Adverse Liver and Renal Outcomes After Initiating SGLT‐2i and GLP‐1RA Therapy Among Patients With Diabetes and MASLD

This large, population‐based, retrospective cohort study was conducted using the TriNetX research network (Cambridge, MA, USA). TriNetX is a federated multicenter research network that provides real‐time access to an anonymized data set from participating healthcare organizations' electronic health records (EHR). Details of the data source, quality checks, and diagnosis codes used (according to predefined International Classification of Diseases, Clinical Modification [ICD‐9 and 10] codes) for patient selection are described in eMethods. Details of the TriNetX network are described in previous studies [12, 13]. We followed the strengthening the reporting of observational studies in epidemiology (STROBE) reporting guideline.

We identified all adult (aged ≥ 20 years) patients with MASLD and T2DM who newly started treatment with non‐insulin antidiabetic drugs (SGLT2i, DPP4i, GLP‐1Ras, sulfonylureas, meglitinides, thiazolidinediones, and acarbose) between January 1, 2013, and September 31, 2022. We limited the study cohort to patients who were required to have at least 1 year of follow‐up before cohort entry (i.e., receiving their first antidiabetic prescription). Furthermore, to reduce reverse causality and detection bias, we included only those with more than 1 year of follow‐up after the start of the study.

The identification of MASLD at the study's outset was based on the presence of specific ICD‐9 and 10 codes. Patients diagnosed with other liver diseases with any ICD codes were excluded from the MASLD. Patients were excluded if they met any of the following criteria: chronic liver disease other than MASLD, including alcohol, viral, drug‐induced, autoimmune, and genetic diseases; liver cirrhosis; or clinical diagnosis of hepatic decompensation (such as esophageal varices or ascites); history of excessive alcohol use, alcohol abuse, or alcohol use disorder; or history of alcohol‐related disorders; HIV infection; solid organ transplantation; patients with estimated glomerular filtration rate (eGFR) < 30 mL/min/1.73 m² (within 6 months before the cohort entry) and history of undergoing dialysis treatment. Finally, patients with follow‐up of less than 1 year after the cohort and patients with any history of adverse liver and renal events, or dialysis interventions prior to inclusion in the cohort or prior to the index event were also excluded. We further excluded patients with a history of treatment with insulin before their initial prescription for a non‐insulin antidiabetic drug (since insulin therapy at baseline is considered for patients with advanced disease); female patients with a history of polycystic ovary syndrome or gestational diabetes; or those with type 1 DM. Finally, we excluded the patients who had a diagnosis of cancer, organ transplantation or dialysis, and eGFR < 30 mL/min/1.73 m2 before cohort entry within 6 months. Patients were followed from 1 year after cohort entry until an incident adverse liver and renal outcome, switch to a comparator drug, any cause of death, or end of the study (31, 2022), whichever occurred first.

Study cohort consisted of new users of SGLT2i (canagliflozin, dapagliflozin, or empagliflozin), DPP4i (sitagliptin, saxagliptin, linagliptin, and alogliptin), GLP‐1RAs (dulaglutide, exenatide, liraglutide, lixisenatide, or semaglutide), and other second or third‐line antidiabetic drugs (thiazolidinediones, sulfonylureas, meglitinides, α‐glucosidase inhibitors, insulin, or a combination of antidiabetic drugs; or add‐on or switched to an antidiabetic drug) were included. Cohort entry was defined as the date of the first‐ever prescription for one of the drugs of interest (SGLT2i, DPP4i, GLP‐1RAs or other second‐or‐third‐line antidiabetic drugs) during the study period, and their respective first exposure was defined as the index event. We used a lag of 6 months for all exposures to minimize protopathic bias and allow for a minimum and sufficient latency period after cohort entry [14, 15].

We used a propensity score matching (PSM) method to compare the new users of SGLT2i with new users of GLP‐1RAs and other second‐ third‐line antidiabetic drugs. The PSM was performed using 1:1 to reduce the confounding effects. The covariates were adjusted in the PSM model for a priori‐identified potential confounders, such as age, sex, race/ethnicity, nicotine dependence, body mass index (BMI), T2DM, hypertension, hyperlipidemia, hypercholesterolemia, chronic respiratory disease, chronic renal diseases, blood pressure, CKD, peripheral vascular diseases, diabetes‐related microvascular complications, glomerular diseases, osteoporosis, sleep apnea, abnormal laboratory findings (glycated hemoglobin, serum cholesterol, low‐density lipoprotein [LDL], high‐density lipoprotein [HDL], and triglycerides [TG]), and intake of cardiovascular medications (Table 1). Logistic regression was performed to obtain the propensity scores, and a greedy nearest‐neighbor matching algorithm was used to perform the matching with a caliper of 0.1 pooled standard deviations (SD). The balancing of potential confounding variables was evaluated using standardized mean differences (SMD) with a threshold set a priori at 0.10. We used SMD to measure the magnitude of difference between the groups rather than the p‐value because of their insensitivity to sample size. Logistic regression was performed using both Python (Python Software Foundation, Wilmington, Delaware, United States) and R 3.4.4 software (R Foundation for Statistical Computing, Vienna, Austria) to ensure the outputs matched and the order of the rows in the covariate matrix was randomized to eliminate this bias.

The primary outcome was to assess the incidence of a new onset of adverse liver events as a first incidence, which were categorized as (1) cirrhosis, (2) composite outcome of hepatic decompensation events, and (3) HCC. A composite outcome of hepatic decompensation events was defined as the first occurrence of any one of the following events: variceal hemorrhage, ascites, hepatic encephalopathy [16]. The secondary outcome was to evaluate the incidence of adverse renal outcomes, which were categorized as (1) CKD, (2) composite endpoint of a severe stage of CKD, and (3) Need for renal hemodialysis. A composite outcome of CKD was defined as CKD progression from five stages (stages 1–5), whereas a composite endpoint of a severe stage of CKD was defined as CKD progression stages 4–5.

All analyses were performed using the TriNetX real‐time analytics platform. This approach involves dynamic and immediate data analysis, enabling continuous processing and interpretation of data as it is generated. Categorical variables were compared using the Pearson chi‐square test, and continuous variables were compared using an independent‐sample t‐test. Continuous variables were expressed as mean ± SD, and categorical variables were presented as frequency and percentage. Analyses were performed to examine the rate of adverse CVEs using Cox proportional hazards models. Hazard ratios (HRs) and confidence intervals (CIs), along with tests for proportionality, were calculated using R's Survival package v3.2‐3. The results were validated by comparing them with the output from SAS version 9.4. Patients were censored when the time window ended or the day after the last fact in their record. We utilized a 1:1 propensity matching strategy to establish comparable groups of patients treated with different antidiabetic drugs, including SGLT2i, DPP4i, and GLP1RA. In addition, we used this matching approach aimed to balance the covariates between the groups effectively. To account for clustering within the 1:1 propensity‐matched sample and address the loss of independence among individuals resulting from the matching procedure, we incorporated a robust variance estimator in the Cox regression model [17]. The robust variance estimator is essential in enhancing the accuracy of our analytical approach and ensuring the validity of the study's findings. A priori‐defined two‐sided alpha of < 0.05 was used for statistical significance.

GLP‐1RAs may directly improve hepatic steatosis due to their action on upregulating fatty acid metabolism and insulin signaling pathways, and GLP‐1RAs provide benefit to patients with MASLD and provide histological improvements in steatohepatitis for patients with NASH [18]. Hence, we used GLP1RA as a control in the secondary analysis. In this analysis, we matched new users of SGLT2i to new GLP1RA users on propensity scores.

We conducted two sensitivity analyses to ensure the robustness of our findings due to the heterogeneous nature of study outcomes. The first sensitivity analysis evaluated the influence of different reference groups/disease stages. We compared the use of SGLT2i (second‐or‐third‐line antidiabetic drugs) with second or third‐line antidiabetic therapy to minimize confounding, thereby allowing us to explore the impact of including a reference group with the same disease stage, even though we adjusted for microvascular complications. In the second sensitivity analysis, we estimated the rates of new incidences of study outcomes by excluding patients with outcomes 2 years after the index event. Both analyses were performed using the same methods identical to the primary analysis. Our reporting of the results follows the EQUATOR guidelines.

Table 1 provides a comparison of the baseline characteristics between new users of SGLT2i and those starting on DPP4i in a cohort of patients with MASLD and T2DM. The study flow diagram is shown as Figure 1. Before propensity matching (Figure S1), there were some notable differences between groups. The SGLT2i users were slightly younger on average compared to the DPP4i users (58.8 vs. 60.4 years, Table 1) and had a lower proportion of female patients compared to the DPP4i group (50.9% vs. 57.3%, Table 1). In terms of ethnicity and race, both groups were predominantly White, with a slightly higher percentage of Hispanic or Latino patients in the DPP4i group. Another notable difference was in nicotine dependence, which was more prevalent in the SGLT2i group (16.5%) than in the DPP4i group (13.4%, Table 1). In terms of BMI, the SGLT2i users had a marginally higher average BMI than the DPP4i users (34.7 vs. 33.5). The prevalence of comorbidities such as hypertension, hyperlipidemia, and heart failure was also higher in the SGLT2i group before matching. This was mirrored in the use of related medications; for instance, the SGLT2i users were more likely to be on beta‐blockers, antilipemic agents, and diuretics (Table 1). However, after applying PSM, these initial differences were resolved (Figure S1).

In the SGLT2i group, cirrhosis was observed in 644 patients, compared to 1008 in the DPP4i group, yielding a HR of 0.97 (95% CI: 0.91–0.99, p < 0.001). Regarding hepatic decompensations, the SGLT2i group experienced 569 events, markedly fewer than the 1062 events in the DPP4i group, with an HR of 0.84 (95% CI: 0.76–0.94, p < 0.001, Table 2). Most notably, the incidence of HCC was substantially lower in the SGLT2i group (106 events) compared to the DPP4i group (353 events), with an HR of 0.50 (95% CI: 0.40–0.63, p < 0.001).

CKD was notably lower in the SGLT2i group (6384 events) than in the DPP4i group (8878 events), resulting in an HR of 0.87 (95% CI: 0.84–0.90, p < 0.001, Table 2). When focusing on severe stages of CKD (stages 4–5), the SGLT2i group had 1005 events compared to 2346 in the DPP4i group, with an HR of 0.53 (95% CI: 0.49–0.57, p < 0.001), suggesting a significantly lower progression to severe CKD stages. Furthermore, the need for hemodialysis was markedly lower in the SGLT2i group (112 events) compared to the DPP4i group (415 events), with an HR of 0.38 (95% CI: 0.30–0.46, p < 0.001).

In our ancillary analysis, we compared liver and renal outcomes between new users of SGLT2i and GLP‐1RA, each group comprising 56 398 patients. We observed cirrhosis in 830 patients in the SGLT2i group and 944 in the GLP‐1RA group, leading to an HR of 0.94 (95% CI: 0.86–1.04). Regarding hepatic decompensations, the incidence was comparable between the SGLT2i (740 events) and GLP‐1RA (794 events) groups, with an HR of 1.01 (95% CI: 0.91–1.12). Similarly, the risk of HCC appeared almost identical between the groups, with 133 events in the SGLT2i group and 131 in the GLP‐1RA group, yielding an HR of 1.10 (95% CI: 0.86–1.40, Table 3).

For CKD, the SGLT2i group showed 2568 events compared to 2884 in the GLP‐1RA group, resulting in an HR of 0.92 (95% CI: 0.87–0.97). The difference was more pronounced in the severe stages of CKD, with the SGLT2i group experiencing 695 events versus 960 in the GLP‐1RA group, corresponding to an HR of 0.77 (95% CI: 0.70–0.85). Furthermore, the need for hemodialysis was significantly lower in the SGLT2i group (166 events) compared to the GLP‐1RA group (253 events), with an HR of 0.71 (95% CI: 0.58–0.87).

In the Sensitivity Analysis, we explored liver and renal outcomes between new users of SGLT‐2i versus those on other second‐or‐third‐line anti‐diabetic medications, each group with 55 173 patients (Table 4). In the group treated with SGLT‐2i, there were 921 cases of cirrhosis compared to 1478 in the group on other second‐or‐third‐line medications, resulting in a HR of 0.83 (95% CI: 0.75–0.91). This suggests a significantly reduced risk of cirrhosis associated with SGLT‐2i. For hepatic decompensations, 802 events were recorded in the SGLT‐2i group, lower than the 1289 events in the second‐or‐third‐line group, yielding an HR of 0.86 (95% CI: 0.78–0.94). Additionally, HCC was less prevalent in the SGLT‐2i group with 129 events, compared to 289 events in the second‐or‐third‐line group, with an HR of 0.93 (95% CI: 0.75–0.98) (Table 4).

For CKD, the SGLT‐2i group experienced 2514 events, markedly fewer than the 4816 events in the second‐or‐third line group, with an HR of 0.85 (95% CI: 0.81–0.90). In the context of severe CKD stages, the SGLT‐2i group had 678 events, significantly lower than the 1696 events in the second‐or‐third line group, resulting in an HR of 0.70 (95% CI: 0.64–0.77). In addition, hemodialysis was significantly less common in the SGLT‐2i group, with 145 events compared to 494 in the other drug group, giving an HR of 0.55 (95% CI: 0.46–0.67).

In another Sensitivity Analysis, we compared new users of SGLT‐2 with those using DPP‐4i and excluded outcomes from the first 2 years of treatment for each patient group. In the SGLT‐2i group, cirrhosis was observed in 286 patients, considerably lower than the 595 cases in the DPP‐4i group, resulting in a HR of 0.95 (95% CI: 0.93–0.98). Hepatic decompensations occurred in 254 patients in the SGLT‐2i group, compared to 623 in the DPP‐4i group, yielding an HR of 0.90 (95% CI: 0.78–0.99). Furthermore, the incidence of HCC was lower in the SGLT‐2i group, with 47 events versus 114 in the DPP‐4i group, suggesting a reduced risk with an HR of 0.92 (95% CI: 0.65–0.97). For the composite outcome of CKD, there were 1338 events in the SGLT‐2i group and 2107 in the DPP‐4i group, indicating a reduced CKD progression with an HR of 0.88 (95% CI: 0.82–0.96). In terms of advancing to severe stages of CKD, the SGLT‐2i group showed 506 events, notably fewer than the 806 events in the DPP‐4i group, resulting in an HR of 0.61 (95% CI: 0.53–0.71). Additionally, the need for hemodialysis was significantly lower in the SGLT‐2i group (120 events) compared to the DPP‐4i group (238 events), with an HR of 0.57 (95% CI: 0.43–0.76).