Scientific Reports · Nature Portfolio · 2025

Efficacy and Safety of GLP-1 Receptor Agonists in the Treatment of Type 2 Diabetes Mellitus

A Systematic Review and Network Meta-Analysis · 64 RCTs · 25,572 Participants · Bayesian NMA

Xiaoyu Ren et al. DOI: 10.1038/s41598-025-09807-0

Key Findings

The largest head-to-head comparison integrating data from 64 randomized controlled trials and 25,572 participants via Bayesian network meta-analysis

HbA1c −2.3%

Tirzepatide HbA1c reduction vs. placebo

#1 RANKED

Semaglutide −1.5% · Liraglutide −1.2%

−9.1 kg

Tirzepatide body weight reduction vs. placebo

MAX EFFECT

Semaglutide −2.8 kg · Liraglutide −1.2 kg

64 RCTs

Randomized controlled trials included

LARGEST NMA

25,572 patients · Systematic search through October 2024

GLP-1 receptor agonists (GLP-1 RAs) — including tirzepatide, semaglutide, and liraglutide — represent a new generation of treatments for type 2 diabetes mellitus (T2DM). These agents work by mimicking the incretin hormone GLP-1, stimulating insulin secretion, suppressing glucagon, slowing gastric emptying, and promoting satiety. Beyond glycemic control, they offer meaningful reductions in body weight, a major comorbidity in T2DM.

This study represents the most comprehensive Bayesian network meta-analysis (NMA) to date, integrating 64 randomized controlled trials (RCTs) encompassing 25,572 participants. Using NMA methodology, the researchers were able to compare drugs that have never been directly tested against each other — providing clinically critical indirect comparisons across nine GLP-1 RA formulations and multiple conventional antidiabetic agents.

What is Network Meta-Analysis (NMA)?

Traditional meta-analysis combines data from RCTs that directly compare treatments—for example, "Drug A vs. placebo" or "Drug B vs. placebo." However, if no trial has directly compared Drug A to Drug B, traditional meta-analysis cannot answer "which is better?"

Network Meta-Analysis (NMA) solves this problem. By using data from "A vs. placebo" and "B vs. placebo" trials, NMA can indirectly estimate the relative effect of A vs. B. In this study, 64 RCTs are connected into a network, enabling simultaneous comparison of all GLP-1 agonists against each other—and against placebo—in a single unified analysis.

The primary analysis focused on HbA1c (glycated hemoglobin, the 3-month average blood glucose indicator) and FPG (fasting plasma glucose). Secondary outcomes included body weight, BMI, blood pressure, lipid profiles, and adverse events including hypoglycemia and gastrointestinal symptoms. The results provide evidence-based guidance for individualized treatment selection in clinical practice.

Background

Type 2 Diabetes and GLP-1 Receptor Agonists

Type 2 diabetes mellitus (T2DM) is a major global health challenge. According to the International Diabetes Federation (IDF), the number of adults living with diabetes worldwide is projected to reach 643 million by 2030 and 783 million by 2045. The disease carries enormous human and economic costs, encompassing cardiovascular complications, renal failure, neuropathy, and retinopathy.

GLP-1 receptor agonists have emerged as a transformative class of antidiabetic agents. They act by binding to and activating the glucagon-like peptide-1 receptor, thereby stimulating glucose-dependent insulin secretion from pancreatic beta cells, suppressing inappropriate glucagon secretion, slowing gastric emptying to reduce postprandial glucose spikes, and signaling satiety to the brain. These mechanisms combine to produce both glycemic improvement and clinically significant weight loss.

How GLP-1 Receptor Agonists Work

GLP-1 (Glucagon-Like Peptide-1) is a hormone released by the small intestine after meals. It signals the pancreas only when blood sugar is elevated—the so-called "incretin effect." GLP-1 receptor agonists are drugs that mimic this natural hormone.

Four pillars of action:

Pancreatic β-cells: Stimulate insulin secretion in a glucose-dependent manner (low hypoglycemia risk)
Suppress glucagon from pancreatic α-cells (reduces liver glucose output)
Slow gastric emptying (blunts post-meal blood sugar spikes)
Act on the brain to reduce appetite and promote satiety

Despite the growing number of approved GLP-1 RAs, direct head-to-head trial data comparing these agents to each other — and to conventional antidiabetic drugs such as metformin, sulphonylureas, DPP-4 inhibitors, and SGLT-2 inhibitors — remain limited. Previous meta-analyses have been constrained by smaller sample sizes, narrower drug comparisons, or shorter follow-up periods.

Tirzepatide (Mounjaro/Zepbound) represents a particularly important advance as a dual GIP/GLP-1 receptor agonist — the first in its class. By activating both glucose-dependent insulinotropic polypeptide (GIP) and GLP-1 receptors simultaneously, tirzepatide achieves greater metabolic effects than single-receptor agonists, as demonstrated in the SURPASS clinical trial program.

Tirzepatide's Dual Agonism Explained

Unlike existing GLP-1 receptor agonists, tirzepatide simultaneously activates both GIP and GLP-1 receptors—making it a true "dual agonist." GIP (Glucose-dependent Insulinotropic Polypeptide) is another incretin hormone that also regulates fat tissue metabolism.

GLP-1-only agonists (e.g., semaglutide): activate GLP-1 receptor only
Tirzepatide: co-activates GIP + GLP-1 receptors → stronger glycemic control and weight loss
Weight reduction: tirzepatide −9.1 kg vs. semaglutide −2.8 kg — nearly 3× the difference, explained by GIP co-activation

This study conducted the largest network meta-analysis to date comparing 8 GLP-1 receptor agonist formulations against each other, conventional antidiabetic agents, and placebo — including both direct and indirect comparisons — to provide comprehensive evidence for optimal treatment selection in T2DM.

The PRISMA-NMA reporting guidelines were followed throughout. The systematic search was prospectively registered in the PROSPERO database (CRD42024595773), ensuring methodological transparency and reproducibility. Five major databases were searched through October 2024, capturing the most current evidence including recently approved agents.

The clinical importance of this NMA lies in its ability to inform individualized treatment decisions: which drug offers the best glycemic control for a given patient profile, which carries the lowest risk of hypoglycemia, and how the magnitude of weight loss differs across agents. These are questions that individual RCTs cannot fully answer.

GLP-1 Receptor Agonists Compared

Tirzepatide Semaglutide inj. Semaglutide oral Liraglutide Dulaglutide Albiglutide Lixisenatide Exenatide QW Exenatide BID

Long-Acting vs. Short-Acting

Long-acting: Tirzepatide, Semaglutide, Liraglutide, Dulaglutide, Albiglutide — once-weekly or once-daily dosing
Short-acting: Exenatide BID, Lixisenatide — twice-daily dosing, stronger postprandial effect

Methods

PRISMA-NMA Guidelines · PROSPERO Registration: CRD42024595773

Literature Search and Selection Process

PRISMA flow diagram: 12,074 records identified across 5 databases → 64 RCTs finally included.

Database breakdown: PubMed (n=1,596) · Embase (n=3,496) · Cochrane (n=4,639) · Web of Science (n=2,265) · Chinese databases (n=78) → After deduplication: 8,523 → After title/abstract screening: 151 full-texts reviewed → Final inclusion: 64 RCTs (25,572 patients)

Systematic Literature Search

Searched PubMed, Cochrane Library, Embase, Web of Science, and Chinese databases (CNKI, Wanfang) through October 2024. Only RCTs with parallel-group design were eligible. Two independent reviewers screened all titles, abstracts, and full texts.

RCTs ONLY

Bayesian Network Meta-Analysis

NMA integrates both direct (head-to-head) and indirect (via common comparator) evidence, enabling comparisons between drugs never directly tested against each other. SUCRA (Surface Under the Cumulative Ranking Curve) values rank treatments probabilistically. Consistency between direct and indirect evidence was assessed.

What is SUCRA (Surface Under the Cumulative Ranking Curve)?

When comparing multiple drugs simultaneously in an NMA, we need a probabilistic metric to express "which drug is best." SUCRA is that metric.

Intuitive interpretation: SUCRA = "the expected probability that this drug ranks first among all comparators" (0–100%)

SUCRA 100%: almost certainly the most effective
SUCRA 50%: median rank among comparators
SUCRA 0%: almost certainly the least effective

In this paper, tirzepatide achieves SUCRA 93% for HbA1c reduction—meaning it ranks first in 93% of simulated ranking scenarios across all possible drug combinations.

Bayesian NMA

Cochrane Risk of Bias Assessment

All 64 RCTs were independently evaluated by two reviewers using the Cochrane Risk of Bias (RoB) tool across 7 domains: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting, and other bias.

Cochrane RoB

Eligibility Criteria and Outcomes

Eligible studies were parallel-group RCTs in adults (aged ≥18 years) with confirmed T2DM comparing at least one GLP-1 RA (tirzepatide, semaglutide, liraglutide, dulaglutide, albiglutide, lixisenatide, exenatide QW, or exenatide BID) against placebo or another antidiabetic treatment (insulin, metformin, sulphonylureas, DPP-4 inhibitors, or SGLT-2 inhibitors). Minimum follow-up duration was 12 weeks.

What is HbA1c (Hemoglobin A1c): A measure of average blood glucose over the past 2–3 months. It represents the percentage of hemoglobin (the oxygen-carrying protein in red blood cells) that has glucose attached. Normal range: below 6%; treatment target for diabetes: generally below 7%. This study compares drugs by the change in HbA1c (e.g., −2.3% = a reduction of 2.3 percentage points).

Primary outcomes were changes from baseline in HbA1c (%) and fasting plasma glucose (FPG, mmol/L). Secondary outcomes included body weight (kg), body mass index (BMI, kg/m²), systolic and diastolic blood pressure, total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), triglycerides, and safety outcomes including hypoglycemia, nausea, diarrhea, vomiting, constipation, and decreased appetite.

Statistical analysis employed a Bayesian random-effects NMA model. Heterogeneity was assessed using the I² statistic. Sensitivity analyses were conducted by merging semaglutide injection and oral formulations with conventional drug classes where appropriate, to test the robustness of primary findings. Publication bias was assessed with funnel plots for outcomes with sufficient studies.

Bayesian vs. Frequentist Statistics — Why Bayesian NMA?

Conventional frequentist statistics (p-values, 95% CI) calculate "the probability that data arose by chance." Bayesian statistics combines prior knowledge with new data to directly estimate "the probability distribution of the parameter."

Reasons for using Bayesian NMA:

Comparing many drugs simultaneously makes frequentist multiple-comparison corrections unwieldy
Bayesian NMA estimates all pairwise drug comparisons within a single unified model
The 95% CrI directly states "the true value has 95% probability of falling in this range"—more intuitive than frequentist 95% CI

Primary Results: Glycemic Control

Head-to-head comparison on HbA1c (glycated hemoglobin) and FPG (fasting plasma glucose)

Network Comparison Structure

Network meta-analysis diagram for HbA1c and FPG

Each node represents a treatment; edge thickness indicates the number of direct comparison trials. Placebo is the most common comparator. Panel A: HbA1c network. Panel B: FPG network.

Top 3 Performers vs. Placebo (HbA1c Reduction)

TOP PERFORMER

Tirzepatide

HbA1c reduction:−2.3% (95% CrI: −2.7, −1.9)

FPG reduction:−3.1 mmol/L vs. placebo

Weight change:−9.1 kg vs. placebo

2ND

Semaglutide (injectable)

HbA1c reduction:−1.5% (95% CrI: −1.8, −1.2)

FPG reduction:−2.0 mmol/L vs. placebo

Weight change:−2.8 kg vs. placebo

3RD

Liraglutide

HbA1c reduction:−1.2% (95% CrI: −1.4, −0.96)

FPG reduction:−1.6 mmol/L vs. placebo

Weight change:−1.2 kg vs. placebo

Comparison vs. Placebo (Sensitivity Analysis Forest Plot)

Forest plot: HbA1c sensitivity analysis vs placebo

Vertical axis: individual GLP-1 RA agents. Horizontal axis: mean difference (MD) in HbA1c. Diamond shapes show pooled estimates; horizontal bars represent 95% credible intervals. Agents whose confidence intervals do not cross zero show statistically significant effects.

How to read a forest plot: A point to the LEFT of zero means the drug lowers blood glucose more than placebo (beneficial effect).

If the whiskers (confidence interval) do NOT cross zero, the difference is statistically significant. All major GLP-1 RAs show significant HbA1c reductions vs. placebo.

HbA1c: GLP-1 RAs vs. Conventional Treatments

Forest plot: HbA1c NMA vs multiple comparators

Forest plot comparing GLP-1 RAs against insulin, metformin, sulphonylureas, DPP-4 inhibitors, and SGLT-2 inhibitors. Tirzepatide demonstrates statistically significant HbA1c advantages over all conventional comparators.

All-Drug Comparison: League Table

Upper triangle: FPG comparisons. Lower triangle: HbA1c comparisons. The Tirzepatide (Tir) row shows the greatest improvements across all pairwise comparisons. Values represent mean differences with 95% credible intervals.

Detailed HbA1c Results

In the primary NMA for HbA1c, all GLP-1 RAs demonstrated statistically significant reductions compared to placebo. Tirzepatide showed the largest effect (MD: −2.3%, 95% CrI: −2.7 to −1.9), followed by semaglutide injection (MD: −1.5%), semaglutide oral (MD: −1.4%), liraglutide (MD: −1.2%), dulaglutide (MD: −1.1%), albiglutide (MD: −0.88%), exenatide QW (MD: −0.93%), exenatide BID (MD: −0.82%), and lixisenatide (MD: −0.56%).

Compared to conventional antidiabetic agents, tirzepatide significantly outperformed insulin (MD: −1.5%), metformin (MD: −1.4%), sulphonylureas (MD: −1.7%), DPP-4 inhibitors (MD: −1.6%), and SGLT-2 inhibitors (MD: −1.6%). Semaglutide injection and liraglutide also showed significant advantages over sulphonylureas (MD: −0.91% and −0.58%, respectively).

How to read 95% CrI (Credible Interval): "Tirzepatide HbA1c reduction: −2.3% (95% CrI: −2.7 to −1.9)" means there is a 95% probability that the true effect lies between −2.7% and −1.9%. If this interval does not cross zero, the result is statistically meaningful. A narrower interval indicates higher precision. Analogous to frequentist 95% CI but allows a direct probabilistic interpretation under Bayesian inference.

SUCRA rankings for HbA1c reduction placed tirzepatide first (SUCRA: 97.6%), followed by semaglutide injection (87.2%), semaglutide oral (79.4%), liraglutide (65.1%), and dulaglutide (60.3%). The long-acting GLP-1 RAs consistently ranked above short-acting formulations and conventional treatments, reflecting their superior once-weekly dosing and extended receptor engagement.

Fasting Plasma Glucose (FPG) Results

For FPG reduction, tirzepatide again demonstrated the greatest effect among all treatments (MD: −3.1 mmol/L vs. placebo). Semaglutide injection (MD: −2.0 mmol/L), liraglutide (MD: −1.6 mmol/L), and dulaglutide (MD: −1.5 mmol/L) followed. All long-acting GLP-1 RAs showed significant FPG reductions compared to placebo.

What is FPG (Fasting Plasma Glucose): Blood glucose measured after at least 8 hours of fasting (in mmol/L). Normal range: 3.9–6.0 mmol/L (70–108 mg/dL). While HbA1c reflects average glucose over 3 months, FPG reflects current glycemic control at the time of measurement. GLP-1 RAs improve FPG primarily by suppressing nocturnal hepatic glucose output.

The consistency between direct and indirect evidence was satisfactory for most treatment comparisons. Where inconsistency was identified, it was limited to comparisons with sparse direct evidence, which is expected in any large-scale NMA and does not substantially affect the primary conclusions.

Secondary Results: Weight Loss and Safety

Body Weight Effects

Tirzepatide

−9.1 kg

95% CrI: −11.0 to −7.4 kg

Semaglutide

−2.8 kg

95% CrI: −3.9 to −1.8 kg

Liraglutide

−1.2 kg

95% CrI: −2.2 to −0.12 kg

Note: BMI, blood pressure, and lipid parameters (total cholesterol, HDL-C, LDL-C) showed no statistically significant differences versus placebo across GLP-1 RA treatments in this analysis.

Forest plot for body weight change (kg vs. placebo). Tirzepatide demonstrates the largest weight reduction across all comparisons.

Fasting Plasma Glucose: Full Forest Plot

NMA forest plot for FPG reduction across all GLP-1 RAs and conventional comparators vs. placebo and active comparators.

Safety Profile (Adverse Events)

Gastrointestinal Events

CAUTION

Semaglutide, tirzepatide, liraglutide, and dulaglutide all showed significantly higher rates of nausea, diarrhea, vomiting, constipation, and decreased appetite compared to placebo. These GI effects are class-characteristic for long-acting GLP-1 RAs.

GI symptoms are typically transient, occurring most often during dose titration and improving over time.

Hypoglycemia Risk

DRUG-DEPENDENT

Higher risk: Semaglutide inj. (RR: 4.6, 95% CI: 1.6–10.0) and Exenatide BID (RR: 3.3) vs. placebo
Protective effect: Liraglutide significantly REDUCES hypoglycemia risk vs. conventional drugs — a unique advantage

Treatment Selection Guide

CLINICAL GUIDANCE

Obese T2DM → Tirzepatide (−9.1 kg additional weight benefit)
Normal-weight T2DM → Semaglutide or Liraglutide
Hypoglycemia concern → Liraglutide as first choice
GI disease history → Use semaglutide and liraglutide with caution

Other Secondary Outcomes

Beyond weight, several GLP-1 RAs demonstrated effects on cardiometabolic risk factors. Semaglutide injection showed significant reductions in systolic blood pressure versus some comparators. However, across the full NMA, differences in BMI, total cholesterol, HDL-C, LDL-C, and triglycerides between GLP-1 RAs and placebo did not consistently reach statistical significance, suggesting these outcomes may require larger or longer trials to detect.

For all adverse events analyzed, the safety signals were consistent with known class effects and product labels. No unexpected safety findings emerged from this NMA. The overall benefit-risk profile favored GLP-1 RAs over conventional antidiabetic agents for patients with T2DM, particularly when both glycemic and weight-related outcomes are considered together.

Discussion

Long-Acting vs. Short-Acting GLP-1 RAs

A key finding of this NMA is the consistent superiority of long-acting GLP-1 receptor agonists — particularly tirzepatide and semaglutide — over short-acting formulations in reducing HbA1c. Long-acting agents maintain sustained GLP-1 receptor occupancy, which more effectively suppresses hepatic glucose production overnight and reduces fasting glucose, while short-acting agents primarily blunt postprandial glucose excursions.

Long-Acting vs. Short-Acting GLP-1 RAs — Why Long-Acting Wins

GLP-1 receptor agonists differ by duration of action:

Long-acting (tirzepatide, weekly semaglutide, liraglutide): Occupy GLP-1 receptors continuously for 24+ hours → sustained suppression of nighttime hepatic glucose production → superior FPG reduction
Short-acting (exenatide BID, lixisenatide): Receptor occupancy lasts only hours → stronger gastric-emptying effect → better control of post-meal glucose spikes

This study's finding that long-acting agents outperform short-acting agents on both HbA1c and FPG is consistent with their pharmacological mechanism of action.

Tirzepatide's exceptional performance — the highest SUCRA rankings across HbA1c, FPG, and body weight — is attributable to its unique dual mechanism as both a GLP-1 and GIP receptor agonist. The GIP component enhances insulin secretion and may directly promote adipose tissue lipolysis, contributing to the substantially greater weight loss (−9.1 kg) observed compared to single-receptor GLP-1 RAs (−1.2 to −2.8 kg).

Semaglutide: Injection vs. Oral Formulations

Both injectable and oral semaglutide formulations demonstrated comparable glycemic efficacy (HbA1c reduction: injection −1.5%, oral −1.4%). The oral formulation, which uses the absorption enhancer SNAC (sodium N-[8-(2-hydroxybenzoyl)amino]caprylate) to enable intestinal absorption, provides a needle-free option that may improve adherence in patients with needle phobia or injection fatigue.

What is SNAC (Sodium N-[8-(2-Hydroxybenzoyl)amino]caprylate): An absorption enhancer used in oral semaglutide. GLP-1 RAs are peptides (protein fragments) that are normally broken down by stomach acid, making oral delivery difficult. SNAC raises local pH at the gastric mucosa and increases membrane permeability, enabling semaglutide to be absorbed through the gut wall. This represents a key pharmaceutical innovation that offers patients a needle-free alternative to injection.

The once-weekly dosing schedule of both semaglutide formulations aligns with patient preference data showing strong associations between less frequent dosing and treatment adherence in T2DM. This pharmacokinetic advantage, combined with efficacy comparable to daily-dosing agents, reinforces semaglutide's position as a second-line option behind tirzepatide.

Body Weight and Metabolic Effects

Tirzepatide's weight loss effect (−9.1 kg vs. placebo) is substantially larger than other GLP-1 RAs analyzed and approaches the weight reductions seen with dedicated anti-obesity medications. This finding has important clinical implications for the substantial proportion of T2DM patients with concomitant obesity, for whom weight loss is an equally important therapeutic target alongside glycemic control.

The lack of statistically significant differences in BMI, blood pressure, and lipid parameters (TC, HDL-C, LDL-C) compared to placebo across most GLP-1 RAs is notable and contrasts with findings from some individual trials and earlier meta-analyses. This may reflect the heterogeneity of study populations, varying baseline cardiovascular risk profiles, and the shorter follow-up periods common in glycemic control trials compared to dedicated cardiovascular outcomes trials (CVOTs).

Safety Considerations

The gastrointestinal adverse event profile — dominated by nausea, diarrhea, vomiting, and constipation — is consistent across all long-acting GLP-1 RAs and represents the primary tolerability challenge. These effects are believed to result from slowed gastric motility mediated by GLP-1 receptor activation in the enteric nervous system and are dose-dependent. They typically peak during dose escalation and diminish with continued treatment.

The differential hypoglycemia risk across agents is a clinically meaningful finding. Semaglutide injection's increased hypoglycemia risk (RR: 4.6 vs. placebo) warrants attention, particularly when used in combination with insulin or sulphonylureas. In contrast, liraglutide's significant protective effect against hypoglycemia — a finding that has been observed in some prior cardiovascular outcomes trials — suggests it may be particularly suitable for patients at high hypoglycemia risk.

Study Limitations

This NMA has several limitations. Network sparsity in some treatment comparisons limited the power of inconsistency testing. The analysis used aggregate-level data rather than individual patient data (IPD), which precludes subgroup analyses by age, sex, diabetes duration, or baseline HbA1c. Regional and ethnic variation in pharmacogenomics and dietary patterns may influence outcomes not fully captured by pooling international RCTs.

The risk of bias assessment found moderate-to-low risk overall, but performance bias (participant and personnel blinding) was rated high or unclear in a subset of studies — an inherent challenge in trials of injectable agents. The time horizons of included trials (typically 24–52 weeks) may not fully reflect long-term cardiovascular and renal outcomes that have been demonstrated in dedicated CVOTs for some agents.

STUDY STRENGTHS

Largest NMA in class (64 RCTs, 25,572 patients). PRISMA-NMA compliant. Bayesian random-effects model. Covers all major GLP-1 RAs plus conventional comparators. PROSPERO pre-registered.

LIMITATIONS

Network sparsity · Aggregate data only · Regional variation · Moderate performance bias risk · Short-term follow-up in some trials

CLINICAL IMPLICATION

Personalized medicine approach: match drug selection to patient obesity status, hypoglycemia risk tolerance, injection preference, and cost/access considerations.

Conclusion

GLP-1 receptor agonists significantly improve glycemic control in type 2 diabetes, with long-acting formulations demonstrating consistently greater HbA1c and FPG reductions compared to short-acting agents and conventional antidiabetic drugs. This comprehensive Bayesian NMA of 64 RCTs and 25,572 patients provides the most robust evidence base currently available for comparative treatment selection.

Tirzepatide, semaglutide, and liraglutide emerge as the three most effective GLP-1 RAs for glycemic control. Tirzepatide's dual GIP/GLP-1 mechanism confers exceptional advantages in both HbA1c reduction (−2.3%) and body weight loss (−9.1 kg), making it the preferred option when significant weight reduction is a treatment goal. For patients prioritizing hypoglycemia safety, liraglutide offers a distinctive protective profile.

Treatment decisions should be individualized based on patient-specific factors including obesity status, hypoglycemia risk, cardiovascular risk profile, tolerability of GI adverse effects, dosing convenience (injection vs. oral), and access. The evidence from this NMA supports GLP-1 RAs as a first-line or early combination option in T2DM management, particularly for patients where weight management is a co-equal goal.

Obese T2DM

Tirzepatide

Maximum glycemic control (−2.3% HbA1c) plus the greatest weight loss (−9.1 kg) among all treatments

FIRST LINE

Normal-Weight T2DM

Liraglutide

Strong glycemic efficacy with the lowest hypoglycemia risk among GLP-1 RAs — protective vs. conventional drugs

OPTIMAL SAFETY

Normal-Weight T2DM (Alternative)

Semaglutide

High glycemic efficacy; oral formulation available for needle-averse patients. Note elevated hypoglycemia risk

NOTE: Hypoglycemia risk

Evidence Quality: Bias Risk Assessment

Cochrane Risk of Bias (RoB) framework applied to all 64 included RCTs across 7 domains

Green = Low risk · Yellow = Unclear risk · Red = High risk. Performance bias (participant/personnel blinding) shows some high-risk entries, inherent to injection trials. Overall: moderate-to-low risk evidence base.

Randomization Procedures

LOW RISK DOMINANT

Sequence generation and allocation concealment were judged low-risk in the majority of included RCTs, supporting the integrity of randomization.

Blinding

SOME HIGH RISK

Participant and personnel blinding was rated high-risk or unclear in some studies, as blinding to injectable treatments is inherently challenging. This represents the primary quality concern.

Selective Reporting

LOW RISK DOMINANT

Selective outcome reporting was judged low-risk in most studies. The overall evidence base is considered suitable for drawing comparative conclusions.

Authors

Xiaoyu Ren
Honghao Hua
Yuanqin Wu
Wei Zhang
Xianzhen Long
Yana Bai
Ning Cheng

Publication & Funding

Scientific Reports (Nature Portfolio), 2025
DOI: 10.1038/s41598-025-09807-0
PMC ID: PMC12230154
Competing interests: The authors declare no competing interests.

B2B Content

Any content, beautifully transformed for your organization

PDFs, videos, web pages — we turn any source material into production-quality content. Rich HTML · Custom slides · Animated video.

View Services Contact Us