Comparative cardiovascular outcomes and safety of hypoglycemic drug classes in patients with type 2 diabetes and hypertension: a multicenter cohort analysis

What is currently known about this topic?

Limited real-world evidence on CV safety of hypoglycemic agents in T2D+hypertension.

What is the key research question?

Compare novel vs. traditional agents on MACE/safety in T2D+hypertension.

What is new?

GLP-1 RAs/DPP4is show lower MACE vs. insulin/acarbose; DPP4is reduce CKD risk.

How might this study influence clinical practice?

Prioritize GLP-1 RAs/DPP4is for T2D+hypertension to improve CV outcomes.

Over the past decade, there have been substantial changes in the therapeutic approaches for type 2 diabetes (T2D). Novel hypoglycemic agents, targeting sodium-glucose transporter 2 and glucagon-like peptide-1 receptors, have emerged, offering dual benefits of reducing blood glucose levels and directly lowering cardiovascular risk in patients [1]. Considering the high incidence of cardiovascular events in individuals with hypertension [2], novel hypoglycemic drugs represent a new therapeutic option for patients with concurrent T2D and hypertension.

A series of clinical trials designed to evaluate the cardiovascular safety of sodium-glucose transporter 2 inhibitors (SGLT2is) and glucagon-like peptide-1 receptor agonists (GLP-1 RAs) found that these medications have been shown to reduce major adverse cardiovascular events (MACE), including myocardial infarction (MI), hospitalization for heart failure (HF), and cardiovascular mortality [3]. However, for dipeptidyl peptidase-4 inhibitors (DPP4is), sulfonylureas (SUs), glinides, acarbose, and insulin, which were widely used prior to the introduction of these newer drugs, no similar trials have been conducted to evaluate their cardiovascular efficacy or safety [4]. Nevertheless [3], these drugs continue to be extensively utilized in clinical practice within certain countries and regions, and are recommended as second-line pharmacotherapy for T2D in clinical practice guidelines [5].

To identify existing gaps and limitations in the current literature, we conducted a systematic review. After carefully applying inclusion and exclusion criteria, we identified 16 relevant studies (Supplementary Tables, Table S6), including 11 observational or retrospective studies and 5 systematic reviews and meta-analyses of randomized controlled trials (RCTs). These studies primarily evaluated cardiovascular efficacy and safety outcomes for novel hypoglycemic agents compared with placebo or standard care. However, several important gaps remain in the current evidence. First, existing RCTs have primarily assessed new hypoglycemic agents as add-on therapies rather than comparing them directly with traditional medications, resulting in uncertainties regarding their comparative cardiovascular effectiveness and safety [6, 7]. Second, there is still a notable lack of comprehensive, large-scale head-to-head direct comparisons among novel agents [8]. Finally, current evidence regarding cardiovascular and safety profiles across diverse patient populations is still insufficient, particularly for individuals with both T2D and hypertension [9]. These evidence gaps hinder the development of precise and individualized therapeutic recommendations for this patient group.

Although the 2024 American Diabetes Association (ADA) Standards of Care in Diabetes have explicitly recognized hypertension as a significant risk factor for the occurrence and progression of chronic kidney disease [10], formulating therapeutic recommendations for patients with T2D and concomitant hypertension on the basis of existing clinical evidence still poses several challenges [11].

Inspired by the LEGEND-T2DM (Large-Scale Evidence Generation and Evaluation Across a Network of Databases for Type 2 Diabetes Mellitus) initiative [12], we conducted a study to compare common hypoglycemic drug treatments in patients with T2D and concomitant hypertension by employing a multicenter cohort analysis across observational databases from the Observational Health Data Science and Informatics (OHDSI) distributed data network, which utilizes data from multiple centers and performs integrated comparisons via analytical techniques to mitigate confounding factors [13]. Our study, carefully constructed by experts using a previously validated methodological framework and drawing on existing evidence to minimize bias and improve reproducibility, offers a holistic perspective on the results and their consistency across diverse populations, medications, and outcomes. We report results from comparisons of combination therapies involving two drugs, using data from participating sources through June 2023, covering patients from July 1996 to March 2023.

The coexistence of T2D and hypertension confers a markedly elevated risk of MACE, including MI, stroke, and HF. Consequently, comprehensive management of both glycemic and blood pressure control is imperative in this high-risk population. Recent international guidelines, such as the 2024 European Society of Cardiology/European Association for the Study of Diabetes (ESC/EASD) recommendations, underscore the importance of selecting antihyperglycemic agents with demonstrated pleiotropic cardiovascular benefits for individuals with T2D and concomitant hypertension. While previous investigations have largely evaluated cardiovascular outcomes in broader T2D cohorts, our study specifically focuses on patients with both T2D and hypertension—a subgroup in whom optimal cardiovascular risk reduction is especially critical.

To the best of our knowledge, this is the first study independently initiated and led by China within the OHDSI community, aiming to provide robust evidence on the cardiovascular efficacy and safety of hypoglycemic drugs in this unique patient population. We analyzed a large real-world cohort of 10,507 patients from two hospitals in China who, after receiving metformin monotherapy, escalated to dual therapy by adding one of seven major antihyperglycemic drug classes. All pairwise comparisons among the seven second-line agents were performed across 17 health outcomes, providing comprehensive insights into their comparative cardiovascular effectiveness and safety in patients with T2D and comorbid hypertension. The majority of efficacy comparisons revealed no significant differences between the drug classes.

Notably, we found that patients who initiated DPP4is had significantly lower risks for 3-point MACE and 4-point MACE compared to new users of several other dual therapies, including metformin combined with insulin, acarbose, or SUs. However, other drug combinations, such as those GLP-1 RAs or glinides group, showed no statistically significant difference in MACE risk compared to DPP4is group. Similar findings were corroborated in the CAROLINA (Cardiovascular Outcome Study of Linagliptin Versus Glimepiride in Patients with T2D) trial, further substantiating the results of the present study [25]. These findings suggest that DPP4i may be a preferred choice for combination therapy with metformin, offering cardiovascular benefits compared to insulin, SUs and acarbose, while exhibiting comparable efficacy to GLP-1 RAs and glinides.

GLP-1 RAs, a key focus of our study, also performed well in terms of efficacy. Although previous studies have demonstrated the cardiovascular benefits of GLP-1 RAs [1, 3], our real-world comparison between GLP-1 RAs and DPP4is in patients with comorbid T2D and hypertension did not reveal a statistically significant difference in the primary cardiovascular outcomes. This is attributable to the distinct characteristics of this high-risk subgroup and the relatively limited sample size of GLP-1 RAs users in our cohort. Our findings are consistent with the body of evidence from cardiovascular outcome trials, where these agents were evaluated in comparison to placebo [26].

The limited sample size of the SGLT2is group at FAHZU reflects a form of clinical inertia. Diabetologists have identified this clinical inertia as a major barrier preventing patients who meet guideline-based eligibility criteria from receiving optimal SGLT2i therapy [27]. In the analysis from JSPH, SGLT2is were associated with a reduced risk of 3-point MACE (HR: 0.69 [95% CI 0.52–0.92]) and 4-point MACE (HR: 0.75 [95% CI 0.58–0.97]) compared to acarbose, with this benefit being particularly pronounced in reducing the risk of ischemic stroke. However, no significant advantage was observed when SGLT2is were compared with other agents, warranting further investigation.

SUs were associated with a reduced risk of both 3-point and 4-point MACE compared to insulin, with the difference in 4-point MACE remaining statistically significant after FDR correction. Additionally, acarbose showed a trend toward an increased risk of 4-point MACE when compared to SUs-based dual therapy. However, this difference did not reach statistical significance after FDR correction. Class-level comparisons and assessments across large and diverse populations likely contribute to these observed differences. Moreover, these findings do not suggest an absolute benefit from SUs but might even reflect a relative harm from acarbose, as suggested in prior studies [28–32].

Currently, there are few ongoing or planned trials addressing this knowledge gap. Furthermore, a limited number of observational studies have employed singular data sources to assess the cardiovascular outcomes associated with the aforementioned therapeutic agents [33]. The associations observed in the current study, particularly those involving DPP4is, differ from the findings of other studies that used smaller sample sizes. These studies might have been compelled to directly compare drug combinations against SUs due to their limited study population. However, this approach could have introduced challenges in isolating the specific effects of individual drugs within the combinations [34].

In terms of safety, the study findings reveal that DPP4is are at a comprehensive disadvantage in the prevention of atherosclerotic cardiovascular events when compared head-to-head with various hypoglycemic agents. This conclusion aligns with the results of previous real-world studies [35]. Research on the role of DPP4is in atherosclerotic cardiovascular events is limited. Based on the available evidence, we can currently only suggest that DPP4is may not be the preferred choice for the prevention of atherosclerotic cardiovascular events. However, compared to acarbose and glinides, DPP4is demonstrated superior performance in reducing the risk of chronic kidney disease. Our findings provide further validation, from a different perspective, of the SAVOR-TIMI 53 trial—the largest and longest-term efficacy study to date investigating the renal effects of DPP4is [36]. Patients who initiated insulin therapy exhibited reduced risks of inflammatory polyarthritis. Studies have demonstrated that the use of insulin can prevent abnormalities in immune cell regulation, thereby decreasing the production of inflammatory factors and subsequently improving the condition of bones and joints [37]. Notably, our findings indicate that glinides were associated with a reduced risk of heart disease, which is consistent with the results of previous large-scale cohort studies [38]. However, similar outcomes have not been observed in clinical trials investigating glinides agents. We hypothesize that this discrepancy may be attributed to the inhibitory effects of glinides agents on cardiovascular KATP channels, as different insulin secretagogues exhibit varying degrees of selectivity for cardiovascular KATP channels [39].

Furthermore, our findings suggest that T2D patients with hypertension and insulin exposure have a significantly lower risk of developing insomnia disorders compared to those treated with DPP4is and SUs. Several Mendelian randomization studies have identified an association between insomnia disorders and insulin resistance [40, 41]. However, the impact of insulin injection on sleep remains to be elucidated.

With respect to urinary tract infections, acarbose users exhibited the highest risk, which was not unexpected. However, it is noteworthy that GLP-1 RAs users had a significantly lower risk of urinary tract infections, a finding that is consistent with the evidence from previous cohort studies [42]. We also found that insulin users had a higher risk of hepatic failure. We hypothesize that long-term insulin injection may induce or exacerbate insulin resistance, potentially contributing to chronic injury and fibrosis [43].

Overall, our findings provide important insights to inform individualized treatment decisions in patients with T2D and hypertension. In terms of cardiovascular effectiveness, DPP4is, GLP-1 RAs, and glinides consistently outperformed insulin and acarbose, with SUs also showing superiority over insulin but being less effective than DPP4is. These results support prioritizing DPP4is, GLP-1 RAs, and glinides as second-line agents in this high-risk population. However, differences in safety profiles across drug classes provide important considerations for individualized treatment selection. Among these agents, DPP4is notably demonstrated a favorable impact on chronic kidney disease, in addition to their benefits in 3-point MACE and 4-point MACE, reinforcing their value in this high-risk population. Nevertheless, they were also associated with higher risks of coronary atherosclerotic disease and insomnia compared to insulin and other agents, while showing a lower risk of hepatic failure. Compared to DPP4is, glinides were associated with lower risks of coronary atherosclerotic disease, hypertensive heart disease, and urinary tract infections, suggesting potential benefits in patients with cardiovascular or infection susceptibility. Despite these benefits, glinides were also linked to increased risks of chronic kidney disease and inflammatory polyarthritis, indicating they may not be optimal for patients with renal impairment or autoimmune inflammatory conditions. GLP-1 RAs demonstrated a favorable safety profile, particularly with a significantly reduced risk of urinary tract infections compared to acarbose, DPP4is, and SUs. No significant increase in other adverse events was observed, suggesting GLP-1 RAs may be a well-tolerated option for patients with susceptibility to genitourinary infections. Collectively, these findings emphasize the importance of balancing cardiovascular efficacy and safety when selecting second-line therapies, and highlight the value of tailoring treatment choices based on individual comorbidity profiles, including cardiovascular, renal, immune, and neuropsychiatric risks.

This study offers several notable strengths. First, it provides a comprehensive, head-to-head comparison of commonly used novel hypoglycemic agents and traditional therapies within a unified analytical framework, thereby offering robust evidence to support individualized treatment strategies for patients with T2D and hypertension. Second, the study population predominantly comprises Asian individuals, addressing a critical gap in the existing literature related to racial and ethnic representation and enhancing the generalizability of our findings to this understudied population. Third, by leveraging large-scale, real-world electronic health record data, this study complements evidence derived from RCTs. While RCTs, conducted under controlled conditions with selected populations, remain the gold standard for establishing efficacy and causal inference, our analysis provides crucial estimates of real-world effectiveness, reflecting outcomes in routine clinical practice and helping to fill evidence gaps relevant to broader patient populations and heterogeneous healthcare settings. Finally, advanced statistical techniques, including propensity score adjustment, were employed to control for potential confounding factors, thereby strengthening the internal validity and interpretability of the results.

However, there are still some limitations. First, the study lacked data on all-cause mortality and certain clinically important adverse events, such as severe hypoglycemia and fractures, due to institutional data-sharing restrictions, rarity of events, and likely underreporting. This hampers comparability with trial-based survival data and may weaken the robustness of outcome-specific analyses, highlighting the need for further investigation. Second, during the pooled analysis, no relevant outcomes were observed in the SGLT2is group, although their cardiovascular benefits have been confirmed in RCTs. This discrepancy between established efficacy and observed real-world effectiveness warrants further exploration in future investigations. Third, given the relatively limited number of events in the GLP-1 RAs group, particularly in the FAHZU center, the corresponding findings should be interpreted with caution. These results are considered exploratory in nature but may still provide valuable preliminary insights to guide future research. Fourth, although advanced statistical methods were applied to adjust for available covariates, the absence of key clinical factors such as diabetes duration and concomitant medications may have introduced residual confounding. Prospective studies are needed to address these gaps and validate our findings. Therefore, caution is warranted when interpreting the results. In addition, we acknowledge the potential for indication bias, especially in comparisons involving insulin. Since insulin is often prescribed to patients with more advanced or poorly controlled diabetes, their higher cardiovascular risk may reflect underlying disease severity rather than a direct effect of insulin. Despite PSM and statistical adjustments, residual indication bias may persist and should be considered when interpreting these findings. Lastly, these analyses are subject to limitations arising from site-specific treatment patterns and the inherent challenges in accurately ascertaining exposure and outcomes across heterogeneous data sources, which may introduce bias into the observed associations between drug classes and clinical outcomes. Collectively, these limitations highlight the need for large-scale comparative effectiveness studies using harmonized real-world data alongside ongoing RCTs to provide a more complete understanding of treatment outcomes.

Our large, rigorous, multinational network study of patients with T2D and hypertension demonstrated that DPP4is, GLP-1 RAs, and glinides significantly reduced cardiovascular risk. All three agents were more effective than insulin and acarbose. While SUs outperformed insulin, they were less effective than DPP4is. No statistically significant differences were observed among the other drug groups. These findings highlight the importance of prioritizing DPP4is, GLP-1 RAs, and glinides as second-line agents for T2D patients with hypertension.

Below is the link to the electronic supplementary material.

We would like to express our sincere gratitude to all those who contributed their expertise and support to this study. We are especially grateful to colleagues from the First Affiliated Hospital with Nanjing Medical University, Nanjing, Jiangsu, China, including Dr. Wensong Zhang, PhD, Associate Professor, Department of the Core Facility; Dr. Shan Lu, MD, Chief Physician, Department of Nutrition; Dr. Yuan He, MD, Chief Physician, Department of Cardiology; Dr. Xiaoxuan Qin, MD, Chief Physician, Department of Neurology; Dr. Yifang Hu, MD, Chief Physician, Department of Geriatric Endocrinology; Mr. Xiaoliang Zhang, Engineer, Department of Information; Ms. Jiyue Fan, Engineer, Department of Information; and Ms. Jiani Xu, Statistician, Center for Data Management. We also thank our colleagues from Nanjing Medical University, Nanjing, Jiangsu, China, including Dr. Feng Chen, PhD, Professor, Department of Biostatistics; Ms. Dan Wang, MD Candidate, the First Clinical Medical College; Ms. Jingyu Wang, Master, Department of Epidemiology; Mr. Youyuan Shui, Master’s Candidate, Department of Medical Informatics; and Mr. Jian Cai, Master’s Candidate, Department of Medical Informatics.In addition, we acknowledge the valuable support of Dr. Yue Gu, MD, Chief Physician, Department of Orthopedics, and Dr. Han Liu, MD, Department of Anesthesiology, from the Second Affiliated Hospital of Nanjing Medical University, Nanjing, Jiangsu, China. We are grateful to Dr. Yaoqi Ge, MD, Chief Physician, Department of General Practice, at Zhongda Hospital Affiliated with Southeast University, Nanjing, Jiangsu, China; Dr. Wei Feng, MD, Engineer, Department of Information, at Wuxi People’s Hospital, Wuxi, Jiangsu, China; Mr. Jiaqi Wang, Engineer, and Ms. Shuang Ma, Engineer, Research Center for Data Hub and Security, Zhejiang Lab, Kechuang Avenue, Zhongtai Sub-district, Hangzhou, Zhejiang, China; Mr. Hongjian Zhang, Engineer, Department of Medical Engineering, the First Affiliated Hospital of Soochow University, Suzhou, Jiangsu, China; and Ms. Mengdie Xu, Research Intern, Research and Development Platform Management Office, Nanjing Drum Tower Hospital, Nanjing, Jiangsu, China.We are deeply grateful for their generous support, insightful feedback, and technical assistance, all of which were essential to the successful completion of this work.

Zy.W. drafted the manuscript with the assistance of Y.L. and J.L. X.Z., Zy.W., and W.X. contributed to the conception and design of the study. W.X. and Y.W. performed the statistical analyses. Y.T., Zm.W., S.J., and C.Q. contributed to data collection. W.L. and S.S. contributed to the revision and proofreading of the manuscript. All authors contributed to data compilation, analysis, interpretation, and manuscript writing. All authors have read and approved the final manuscript.

This study was supported by the National Key Research and Development Program (2023YFC3605800), the National Science and Technology Major Project (2023ZD0509206, 2024ZD0531800), the National Natural Science Foundation of China Youth Science Fund (82404291), the Major Special Projects of Jiangsu Provincial Department of Science and Technology (BK20243054), the Basic Research Program of Jiangsu Province (BK20241308), the Social Development Plan of the Provincial Department of Science and Technology in Jiangsu Province (BE2023781), and the Leading Major Projects in Basic Research at the Forefront of the Institute (QY202403).

The data that support the findings of this study are available from JSPH but restrictions apply to the availability of these data, which were used under license for the current study, and therefore are not publicly available. Data are however available from the co-authors upon reasonable request and with permission of JSPH.

Xin Zhang, Email: zx6800@njmu.edu.cn.

Jingsong Li, Email: ljs@zju.edu.cn.

Yun Liu, Email: liuyun@njmu.edu.cn.