-related mortality, this recommendation remains weak.8-10 Novel antidiabetic agents -glucagon-like peptide-1 receptor agonists (GLP-1 RA), sodium-glucose cotransporter- 2 inhibitors (SGLT-2i), and dipeptidyl peptidase-4 inhibitors (DPP-4i)- are now recommended as second or third-line treatments when metformin alone fails to achieve adequate glycemic control.11 The use of these new antidiabetic agents has been shown to improve hepatic steatosis and inflammation in patients with steatotic liver disease (SLD).12-14 Moreover, based on clinical trial evidence demonstrating cardiovascular benefits in patients with DM, the American College of Cardiology and the American Diabetes Association recommend GLP-1 RAs and SGLT-2i as treatment options for diabetic patients with cardiovascular disease.11,15 However, clinical studies evaluating the protective effects of novel antidiabetic agents on mortality and liver-related outcomes, including HCC, cirrhosis, and hepatic decompensation (Supplementary Table 1), remain limited. Moreover, there are no standardized treatment guidelines for diabetic patients with liver disease and heterogeneity in study populations across studies has led to inconsistent results.16-21 Therefore, this review aimed to provide a comprehensive summary of liver-related prognosis associated with these new second-line antidiabetic agents in diabetic patients.

GLUCAGON-LIKE PEPTIDE-1 RECEPTOR AGONISTS
GLP-1 RAs indirectly affect the liver by influencing hepatic insulin resistance, peripheral plasma insulin and glucose levels, and reducing lipotoxicity (Fig. 1).22 They enhance insulin secretion and inhibit glucagon release, leading to a reduction in de novo lipogenesis and triglyceride secretion, decreased gluconeogenesis, and improved glucose uptake.23 These pleiotropic effects of GLP-1 RAs improve hepatic inflammation and steatosis, highlighting their benefits beyond glycemic control in diabetic patients.24 The use of GLP-1 RAs in diabetic patients has been reported to exert a protective effect against mortality and liver-related outcomes, including cirrhosis, hepatic decompensation, and HCC.

GLP-1 RA and risk of HCC
Several nationwide observational studies have demonstrated an association between GLP-1 RA use and reduced HCC risk in patients with type 2 diabetes mellitus (T2DM) (Table 1). In a large-scale study, Wang et al.16 demonstrated significant chemopreventive effects of GLP-1 RAs across various comparator groups (insulin, sulfonylurea [SU], metformin, DPP-4i, SGLT-2i, thiazolidinedione [TZD]) and stages of SLD. Among all DM patients, GLP-1 RA users showed chemopreventive effects against HCC compared to users of insulin (adjusted hazard ratio [aHR], 0.20; 95% confidence interval [CI], 0.14-0.31), SU (aHR, 0.78; 95% CI, 0.65-0.93), metformin (aHR, 0.99; 95% CI, 0.79-1.27), and DPP-4i (aHR, 0.89; 95% CI, 0.76-1.05). The protective association was particularly evident in patients with prior MASLD, steatohepatitis, or cirrhosis, and remained consistent when comparing combination therapies with monotherapy regimens, suggesting a possible independent effect of GLP-1 RAs. This study showed that GLP-1 RAs have chemopreventive effects regardless of SLD severity, though it remains unclear whether this applies to T2DM patients without pre-existing liver disease. Additionally, the proportion of baseline liver disease in each comparator group was not reported, limiting assessment of potential confounding despite propensity score matching (Supplementary Table 2).
However, in patients without pre-existing liver disease, GLP-1 RA use did not significantly reduce HCC risk in the study of Yen et al.25 Although the underlying reason for this absence of significant difference was not explicitly discussed, the relatively short follow-up duration (median, 2.1 years) may have contributed to the null result. Conversely, Yang et al.26 reported significantly lower HCC risk in GLP-1 RA users compared to long-acting insulin users (subdistribution HR, 0.47; 95% CI, 0.24-0.93). This finding supports possible anti-inflammatory and antifibrotic effects, although exclusion of underlying liver disease was not clearly stated. Rashid et al.27 focusing on older adults with alcohol-associated liver disease (ALD), found no statistically significant reduction in HCC risk between GLP-1 RA and non-GLP-1 RA users (aHR, 0.32; 95% CI, 0.06-1.63). This suggests a potentially limited role for GLP-1 RAs in non-metabolic liver injury, as the predominant injury pathway in ALD involves oxidative stress, acetaldehyde toxicity, and inflammation-pathways that may not be as responsive to the metabolic improvements induced by GLP-1 RAs. Similarly, studies by Yen et al.18 and Kanwal et al.28 in diabetic patients with cirrhosis showed trends toward reduced HCC risk but lacked statistical significance, likely due to small numbers of HCC events (40 and 24 cases, respectively) and limited follow-up duration (mean, 3.28; mean, 2.96).
In summary, most large-scale studies support a potential chemopreventive effect of GLP-1 RAs in T2DM patients, particularly those with underlying MASLD or fibrosis. However, heterogeneity in comparator drugs, baseline liver status, and study design limits causal inference, highlighting the need for prospective studies with longer follow-up periods.

GLP-1 RAs and risk of cirrhosis and hepatic decompensation
GLP-1 RAs have shown protective effects against cirrhosis and hepatic decompensation in patients with T2DM (Table 1), although definitions of these outcomes and comparator drug classes vary across studies (Supplementary Table 1).
Regarding cirrhosis risk, Yang et al.26 reported lower risk with GLP-1 RAs compared to long-acting insulin in T2DM patients (aHR, 0.59; 95% CI, 0.43-0.81), while Kanwal et al.28 studying T2DM patients with MASLD (cohort 1), found lower cirrhosis risk compared to DPP-4i (aHR, 0.86; 95% CI, 0.75-0.98). In contrast, Yang et al.21 observed a higher, but non-significant, cirrhosis risk compared to TZDs among older T2DM patients without cirrhosis (aHR, 1.34; 95% CI, 0.82-2.20).
For hepatic decompensation, multiple studies demonstrated reduced risk with GLP-1 RA use. Wang et al.16 reported benefits particularly in patients without underlying SLD, and Rashid et al.27 found that GLP-1 RAs significantly reduced hepatic decompensation risk in older T2DM patients with ALD compared to non-GLP-1 RAs (aHR, 0.56; 95% CI, 0.36-0.86). In advanced liver disease settings, GLP-1 RAs showed either comparable or superior outcomes versus SGLT-2i, DPP-4i, or SU. Simon et al.29 reported comparable risk between GLP-1 RA and SGLT-2i users in T2DM patients with cirrhosis, but GLP-1 RAs showed lower risk compared to DPP-4i or SU. Similarly, Yen et al.18 and Elsaid et al.30 found that GLP-1 RAs significantly reduced hepatic decompensation risk compared to non-GLP-1 RA antidiabetic agents. However, Kanwal et al.28 found no significant difference compared to DPP-4i users in patients with MASLD cirrhosis (cohort 2). These inconsistent findings likely reflect heterogeneity in study populations, largely influenced by variations in underlying liver disease status and comparator drug classes.

GLP-1 RAs and risk of mortality outcomes
Several observational studies have linked GLP-1 RA use to reduced all-cause and cause-specific mortality in T2DM patients (Table 1). In patients with MASLD but without cirrhosis (cohort 1).28, GLP-1 RAs showed a trend toward lower overall mortality compared to DPP-4i users, though this did not reach statistical significance (aHR, 0.89; 95% CI, 0.81-0.98).28, although no significant benefit was seen in those with cirrhosis (cohort 2). Similarly, Krishnan et al.31 reported comparable overall mortality risk between GLP-1 RA and SGLT-2i users (aHR, 1.06; 95% CI, 0.97-1.15).
Yen et al.25 demonstrated significant reductions in overall (aHR, 0.48; 95% CI, 0.43-0.53), cardiovascular (aHR, 0.57; 95% CI, 0.45-0.72), and liver-related mortality (aHR, 0.32; 95% CI, 0.13-0.75) among GLP-1 RA users compared to non-GLP-1 RA users. For patients with cirrhosis, Yen et al.18 also demonstrated significantly lower overall mortality risk compared to non-GLP-1 RA users (aHR, 0.47; 95% CI, 0.32-0.69). However, the causal mechanisms underlying the association between GLP-1 RAs and reduced mortality outcomes remain unclear. Moreover, evidence regarding liver-specific mortality is limited, underscoring the need for further prospective investigation.

SODIUM-GLUCOSE COTRANSPORTER-2 INHIBITOR
SGLT-2i acts on the proximal renal tubules, causing glycosuria and thus leading to a reduction in hemoglobin A1c by average 0.6-1.2% independently of use of insulin (Fig. 2).32 The US Food and Drug Administration has approved four SGLT-2i -canagliflozin, dapagliflozin, empagliflozin, and ertugliflozin- as an adjunct to diet and exercise to improve glycemic control in adults with DM. Beyond glycemic control, SGLT-2i have been reported to improve hepatic steatosis and inflammation, thereby reducing the risk of progression to cirrhosis.32 Furthermore, SGLT-2i have the potential to improve portal hypertension through mechanisms similar to those observed in heart failure.33 They promote glycosuria and modulate the renin-angiotensin-aldosterone system (RAAS) by suppressing sodium reabsorption in the proximal tubule, leading to reduced renin secretion and attenuated RAAS activity.34 Theoretically, these mechanisms support the use of SGLT-2i in diabetic patients with cirrhosis. Due to the hepatoprotective effects of SGLT-2i, interest in their potential benefits beyond glucose-lowering has been steadily increasing. Table 2 summarizes clinical studies evaluating the association of SGLT-2i use and liver-related, and mortality outcomes.

SGLT-2i and risk of HCC
SGLT-2i have shown potential chemopreventive effects against HCC in patients with T2DM. Bea et al.35 demonstrated lower HCC risk with SGLT-2i compared to DPP-4i in the general T2DM population (weighted HR, 0.81), particularly among younger patients (age, <50 years), females, patients without a history of cirrhosis or viral hepatitis (hepatitis B virus [HBV] or hepatitis C virus [HCV]), and those with a fatty liver index ≥60. Bea et al.36 subsequently extended the study period and focused on older patients (age, ≥40 years) with T2DM and MASLD, comparing HCC risk between SGLT-2i users and those using GLP-1 RAs or TZDs. Compared to TZD and GLP-1 RA users, SGLT-2i users showed reduced HCC risk (vs. TZD aHR, 0.91; vs. GLP-1 RA aHR, 0.75). However, statistical significance was lost when patients with impaired liver function were excluded (vs. TZD aHR, 1.02; vs. GLP-1 RA aHR, 1.12). This suggests that the chemopreventive effect of SGLT-2i is primarily mediated through metabolic mechanisms, including improved fluid balance, enhanced hepatic lipolysis via glucagon secretion, and reduced hepatic steatosis and fibrosis.
The beneficial effects of SGLT-2i on HCC risk remain consistent across other studies involving patients with T2DM and MASLD.37-40 In a US population-based study, Mao et al.38 found that SGLT-2i use reduced HCC risk in patients with diabetic MASLD (HR, 0.76), with synergistic effects observed when combined with metformin (HRs, 0.58-0.79). Mendelian randomization analyses from European cohorts and a Korean nationwide study.37 further supported these findings, demonstrating significantly lower HCC risk with SGLT-2i compared to DPP-4i (aHR, 0.68; P=0.02). Conversely, Kawaguchi et al.39 and Shen et al.40 showed no significant differences in HCC risk between SGLT-2i and DPP-4i users.
SGLT-2i demonstrates chemopreventive effects against HCC even in diabetic patients with chronic viral liver diseases. Lee et al.41 and Cho et al.19 showed chemopreventive effects of SGLT-2i against HCC compared to non-SGLT-2i users in patients with chronic hepatitis B (CHB) and chronic viral hepatitis, respectively (HR, 0.54; aHR, 0.43; P<0.05). In contrast, Bea et al.35 did not confirm these protective effects in viral hepatitis subgroups (HBV HR, 1.06; HCV HR, 1.42). These findings suggest that the chemopreventive effects of SGLT-2i against HCC may vary depending not only on the underlying liver disease but also on other metabolic factors influencing HCC development. However, due to population heterogeneity across studies, these findings should be interpreted with caution.

SGLT-2i and risk of cirrhosis and hepatic decompensation
The hepatoprotective effects of SGLT-2i have been primarily studied in MASLD populations. In Chung et al.37, SGLT-2i users showed significantly lower risks of cirrhosis (aHR, 0.86; P=0.004) and cirrhosis-related complications (aHR, 0.88; P=0.03) compared to DPP-4i users. In another Korean study, Bea et al.36 found that SGLT-2i reduced hepatic decompensation risk compared to TZD (aHR, 0.77) but showed similar risk when compared to GLP-1 RA (aHR, 0.93). Recently, Yen et al.42 evaluated the protective effects of SGLT-2i against overall cirrhosis, decompensated cirrhosis, and liver failure in T2DM patients without underlying liver diseases beyond SLD. SGLT-2i users demonstrated significantly lower risks of both compensated and decompensated cirrhosis (overall aHR, 0.69-0.82; decompensated aHR, 0.70-0.80) and liver failure (aHR, 0.68- 0.78) compared to other antidiabetic drugs (DPP-4i, GLP-1 RA, pioglitazone, and SU). However, findings from Western populations have been less consistent. Shen et al.40 found no significant differences in cirrhosis and hepatic decompensation risks compared to DPP-4i users, while Yang et al.21 reported mixed results depending on the comparators.
The hepatoprotective effects of SGLT-2i against hepatic decompensation also varied according to the presence of underlying cirrhosis. Saffo et al.43 showed no benefit for ascites risk in patients with cirrhosis compared to DPP-4i users. In contrast, Bea et al.36 suggested greater benefits in non-cirrhotic patients, with effects varying by comparator drug class (TZD vs. GLP-1 RA). These discrepancies likely stem from heterogeneity in patient demographics, disease severity, adjusted covariates (Supplementary Table 2), and outcome definitions, notably, Saffo et al.43 considered only ascites as an endpoint (Supplementary Table 1).

SGLT-2i and risk of mortality outcomes
The mortality benefits of SGLT-2i are supported by both randomized controlled trials (RCTs) and observational data. In a Bayesian meta-analysis of 53 RCTs, empagliflozin, canagliflozin, and dapagliflozin showed reductions in overall and cardiovascular mortality, with empagliflozin demonstrating the strongest effect (relative risk ratio, 0.79 and 0.78, respectively) using placebo as reference.44 Real-world studies similarly support mortality reduction. Chou et al.45 found that SGLT-2i users had significantly lower risks of overall mortality (aHR, 0.30) and cancer-related mortality (aHR, 0.31; P<0.001) compared to DPP-4i users. Similarly, Bea et al.46 demonstrated that SGLT-2i users were associated with lower risks of overall mortality (aHR, 0.63) and liver-related mortality (aHR, 0.64) compared to DPP-4i users. However, Chung et al.37 did not observe differences in liver-related mortality between SGLT-2i and DPP-4i users, despite utilizing the same population-based cohort as Bea et al.46 These inconsistencies may be driven by differences in comparator drug classes, adjusted covariates, and follow-up duration (median, 4.9 years vs. 266.6 days) across studies. In Yen et al.42, SGLT-2i users had lower liver-related mortality than those on DPP-4i (aHR, 0.53) or GLP-1 RA (aHR, 0.46), but not versus pioglitazone or SU. Additionally, SGLT-2i users had significantly lower overall mortality risk compared to all comparator drug classes (aHR, 0.50-0.60), although there was no significant difference in liver-related mortality compared to the non- SGLT-2i group.
Overall, although SGLT-2i consistently reduce overall mortality, findings on liver-related mortality remain inconsistent. These discrepancies may be explained by population heterogeneity, limited numbers of events, and methodological limitations of retrospective studies, including diagnostic misclassification and short follow-up periods.

DIPEPTIDYL PEPTIDASE-4 INHIBITORS
DPP-4 exhibits various physiological properties, including the promotion of lipid accumulation, inactivation of incretins (GLP-1 and GLP-2), resistance to anti-cancer drugs, and involvement in extracellular matrix binding and degradation (Fig. 3). It is also implicated in immunological processes, such as T cell proliferation, globulin production, and the secretion of interleukin-2 and interferons. Increased DPP-4 expression is also found in liver tissues and serum from rats and humans with either cirrhosis or HCC.47,48 The human liver stem cells express DPP-4, but not CD34 and CD45, which are hematopoietic stem cell and endothelial progenitor cell markers.49 This indicates that DPP-4 may be involved in the regeneration in chronically inflamed liver. Furthermore, DPP-4 inhibition suppresses tyrosine kinase in human hepatoma cells, resulting in anti-apoptotic effects.50 Through this mechanism, DPP-4 inhibitors not only improve glucose intolerance and hepatic steatosis but may also contribute to the suppression of hepatic fibrosis and hepatocarcinogenesis. Table 3 summarizes clinical studies investigating effects of DPP-4i on liver-related outcomes and mortality.

DPP-4i and risk of HCC
The chemo-preventive potential of DPP-4i against HCC has been less frequently studied, as these agents are typically used as comparator treatments rather than primary exposures. Moreover, existing studies report conflicting results regarding their protective effects.
Several studies found no significant difference in HCC risk between DPP-4i users and non-users. Yen et al.20 and Na et al.51 both demonstrated comparable HCC incidence rates between these groups. However, when compared to TZD users, DPP-4i users showed modestly increased HCC risk (aHR, 1.12; P=0.04), particularly among long-term users. In contrast, studies focusing on patients with viral hepatitis reported protective effects. Hsu et al.52 and Chen et al.53 found that DPP-4i users had significantly reduced HCC risk compared to non-users among T2DM patients with concurrent viral hepatitis. Hsu et al.52 demonstrated a dose- and duration-dependent protective effect, proposing that DPP-4i may counteract the upregulated DPP-4 expression observed in hepatocytes of hepatitis B or C patients, thereby modulating C-X-C motif chemokine ligand 10 activity and potentially reducing viral-mediated hepatic inflammation.48,52 These contradictory findings suggest that the hepatoprotective effects of DPP-4i may be context-dependent, particularly influenced by underlying liver disease etiology and patient population characteristics.

DPP-4i and risk of cirrhosis and hepatic decompensation
Despite antifibrotic and antitumor effects observed in preclinical studies, clinical evidence for DPP-4i in preventing cirrhosis or hepatic decompensation remains inconsistent. Yang et al.21 found no significant difference in cirrhosis risk between DPP-4i and TZD users (aHR, 1.15), though the authors noted imprecise estimates due to short exposure duration. Conversely, Na et al.51 reported significantly higher cirrhosis risk with DPP-4i compared to TZD (aHR, 1.27; P<0.001), while other comparator groups showed no significant differences. Both studies suggested DPP-4i were inferior to TZD for liver outcomes, potentially due to the neutral effects of DPP-4i in nonalcoholic fatty liver disease (NAFLD).54 This hypothesis is supported by evidence of elevated serum DPP-4 levels in nonalcoholic steatohepatitis and positive associations between DPP-4 expression and histological severity.55,56 In terms of hepatic decompensation, Yen et al.20 demonstrated that DPP-4i accelerated both hepatic decompensation and hepatic failure incidence (aHR, 1.35 for both) in T2DM patients with compensated cirrhosis compared to non-DPP-4i users. The proposed mechanism involves increased incretin levels in splanchnic circulation, potentially promoting portal hypertension.57 These findings raise safety concerns regarding DPP-4 inhibitor use in cirrhotic patients.
Across comparative studies, DPP-4i consistently showed inferior hepatic outcomes compared to SGLT-2i and GLP-1 RA.16,21,28,29,37,39,40,43,46 However, interpretation requires caution given differences in study populations, designs, and limited follow- up duration (e.g., 0.40-0.66 years).21, which may be insufficient for assessing long-term liver outcomes.

DPP-4i and mortality outcomes
Evidence regarding DPP-4i effects on mortality remains limited and contradictory. Yen et al.20 found no significant difference in overall mortality risk compared to non-DPP-4i users among T2DM patients with cirrhosis. The high number of overall deaths (DPP-4i, n=246; non-DPP-4i, n=230) relative to specific liver-related events suggests competing mortality causes may have obscured liver-specific effects. In contrast, Cristiano et al.58 reported significantly reduced overall mortality in DPP-4i users within a broader T2DM population using Veterans Affairs data (odds ratio, 0.41). The authors hypothesized this benefit might relate to improved cancer survival, though this was not directly evaluated. These conflicting results likely reflect differences in patient populations, follow-up duration, and disease severity. The Yen et al.20 study focused specifically on patients with compensated cirrhosis and higher prevalence of viral hepatitis and alcoholic liver disease, suggesting DPP-4i effects may vary by underlying liver disease severity. This underscores the need for individualized treatment approaches based on hepatic status and comorbidity profiles.

CURRENT LIMITATIONS AND FUTURE DIRECTIONS
Despite growing evidence suggesting hepatoprotective effects of novel antidiabetic agents -particularly GLP-1 RA and SGLT-2i- several limitations hinder definitive conclusions. Most existing studies rely on retrospective administrative data that lack key clinical information such as liver disease severity, treatment adherence, and biochemical response. Variability in study populations, comparators, and outcome definitions further limits evidence synthesis. Additionally, liver-related outcomes such as HCC and liver-related death are rare events with often inadequate follow-up periods. The evidence is particularly limited for patients with existing liver disease, where DPP-4i show inconsistent results.
Randomized trials with standardized definitions and extended follow-up are needed to establish causality. Mechanistic studies would clarify the underlying pathways and support individualized treatment strategies.

CONCLUSION
Novel second-line antidiabetic agents, including GLP-1 RA, SGLT-2i, and DPP-4i, exhibit distinct hepatic profiles in patients with T2DM (Fig. 4). Among these agents, GLP-1 RA demonstrate potential chemo-preventive effects, particularly in patients with MASLD or hepatic fibrosis. These agents also appear to reduce hepatic decompensation risk in advanced liver disease. Similarly, SGLT-2i show consistently beneficial effects on hepatic outcomes, including reduced HCC risk, with efficacy extending across a broader spectrum of liver conditions including chronic viral hepatitis. In contrast, DPP-4i exhibit limited and inconsistent evidence for hepatoprotective effects, especially in high-risk populations, having been primarily studied as comparator agents rather than therapeutic interventions.
These class-specific differences indicate that antidiabetic drug selection should be individualized based on hepatic phenotype, metabolic profile, and coexisting liver disease severity. Future prospective studies with standardized liver-specific endpoints are essential to validate these differential effects and establish evidence-based personalized treatment algorithms.