Abstract
Limited head-to-head studies compare established glucagon-like peptide-1 receptor agonists (GLP-1 RAs), dipeptidyl peptidase-4 inhibitors (DPP-4is), and sodium-glucose co-transporter-2 inhibitors (SGLT-2is) to insulin in the management of uncontrolled type 2 diabetes mellitus (T2DM). This systematic review and meta-analysis evaluated the efficacy and safety of traditional GLP-1 RAs, DPP-4is, and SGLT-2is compared with insulin. Comprehensive searches were conducted in the Cochrane Database, PubMed, MEDLINE, ClinicalTrials.gov, and EMBASE for publications from January 2010 to June 2022, with an additional search extended through June 2025 to capture newly published studies. Randomized controlled trials (RCTs) comparing insulin with established GLP-1 RAs, DPP-4is, or SGLT-2is were included. Thirteen trials involving 5,807 participants were identified. Nine trials compared GLP-1 RAs to insulin, four compared DPP-4is to insulin, and one examined SGLT-2i combined with DPP-4i versus insulin. Compared with insulin, traditional non-insulin agents were associated with greater reductions in hemoglobin A1c (HbA1c) (mean difference (MD) = -0.27, 95% confidence interval (CI) -0.5 to -0.03), body weight (MD = -3.27, 95% CI -4.16 to -2.38), and systolic blood pressure (MD = -3.55, 95% CI -4.92 to -2.17). Insulin use carried a higher relative risk of hypoglycemia (risk ratio (RR) = 2.24, 95% CI 1.88-2.67). Subgroup analyses showed that GLP-1 RAs were superior to insulin in reducing HbA1c and hypoglycemic events, whereas DPP-4is achieved similar glycemic control with improved safety profiles. These findings suggest that established GLP-1 RAs, DPP-4is, and SGLT-2is offer superior or comparable efficacy with better safety than insulin in insulin-naïve patients with uncontrolled T2DM. Categories: Endocrinology/Diabetes/Metabolism Keywords: dpp-4 inhibitors, glp-1 receptor agonists, insulin, meta-analysis, non-insulin diabetes medications, sglt-2 inhibitors, systematic review, type 2 diabetes mellitus Diabetes mellitus (DM) affects more than 38.4 million people in the United States, or approximately 11.6% of the US population [1,2]. Type 2 diabetes mellitus (T2DM) accounts for 90% to 95% of the population with DM [1,2]. T2DM is associated with a relative or complete impairment in insulin secretion, along with varying degrees of peripheral resistance to insulin action [3]. The beta-cell function is typically half that of healthy individuals at the time of diagnosis [3]. With disease progression, insulin output declines further, and many patients ultimately require treatment with exogenous insulin to maintain glycemic control [4]. Although insulin is highly efficacious, it can have severe side effects. Hypoglycemia and weight gain frequently complicate both the initiation and optimization of insulin therapy [4,5]. The leading cause of morbidity and mortality in patients with DM remains cardiovascular (CV) disease, which is estimated to be two to four times more common in patients with DM [6,7]. The risk of developing an adverse CV event in patients with an HbA1c ≥ 8% is 16% higher than in patients with an HbA1c between 6% and 8% [8]. The older antidiabetic agents, including metformin, thiazolidinediones (TZDs), sulfonylureas (SUs), meglitinides, and insulin, have been in use for years to manage T2DM [9-11]. The newer classes of medications, including dipeptidyl peptidase-4 inhibitors (DPP-4is), glucagon-like peptide-1 receptor agonists (GLP-1 RAs), and sodium-glucose cotransporter-2 inhibitors (SGLT-2is), have become increasingly common over the last few years due to their CV and renal protective effects [12]. GLP-1 RAs work by stimulating the release of endogenous insulin in the presence of elevated blood glucose (BG) levels, thereby reducing both fasting and postprandial BG levels [5]. GLP-1 RAs can also promote weight loss by causing early satiety [13,14]. DPP-4is modulate fasting BG, postprandial BG, and HbA1c levels by decreasing the inactivation of incretins such as GLP-1 and glucose-dependent insulinotropic polypeptide. This stimulates 1 2 3 4 Open Access Review Article How to cite this article Ahmed A, Tan Z, Abd El-Radi W, et al. (September 12, 2025) Insulin Versus Established GLP-1 Receptor Agonists, DPP-4 Inhibitors, and SGLT-2 Inhibitors for Uncontrolled Type 2 Diabetes Mellitus: A Systematic Review and Meta-Analysis of Randomized Controlled Trials. Cureus 17(9): e92175. DOI 10.7759/cureus.92175 the release of insulin in a glucose-dependent manner [13,15]. SGLT-2is inhibit the reabsorption of glucose from the proximal tubule of the kidney, causing glycosuria [16]. In one open-label cohort study, sitagliptin (DPP-4i) was found to reduce HbA1c by 3% from a baseline HbA1c of 11.8% [17]. Similarly, a 2% reduction in HbA1c from a baseline of 9.1% was seen with dapagliflozin (SGLT-2i) [18]. Several studies have compared basal insulin to GLP-1 RAs [19]. One such study showed that exenatide use was associated with a greater reduction in HbA1c compared to insulin glargine in patients with a mean baseline HbA1c of ≥ 9.0% [20]. The therapeutic landscape for T2DM has continued to evolve, with newer agents, such as dual incretin receptor agonists (e.g., tirzepatide), showing promising results in recent clinical trials. However, the established glucose-lowering agents, including GLP-1 RAs (exenatide, liraglutide, dulaglutide, semaglutide), DPP-4is (sitagliptin, saxagliptin, linagliptin, alogliptin), and SGLT-2is (canagliflozin, dapagliflozin, empagliflozin, ertugliflozin), remain the most widely available and prescribed non-insulin therapies globally. Understanding their comparative effectiveness against insulin remains crucial for clinical decision making, particularly in healthcare settings where newer agents may not be accessible due to cost or availability constraints. Every few years, the diabetes community reevaluates and updates the current recommendations for T2DM treatment to reflect new information from clinical research and practice. The current guidelines of the American Association of Clinical Endocrinologists (AACE) and the American Diabetes Association (ADA) recommend considering the initiation of insulin for patients with T2DM who have HbA1c levels exceeding 9.0% or 10%, respectively [21,22]. However, these recommendations are based on expert opinion rather than randomized controlled trials (RCTs). The 2025 ADA Standards of Care emphasize individualized pharmacologic approaches that address both glycemic and weight goals, with a preferential use of agents that reduce the risk of CV and kidney disease [23]. The 2022 ADA/European Association for the Study of Diabetes (EASD) consensus report advocates for a holistic, person-centered approach that considers weight management as integral to diabetes care, with specific recommendations for established and newer glucose lowering medications [24]. Given the efficacy and additional benefits associated with the use of GLP-1 RAs, DPP-4is, and SGLT-2is, should these medications be the initial preferred treatments for patients with an HbA1c > 9% over insulin? Several trials and reviews have compared individual GLP-1 RAs, DPP-4is, or SGLT-2is to insulin in poorly controlled T2DM patients. However, these studies typically evaluated single agents rather than examining the broader class effects of established non-insulin therapies. To our knowledge, no comprehensive meta analysis has systematically evaluated these established glucose-lowering medications as distinct classes compared to insulin. Therefore, we conducted this systematic review and meta-analysis to compare the efficacy and safety of traditional GLP-1 RAs, DPP-4is, and SGLT-2is to insulin in poorly controlled T2DM, providing evidence to guide clinical practice with widely available therapeutic options. Our secondary aims were to investigate the effects of these medications on body weight and systolic blood pressure (SBP). Information Sources and Search Strategy All published trials from January 2010 to June 2022 were searched on PubMed, MEDLINE, Embase, the Cochrane Central Register of Controlled Trials, and ClinicalTrials.gov. A secondary comprehensive search was conducted from June 2022 to June 2025 to capture any newly published studies. The search focused on established non-insulin diabetes medications available during this period, including traditional GLP-1 RAs (exenatide, liraglutide, dulaglutide, semaglutide, lixisenatide, albiglutide, taspoglutide), DPP-4is (sitagliptin, saxagliptin, linagliptin, alogliptin, vildagliptin), and SGLT-2is (canagliflozin, dapagliflozin, empagliflozin, ertugliflozin). Dual incretin receptor agonists (such as tirzepatide) were excluded because they represent a mechanistically distinct class that warrants separate analysis. The search strategy combined both keywords and MeSH terms. The following terms were applied: (“T2DM”[MeSH] OR “T2DM” OR “type 2 diabetes”) AND (“insulin” OR “insulin therapy”[MeSH] OR insulin glargine OR insulin degludec OR insulin detemir OR NPH insulin OR basal insulin) AND (“GLP-1 receptor agonists”[MeSH] OR “GLP-1 receptor agonists” OR exenatide OR liraglutide OR dulaglutide OR semaglutide OR lixisenatide OR albiglutide OR taspoglutide) OR (“DPP-4 inhibitors”[MeSH] OR “DPP-4 inhibitors” OR sitagliptin OR saxagliptin OR linagliptin OR alogliptin OR vildagliptin) OR (“SGLT2 inhibitors”[MeSH] OR “SGLT2 inhibitors” OR canagliflozin OR dapagliflozin OR empagliflozin OR ertugliflozin). Selection Criteria According to our protocol's criteria, all RCTs published in English were included. All participants were adultswith T2DM with a baseline mean HbA1c ≥ 8% at the start of the trial. The included trials had an intervention duration of at least 24 weeks. All included trials reported HbA1c at the end of 24-30 weeks. Trials involving pregnant or breastfeeding participants were excluded. Comparisons were made between the first group (basal insulin, basal-bolus insulin regimens, and premixed insulins) and the second group (DPP 4is, GLP-1 RAs, or SGLT-2is). Trials that compared medications within the same class were excluded. In each comparison, background treatment was defined as the antidiabetic medications used in both the intervention and control groups after randomization. Eligible background therapy was either no background treatment or metformin-based background treatment (metformin only or metformin plus another antidiabetic medication). After screening, two independent reviewers (AA, WA) screened titles and abstracts and examined full texts of potentially eligible trials. Disagreements were resolved through discussion with a third reviewer (ZT). Data Extraction Outcomes and Quality Assessment Data were extracted from the full texts of the published studies included. Two reviewers extracted the data independently. A pair of reviewers assessed risk of bias (WA, ZT); discrepancies were resolved with a third reviewer (AA). For each medication, outcome data were merged from all approved doses into a single intervention group. If outcomes were reported at multiple time points, results were extracted at 24-30 weeks or six months. The change from baseline HbA1c was considered the primary outcome. Secondary outcomes were hypoglycemic events, body weight change, and SBP change. Extracted data included participant numbers, demographics, trial duration, and the type of antidiabetic medication used. Additional data included the mean change in HbA1c (SD), the mean change in body weight (SD), and the frequency of hypoglycemic episodes. Data Synthesis and Analysis The efficacy of non-insulin agents versus insulin was compared in the included RCTs. Outcomes compared included HbA1c change, hypoglycemic events, body weight change, and SBP change at 24-30 weeks. HbA1c, SBP, and body weight were analyzed as continuous variables and reported as absolute mean differences. Safety was assessed by risk of hypoglycemia, analyzed as a dichotomous variable, and expressed as a risk ratio. Data were analyzed using a random-effects model with the inverse variance method. Heterogeneity was assessed with chi-square and tau-square tests. All analyses were conducted in RevMan (version 5.4).
Showing the abstract — retrieve the full paper via the Exa API.