The use of GLP-1 receptor agonists and co-agonists in adults without diabetes: a systematic review and network meta-analysis protocol

Abstract

Background

Despite the increasing clinical use of glucagon-like peptide-1 receptor agonists (GLP-1 RAs), head-to-head evidence across these agents remains limited. This systematic review and network meta-analysis (NMA) will assess the comparative efficacy and safety of GLP-1 RAs and novel co-agonists for weight loss among adults without diabetes.

Methods

We will systematically search the PubMed, Ovid, and Cochrane CENTRAL databases to identify randomized controlled trials (RCTs) comparing a GLP-1 RA/co-agonist to placebo or another GLP-1 RA/co-agonist. We will restrict inclusion to RCTs in adults with overweight or obesity; those including participants with diabetes, specific comorbidities or diseases, or history of bariatric surgery will be excluded. The primary outcome will be weight loss, expressed as relative change from baseline, assessed at 6 months (± 4 weeks) and 1–1.5 years (± 4 weeks). Secondary outcomes will include absolute body weight change, total adverse events, gastrointestinal adverse events, serious adverse events, and death. Pairwise meta-analyses and frequentist NMAs will be conducted using a random-effects model. Treatment rankings for efficacy and safety outcomes will be generated using the surface under the cumulative ranking curve. The validity of the results and the assumptions underlying the analyses will be evaluated using a local and global approach. Study-level quality will be assessed using the Cochrane Risk of Bias (RoB) 2 tool and overall network quality will be assessed using the RoB-NMA tool.

Discussion

We aim to provide an up-to-date synthesis of RCTs assessing the weight loss effects of GLP-1 RAs and co-agonists among adults without diabetes to support future clinical decision-making, guideline development, and policy decisions.

Systematic review registration

PROSPERO CRD420251009368

Supplementary Information

The online version contains supplementary material available at 10.1186/s13643-026-03153-6.

Background

Obesity affects over 850 million adults worldwide and is associated with elevated risks of cardiovascular disease, diabetes, and premature mortality [1–3]. While lifestyle modification remains the cornerstone of obesity management [4], weight loss is often modest and difficult to sustain in this population [5, 6]. Bariatric surgery remains the most effective intervention for achieving sustained weight loss, with average losses of 20–30% depending on procedure type [7]. However, it is invasive, not suitable for all patients, and typically reserved for individuals with severe obesity. Pharmacological therapies have therefore emerged as an important adjunct to lifestyle changes. Among these therapies, glucagon-like peptide-1 receptor agonists (GLP-1 RAs) have shown substantial weight loss effects, independent of glycemic control [8]. Initially approved for type 2 diabetes, GLP-1 RAs such as liraglutide and semaglutide are now indicated for chronic weight management [9, 10]. More recently, tirzepatide, a dual GLP-1/glucose-dependent insulinotropic (GIP) polypeptide co-agonist, received regulatory approval for obesity [11]. Novel agents, including the triple-agonist retatrutide (GLP-1/GIP/glucagon RA) and oral agonist orforglipron (GLP-1 RA), have also shown promising efficacy in early-phase trials [12, 13].

Our recent systematic review of 26 randomized controlled trials (RCTs; n = 15,491) found that GLP-1 RAs and co-agonists consistently led to greater weight loss than placebo [8]. While we identified numerical differences in weight loss across agents, heterogeneity in study design precluded formal synthesis. As this is a rapidly evolving area of research and novel GLP-1-based agents are in development, it is likely that additional trials have been published since our last search was conducted in October 2024. While some head-to-head comparisons between approved agents have been published [14, 15], data on the relative efficacy and safety of both pre-market and approved agents among adults without diabetes remain limited. To address these gaps, we propose a network meta-analysis (NMA) of RCTs comparing the use of GLP-1 RAs and co-agonists among adults with overweight or obesity and without diabetes.

Methods

This systematic review and NMA will adhere to the PRISMA 2020 and the NMA extension of PRISMA guidelines to ensure transparency, consistency, and quality of reporting [16, 17]. This protocol has been registered with PROSPERO (CRD420251009368). Any amendments to this protocol will be reported in the final publication of study results. The completed PRISMA-P checklist for study protocols is provided as Additional file 1.

Characteristics of studies

Eligible studies will be parallel-group RCTs. The level of blinding (e.g., open-label, single-blind, or double-blind) may vary across trials and will be recorded as part of the risk of bias (RoB) assessment.

Characteristics of participants

Included RCTs will randomize adult participants with overweight (body mass index [BMI] ≥ 27 kg/m² and ≥ 1 weight-related comorbidity, e.g., hypertension or dyslipidemia) or obesity (BMI ≥ 30 kg/m²) and without diabetes. Studies conducted exclusively in populations with specific comorbidities or diseases (e.g., established cardiovascular disease, polycystic ovary syndrome, chronic obstructive pulmonary disorder, etc.) will be excluded to ensure that our analysis remains focused on adults with general weight-related health challenges, to avoid confounding medical comorbidities associated with obesity as a complex disease. Trials reporting mixed populations (e.g., with and without diabetes) will be included if subgroup data are available for participants meeting our inclusion criteria.

Interventions

The interventions of interest will be GLP-1 RAs and co-agonists (e.g., semaglutide, liraglutide, tirzepatide, retatrutide, orforglipron) at the highest test dose (for pre-market agents) or the approved weight management dose (for commercially available agents). Comparators will include placebo or another GLP-1 RA or co-agonist. Trials using active comparators outside the GLP-1 class (e.g., orlistat, phentermine-topiramate, naltrexone-bupropion) will be excluded to preserve a mechanistic consistency across comparisons. These non-GLP-1 agents differ substantially in their mechanisms of action and are less efficacious [18], which may introduce heterogeneity and compromise clinical interpretability. Bariatric surgery trials will also be excluded, as surgical interventions differ in patient selection (typically reserved for individuals with BMI ≥ 40 kg/m²), treatment goals, and intensity [19]. The inclusion of surgical trials may introduce variability in baseline characteristics and outcome expectations. In addition, bariatric surgery naturally elevates endogenous GLP-1 levels which may introduce mechanistic confounding [20]. Trials that include lifestyle interventions (e.g., diet, exercise, behavioral counseling) will only be eligible if such interventions are applied uniformly across all study arms, as they are typically embedded in standard care and help ensure that observed effects can be attributed to the pharmacological therapy [21].

Outcome measures

The primary endpoint will be change in body weight, expressed as relative change from baseline, assessed at two predefined timepoints: 6 months (± 4 weeks) and 1–1.5 years (± 4 weeks), as reported in each eligible trial. Secondary endpoints will include absolute body weight change, total adverse events, gastrointestinal adverse events, adverse events leading to treatment discontinuations, total serious adverse events, severe gastrointestinal events, biliary disorders, pancreatitis, psychiatric disorders, and death, as defined by the original study investigators.

Search strategy and study selection

Studies will be identified through systematic searches of MEDLINE (via PubMed), EMBASE (via Ovid), and Cochrane CENTRAL from database inception to May 13, 2025. The detailed search strategy is provided in Table 1, which includes specific drug names identified using the PATENTSCOPE database [22]. Additional sources will include clinicaltrials.gov and reference lists of included studies and relevant systematic reviews. No restrictions will be placed on the language of publication. For articles in non-English languages, we will use a combination of machine translation (Google Translate or DeepL) and review by bilingual team members to assess eligibility and extract data. The time period will be from database inception to the date of the search. The RCT search hedge from Chapter 4.S1 of the Cochrane Handbook for Systematic Reviews of Interventions will be used [23].

Table 1.

Detailed search strategy for systematic review and network meta-analysis of GLP-1 RAs and co-agonists

Panel I. MEDLINE via PubMed
Search	Query
#1	"Glucagon like peptide 1 receptor agonists"[MeSH Terms] OR "GLP-1 receptor agonist*" OR "GLP-1 RA" OR "GLP RAs" OR "glucagon-like peptides" OR “Glucagon receptor agonist*" OR "Glucose-dependent insulinotropic polypeptide agonist*" OR “GIP agonist*"
#2	"semaglutide" OR "liraglutide" OR "exenatide" OR "dulaglutide" OR "lixisenatide" OR "albiglutide" OR "beinaglutide" OR "loxenatide" OR "tirzepatide" OR "retatrutide" OR "orforglipron" OR "efpeglenatide"
#3	"Overweight” OR “Obes*" OR "Obesity"[MeSH Terms]
#4	((randomized controlled trial[pt]) OR (controlled clinical trial[pt]) OR (randomized[tiab] OR randomised[tiab]) OR (placebo[tiab]) OR (drug therapy[sh]) OR (randomly[tiab]) OR (trial[tiab]) OR (groups[tiab])) NOT (animals[mh] NOT humans[mh])
#5	#1 AND #2 AND #3 AND #4
Panel II. Embase via Ovid
Search	Query
#1	exp glucagon like peptide 1 receptor agonist/or exp glucagon receptor agonist/or exp gip agonist/or albiglutide/or beinaglutide/or dulaglutide/or efpeglenatide/or exenatide/or liraglutide/or lixisenatide/or loxenatide/or orforglipron/or retatrutide/or semaglutide/or tirzepatide/
#2	(overweight* or obes**).mp. or exp obesity/[mp = title, abstract, heading word, drug trade name, original title, device manufacturer, drug manufacturer, device trade name, keyword, floating subheading word, candidate term word]
#3	crossover-procedure/or double-blind procedure/or randomized controlled trial/or single-blind procedure/or (random* or factorial* or crossover* or cross over* or placebo* or (doubl* adj blind*) or (singl* adj blind*) or assign* or allocat* or volunteer*).tw
#4	1 and 2 and 3
Panel III. Cochrane CENTRAL
Search	Query
#1	"Glucagon like peptide 1 receptor agonist" OR "GLP-1 Receptor Agonist" OR "GLP-1 RA" OR "GLP RAs" OR "glucagon-like peptides" OR "Glucagon receptor agonist" OR "Glucose-dependent insulinotropic polypeptide agonist" OR "GIP agonist" OR "semaglutide" OR "liraglutide" OR "exenatide" OR "dulaglutide" OR "lixisenatide" OR "albiglutide" OR "beinaglutide" OR "loxenatide" OR "tirzepatide" OR "retatrutide" OR "orforglipron" OR "efpeglenatide"
#2	"Overweight" OR "obesity"
#3	MeSH descriptor: [Obesity] explode all trees
#4	#1 AND (#2 OR #3)
#5	#4 in Trials (Cochrane Content Type)

Studies identified by our search will be imported into Covidence [24], a systematic review software, where duplicate citations will be removed. Titles and abstracts will then be screened in duplicate by two independent reviewers for eligibility using pre-determined inclusion/exclusion criteria (Table 2). The full text of any reference considered potentially eligible by either reviewer will be retrieved for full-text review. The full text of all potentially eligible articles will be reviewed similarly by two reviewers, with disagreements resolved by consensus or a third reviewer, if necessary. The studies meeting all inclusion and no exclusion criteria will be included in the qualitative synthesis.

Table 2.

Inclusion and exclusion criteria for systematic review and network meta-analysis of GLP-1 RAs and co-agonists

Panel 1. Inclusion criteria
#1	RCT compared at least one GLP-1 RA or co-agonist with placebo or another active GLP-1 RA/co-agonist
#2	Participants were adults with overweight (BMI ≥ 27 kg/m2 and ≥ 1 weight-related comorbidity, e.g., hypertension or dyslipidemia) or obesity (BMI ≥ 30 kg/m2) and without diabetes
#3	Weight loss outcomes were reported at a follow-up duration of 24 weeks (with a grace period of ± 4 weeks) or 1–1.5 years (with a grace period of ± 4 weeks) given the relative plateauing of weight loss observed
#4	Weight loss outcomes were reported from baseline
Panel 2. Exclusion criteria
#1	RCT was conducted in individuals with diabetes, specific comorbidities unrelated to weight (e.g., established cardiovascular disease, polycystic ovary syndrome, chronic obstructive pulmonary disorder), or a prior history of bariatric surgery
#2	RCT used a comparator that is not a GLP-1 based therapy (e.g., orlistat, naltrexone-bupropion), bariatric surgery, or lifestyle intervention not applied uniformly across arms
#3	RCT had a follow-up duration of less than 20 weeks
#4	Publication reported post-treatment long-term safety outcomes
#5	Publication is a review or editorial without original data
#6	Publication is a non-randomized study, crossover trial, or observational study
#7	Publication is a case report or case study
#8	Publication is an abstract or conference proceeding

Data extraction

Data will be extracted independently by two reviewers using pre-tested forms in Covidence. Disagreements will be resolved by consensus or a third review. Data that will be collected is provided in Table 3. Outcome data will be extracted at 6 months (± 4 weeks) and 1–1.5 years (± 4 weeks). If outcomes are reported at multiple timepoints, we will only use those falling within the prespecified windows. Studies without data within either time window will be excluded.

Table 3.

Data items to be collected in systematic review and network meta-analysis of GLP-1 RAs and co-agonists

General study characteristics	Publication year, list of authors, trial phase, sample size (overall and by treatment group), countries of enrollment, and description of population
Baseline participant characteristics	Age, sex, body weight, body mass index, waist circumference, systolic and diastolic blood pressure
Intervention characteristics	Method of administration, frequency, dosage, duration of treatment, and description of lifestyle interventions
Weight loss outcomes	Relative or absolute change in body weight from baseline with means and standard deviations or medians and inter-quartile ranges
Safety outcomes	Count data for total adverse events, gastrointestinal adverse events, adverse events leading to treatment discontinuation, total serious adverse events, severe gastrointestinal events, biliary disorders, pancreatitis, psychiatric disorders, and death

Quality assessment and certainty of evidence

The Cochrane RoB (Risk of Bias) 2.0 tool [25] will be used to assess study quality and the RoB-NMA tool [26] will be used to assess overall network quality. Two reviewers will independently evaluate each study and the final NMA, and disagreements will be resolved by consensus or a third review. Quality assessment results will be reported in the manuscript. All eligible studies will be included in the manuscript, regardless of study quality. Certainty of evidence across the network will be assessed using the CINeMA (Confidence in Network Meta-Analysis) framework.

Main summary measures

The primary summary measures will include weighted mean differences (WMDs) for continuous weight loss outcomes and relative risks (RRs) for binary safety outcomes. These measures will be reported with their respective 95% confidence intervals (CIs).

Data synthesis and meta-analysis

We will first conduct pairwise meta-analyses for all direct treatment comparisons in the included studies. These pairwise comparisons will serve as the foundation for subsequent NMA. DerSimonian and Laird random-effects models with inverse variance weighting and the Jackson and modified Knapp–Hartung method extensions will be used to pool crude continuous data across studies and obtain pooled WMDs and corresponding 95% CIs or crude count data for RRs and corresponding 95% CIs [27–29]. For binary safety outcomes with zero events in one or both study arms, we will apply a standard continuity correction (the addition of 0.5 to all cells of the 2 × 2 table) to allow calculation of RRs. When events are too sparse for quantitative synthesis, results will be summarized narratively.

Following the pairwise analyses, an NMA will be used to integrate both direct and indirect evidence across the network of interventions. The graph-theoretical approach by Rücker will be utilized to estimate the relative efficacy of treatments [30], even if they have not been directly compared in the included studies. Network geometry will be explored graphically. Treatments will be ranked using surface under the cumulative ranking curve (SUCRA) probabilities [31]. SUCRA rankings will be presented alongside effect estimates and certainty assessments and will not be interpreted in isolation.

Analyses will prioritize intention-to-treat results; if only per-protocol results are available, we will contact authors for missing data or use per-protocol results with appropriate documentation. Missing standard errors or CIs will be imputed using recommended formulas from the Cochrane Handbook [32]. Pairwise meta-analyses will be conducted using the meta package in R (version 3.2.1). NMAs will be performed using the netmeta package [33].

To evaluate the validity of the results and the assumptions underlying the analyses, the following assessments will be conducted:

• Heterogeneity will be evaluated using I² and Tau² statistics.

• Consistency will be evaluated through node-splitting analyses and inconsistency models. Node-splitting analyses will compare direct and indirect evidence at individual nodes to identify local inconsistencies. Global inconsistency will be assessed using design-by-treatment interaction models and residuals in inconsistency plots will be examined to identify broader network inconsistencies.

• Transitivity will be evaluated by confirming the comparability of populations, interventions, and outcomes across the network, including key characteristics such as baseline BMI, age, sex, intensity of background lifestyle interventions, titration schedules, and attrition rates.

Additional analyses

Subgroup analyses or meta-regression may be conducted to examine potential sources of heterogeneity and explore potential effect modifiers relevant to transitivity. Study-level characteristics that may be assessed are dose variations, use of combination therapy (e.g., co-intervention with behavioral or lifestyle changes provided to both arms), and single versus dual or triple-agonists. In sensitivity analyses, we will repeat our analyses by using a fixed-effects model and excluding studies with high RoB. The impact of our broader follow-up windows will be examined by repeating analyses restricted to trials reporting outcomes at shorter ranges (24–28 weeks and 68–72 weeks), where feasible. For binary safety outcomes, we will also conduct a sensitivity analysis where a different continuity correction (0.1) is used.

Assessment of publication bias

Publication bias will be assessed using methods tailored for NMA. Funnel plots will be adapted as comparison-adjusted funnel plots to examine asymmetry within the network. If the number of included studies exceeds 10, a regression-based approach specific to NMA will be applied to test for small-study effects. If publication bias is detected, network-wide adjustment or Bayesian approaches will be employed to estimate publication bias-adjusted treatment effects. If the number of included RCTs is fewer than 10, we will not quantitatively assess publication bias but will qualitatively explore potential bias through evaluation of network geometry and consistency between direct and indirect evidence. This limitation will be mentioned in the discussion if warranted.

Discussion

GLP-1 RAs and co-agonists have transformed the treatment landscape for obesity. Clinical trials and real-world data have shown that these agents can induce weight loss of 10–25% from baseline in individuals without diabetes—effects that could previously only be achieved through bariatric surgery [8, 34]. These outcomes have prompted several clinical practice guidelines, including those by Obesity Canada, the Obesity Society and the European Congress of Obesity, to recommend GLP-1 RAs as a treatment option for individuals with obesity who do not adequately respond to lifestyle interventions [35–37]. However, despite the widespread uptake of GLP-1 RAs, questions remain regarding the comparative benefits and harms of individual agents within this therapeutic class, particularly among people without diabetes.

Although GLP-1 RAs were initially developed for glycemic control in type 2 diabetes, their use has expanded substantially in populations without diabetes due to evidence from large RCTs [8]. For example, semaglutide 2.4 mg was associated with 12.4% mean weight loss at 68 weeks in the STEP-1 trial [38], while the dual-agonist tirzepatide 15 mg led to 17.8% mean weight loss at 72 weeks in SURMOUNT-1 [39]. More recently, the phase 2 trial of the triple-agonist retatrutide reported up to 22.1% weight loss at 48 weeks [12]. These findings suggest that newer co-agonists may result in greater weight loss than earlier GLP-1 RAs. However, these estimates are derived from separate placebo-controlled trials, and although a few head-to-head trials between approved agents have been conducted [14, 15], comparative data involving pre-market therapies remain limited, making it unclear whether differences in outcomes reflect true pharmacological variation or study design.

Previous reviews have attempted to synthesize the evidence base of GLP-1 RAs and related therapies [40–43]. However, most have included mixed populations with and without diabetes, limiting their generalizability to individuals with obesity alone. Additionally, prior syntheses were largely restricted to placebo-controlled or pairwise comparisons and did not account for the growing number of novel agents such as dual and triple-agonist or oral formulations [8]. This knowledge gap is clinically relevant given the differences across agents in delivery mechanisms (subcutaneous or oral), dose escalation protocols, and study-reported safety outcomes [8]. Gastrointestinal adverse events are common across GLP-1 RAs and co-agonists, but differences in frequency or severity between agents remain uncertain. While some trials have reported higher rates of nausea or vomiting with certain therapies, inconsistent definitions, pooled reporting, and heterogeneity in trial design limit cross-study comparisons. Without head-to-head comparisons, it remains unclear whether observed differences reflect true variation or are attributable to other factors such as dose, titration schedule, or study duration.

Findings from this study will have important implications for clinical care and health policy. Clinicians will gain a clearer understanding of how available agents compare in terms of weight loss and tolerability, helping to guide personalized treatment decisions. Current guidelines list multiple agents as options but do not offer agent-level recommendations [4, 35, 36, 44]. The proposed NMA may help fill this gap by providing comparative estimates that can be used to inform future updates to treatment algorithms. The results may also inform payer decisions regarding coverage and reimbursement. Given the high cost and ongoing supply constraints of GLP-1-based therapies [45–47], health systems are increasingly being asked to prioritize access based on clinical value. By providing comparative evidence on efficacy and safety, this project will support more transparent and evidence-based coverage decisions. This study may also help identify areas where future trials are needed, particularly as GLP-1-based therapies continue to expand into new indications, including cardiovascular prevention [48], heart failure [49], and liver disease [50, 51]. By highlighting agents or comparisons for which evidence is weak or inconsistent, our findings may help shape the next generation of RCTs or real-world effectiveness studies. With the increasing use of GLP-1 RAs and co-agonists, and the availability of newer agents, understanding their relative benefits for weight loss in people without diabetes will remain a foundational question.

Abbreviations

BMI

Body mass index

CI

Confidence interval

GLP-1

Glucagon-like peptide-1

NMA

Network meta-analysis

PRISMA

Preferred Reporting Items for Systematic Reviews and Meta-Analyses

RA

Receptor agonist

RCT

Randomized controlled trial

RoB

Risk of bias

RR

Relative risk

SUCRA

Surface under the cumulative ranking curve

WMD

Weighted mean difference

Authors’ contributions

AM contributed to the study conception and design, registered the protocol with PROSPERO, and drafted the manuscript. KBF and MJE were involved in the study conception and design, supervised the study, and provided methodological or clinical expertise. PR provided statistical expertise. MAT, OHYY, and TMP provided clinical expertise. All authors have critically revised the manuscript for important intellectual content and approved its final version.

Funding

TMP is a Fond de recherche du Québec–Santé (FRQS) research scholar. MJE holds a James McGill Professor award from McGill University. KBF is supported by a FRQS mérite award and a William Dawson Scholar award from McGill University (Montreal, QC, Canada). These funding sources will not be involved in the conduct of this study, interpretation of results, or the preparation of the final study for publication.

Data availability

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

MAT has received speaker honoraria from Novo Nordisk, Eli Lilly, Boehringer-Ingelheim, and Sanofi. KBF has received speaker honoraria from Regeneron and Statlog, unrelated to the present work. All other authors declare that they have no competing interests.