Evidence reference only. Not medical advice, not a dosing guide, and not a recommendation to use any drug.
Echo The Echo Compendium
<- All drugs

GLP-1 RA class

MARKETED

incretin / GLP-1-based

90
graded findings
13
effect domains
Evidence spread
High evidence 4Moderate evidence 27Low evidence 41Very low evidence 18

GLP-1 RA class is shown as a marketed evidence record where applicable. Findings describe observed research and regulatory records; they do not provide treatment advice.

GLP-1 RA class MARKETED
Safety signals Decrease
High evidence

FOLK-CLAIM VERDICT ('reduced inflammation / joint pain relief / skin improvements'): MIXED, and the mechanism-split is the whole story. (1) Knee-OA pain relief = TRUE...

Bliddal H et al. Semaglutide in knee osteoarthritis with obesity (STEP-9). N Engl J Med 2... Source
Full findingFOLK-CLAIM VERDICT ('reduced inflammation / joint pain relief / skin improvements'): MIXED, and the mechanism-split is the whole story. (1) Knee-OA pain relief = TRUE but predominantly WEIGHT-LOSS-mediated (less joint load; STEP-9 real and large, but tracks weight and the placebo arm also improved - not proven direct chondroprotection). (2) Systemic inflammation / CRP reduction = TRUE and PARTLY DIRECT (CRP falls somewhat beyond weight loss in SELECT), though magnitude still mostly adiposity-coupled. (3) Skin (psoriasis, hidradenitis) = UNPROVEN (small/uncontrolled, weight-confounded).
PopulationSemaglutide knee-OA RCT (STEP-9) + reta knee-OA phase-3 topline (TRIUMPH-4) + CRP data (SELECT/HFpEF) + small skin series.
Fundingindustry - Novo Nordisk (trial sponsor; disclosed)
Scope limitsknee-OA / obesity population; composite finding also carries press-topline (TRIUMPH-4) and small-series strands separately graded
Comparatorsplacebo
GLP-1 RA class MARKETED
Safety signals Decrease
Moderate evidence

A 2018 Bayesian network meta-analysis of 54 RCTs (49,602 T2D patients) found GLP-1 RAs were associated with decreased fracture risk versus placebo - an independent...

Zhang YS, Weng WY, Xie BC, et al. Glucagon-like peptide-1 receptor agonists and fracture ... Source
Full findingA 2018 Bayesian network meta-analysis of 54 RCTs (49,602 T2D patients) found GLP-1 RAs were associated with decreased fracture risk versus placebo - an independent net-neutral-to-protective fracture pole, separate from the contested anabolic-BMD claim.
PopulationGLP-1 RA fracture network meta-analysis (Zhang et al., Osteoporos Int 2018; 54 RCTs, T2D, PROSPERO CRD42018094433).
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsSCOPE LIMIT: T2D RCTs, PREDATES the high-dose obesity agents (semaglutide 2.4 mg / tirzepatide / retatrutide) - does not speak to rapid-weight-loss bone effects. The 'exenatide ranked safest' sub-claim (and the agent ranking generally) is NOT established - encoded is only the NET protective direction, not the ranking (which rests on sparse fracture events and acknowledged network inconsistency). Provides an INDEPENDENT fracture pole so the net-neutral/protective read no longer rests solely on the contested PMID39985672. Opposes the older-adult-harm pole (C-CLASS-BONE-FRAC-ELDERLY-01) and the unverified SELECT 75+ concern (OQ-BMD-B).
Comparatorsplacebo; other anti-hyperglycaemic drugs
GLP-1 RA class MARKETED
Safety signals Mixed
Moderate evidence

Peri-operative aspiration meta-analysis: GLP-1 RA use associated with sharply increased residual gastric contents despite appropriate fasting, but NOT with increased...

Elkin J et al., Anaesthesia 2025 Source
Full findingPeri-operative aspiration meta-analysis: GLP-1 RA use associated with sharply increased residual gastric contents despite appropriate fasting, but NOT with increased pulmonary aspiration; withholding at least one dose lowered residual-contents odds.
PopulationMeta-analysis, 28 observational studies, fasted patients undergoing anaesthesia/sedation
Fundingacademic/independent - investigator (Australian Government Research Training Program Scholarship); no industry study sponsor
Scope limitsCentral anaesthesia tension: retained contents up ~6-fold yet no measurable rise in actual aspiration (low/very-low GRADE). Underlies ASA/multi-society pre-procedure hold guidance.
Comparatorsnon-GLP-1 RA exposure
GLP-1 RA class MARKETED
Safety signals Increase
Moderate evidence

Peri-procedural meta-analysis: pre-procedural GLP-1 RA increased pre-procedural GI symptoms and elevated residual gastric content vs control, with no significant...

do Nascimento TS et al., Perioper Med (Lond) 2024 Source
Full findingPeri-procedural meta-analysis: pre-procedural GLP-1 RA increased pre-procedural GI symptoms and elevated residual gastric content vs control, with no significant difference in PONV or 30-day mortality.
PopulationMeta-analysis, 14 RCTs/observational, 2143 adults, elective surgery/procedures
Fundingacademic/independent - self-funded (no external sponsorship)
Scope limitsConcordant with Elkin on residual gastric content; also reports peri-procedural glycaemic benefit.
Comparatorscontrol (no GLP-1 RA)
GLP-1 RA class MARKETED
Safety signals Increase
Moderate evidence

NET VERDICT (peri-operative aspiration) - the most clinically live procedural signal: by delaying gastric emptying these agents leave residual gastric contents despite...

Elkin J et al., Anaesthesia 2025 Source
Full findingNET VERDICT (peri-operative aspiration) - the most clinically live procedural signal: by delaying gastric emptying these agents leave residual gastric contents despite guideline fasting, raising a theoretical aspiration risk under anaesthesia/sedation. Meta-analysis robustly demonstrates the RETAINED-CONTENTS surrogate but does NOT demonstrate increased actual pulmonary aspiration. Operationally live because multi-society/ASA guidance now recommends mitigation.
PopulationPatients on GLP-1RAs undergoing anaesthesia/endoscopy/sedation; meta-analyses of peri-procedural cohorts + gastric-ultrasound studies; semaglutide/liraglutide-era data; no reta-direct.
Fundingacademic/independent - investigator (Australian Government Research Training Program Scholarship); no industry study sponsor
Scope limitsGRADE SPLIT: retained-gastric-contents surrogate = moderate (consistent, ~6-fold); actual pulmonary-aspiration hard outcome = low/very-low and NOT demonstrated to increase; guidance pole = expert multi-society consensus (moderate, not an agency label). Both poles: ANCHOR = Elkin (ANAES17) + do Nascimento (ANAES18) + Wu (ASP09); GUIDANCE = the new Kindel row; CAVEAT = narrative reviews call data preliminary, recommend gastric ultrasound (ANAES19), flag the drug-AE-vs-surgical-complication misidentification trap. RETA POSITION: reta is GI-dominant and shares the delayed-gastric-emptying mechanism, so the retained-contents class-transfer expectation applies; no reta-direct peri-operative data. VALIDATOR: confirm PMID 40230298 (retained contents ~6-fold, no aspiration increase).
Comparatorsnon-GLP-1-RA controls; fasted controls; dose-withheld vs continued
GLP-1 RA class MARKETED
Safety signals Not directional
Moderate evidence

Multi-society clinical practice guidance for safe peri-operative use of GLP-1 RAs: individualised risk-stratified management rather than fixed cessation, balancing...

Kindel TL, Wang AY, Wadhwa A et al. (multisociety), Clin Gastroenterol Hepatol, 2024 Source
Full findingMulti-society clinical practice guidance for safe peri-operative use of GLP-1 RAs: individualised risk-stratified management rather than fixed cessation, balancing aspiration-risk mitigation against the harms of interrupting therapy. Operationalises the earlier ASA October-2023 consensus (which advised holding GLP-1 RAs pre-procedure).
PopulationAdults on GLP-1 RAs presenting for elective procedures/anaesthesia/endoscopy.
Fundingprofessional-society guidance - no external/commercial funding (AGA and co-sponsoring societies; author consulting COIs disclosed)
Scope limitsno outcome data yet (ongoing)
GLP-1 RA class MARKETED
Safety signals Increase
Moderate evidence

NET VERDICT (gallbladder / biliary) - one of the FEW high-confidence real GI safety signals: randomised-trial meta-analysis shows GLP-1-based therapy increases the...

He L et al., JAMA Intern Med 2022 Source
Full findingNET VERDICT (gallbladder / biliary) - one of the FEW high-confidence real GI safety signals: randomised-trial meta-analysis shows GLP-1-based therapy increases the composite of gallbladder/biliary disease, concentrated at higher doses, longer durations and in weight-loss trials. Leading mechanism is rapid weight loss + reduced gallbladder motility, so it is expected to track with weight-loss magnitude.
PopulationPooled randomised participants across GLP-1RA trials (T2D and obesity) + CV-outcomes RCT-meta + obesity head-to-head meta. No retatrutide-direct biliary endpoint.
Fundingacademic-Chinese public (NSFC / Beijing NSF / CAMS)
Scope limitsGRADE: high-confidence real (RCT-meta, multiple concordant). STRONG pole = He composite RR ~1.37 (+37%) with dose/weight-loss gradient (BIL01) + Galli (BIL04) + Zeng (BIL02); NUANCE/BRAKE = tirzepatide showed NO significant biliary excess in obesity-only (BIL03) - a genuine brake on the reta-amplification expectation, since reta is mechanistically closer to the dual/triple agents than to semaglutide; a semaglutide bile-duct-cancer VigiBase signal is very-low/contested (BIL15). RETA POSITION: reta is the most weight-loss-potent agent and the signal scales with weight loss, so shared (plausibly greater) biliary risk is a reasonable class-transfer expectation - expectation, NOT observation (no reta-direct endpoint). VALIDATED: PMID 35344001 (He JAMA Intern Med 2022) confirms composite RR 1.37 (95% CI 1.23-1.52); the ~1.27/+26% figure is the cholelithiasis sub-component, corrected.
Comparatorsplacebo; active comparators; semaglutide; tirzepatide
GLP-1 RA class MARKETED
Safety signals No change
Moderate evidence

REFUTING/NULL pole at class level: meta-analysis of 28 observational studies found GLP-1 RAs did NOT increase NAION, glaucoma, or retinopathy vs other antidiabetics.

Anatriello A et al., Front Pharmacol 2026 Source
PopulationMeta-analysis, 28 observational studies (6 semaglutide, 22 all GLP-1 RA) in T2D, 2006-2025
Fundingacademic/independent - no financial support received
Scope limitsHigh heterogeneity (I2 ~89%); pooled estimate null. Counterpole to signal records.
Comparatorsother antidiabetic treatments
GLP-1 RA class MARKETED
Safety signals Decrease
Moderate evidence

Network meta-analysis (body composition): potent agents produce greatest fat-mass loss but significant ABSOLUTE lean-mass loss; RELATIVE (% baseline) lean mass...

Karakasis P et al., Metabolism 2024 Source
Full findingNetwork meta-analysis (body composition): potent agents produce greatest fat-mass loss but significant ABSOLUTE lean-mass loss; RELATIVE (% baseline) lean mass unchanged. Tirzepatide/semaglutide most effective for weight/fat, among least effective at preserving lean mass.
PopulationNetwork meta-analysis, 22 RCTs, 2258 participants (diabetes and/or overweight/obesity), through Nov 2024
Fundingacademic / independent - no specific grant funding
Scope limitsAbsolute vs relative lean-mass framings both recorded. Liraglutide was the only GLP-1 RA achieving significant weight loss without significant lean-mass reduction. Cross-links body-composition.
Comparatorsplacebo; liraglutide; tirzepatide; semaglutide
GLP-1 RA class MARKETED
Safety signals No change
Moderate evidence

NET VERDICT (pancreatitis / pancreatic cancer) - largely REASSURED at trial level: randomised and meta-analytic evidence shows no clear increase in pancreatitis (any...

Wen J et al., Endocrinol Diabetes Metab 2025 Source
Full findingNET VERDICT (pancreatitis / pancreatic cancer) - largely REASSURED at trial level: randomised and meta-analytic evidence shows no clear increase in pancreatitis (any nominal excess attenuates on stratification) and no overall pancreatic-cancer association. Against this sits a persistent FAERS reporting signal (reporting != incidence). Net: not a confirmed real signal at trial level, with the FAERS pole retained as a low-grade caveat.
PopulationPooled randomised trial participants (the Wen 2025 meta reportedly INCLUDES retatrutide); CV-outcomes trial populations; FAERS.
Fundingnone-declared (independent meta-analysis)
Scope limitsGRADE: REASSURING at RCT-meta (moderate evidence of no clear increase) vs very-low FAERS reporting signal. NULL pole = Wen (PANC05) + Galli (BIL04); SIGNAL pole = FAERS liraglutide pancreatitis (PV07), semaglutide pancreatic-cancer (PV06), + the residual with-background-medication pancreatic-cancer stratum in Wen - all reporting != incidence. RETA POSITION: reta is reportedly INCLUDED in Wen's pooled trials, so it shares the reassuring trial-level pole DIRECTLY (not merely class-transfer); no reta-specific pancreatitis excess. VALIDATOR: confirm PMID 40988099 (Wen 2025) + whether it includes retatrutide.
Comparatorsplacebo; active comparators; background-medication strata
GLP-1 RA class MARKETED
Safety signals No change
Moderate evidence

Pooled meta-analyses found NO significant increase in DKA or severe hypoglycaemia with GLP-1 RAs in T1D - but the DKA estimate is VERY LOW certainty and this null does...

Kateel R, Parida A, Chogtu B, Holla SN. Safety of GLP-1 receptor agonists in type 1 diabe... Source
Full findingPooled meta-analyses found NO significant increase in DKA or severe hypoglycaemia with GLP-1 RAs in T1D - but the DKA estimate is VERY LOW certainty and this null does NOT overturn the dose-dependent ketosis signal in the individual ADJUNCT trials.
PopulationGLP-1 RA T1D safety meta-analysis (Kateel et al., 25 studies [23 RCTs + 2 observational]; corroborated by Tan et al. PMC10797415).
Fundingacademic/government (Kasturba Medical College, Manipal; Dept of Health Research, India)
Scope limitsMUST NOT READ AS REASSURANCE: the DKA RR 0.60 is VERY LOW certainty with wide CIs (statistical imprecision); the pool DILUTES the high-dose/liraglutide ketosis signal across lower doses and non-liraglutide agents. It does NOT overturn the dose-dependent hyperglycaemia-with-ketosis seen at liraglutide 1.8mg in the individual trials (C-LIRA-T1D-SAFETY-02). Tan's DKA OR 1.38 is numerically higher but NON-significant. The honest read is UNSETTLED at high dose, not safe. Routed to OQ-T1D-D.
Comparatorsplacebo (insulin background)
GLP-1 RA class MARKETED
Safety signals Mixed
Moderate evidence

A class-level meta-analysis found GLP-1 RAs reduce serum uric acid before-versus-after treatment but NOT significantly versus placebo, and less than active comparators...

Najafi S, Bahrami M, Butler AE, Sahebkar A. The effect of glucagon-like peptide-1 recepto... Source
Full findingA class-level meta-analysis found GLP-1 RAs reduce serum uric acid before-versus-after treatment but NOT significantly versus placebo, and less than active comparators including SGLT2 inhibitors - so the apparent urate-lowering is not an established placebo-controlled drug effect at the class level.
PopulationGLP-1 RA class meta-analysis (Najafi et al., 17 studies; RCTs, observational and uncontrolled studies pooled).
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsSCOPE LIMIT: predates high-dose agents (tirzepatide/retatrutide) and the comparator estimate pools four drug classes. The pre-post drop is confounded by weight loss/regression; the placebo-controlled NS result is the honest class verdict. Two-directional (pre-post down vs placebo NS vs comparator-worse) -> mixed.
Comparatorsplacebo; insulin; metformin; SGLT-2 inhibitor; DPP-4 inhibitor
GLP-1 RA class MARKETED
Safety signals Increase
Low evidence

In adults aged 65+ with T2D, GLP-1 RA initiation carried a modestly increased fragility-fracture risk versus other glucose-lowering drugs.

GLP-1 RA and fragility fracture in adults >=65 (cohort, n=46,177), J Clin Endocrinol Meta... Source
PopulationPeer-reviewed primary (women's-health wave)
Fundingacademic / investigator (not separately extracted)
Scope limitsModest and uncertain: observational, active-comparator-confounded (SGLT2i carries its own fracture history), CI lower bound hugs 1.01. SHIFTS the verdict off neutral-reassurance but does NOT establish a fracture excess - not a fracture alarm.
GLP-1 RA class MARKETED
Safety signals Not directional
Low evidence

GLP-1 RA-associated hair loss is markedly female-predominant (men about two-thirds less likely to report it), and more severe with tirzepatide.

GLP-1 RA hair-loss sex-skew (cross-sectional, n=254), J Cosmet Dermatol 2026 Source
PopulationPeer-reviewed primary (women's-health wave)
Fundingacademic / investigator (not separately extracted)
Scope limitsSelf-report, single-country; the female predominance is partly adjusted for the cohort skew. Quantifies the sex-skew beyond the SURMOUNT-1 ~7% vs ~0.5% datum. DEEPENS the alopecia row.
GLP-1 RA class MARKETED
Safety signals Decrease
Low evidence

In a Danish nationwide register of people with type-2 diabetes and hereditary haemochromatosis, each year of GLP-1 receptor agonist exposure was associated with...

Bain SC, Hansen TK et al. Glucose-lowering drugs and serum ferritin in haemochromatosis a... Source
Full findingIn a Danish nationwide register of people with type-2 diabetes and hereditary haemochromatosis, each year of GLP-1 receptor agonist exposure was associated with roughly 30% lower serum ferritin - i.e. GLP-1 RAs are associated with LOWER iron stores, not higher.
PopulationDanish nationwide register (observational cohort), 439 people with hereditary haemochromatosis + type-2 diabetes, 6,529 ferritin measurements 2008-2021; only ~36 GLP-1-exposed contributing ~366 samples.
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitspossible confounding by indication
Comparatorsno GLP-1 RA treatment; SGLT-2 inhibitor
GLP-1 RA class MARKETED
Safety signals Increase
Low evidence

In the same register, SGLT-2 inhibitors RAISED serum ferritin, so the headline GLP-1-vs-SGLT-2 gap (~40% at HH year 4) is partly a comparator artefact: part of the...

Bain SC, Hansen TK et al. Glucose-lowering drugs and serum ferritin in haemochromatosis a... Source
Full findingIn the same register, SGLT-2 inhibitors RAISED serum ferritin, so the headline GLP-1-vs-SGLT-2 gap (~40% at HH year 4) is partly a comparator artefact: part of the apparent GLP-1 'lowering' is relative to a drug that raises ferritin, which overstates GLP-1 lowering versus no treatment.
PopulationDanish nationwide register (observational cohort), people with hereditary haemochromatosis + type-2 diabetes; SGLT-2 inhibitor exposure arm.
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitspossible confounding by indication
ComparatorsGLP-1 RA; no treatment
GLP-1 RA class MARKETED
Safety signals Not directional
Low evidence

The register study did NOT establish a mechanism for the lower ferritin; the authors flag possible mediation via glycaemic control and body weight, and the model did...

Bain SC, Hansen TK et al. Glucose-lowering drugs and serum ferritin in haemochromatosis a... Source
Full findingThe register study did NOT establish a mechanism for the lower ferritin; the authors flag possible mediation via glycaemic control and body weight, and the model did not adjust for weight loss, CRP/inflammation or HbA1c - so the association cannot yet be attributed to a direct iron-handling effect.
PopulationDanish nationwide register (observational cohort), people with hereditary haemochromatosis + type-2 diabetes.
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitspossible confounding by indication
Comparatorsno treatment
GLP-1 RA class MARKETED
Safety signals Decrease
Low evidence

A narrative review reports that GLP-1 RA users have 26-30% lower ferritin than SGLT2i comparators with frequent iron depletion, attributing it to appetite suppression,...

Urbina et al., GLP-1 receptor agonists and iron status, Clin Obes 2026;16:e70070 Source
Full findingA narrative review reports that GLP-1 RA users have 26-30% lower ferritin than SGLT2i comparators with frequent iron depletion, attributing it to appetite suppression, delayed gastric emptying and altered absorption.
PopulationNarrative review (Urbina et al., Clin Obes 2026) summarising GLP-1 RA iron-status literature.
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsSecondary / narrative source; rests largely on the single Bain register. Confidence medium. Review, not primary data.
ComparatorsSGLT-2 inhibitor
GLP-1 RA class MARKETED
Safety signals No change
Low evidence

The lowering effect is NOT universal: a 700-patient retrospective GLP-1 analogue cohort found no significant change in ferritin (haemoglobin fell 0.2 g/dL and 8.4%...

Effect of GLP-1 receptor agonists on haematological and iron indices: a retrospective coh... Source
Full findingThe lowering effect is NOT universal: a 700-patient retrospective GLP-1 analogue cohort found no significant change in ferritin (haemoglobin fell 0.2 g/dL and 8.4% developed anaemia), which is counter-evidence to the class-wide ferritin-lowering claim.
Population700-patient retrospective observational cohort of GLP-1 analogue users (Saudi Med J 2025); ferritin a secondary endpoint.
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsFerritin was a secondary endpoint with no p-value or magnitude reported; possible underpowering. Counter-evidence to C-CLASS-IRON-01. Observational cohort.
Comparatorsbaseline
GLP-1 RA class MARKETED
Safety signals Not directional
Low evidence

Interpretive key bounding all direction findings: ferritin is a positive acute-phase reactant, so an elevated value may reflect inflammation not iron stores, and a...

Urbina et al., GLP-1 receptor agonists and iron status, Clin Obes 2026;16:e70070 Source
Full findingInterpretive key bounding all direction findings: ferritin is a positive acute-phase reactant, so an elevated value may reflect inflammation not iron stores, and a fall during weight loss / reduced steatosis on these drugs may partly reflect resolving inflammation rather than only true iron depletion.
PopulationNarrative review (Urbina et al., Clin Obes 2026) plus acute-phase-reactant physiology (AASLD Liver Fellow Network; PMC2893236/PMC5878890/PMC5223018).
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsInterpretive key; bounds the meaning of every direction finding in this cluster. Note AASLD Liver Fellow Network and PMC2893236/PMC5878890/PMC5223018. Review-level.
Comparatorsnot applicable
GLP-1 RA class MARKETED
Safety signals Decrease
Low evidence

The same narrative review reports vitamin-D deficiency as the most common micronutrient abnormality on GLP-1 RA therapy (7.5% at 6 mo / 13.6% at 12 mo in an n=69...

Urbina et al., GLP-1 receptor agonists and iron status, Clin Obes 2026;16:e70070 Source
Full findingThe same narrative review reports vitamin-D deficiency as the most common micronutrient abnormality on GLP-1 RA therapy (7.5% at 6 mo / 13.6% at 12 mo in an n=69 sub-study) and intakes below the DRI for calcium and iron in >60% of patients - but these are intake/small-substudy data, not serum status.
PopulationNarrative review (Urbina et al., Clin Obes 2026) summarising GLP-1 RA micronutrient-status literature; the vitamin-D-deficiency rates from an n=69 sub-study.
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsCONFIDENCE medium; low-grade (narrative review of INTAKE not serum). Same source + domain as the iron finding C-CLASS-IRON-04 (bumps Urbina n_findings to 4). B12, folate and magnesium are NOT verified by this source - explicit gaps (routed to OQ-BMD-C). Do not read the 7.5%/13.6% deficiency rates as a cohort-wide serum prevalence.
ComparatorsSGLT-2 inhibitor
GLP-1 RA class MARKETED
Safety signals Mixed
Low evidence

In a large nationwide cohort, periconceptional GLP-1 exposure raised preterm-birth risk only among diabetes-indication users, not weight-management users - implicating...

Periconceptional GLP-1 exposure and pregnancy outcomes (Danish nationwide cohort), Hum Re... Source
Full findingIn a large nationwide cohort, periconceptional GLP-1 exposure raised preterm-birth risk only among diabetes-indication users, not weight-management users - implicating the underlying disease rather than the drug.
PopulationPeer-reviewed primary (women's-health wave)
Fundingacademic / investigator (not separately extracted)
Scope limitsFills the non-diabetic-obese pregnancy gap the prior T2D-only cohort missed. Observational, 529 exposed, residual confounding by glycaemic control; preterm birth not malformation. DEEPENS the class pregnancy rows.
GLP-1 RA class MARKETED
Safety signals No change
Low evidence

NET VERDICT (bone / fracture) - NEUTRAL-to-mildly-negative and LOW-grade: GLP-1R agonists show neutral-to-mildly-negative (non-significant) BMD/turnover effects, but...

Consolidation of GLP-1RA/incretin bone-health appraisal with the rapid-weight-loss/sarcop...
Full findingNET VERDICT (bone / fracture) - NEUTRAL-to-mildly-negative and LOW-grade: GLP-1R agonists show neutral-to-mildly-negative (non-significant) BMD/turnover effects, but fracture risk does NOT appear increased at clinically relevant doses; dual/triple incretin/amylin analogues may even be bone-positive (mostly preclinical). The live concern is RAPID weight loss driving sarcopenia and downstream fracture risk, not a direct osteotoxic drug effect.
PopulationBone-health appraisal/review across GLP-1RA + incretin analogues; preclinical for dual/triple bone-positive signals.
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitssurrogate/exploratory endpoint
Comparatorsplacebo; post-bariatric-surgery populations; lifestyle weight loss
GLP-1 RA class MARKETED
Safety signals No change
Low evidence

Bone-health appraisal: GLP-1R agonists show neutral-to-mildly-negative (non-significant) effects on bone turnover/BMD, but fracture risk does NOT appear increased at...

Anastasilakis AD et al., Diabetes Obes Metab 2025 Source
Full findingBone-health appraisal: GLP-1R agonists show neutral-to-mildly-negative (non-significant) effects on bone turnover/BMD, but fracture risk does NOT appear increased at clinically relevant doses; dual/triple incretin and amylin analogues may be bone-positive (mostly preclinical).
PopulationCritical-appraisal review of anti-obesity medications and bone metabolism (clinical + preclinical)
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsDistinguishes pharmacological weight loss from post-bariatric bone loss/fracture increase. Concurring earlier review: Herrou/Paccou PMID:37999750.
GLP-1 RA class MARKETED
Safety signals Decrease
Low evidence

Vs insulin, GLP-1RA initiation associated with LOWER liver and pancreatic cancer risk and no increased risk for most other major cancers, in a large T2D cohort.

Rashid Z et al., J Gen Intern Med 2026 Source
PopulationRetrospective cohort, IBM-MarketScan, T2D 2013-2021, N=106,088, weighted vs insulin
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitspossible confounding by indication
Comparatorsinsulin
GLP-1 RA class MARKETED
Safety signals Mixed
Low evidence

In obesity, GLP-1RA use associated with LOWER overall cancer risk and reduced endometrial/ovarian/meningioma cancer, but a marginally non-significant INCREASED kidney...

Dai H et al., JAMA Oncol 2025 Source
Full findingIn obesity, GLP-1RA use associated with LOWER overall cancer risk and reduced endometrial/ovarian/meningioma cancer, but a marginally non-significant INCREASED kidney cancer risk.
PopulationTarget-trial emulation, OneFlorida+ EHR 2014-2024, 86,632 matched adults with obesity
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsPer PubMed. Kidney-cancer signal echoes Wang 2024 (HR 1.54 vs metformin); recorded as emerging/contested, not a verdict.
Comparatorsnon-users eligible for anti-obesity medication
GLP-1 RA class MARKETED
Safety signals Mixed
Low evidence

Mechanistic 'double-edged sword' review: incretin therapies linked in some studies to increased pancreatic/thyroid/cholangiocarcinoma/colorectal cancer (esp....

Mavaddat H et al., Mol Biol Rep 2025 Source
Full findingMechanistic 'double-edged sword' review: incretin therapies linked in some studies to increased pancreatic/thyroid/cholangiocarcinoma/colorectal cancer (esp. genetically predisposed), while showing anticancer effects in prostate, breast, ovarian and others.
PopulationNarrative review of clinical cohorts + in vivo/in vitro
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsPer PubMed. Opposing-direction hypotheses (beta-cell proliferation concern vs antitumour mechanisms); both poles.
GLP-1 RA class MARKETED
Safety signals Mixed
Low evidence

NET VERDICT (thyroid C-cell / MTC) - a RODENT-based FDA boxed warning and class contraindication (personal/family MTC history, or MEN 2) coexists with a...

Consolidation of the FDA semaglutide/tirzepatide boxed warning with human cohort + pharma...
Full findingNET VERDICT (thyroid C-cell / MTC) - a RODENT-based FDA boxed warning and class contraindication (personal/family MTC history, or MEN 2) coexists with a NULL-to-unconfirmed HUMAN signal. The regulatory contraindication stands as a labelling FACT while no human MTC excess is confirmed - the rodent C-cell tumourigenic effect does NOT translate (human calcitonin screening negative; null cohorts dominate over reporting-driven signal poles).
PopulationRodent toxicology (mice/rats) underpinning labels; human pharmacovigilance + national active-comparator cohorts (SNDS, Scandinavian, Korean) + FAERS; prospective calcitonin screening (liraglutide programme).
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsanimal data; human relevance uncertain; identifier not fully verified
Comparatorsplacebo; DPP-4 inhibitors; other antidiabetics; no-treatment cohorts
GLP-1 RA class MARKETED
Safety signals Increase
Low evidence

SIGNAL pole: French SNDS nested case-control found GLP-1 RA use for 1-3 years associated with increased all-thyroid-cancer AND medullary thyroid cancer risk.

Bezin J et al., Diabetes Care 2023 Source
Population2562 thyroid-cancer cases vs 45,184 controls; T2D on second-line antidiabetics 2006-2018, France
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitspossible confounding by indication
ComparatorsGLP-1 RA non-users among T2D
GLP-1 RA class MARKETED
Safety signals No change
Low evidence

REFUTING pole: Scandinavian three-country active-comparator cohort found no substantially increased thyroid cancer risk vs DPP-4 inhibitors; MTC estimate...

Pasternak B et al., BMJ 2024 Source
Full findingREFUTING pole: Scandinavian three-country active-comparator cohort found no substantially increased thyroid cancer risk vs DPP-4 inhibitors; MTC estimate imprecise/null on point estimate.
Population145,410 GLP-1 RA vs 291,667 DPP-4i initiators; Denmark/Norway/Sweden 2007-2021; mean follow-up 3.9 yr
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsNULL/refuting pole against Bezin; upper CI consistent with no more than 31% relative increase.
ComparatorsDPP-4 inhibitors; SGLT2 inhibitors
GLP-1 RA class MARKETED
Safety signals No change
Low evidence

Second NULL pole: Korean nationwide cohort found no association with new-onset cancers incl. pancreatic and MTC; thyroid point estimate >1 but NS.

Kim M et al., Diabetes Metab J 2024 Source
Population7827 propensity-matched (GLP-1 RA / DM control / non-DM control); Korean NHIS 2004-2021; median follow-up 8 yr
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsLonger median follow-up (8 yr).
ComparatorsDM controls; non-DM controls
GLP-1 RA class MARKETED
Safety signals Mixed
Low evidence

NET VERDICT (reproductive) - three threads, distinct grades. (1) ORAL hormonal contraceptive efficacy is REDUCED by tirzepatide (delayed gastric emptying) - a real...

Consolidation of the tirzepatide oral-contraceptive labelled interaction with GLP-1RA fer...
Full findingNET VERDICT (reproductive) - three threads, distinct grades. (1) ORAL hormonal contraceptive efficacy is REDUCED by tirzepatide (delayed gastric emptying) - a real LABELLED interaction (barrier/non-oral advised), labelled for tirzepatide but not injectable semaglutide. (2) FERTILITY: GLP-1RAs improve ovulatory function in obesity/PCOS ('Ozempic babies'), raising conception/unintended-pregnancy risk. (3) PREGNANCY: contraindicated - animal repro-tox shows adverse offspring outcomes, while human cohorts show NO large major-malformation excess vs insulin (reassuring but confounded, not exonerating).
PopulationTirzepatide label (PK interaction); GLP-1RA ovulation/PCOS literature; animal reproductive-toxicology + a multinational human pregnancy-exposure cohort (T2D).
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsanimal data; human relevance uncertain; identifier not fully verified
Comparatorsinsulin (pregnancy cohort); non-oral contraceptive methods; pre-treatment baseline (ovulation)
GLP-1 RA class MARKETED
Safety signals No change
Low evidence

In a large real-world cohort, GLP-1 RAs were gout-neutral on clinical OUTCOMES while SGLT2 inhibitors reduced incident gout and outperformed GLP-1 RAs head-to-head -...

Preston FG et al. SGLT2 inhibitors, but not GLP-1 receptor agonists, reduce incidence of ... Source
Full findingIn a large real-world cohort, GLP-1 RAs were gout-neutral on clinical OUTCOMES while SGLT2 inhibitors reduced incident gout and outperformed GLP-1 RAs head-to-head - the biochemical urate-lowering did not translate into a gout-incidence benefit for GLP-1 RAs.
PopulationTriNetX federated EHR cohort (Preston et al.), T2D on metformin/insulin +/- SGLT2i or GLP-1 RA, propensity-matched, 5-yr gout incidence.
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsObservational/EHR, coding-based gout outcomes, confounding by indication. The gout-NEUTRAL summary sub-claim for GLP-1 RAs was 2-1 (not unanimous) - encoded as no-change with that caveat. SGLT2i contrast pole is the genuinely uricosuric comparator.
Comparatorsmetformin; insulin; SGLT-2 inhibitor
GLP-1 RA class MARKETED
Safety signals Increase
Low evidence

A 4-patient case series reported paradoxical transient hyperuricaemia and acute gout flares during rapid GLP-1/GIP-driven weight loss, including a recurrent flare...

Chaaya JA et al. Transient increase in serum uric acid and gout attacks after weight-loss... Source
Full findingA 4-patient case series reported paradoxical transient hyperuricaemia and acute gout flares during rapid GLP-1/GIP-driven weight loss, including a recurrent flare despite urate-lowering therapy and normal uric acid - a hypothesis-generating counter-signal to the net urate-lowering RCT data.
PopulationAACE Endocrinology & Diabetes 2026 observational case series / uncontrolled cohort (Chaaya et al.); 4 adults on tirzepatide or semaglutide with substantial weight loss.
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsVery low grade: n=4, uncontrolled, hypothesis-generating; proposed mechanism is ketosis-related renal urate retention during rapid loss. EXCLUDED sub-claim: the '3/4 rose within 3-5 months regardless of baseline urate' statement was REFUTED 0-3 in adversarial review and is deliberately NOT encoded. Contrasts with the net urate-lowering RCT pole.
Comparatorsnone (uncontrolled observational case series)
GLP-1 RA class MARKETED
Safety signals Increase
Low evidence

A real-world cohort found a small but statistically significant HIGHER gout/colchicine signal in GLP-1 users versus non-users - not protective - consistent with a...

Lam DCH et al. Assessing gout risk associated with GLP-1 therapy in obese patients with t... Source
Full findingA real-world cohort found a small but statistically significant HIGHER gout/colchicine signal in GLP-1 users versus non-users - not protective - consistent with a transient-flare or detection-bias effect rather than causal urate-raising.
PopulationTriNetX observational cohort (Lam et al.), overweight/obese T2D, GLP-1 users vs non-users.
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsObservational, confounding by indication, absolute difference only 0.2 percentage points; colchicine is an indirect flare proxy (the colchicine sub-claim was 2-1). Consistent with the transient-flare phenomenon / detection bias, NOT demonstrated causal urate-raising. Both-poled against the net urate-lowering RCT data.
Comparatorsnon-users
GLP-1 RA class MARKETED
Safety signals Mixed
Low evidence

In mechanistic renal-physiology trials, acute exenatide transiently raised renal urate excretion without lowering plasma urate, and prolonged liraglutide/lixisenatide...

Tonneijck L et al. Effect of immediate and prolonged GLP-1 receptor agonist administratio... Source
Full findingIn mechanistic renal-physiology trials, acute exenatide transiently raised renal urate excretion without lowering plasma urate, and prolonged liraglutide/lixisenatide produced no urate change - indicating only a marginal, transient tubular effect, not chronic urate-lowering.
PopulationGLP-1 RA renal-physiology trials (Tonneijck et al., post hoc of 4 controlled phase-4 trials; Study-A n=9 healthy, Study-B n=52, Study-C n=36, Study-D n=35 T2D).
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsSmall mechanistic cohorts, normal-urate/low-CV-risk T2D; exploratory (no sample-size calculation). Two-directional: transient uricosuria (excretion up) but plasma urate unchanged/slightly up -> mixed. Authors conclude urate is unlikely to mediate GLP-1 RA cardio-renal benefit. Design = observational cohort -> low (NOT promoted to moderate).
Comparatorsplacebo; insulin glulisine
GLP-1 RA class MARKETED
Safety signals Increase
Very low evidence

NET VERDICT (acute kidney injury) - a LOW/very-low-grade pharmacovigilance signal that is mostly VOLUME-DEPLETION-mediated, i.e. secondary to dehydration from GI...

Consolidation of GLP-1RA AKI pharmacovigilance (volume-depletion mechanism) + comparative...
Full findingNET VERDICT (acute kidney injury) - a LOW/very-low-grade pharmacovigilance signal that is mostly VOLUME-DEPLETION-mediated, i.e. secondary to dehydration from GI adverse events (vomiting/diarrhoea/poor intake), NOT a primary nephrotoxic effect. Tirzepatide reported less than semaglutide (reporting, not incidence). Higher risk in pre-existing renal impairment. This is the AE signal; longer-term renal OUTCOMES are favourable and live in the renal domain.
PopulationFAERS pharmacovigilance + case literature; comparative tirz-vs-sema FAERS.
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsidentifier not fully verified
Comparatorssemaglutide (for tirzepatide comparison); background FAERS reporting
GLP-1 RA class MARKETED
Safety signals Increase
Very low evidence

GLP-1RA-associated acute kidney injury reported in pharmacovigilance/case literature, proposed mechanism dehydration from vomiting/diarrhoea/poor intake; renal...

Dong & Sun, Front Endocrinol 2022 (GLP-1RA AKI FAERS disproportionality) Source
Full findingGLP-1RA-associated acute kidney injury reported in pharmacovigilance/case literature, proposed mechanism dehydration from vomiting/diarrhoea/poor intake; renal disorders the second-largest AE category after GI in one case series.
PopulationPost-marketing pharmacovigilance + case reports/review
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsdisproportionality: reporting != incidence/causality
GLP-1 RA class MARKETED
Safety signals Mixed
Very low evidence

NET VERDICT (broader cancer) - observational cohorts (confounded by indication) lean reassuring, with a protective pattern for obesity-associated malignancies vs...

Consolidation of GLP-1RA cancer-association cohorts (vs insulin/metformin) with FAERS tum...
Full findingNET VERDICT (broader cancer) - observational cohorts (confounded by indication) lean reassuring, with a protective pattern for obesity-associated malignancies vs insulin (lower liver/pancreatic and 10/13 obesity-associated cancers; lower overall incl. endometrial/ovarian/meningioma in obesity). This is NOT a protective claim and NOT a net-harm verdict: it sits against a contested marginal kidney-cancer signal, no reduction for postmenopausal breast/thyroid, residual pancreatic pharmacovigilance noise, and a mechanistic double-edged-sword hypothesis.
PopulationLarge T2D and obesity observational cohorts + FAERS + mechanistic review.
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitspossible confounding by indication
Comparatorsinsulin; metformin; DPP-4 inhibitors; non-users
GLP-1 RA class MARKETED
Safety signals Increase
Very low evidence

FOLK-CLAIM VERDICT ('hair loss/shedding around month 3-4'): TRUE that it happens / OVERSTATED as a drug toxicity. Alopecia is a real, RCT-replicated signal (RR ~3.25),...

Cheng & Chang. GLP-1 receptor agonists and hair loss: a meta-analysis. Diabetes Res Clin ... Source
Full findingFOLK-CLAIM VERDICT ('hair loss/shedding around month 3-4'): TRUE that it happens / OVERSTATED as a drug toxicity. Alopecia is a real, RCT-replicated signal (RR ~3.25), and the month 3-4 timing is textbook TELOGEN EFFLUVIUM. But the weight of evidence attributes it predominantly to RAPID WEIGHT LOSS + caloric/nutritional deficit (the classic trigger), NOT a direct drug-on-follicle toxicity - i.e. the drug is largely the vehicle for rapid weight loss. It is REVERSIBLE, dose-/weight-loss-magnitude-dependent and female-biased. Mitigation overlaps the muscle-loss fix (adequate protein + micronutrients: iron, zinc, B12, vitamin D).
PopulationMeta-analysis (7 RCTs) + SURMOUNT-1 trial alopecia rates + FAERS disproportionality + dermatology systematic reviews.
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsdisproportionality: reporting != incidence/causality
Comparatorsplacebo
GLP-1 RA class MARKETED
Safety signals Mixed
Very low evidence

NET VERDICT (gastroparesis / ileus / intestinal obstruction) - a WEAK, low-grade signal: the class delays gastric emptying (on-target, prominent in tolerability) and...

GI Safety Assessment of GLP-1 RAs from FAERS, Diagnostics (MDPI) 2024 Source
Full findingNET VERDICT (gastroparesis / ileus / intestinal obstruction) - a WEAK, low-grade signal: the class delays gastric emptying (on-target, prominent in tolerability) and spontaneous reports surface ileus/obstruction events, but randomised-trial-level evidence does NOT confirm elevated mechanical-obstruction risk and dedicated FAERS work found only weak/non-significant signals. Honest net: low-grade, trial-unconfirmed.
PopulationFAERS spontaneous-report databases + a randomised-trials network meta-analysis for the obstruction endpoint. No reta-direct.
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsdisproportionality: reporting != incidence/causality
Comparatorsnon-GLP-1-RA reports (FAERS reference); placebo/active (network meta-analysis)
GLP-1 RA class MARKETED
Safety signals Mixed
Very low evidence

FOLK-CLAIM VERDICT ('libido changes, both up and down'): UNPROVEN / MIXED - no controlled evidence of a direct GLP-1 libido effect in either direction. The only...

Narrative review of GLP-1 RAs and sexual function (men and women): dulaglutide crossover ... Source
Full findingFOLK-CLAIM VERDICT ('libido changes, both up and down'): UNPROVEN / MIXED - no controlled evidence of a direct GLP-1 libido effect in either direction. The only RCT-level human datapoint (dulaglutide crossover) was null; FAERS was non-significant; the plausible 'up' direction is confounded by weight-loss-driven testosterone normalisation and improved body image, and the 'down' direction lacks a significant controlled signal. Bidirectional anecdotes are genuine experiences but do not establish a pharmacological effect.
PopulationOne null dulaglutide RCT (lean men) + FAERS disproportionality + rodent data + online self-report; weight-loss/testosterone confound.
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsanimal data; human relevance uncertain; disproportionality: reporting != incidence/causality
Comparatorsplacebo
GLP-1 RA class MARKETED
Safety signals Mixed
Very low evidence

FAERS tumour disproportionality (2004-2021): NO overall increase at system-organ-class level, but significant signals for MTC, papillary thyroid cancer, malignant...

Yang Z et al., Front Pharmacol 2022 Source
Full findingFAERS tumour disproportionality (2004-2021): NO overall increase at system-organ-class level, but significant signals for MTC, papillary thyroid cancer, malignant pancreatic neoplasms and islet-cell neoplasms.
Population8718 GLP-1 RA tumour reports; FAERS Q1 2004-Q2 2021; disproportionality
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsdisproportionality: reporting != incidence/causality
ComparatorsFAERS background
GLP-1 RA class MARKETED
Safety signals Increase
Very low evidence

FAERS: liraglutide carried the greatest pancreatitis reporting signal of the class; semaglutide greatest for nausea/vomiting/diarrhoea/constipation.

Liu L et al., Front Endocrinol 2022 Source
PopulationFAERS Jan 2018-Sep 2022; 21,281 GI reports; disproportionality
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsdisproportionality: reporting != incidence/causality
ComparatorsFAERS background
GLP-1 RA class MARKETED
Safety signals Mixed
Very low evidence

CROSS-SIGNAL STRUCTURE (the dossier spine) - the class safety signals ranked by EVIDENCE GRADE, not by alarm. HIGH/REAL (RCT or consistent replication):...

Structural synthesis over the safety-other corpus (the per-signal consolidations); the do...
Full findingCROSS-SIGNAL STRUCTURE (the dossier spine) - the class safety signals ranked by EVIDENCE GRADE, not by alarm. HIGH/REAL (RCT or consistent replication): gallbladder/biliary (weight-loss-coupled); rapid-glucose-lowering early diabetic-retinopathy worsening (SUSTAIN-6 HR 1.76; note steady-state pooled DR is NULL, OR 1.04, TSA-sufficient); peri-operative retained-gastric-contents SURROGATE (the surrogate is robust ~6-fold but actual pulmonary aspiration is unproven). CLEARED/NULL in humans: medullary thyroid carcinoma (rodent-basis boxed warning, no human MTC signal); suicidality (regulatory-cleared). DEBATED: NAION (cohorts split; RCT meta OR 3.92 but TSA-insufficient; label 'very rare'). WEAK/pharmacovigilance-only (reporting != incidence): pancreatitis, ileus, acute kidney injury (mostly volume-depletion-mediated).
PopulationClass-wide across GLP-1 mono, GIP/GLP-1 dual, triple, and amylin co-agonists; strength/population vary by signal.
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsanimal data; human relevance uncertain; post-hoc (not prespecified); surrogate/exploratory endpoint
Comparatorsplacebo; other antidiabetics; SGLT2i; non-GLP-1RA
GLP-1 RA class MARKETED
Body composition Mixed
Moderate evidence

Network meta-analysis (22 RCTs, 2258 participants): GLP-1RAs reduced absolute lean mass by ~0.86 kg (~25% of weight lost), but the RELATIVE proportion of body weight...

Karakasis P et al., Metabolism 2024 Source
Full findingNetwork meta-analysis (22 RCTs, 2258 participants): GLP-1RAs reduced absolute lean mass by ~0.86 kg (~25% of weight lost), but the RELATIVE proportion of body weight that is lean was unaffected. The same dataset supports both poles of the debate: 'relative lean unchanged' feeds the adaptive reading; the absolute -0.86 kg and the tirz/sema-least-preserving ordering feed the caution reading.
PopulationSystematic review + network meta-analysis, 22 RCTs, N=2258, adults with diabetes and/or overweight/obesity; DXA/BIA.
Fundingacademic / independent - no specific grant funding
Scope limitsNEW. The numeric backbone both poles cite; meta-analysis graded moderate by default. Entirely a MASS meta-analysis: no strength/gait/SPPB pooled because the component trials did not measure them. Consistent with the Annals review (C-CLASS-BODYCOMP-01, ~25-28% lean fraction). COROLLARY (Council Methodologist): 'relative lean unchanged' is a COROLLARY of the fixed lean-fraction-of-weight-lost property (cf. SURMOUNT-1 placebo parity), NOT an independent adaptive signal, and does not address absolute lean-kg or function. Cross-ref C-CLASS-BODYCOMP-01.
Comparatorsplacebo; liraglutide; semaglutide; tirzepatide; dulaglutide; exenatide
GLP-1 RA class MARKETED
Body composition Decrease
Moderate evidence

FOLK-CLAIM VERDICT ('I'm losing MUSCLE not just fat; eat protein and lift'): MIXED, leaning TRUE-but-reframed. Lean-mass loss is real (~25-40% of weight lost is...

Batsis JA et al., Ann Intern Med, 2026 Source
Full findingFOLK-CLAIM VERDICT ('I'm losing MUSCLE not just fat; eat protein and lift'): MIXED, leaning TRUE-but-reframed. Lean-mass loss is real (~25-40% of weight lost is fat-free mass), but the 'losing muscle' framing overstates it - most is obligatory for any large weight loss, the split mirrors placebo/lifestyle, and the LEAN FRACTION is NOT worse with the more potent agents (incl. retatrutide), only the absolute kg. The protein+resistance-training rule is mechanistically sound but its GLP-1-specific trial evidence is MIXED. The genuine concern is functional (sarcopenia in older/sarcopenic-obese patients) and is UNDER-MEASURED (no trial assessed physical function).
PopulationSystematic review (35 RCTs) + DXA substudies of SURMOUNT-1 (tirzepatide) and the reta phase-2 T2D trial.
Fundingacademic-none/no external funding (independent)
Scope limitsGRADE moderate (systematic review of RCT DXA data). MATCHED to the D2 lean-mass thread; this row adds the folk-verdict + the reframe (fraction not worse with potency; function under-measured) + the new systematic-review anchor. Belief-vs-reality: the scale drop is ~25-40% fat-free mass (expected, similar to placebo/lifestyle), mostly benign with composition improving - but the older/sarcopenic functional risk is real and untested, and protein+lifting is the right hedge. Reta loses most weight -> most ABSOLUTE lean kg, but not a worse fraction. Cross-ref the D2 body-composition rows.
Comparatorsplacebo; lifestyle weight loss
GLP-1 RA class MARKETED
Body composition No change
Moderate evidence

In T2D-only RCTs, GLP-1 RAs showed no fracture excess (RR 0.80, NS) and no detectable BMD loss (BMD point estimates modestly higher than control) - a...

GLP-1 RA bone density and turnover (25-RCT meta-analysis), Acta Diabetol 2025 Source
Full findingIn T2D-only RCTs, GLP-1 RAs showed no fracture excess (RR 0.80, NS) and no detectable BMD loss (BMD point estimates modestly higher than control) - a reassurance-of-absence in diabetic bone, NOT a demonstrated bone-building effect. This does NOT generalise to obesity/weight-loss populations, where the direct double-blind Hansen RCT and DXA cohorts show resorption-dominant BMD loss.
PopulationPeer-reviewed primary (women's-health wave)
Fundingacademic / investigator-initiated; no industry sponsor found in indexed metadata (Europe PMC fundingList empty)
Scope limitsREMEDIATED (WI-BMD-01): re-scoped from a near-anabolic class reading to a CONTESTED, T2D-meta-analytic-specific pole. T2D bone tends high-density/low-quality, so a meta of T2D RCTs can report 'no loss / modestly higher BMD' while the direct double-blind weight-loss RCT (Hansen, C-CLASS-BONE-MECH-RCT-01: P1NP null, CTX up, lumbar/hip aBMD down) and obesity cohorts (C-CLASS-BONE-BMD-COHORT-01) show loss. The discordance is explained by population (T2D vs obesity) and weight-loss magnitude; the higher-BMD reading must NOT be transferred to the rapid-weight-loss obesity setting. The fracture RR 0.80 NS is real and retained.
GLP-1 RA class MARKETED
Body composition Decrease
Moderate evidence

In a 52-week double-blind RCT in adults with increased fracture risk (mostly postmenopausal), semaglutide 1.0 mg did NOT raise bone formation (P-PINP null) but RAISED...

Hansen MS, Wolfel EM, Jeromdesella S, et al. Once-weekly semaglutide versus placebo in ad... Source
Full findingIn a 52-week double-blind RCT in adults with increased fracture risk (mostly postmenopausal), semaglutide 1.0 mg did NOT raise bone formation (P-PINP null) but RAISED bone resorption (CTX) and LOWERED lumbar-spine and total-hip BMD versus placebo - a resorption-dominant pattern, not bone-building.
PopulationSemaglutide bone phase-2 double-blind RCT (Hansen et al., EClinicalMedicine 2024, NCT04702516); 64 adults at increased fracture risk (T-score < -1.0 and/or recent low-energy fracture), two Danish hospitals.
Fundingacademic/investigator-initiated (Region of Southern Denmark + Novo Nordisk Foundation + Gangsted Foundation; Novo Nordisk supplied study drug only, not sponsor)
Scope limitsTHE direct double-blind DXA/turnover RCT counterweight to the T2D meta (C-CLASS-BONE-BMD-META-01 / -TURNOVER-01): shows resorption-dominant loss, NOT the meta's favourable-turnover/higher-BMD reading. Graded moderate (phase-2), but it is a direct double-blind randomised DXA + bone-turnover measurement, methodologically stronger than the meta's pooled T2D RCTs for the obesity/weight-loss question. UNSETTLED whether drug-intrinsic or weight-loss-mediated - NO diet-only comparator arm, and the authors attribute the CTX rise to the accompanying weight loss. Postmenopausal/modest n; not an obesity-dose (1.0 mg) trial. Corrects and re-scopes the contested anabolic pole.
Comparatorsplacebo
GLP-1 RA class MARKETED
Body composition Mixed
Moderate evidence

In T2D, the meta reports a favourable-looking turnover shift (beta-CTX down, formation markers up), but this is CONTESTED as a generalisable bone-building effect -...

GLP-1 RA bone density and turnover (25-RCT meta-analysis), Acta Diabetol 2025 Source
Full findingIn T2D, the meta reports a favourable-looking turnover shift (beta-CTX down, formation markers up), but this is CONTESTED as a generalisable bone-building effect - likely low-turnover diabetic-bone normalisation. It is directly contradicted by the Hansen double-blind RCT in weight-losing adults (CTX UP, P1NP null) and so must NOT be read as a class anabolic effect.
PopulationPeer-reviewed primary (women's-health wave)
Fundingacademic / investigator-initiated; no industry sponsor found in indexed metadata (Europe PMC fundingList empty)
Scope limitsREMEDIATED (WI-BMD-01): the favourable-turnover reading is re-scoped to a CONTESTED T2D-specific pole and explicitly REFUTED as a generalisable anabolic effect by the WI-BMD-01 deep-research pass. Most plausibly normalising low-turnover diabetic bone, NOT bone-building; directly OPPOSED by the direct double-blind weight-loss RCT (Hansen, C-CLASS-BONE-MECH-RCT-01: resorption marker CTX UP, formation marker P1NP null, BMD down) and the obesity BMD-loss cohorts. Do NOT transfer to the rapid-weight-loss obesity setting.
GLP-1 RA class MARKETED
Body composition Mixed
Low evidence

Narrative review of human studies on GLP-1 therapies and sarcopenia: results are mixed - some studies emphasise excessive skeletal-muscle-mass loss while others argue...

Scheen AJ, Expert Opin Drug Saf, 2026 Source
Full findingNarrative review of human studies on GLP-1 therapies and sarcopenia: results are mixed - some studies emphasise excessive skeletal-muscle-mass loss while others argue for a protective effect via improved muscle quality (less myosteatosis); no definitive conclusion on net detriment or benefit to skeletal muscle.
PopulationHuman studies of GLP-1-based therapies assessing skeletal-muscle mass, strength/function and structure/quality
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsExplicitly preserves BOTH poles of the sarcopenia question (harm vs muscle-quality benefit). Good neutrality anchor for the muscle-quality column. Echoes the SCWD/FDA workshop (PMID 41362110, DOI 10.1002/jcsm.70147) on endpoint challenges.
GLP-1 RA class MARKETED
Body composition Not directional
Low evidence

CLASS GAP (the absence is the finding): direct measurement of muscle FUNCTION (grip strength, gait speed, SPPB, chair-stand) is essentially ABSENT from the pivotal...

Dubin RL, Heymsfield SB, Ravussin E, Greenway FL, Diabetes Obes Metab, 2024 (28-trial gap... Source
Full findingCLASS GAP (the absence is the finding): direct measurement of muscle FUNCTION (grip strength, gait speed, SPPB, chair-stand) is essentially ABSENT from the pivotal incretin weight-loss trials; body composition is captured almost entirely as DXA/MRI MASS. Multiple reviews (Neeland 2024; Dubin/Heymsfield 2024; Conte/Hall/Klein; Drucker 2024) converge that strength/function/mobility were not assessed and call for them. Authoritative safety reviews (Drucker, Diabetes Care 2024, PMID 38843460) list muscle strength as an under-characterised SAFETY domain for the class, including retatrutide, survodutide and MariTide.
PopulationClass-wide gap across STEP / SURMOUNT / SUMMIT / SUSTAIN body-composition substudies; synthesised in reviews + one protocol-stage older-adult RCT.
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsTHE ABSENCE IS THE FINDING. Dubin reviewed 28 GLP-1RA trials, all DXA mass, no functional testing, and call for MRI + functional testing. The two HF-PRO signals (6MWD, KCCQ) are folded into the magnitude with an explicit NOT-muscle-strength tag, so they are captured without miscategorising HF symptom relief as muscle function. Drucker PMID 38843460 folded here. Class-level counterpart to the reta-specific gap (C-RETATRUTIDE-BODYCOMP-03). The folded 6MWD/KCCQ are heart-failure-STATUS PROs confounded by dyspnoea/fluid/symptom relief - NOT gait-speed or strength - and do NOT narrow the muscle-function gap (Council Evidence-grader).
GLP-1 RA class MARKETED
Body composition Not directional
Low evidence

FOLK-CLAIM VERDICT ('Ozempic face'): MIXED - the gaunt/hollow/aged/sagging face is REAL (TRUE), but the attribution to the DRUG is FALSE: it is a WEIGHT-LOSS-mediated...

Barone M et al. GLP-1 receptor agonists and skin quality: a systematic review (incl. faci... Source
Full findingFOLK-CLAIM VERDICT ('Ozempic face'): MIXED - the gaunt/hollow/aged/sagging face is REAL (TRUE), but the attribution to the DRUG is FALSE: it is a WEIGHT-LOSS-mediated effect (non-selective loss of facial fat pads + dermal laxity that fails to retract, worse with age), not a drug-specific facial toxicity. The same occurs with any rapid/large weight loss including bariatric surgery and diet. NOT semaglutide-specific (named for first-mover fame); class-wide - and tirzepatide/retatrutide likely produce MORE facial change purely via GREATER weight loss. Folk preventatives (slower loss, more protein, hydration, fillers/microneedling) are mechanistically coherent with the volume-loss explanation.
PopulationDermatology + plastic-surgery systematic reviews of case reports/cohorts + authoritative clinical commentary (Cleveland Clinic); bariatric-surgery facial-change cohorts as the weight-loss-generic comparator.
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsGRADE low: dermatology/plastic-surgery review of mostly case reports + commentary; facial appearance has NEVER been a controlled GLP-1 trial endpoint, and no head-to-head drug-vs-equivalent-diet-weight-loss facial comparison exists. The directional verdict (weight-loss-mediated, not drug-specific) is consistent and well-supported; a minority direct-drug hypothesis (Kruglikov 2025, dermal white-adipose effect) is unproven. BELIEF-VS-REALITY: people believe 'the drug gives you the face'; reality - losing a lot of weight fast gives you the face, and the drug is just an efficient way to do that. Folk-claim frequency: EXTREMELY COMMON as a cultural term.
Comparatorsbariatric-surgery weight loss; diet weight loss
GLP-1 RA class MARKETED
Body composition Decrease
Low evidence

In an older mostly-female cohort, semaglutide/tirzepatide-associated bone loss tracked weight loss and was greater in non-diabetic patients.

Semaglutide/tirzepatide skeletal effects (DXA cohort, n=255), J Clin Endocrinol Metab 2026 Source
PopulationPeer-reviewed primary (women's-health wave)
Fundingacademic / investigator (not separately extracted)
Scope limitsWEIGHT-LOSS-MAGNITUDE-MEDIATED, DM-status-modified; implies the T2D meta understates loss in obesity/non-diabetic use. Retrospective, single-centre.
GLP-1 RA class MARKETED
Body composition No change
Low evidence

In a 20-week pilot RCT in older adults (n=20), whole-body and regional BMD and bone-turnover markers did not change significantly, but a post-hoc correlation linked...

Dinkla L, Beavers KM, Robbins R, ... Trejo J, Stepanenko A, Cortes TM, et al. Bone minera... Source
Full findingIn a 20-week pilot RCT in older adults (n=20), whole-body and regional BMD and bone-turnover markers did not change significantly, but a post-hoc correlation linked greater weight loss to greater pelvis bone loss (r=0.62) - hypothesis-generating support for weight-loss-mediated loss, not proof.
PopulationOlder-adult GLP-1 bone pilot (Trejo/Cortes et al., Front Aging 2025, NCT05786521); 20 community-dwelling adults aged 65+ with prediabetes/T2D + overweight/obesity, UT Health San Antonio.
Fundingacademic/government (NIH/NIA + San Antonio VA GRECC + Colorado CCTSI; no industry sponsor)
Scope limitsVERY LOW/LOW grade (audit-downgraded from the open-label-RCT rule grade): n=20, pilot, unblinded assessors, whole-body BMD NS, the directional bone-loss trend did not reach significance, and the supportive r=0.62 is a single post-hoc regional correlation. Supports the weight-mediated reading WITHOUT establishing it; do not over-read the r=0.62. Note the paper itself REPEATS the unverified '~5-fold hip/pelvic fracture in 75+ (SELECT)' figure in its discussion - that magnitude is NOT encoded here (open question OQ-BMD-B).
Comparatorslifestyle counselling alone
GLP-1 RA class MARKETED
cns-neuropsychiatric Decrease
High evidence

FOLK-CLAIM VERDICT ('smoking/nicotine cessation or reduction'): UNPROVEN-but-PROMISING (do not call it TRUE). One small positive pilot RCT (exenatide) + a consistent...

Yammine L, Green CE, Kosten TR et al., Nicotine Tob Res, 2021 Source
Full findingFOLK-CLAIM VERDICT ('smoking/nicotine cessation or reduction'): UNPROVEN-but-PROMISING (do not call it TRUE). One small positive pilot RCT (exenatide) + a consistent semaglutide EHR signal + preclinical nicotine-reward data make it biologically coherent and encouraging, but there is no adequately powered confirmatory trial (one is ongoing). Likely a class effect but unproven as such; the EHR data are observational (no actual smoking-status data).
PopulationPilot exenatide+patch RCT (n=84) + semaglutide EHR target-trial emulation (n=222,942); confirmatory RCT pending.
Fundingacademic / NIH - UTHealth Center for Clinical and Translational Sciences (NIH CTSA) + PARTNERS Research Awards; no industry sponsor (prior 'AstraZeneca' inference was WRONG)
Scope limitssmall sample (N~84)
Comparatorsplacebo; nicotine patch alone
GLP-1 RA class MARKETED
cns-neuropsychiatric Decrease
Low evidence

ADDICTION (opioid/cross-substance): in large EHR cohorts of patients with opioid or alcohol use disorder, GIP/GLP-1 receptor agonist prescriptions were associated with...

Qeadan F, McCunn A, Tingey B, Addiction, 2025 (+ exploratory OUD+AUD cohort, Commun Med 2... Source
Full findingADDICTION (opioid/cross-substance): in large EHR cohorts of patients with opioid or alcohol use disorder, GIP/GLP-1 receptor agonist prescriptions were associated with lower rates of opioid overdose and alcohol intoxication. The opioid evidence is OBSERVATIONAL-ONLY (no OUD RCT exists).
PopulationQeadan EHR cohorts (Oracle/Cerner, >100M): N=503,747 OUD + 817,309 AUD (2014-2022); plus an exploratory N=107,217 OUD+AUD-comorbid cohort.
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsfew events/wide CI; possible confounding by indication; surrogate/exploratory endpoint
Comparatorsno GIP/GLP-1 RA prescription
GLP-1 RA class MARKETED
cns-neuropsychiatric Decrease
Low evidence

NEURODEGENERATION - the EPIDEMIOLOGY pole (in tension with the negative RCTs): in a Scandinavian active-comparator cohort, GLP-1RA use was associated with lower...

Engstrom A, Svanstrom H, Hviid A et al., Diabetes Obes Metab, 2026 (Scandinavian PD cohort) Source
Full findingNEURODEGENERATION - the EPIDEMIOLOGY pole (in tension with the negative RCTs): in a Scandinavian active-comparator cohort, GLP-1RA use was associated with lower incident Parkinson's than sulfonylurea use. This is a lower-INCIDENCE association in diabetics (exposure 73% liraglutide), NOT disease modification - the randomised tests (exenatide PD ph3, EVOKE) were NEGATIVE, and reverse causation (prodromal disease reducing drug initiation) is a MATERIAL threat. The mechanistic rationale (brain GLP-1 receptors; reduced amyloid/tau, neuroinflammation; improved central insulin signalling in rodent models) is preclinical and did NOT translate to disease modification in the human RCTs.
PopulationEngstrom Scandinavian cohort (DK/NO/SE registers): N=158,961 new GLP-1RA vs 188,065 new sulfonylurea users, age >=45; exposure dominated by liraglutide (73%).
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsanimal data; human relevance uncertain; possible confounding by indication
Comparatorssulfonylureas
GLP-1 RA class MARKETED
cns-neuropsychiatric Not directional
Low evidence

FOLK-CLAIM VERDICT ('one drink hits like three' / wrecked alcohol tolerance, described as near-universal for retatrutide): NOT SUPPORTED, likely FALSE as stated - and...

Quddos F et al. GLP-1 receptor agonists and acute alcohol pharmacokinetics/subjective eff... Source
Full findingFOLK-CLAIM VERDICT ('one drink hits like three' / wrecked alcohol tolerance, described as near-universal for retatrutide): NOT SUPPORTED, likely FALSE as stated - and DISTINCT from the real, separately-evidenced reduced-craving effect. The only human alcohol-challenge data show LOWER blood-alcohol and LESS intoxication per drink, and the gastric-emptying physiology points the same way. The lived experience is better explained by drinking on a near-empty stomach, far lower desire (stopping at one), GI upset making a drink unpleasant fast, and (chronically) a smaller body-water volume from weight loss - none of which is 'higher blood alcohol per drink'. NO retatrutide-specific human alcohol data exist; the reta-specificity is a community-reporting artefact.
PopulationOne human alcohol-challenge PK pilot (n=20, non-randomised); gastric-emptying/alcohol-PK physiology literature; one rat alcohol-interoception study (incl. retatrutide).
Fundingacademic / investigator-initiated (Quddos et al., alcohol-challenge pilot)
Scope limitsanimal data; human relevance uncertain; small sample (N~20)
Comparatorsnon-GLP-1 controls
GLP-1 RA class MARKETED
cns-neuropsychiatric Decrease
Low evidence

FOLK-CLAIM VERDICT ('reduces compulsive behaviours - shopping, gambling, nail-biting, skin-picking, hair-pulling'): MIXED - the food/alcohol reward base is real and...

Klausen MK, Jensen ME, Moller M et al., JCI Insight, 2022 Source
Full findingFOLK-CLAIM VERDICT ('reduces compulsive behaviours - shopping, gambling, nail-biting, skin-picking, hair-pulling'): MIXED - the food/alcohol reward base is real and mechanistically grounded (MATCHED to D5), but the GENERALISATION to non-consummatory compulsions (gambling, shopping, BFRBs) is UNPROVEN: zero RCTs, social-media/self-report only, plausible shared mesolimbic substrate but not demonstrated. This is belief running ahead of the evidence - and there is a counter-pole (a semaglutide-associated manic-episode case report), so the neuropsychiatric effect is not uniformly 'calming'.
PopulationExenatide AUD RCT (reward base) + a Reddit/social-media compulsive-behaviour study + a skin-picking pharmacology review (GLP-1 absent); no controlled trial in gambling/shopping/BFRBs.
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitssmall sample (N~127)
Comparatorsplacebo
GLP-1 RA class MARKETED
cns-neuropsychiatric Decrease
Very low evidence

ADDICTION MECHANISM (RODENT, not human): GLP-1RAs act on the mesolimbic reward circuit - semaglutide reaches the nucleus accumbens and reduces alcohol-evoked dopamine...

Aranas C et al., EBioMedicine, 2023 (semaglutide/alcohol/NAcc dopamine); Herman RJ et al.... Source
Full findingADDICTION MECHANISM (RODENT, not human): GLP-1RAs act on the mesolimbic reward circuit - semaglutide reaches the nucleus accumbens and reduces alcohol-evoked dopamine release (the canonical reward signal) and intake/relapse; liraglutide reduces nicotine self-administration and relapse-like reinstatement. A biologically-plausible substrate (in rodents) for the human craving/consumption signals - plausibility only, NOT shown to be the operative human mechanism, and it does NOT raise the grade of the human observational rows it is cross-referenced from.
PopulationRats and mice (Aranas semaglutide-alcohol; Herman liraglutide-nicotine); self-administration / microdialysis / locomotion models.
Fundingacademic (Aranas/Herman rodent neuroscience groups; no sponsor disclosure extracted)
Scope limitsanimal data; human relevance uncertain
Comparatorsvehicle/saline
GLP-1 RA class MARKETED
cns-neuropsychiatric No change
Very low evidence

PSYCHIATRIC SAFETY - consolidated verdict (cross-refs the existing safety-other MH rows, not re-gathered): the suicidality/self-harm signal was INVESTIGATED by...

Consolidation of existing safety-other MH rows; new RCT-pooled suicidality meta-analysis ... Source
Full findingPSYCHIATRIC SAFETY - consolidated verdict (cross-refs the existing safety-other MH rows, not re-gathered): the suicidality/self-harm signal was INVESTIGATED by regulators (EMA PRAC, FDA) and LARGELY CLEARED in RCT and meta-analytic data (FDA removed the suicidal-ideation warning from liraglutide/semaglutide-2.4/tirzepatide labels), with residual pharmacovigilance (FAERS/VigiBase) noise and confounding-by-indication in observational data; RCT depression/anxiety symptoms did not increase.
PopulationAdults with T2D/obesity across the class; regulatory reviews, the STEP/SURMOUNT RCT programmes, RCT + observational meta-analyses (27-RCT pool; 82-study synthesis), pharmacovigilance databases.
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitspossible confounding by indication
Comparatorsplacebo; active comparators; non-GLP-1 anti-obesity medications
Thin or bounded evidence here means uncertainty remains visible. It is not evidence that an effect has been ruled out.
GLP-1 RA class MARKETED
Pharmacology and mechanism No change
Low evidence

In a single small human pilot, patients' baseline gut microbiome differed between those who did and did not respond glycaemically to GLP-1 receptor agonists, and an...

Tsai CY, Lu HC, Chou YH, Liu PY, Chen HY, Huang MC, Lin CH, Tsai CN. Gut Microbial Signat... Source
Full findingIn a single small human pilot, patients' baseline gut microbiome differed between those who did and did not respond glycaemically to GLP-1 receptor agonists, and an in-sample model predicted response well - but this is correlational, single-centre, not externally validated, and says nothing about GI tolerability.
Population52 type-2-diabetes patients on liraglutide (n=22) or dulaglutide (n=30); single-centre Taiwanese pilot; baseline 16S microbiome vs glycaemic response (n=52).
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitssmall sample (n=52); single-centre pilot; correlational (baseline-predicts-response); in-sample AUROC (not externally validated)
Comparatorsresponders vs non-responders (within-cohort)
GLP-1 RA class MARKETED
Pharmacology and mechanism No change
Very low evidence

In rodents, liraglutide shifts gut-microbiota composition in a broadly consistent direction (enriching Akkermansia, Bifidobacterium, Lactobacillus), and semaglutide...

Zhao L, Qiu Y, Zhang P, Wu X, Zhao Z, Deng X, Yang L, Wang D, Yuan G. Gut microbiota medi... Source
Full findingIn rodents, liraglutide shifts gut-microbiota composition in a broadly consistent direction (enriching Akkermansia, Bifidobacterium, Lactobacillus), and semaglutide does similarly, but the specific taxa are heterogeneous and the evidence is 16S-only, small, and confounded by reduced food intake.
PopulationHigh-fat-diet C57BL/6 mice, liraglutide (Zhao 2022 Front Nutr; 16S); semaglutide corroboration encoded separately (Feng 2024 PeerJ mice).
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsanimal data; human relevance uncertain; 16S compositional only
Comparatorshigh-fat-diet control; normal-chow control
GLP-1 RA class MARKETED
Pharmacology and mechanism Not directional
Very low evidence

Across the whole compositional literature, no study demonstrates that the gut-microbiome change CAUSES the weight loss from GLP-1 receptor agonists: the designs are...

Zhao L, Chen Y, Xia F, Abudukerimu B, Zhang W, Guo Y, Wang N, Lu Y. A Glucagon-Like Pepti... Source
Full findingAcross the whole compositional literature, no study demonstrates that the gut-microbiome change CAUSES the weight loss from GLP-1 receptor agonists: the designs are associational (16S only), confounded by reduced food intake, and despite some 'mediates' titles, causal weight-loss mediation is unestablished for every agent.
PopulationCompositional GLP-1-RA microbiome studies reviewed as a body (rodent: Wang 2016, Zhao 2018, Zhao 2022, Feng 2024; human: Chen 2025 n=15) for any weight-loss causal/mediation design - none present.
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsanimal data; human relevance uncertain; 16S compositional only
Comparatorsn/a (structural across the compositional corpus)
GLP-1 RA class MARKETED
Pharmacology and mechanism Increase
Very low evidence

Gut microbes make short-chain fatty acids that causally stimulate the body's OWN GLP-1 and PYY (proven with FFAR2/FFAR3-knockout rodents) - but this...

Psichas A, Sleeth ML, Murphy KG, et al. (Frost G, senior). The short chain fatty acid pro... Source
Full findingGut microbes make short-chain fatty acids that causally stimulate the body's OWN GLP-1 and PYY (proven with FFAR2/FFAR3-knockout rodents) - but this endogenous-incretin pathway is a DISTINCT question from how injected GLP-1 receptor agonists work; it does not show the microbiome mediates the drugs' weight loss.
PopulationFFA2/FFAR2 (and FFAR3) knockout mice + Wistar rats; primary murine colonic crypt cultures and in-vivo intra-colonic propionate infusion (Psichas 2015 Int J Obes; Tolhurst 2012 Diabetes, Gribble/Reimann).
Fundingacademic/independent (Imperial College London; MRC/Wellcome; Frost/Bloom lab)
Scope limitsanimal data; human relevance uncertain; mechanistic (endogenous-incretin axis, not injected-drug mechanism)
Comparatorswild-type; FFA2(-/-) / FFAR3(-/-) knockout
GLP-1 RA class MARKETED
Pharmacology and mechanism Increase
Very low evidence

A second knockout study confirms that gut-microbial short-chain fatty acids stimulate the body's own GLP-1 through the FFAR2 receptor - reinforcing the...

Tolhurst G, Heffron H, Lam YS, et al. (Gribble FM, Reimann F, senior). Short-chain fatty ... Source
Full findingA second knockout study confirms that gut-microbial short-chain fatty acids stimulate the body's own GLP-1 through the FFAR2 receptor - reinforcing the endogenous-incretin pathway, which is distinct from the mechanism of injected GLP-1 receptor agonists.
Populationffar2(-/-) and ffar3(-/-) knockout mice + mixed primary murine colonic cultures; in-vitro and in-vivo SCFA-triggered GLP-1 secretion (Tolhurst 2012 Diabetes; Gribble/Reimann, Cambridge).
Fundingacademic/independent (Cambridge; Wellcome Trust/MRC; Gribble & Reimann lab)
Scope limitsanimal data; human relevance uncertain; mechanistic (endogenous-incretin axis, not injected-drug mechanism)
Comparatorswild-type; ffar2(-/-) / ffar3(-/-) knockout
GLP-1 RA class MARKETED
special-populations Not directional
High evidence

No GLP-1 receptor agonist is approved for type-1 diabetes - every use is off-label, consistent with the pivotal-trial investigators concluding the safety signals limit...

Mathieu C, Zinman B, Hemmingsson JU, et al. Efficacy and Safety of Liraglutide Added to I... Source
Full findingNo GLP-1 receptor agonist is approved for type-1 diabetes - every use is off-label, consistent with the pivotal-trial investigators concluding the safety signals limit clinical use. This is inferred from the trial conclusions, not a primary regulator document.
PopulationLiraglutide pivotal T1D RCTs (Mathieu ADJUNCT ONE; Ahren ADJUNCT TWO); regulatory-status inferred, not primary-sourced this pass.
Fundingindustry - Novo Nordisk (sponsor; NCT01836523)
Scope limitsCONFIDENCE MEDIUM / INFERRED: regulator status derived from the trialists' 'limiting clinical use' conclusion and the lack of any T1D indication, NOT a primary ADA/EASD/NICE/EMA/FDA citation. Flagged to OQ-T1D-C (add a primary guideline/regulator doc).
Comparatorsplacebo (insulin background)
GLP-1 RA class MARKETED
special-populations Decrease
Moderate evidence

Pooled RCT data found GLP-1 RAs modestly lower total testosterone and BMI in obese PCOS women.

GLP-1 RA and androgens in PCOS (meta-analysis, 4 RCTs), J Diabetes Complications 2024 Source
PopulationPeer-reviewed primary (women's-health wave)
Fundingacademic / investigator - no industry sponsor disclosed
Scope limitsSmall/fragile pool (p=0.03, mostly liraglutide, total-T only); any drop is plausibly SHBG/weight-mediated, not a demonstrated direct anti-androgen effect. CONTESTED by a broader meta finding androgens unchanged - both poles stand.
GLP-1 RA class MARKETED
special-populations No change
Moderate evidence

A broader PCOS RCT meta-analysis found GLP-1 RAs improve weight and insulin resistance but do NOT change circulating androgens.

GLP-1 RA in PCOS, broad androgen panel (meta-analysis), Sci Rep 2025 Source
PopulationPeer-reviewed primary (women's-health wave)
Fundingacademic-Guangxi colleges teacher research grant (2023KY0576)
Scope limitsNet verdict across the two metas: metabolic improvement yes; androgen effect uncertain/small and at most SHBG/weight-mediated. Both-poled with the androgen-down meta.
GLP-1 RA class MARKETED
special-populations Decrease
Moderate evidence

A 2025 meta-analysis of 25 RCTs confirms a modest class-level benefit of GLP-1 RAs added to insulin in T1D (HbA1c about -0.23%, weight -3.9kg, insulin -5.7 U/day), but...

Rebelos E, Anastasiou IA, Tentolouris N, Karagiannis T, Tsapas A, Ferrannini E, Liakos A.... Source
Full findingA 2025 meta-analysis of 25 RCTs confirms a modest class-level benefit of GLP-1 RAs added to insulin in T1D (HbA1c about -0.23%, weight -3.9kg, insulin -5.7 U/day), but the pooled estimate is effectively liraglutide.
PopulationGLP-1 RA add-on T1D meta-analysis (Rebelos/Liakos et al., 25 RCTs, 3,224 patients; PMC12678461).
Fundingacademic/independent (Greek/Italian university authors; no manufacturer sponsor)
Scope limitsLIRAGLUTIDE-WEIGHTED: the 'class' estimate is dominated by liraglutide trials - do NOT transfer to other agents (non-liraglutide agents are ineffective on HbA1c, see C-CLASS-T1D-02). The review also found GLP-1 RAs do NOT prevent progressive C-peptide loss and counter-regulatory glucagon stays intact. High heterogeneity for insulin dose/TIR (I2~71%).
Comparatorsplacebo (insulin background)
GLP-1 RA class MARKETED
special-populations Decrease
Moderate evidence

A 2023 meta-analysis corroborates the modest class benefit and shows the HbA1c effect is essentially confined to liraglutide - exenatide, exenatide-ER and albiglutide...

Tan X, Pan X, Wu X, Zheng S, Chen Y, Liu D, Zhang X. Glucagon-like peptide-1 receptor ago... Source
Full findingA 2023 meta-analysis corroborates the modest class benefit and shows the HbA1c effect is essentially confined to liraglutide - exenatide, exenatide-ER and albiglutide had no significant glycaemic effect - underlining that 'class' is effectively liraglutide.
PopulationGLP-1 RA add-on T1D meta-analysis (Tan et al., 11 RCTs, 2,856 patients; PMC10797415).
Fundingacademic/government (Central South University, China; Hunan provincial science foundations)
Scope limitsLiraglutide-specific -0.26% and the exenatide/exenatide-ER/albiglutide NULL were confirmed in the PMC full text (not just abstract). Subgroup: HbA1c effect larger in obese/overweight (-0.43%) than normal-weight (-0.10%, NS) T1D - the only suggestion of phenotype-dependence, against Dejgaard's BMI-null (different cut/method). Reinforces liraglutide-weighting (C-CLASS-T1D-01).
Comparatorsplacebo (insulin background)
GLP-1 RA class MARKETED
Gastrointestinal tolerability Decrease
Moderate evidence

Across the roster, GI AEs cluster during dose escalation and are partially mitigated by slower/lower-starting-dose titration; explicit titration-mitigation evidence...

Synthesis: Jastreboff 2022, Jastreboff 2023, ATTAIN-1, survodutide phase 2 Source
Full findingAcross the roster, GI AEs cluster during dose escalation and are partially mitigated by slower/lower-starting-dose titration; explicit titration-mitigation evidence strongest for retatrutide (2 vs 4 mg start), stated for orforglipron and survodutide.
PopulationSynthesis across SURMOUNT-1, retatrutide phase 2, ATTAIN-1, survodutide phase 2
Fundingindustry - multiple, each disclosed per component: Eli Lilly / Boehringer Ingelheim
Scope limitsCross-cutting observation, not a single-trial result. 'Decrease' denotes mitigation of GI burden, not a drug-effect direction.
Comparatorsplacebo
GLP-1 RA class MARKETED
Gastrointestinal tolerability Increase
Moderate evidence

GLP-1 RAs added to insulin in T1D roughly tripled nausea and vomiting and doubled early treatment withdrawal - a substantial early gastrointestinal burden that eases...

Kateel R, Parida A, Chogtu B, Holla SN. Safety of GLP-1 receptor agonists in type 1 diabe... Source
Full findingGLP-1 RAs added to insulin in T1D roughly tripled nausea and vomiting and doubled early treatment withdrawal - a substantial early gastrointestinal burden that eases after about six months, requiring slow dose titration and careful patient selection.
PopulationGLP-1 RA T1D safety/tolerability meta-analyses (Kateel et al., 25 studies; corroborated by Tan et al. GI-AE OR 2.96, PMC10797415).
Fundingacademic/government (Kasturba Medical College, Manipal; Dept of Health Research, India)
Scope limitsTextbook GLP-1 GI tolerability - kept in tolerability-gi (distinct from the safety-other hypo/ketosis signals). The early-withdrawal RR 2.02 is the clinically meaningful consequence; improves after ~6mo, supporting stepwise titration.
Comparatorsplacebo (insulin background)
GLP-1 RA class MARKETED
Gastrointestinal tolerability Increase
Moderate evidence

FOLK-CLAIM VERDICTS (GI cluster): (1) 'Nausea is near-inevitable' = OVERSTATED - real and the top AE but ~25-44% (not inevitable), mostly mild-moderate, concentrated...

Wharton S et al. Gastrointestinal tolerability of semaglutide 2.4 mg (pooled STEP 1-3). D... Source
Full findingFOLK-CLAIM VERDICTS (GI cluster): (1) 'Nausea is near-inevitable' = OVERSTATED - real and the top AE but ~25-44% (not inevitable), mostly mild-moderate, concentrated at titration. (2) 'Constipation is more common than diarrhoea' = FALSE/MIXED - for semaglutide diarrhoea (29.7%) >= constipation (24.2%), and the balance is DRUG-SPECIFIC (sema/lira constipation-leaning, tirzepatide diarrhoea-leaning). (3) 'Fatigue, especially after each dose increase' = TRUE as a recognised (labelled) AE, but the per-step-spike mechanism is OVERSTATED - fatigue is largely secondary to reduced intake/dehydration/electrolytes during the nausea-prone escalation window, not a proven per-step pharmacological effect.
PopulationPooled STEP 1-3 (semaglutide) GI-tolerability analysis + SURMOUNT (tirzepatide) + a class network meta-analysis.
Fundingindustry - Novo Nordisk A/S (sponsor of the underlying STEP 1-3 trials and the load-bearing Wharton 2022 pooled GI-tolerability analysis)
Scope limitsBelief-vs-reality: nausea is a large-minority, mostly-mild, titration-linked event (not inevitable); the 'these drugs back you up' rule is wrong-way-round for semaglutide and is drug-specific; fatigue is real but intake-mediated. Frequency anchors (self-report) are in C-CLASS-TOLGI-REDDIT-FREQ-01. Cross-ref C-TIRZ-VS-SEMA-TOLGI-01 (tirz-gentler) and the existing tolerability-gi rows. [REASONED CAVEAT + OPEN QUESTION (OQ-FOLK-GI-01), 2026-06-22, NOT a settled finding: the constipation-vs-diarrhoea split likely reflects an ASYMMETRY - constipation is the more DRUG-INTRINSIC outcome (slowed transit -> the colon reabsorbs more water -> hard, dry stool; happens largely independent of diet, even on small intake), whereas diarrhoea is the more DIET/BEHAVIOUR-MODULATED one (the drug, especially tirzepatide's GIP-driven effect on fat handling, primes fat and bile to reach the colon where they trigger a secretory/osmotic flush, but it needs a fatty/difficult-to-digest meal to actually fire). The DRUG component of the diarrhoea tilt is REAL (it appears in the SURMOUNT-5 head-to-head and meta-analyses, not only in self-report), but its MAGNITUDE is diet- and population-modulated - and the real-world tirzepatide diarrhoea signal is further confounded as the drug reaches a broader, less-screened, less-diet-coached population (the same confound that grades the self-report rows very-low). Testable prediction: a high-fat-meal challenge or a diet-controlled comparison should move the diarrhoea rate while barely touching the constipation rate.]
Comparatorsplacebo
GLP-1 RA class MARKETED
Gastrointestinal tolerability Not directional
Low evidence

FOLK-CLAIM VERDICT ('sulfur burps' / rotten-egg belches): UNPROVEN - this is the clearest GENUINE EVIDENCE GAP in the GI cluster. The phenomenon is a widespread,...

No indexed literature on GLP-1-associated sulfur eructation/hydrogen-sulfide belching loc...
Full findingFOLK-CLAIM VERDICT ('sulfur burps' / rotten-egg belches): UNPROVEN - this is the clearest GENUINE EVIDENCE GAP in the GI cluster. The phenomenon is a widespread, recognisable community 'class signal' with a plausible mechanism (delayed gastric emptying -> fermentation -> hydrogen sulfide), but there is NO trial, case series or mechanistic paper formally characterising or quantifying it. Real-world reports + coherent mechanism, but no formal evidence base.
Populationn/a - documents the absence of formal literature; online self-report (eructation ~6.9%) is the only frequency anchor.
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsGRADE very-low (no evidence base; verdict is 'unstudied', not 'refuted'). Belief-vs-reality: patients treat sulfur burps as a known class signature; the science has barely looked. A genuine open research question (the one true gap in the GI set). Cross-ref C-CLASS-TOLGI-FOLKVERDICT-01.
GLP-1 RA class MARKETED
Gastrointestinal tolerability Not directional
Very low evidence

REAL-WORLD ONLINE SELF-REPORT context for the GI/other folk frequencies. This is a SIGNAL-DETECTION / pharmacovigilance-adjacent dataset, NOT prevalence, NOT...

Sehgal NKR, Tronieri J, Ungar L, Guntuku SC. Self-reported side effects of semaglutide an... Source
Full findingREAL-WORLD ONLINE SELF-REPORT context for the GI/other folk frequencies. This is a SIGNAL-DETECTION / pharmacovigilance-adjacent dataset, NOT prevalence, NOT incidence, NOT causal - self-selected communities, no denominator, no controls, no titration timing (the authors state they cannot say GLP-1s cause these symptoms). Use it to gauge how COMMON a BELIEF/report is, never to estimate true rates. The sema-vs-tirz nausea split is a text-mining ratio, not a randomised comparison.
PopulationSelf-selected online communities (Reddit); semaglutide + tirzepatide users; computational NLP classification.
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsGRADE very-low (self-selected online self-report; reporting != incidence). Verified to exist (Nature Health 2026) but NOT PubMed-indexed - findable via Nature/medRxiv preprint (10.64898/2026.03.12.26348253)/arXiv 2603.12341. Captured as the frequency-of-belief layer per the Director's instruction; heavily fenced. Anchors the frequency tags throughout the folk-claims register.
Thin or bounded evidence here means uncertainty remains visible. It is not evidence that an effect has been ruled out.
GLP-1 RA class MARKETED
Appetite and food intake Decrease
Low evidence

FOLK-CLAIM VERDICT ('these drugs silence food noise'): MIXED. The craving/appetite-reduction CORE is TRUE and instrument-measured (real, not placebo or weight-loss...

Dhurandhar EJ et al. Defining and measuring food noise (RAID-FN / FNQ). Nutr Diabetes 202... Source
Full findingFOLK-CLAIM VERDICT ('these drugs silence food noise'): MIXED. The craving/appetite-reduction CORE is TRUE and instrument-measured (real, not placebo or weight-loss halo). BUT (1) 'food noise' as a distinct intrusive-rumination construct was only defined and given instruments in 2025 and has NEVER been a pre-registered trial endpoint - the trials measured cravings/appetite, which overlap but are not identical; (2) 'valued above the weight loss itself' rests entirely on surveys/anecdote - UNPROVEN; (3) drug-specificity is lopsided - robust for semaglutide, mechanistically strong for tirzepatide, THIN and indirect for retatrutide.
PopulationNarrative cross-agent verdict over the craving/appetite evidence (CoEQ/FCI/VAS RCTs; tirz fMRI; reta early/indirect data) + the 2025 food-noise construct literature.
Fundingacademic (Dhurandhar et al.; food-noise construct/measurement paper)
Scope limitssurrogate/exploratory endpoint
Comparatorsplacebo
GLP-1 RA class MARKETED
Appetite and food intake Decrease
Low evidence

GIPR agonism-versus-antagonism paradox: both GIPR agonism (within tirzepatide) and GIPR antagonism (anti-GIPR antibody within maridebart cafraglutide/MariTide) produce...

Douros JD, Mowery SA, Knerr PJ, J Clin Med 2025;14(11):3812 Source
Full findingGIPR agonism-versus-antagonism paradox: both GIPR agonism (within tirzepatide) and GIPR antagonism (anti-GIPR antibody within maridebart cafraglutide/MariTide) produce additive weight loss when combined with GLP-1R agonism; reconciling hypotheses (incretin-crosstalk disinhibition; chronic-agonism desensitisation to functional antagonism; distinct CNS populations) remain genuinely unresolved.
PopulationReview of human/preclinical/genetic evidence
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsCOI-visibility (Indiana Biosciences, ex-Lilly/Novo lineage). Reused from corpus T9-GIPR-04/L1-015. HEADLINE DISCREPANCY held open — both poles recorded, NOT resolved by preference.
ComparatorsGLP-1R agonism (background)
GLP-1 RA class MARKETED
Appetite and food intake Mixed
Very low evidence

Both-poles marker: glucagon's satiety effect is real but modest and the human evidence base is thin/dated (one small human test-meal study plus rodent mechanism); no...

Geary N, Neurosci Biobehav Rev 1990;14(3):323-338 Source
Full findingBoth-poles marker: glucagon's satiety effect is real but modest and the human evidence base is thin/dated (one small human test-meal study plus rodent mechanism); no modern human glucagon-infusion ad-libitum-intake study isolates a robust large anorectic effect, glucagon raises glucose (confounding the satiety read), and the dual-agonist contrast shows the GCGR arm does not carry oxyntomodulin's satiety. Competing pole: glucagon's headline therapeutic role is metabolic (glucose/EE), largely independent of its modest intake effect.
PopulationMulti-species synthesis
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsReused from corpus T9-GCGR-08. Explicit flag-both-poles record. Prevents the GCGR arm being written up as a confident appetite-suppressor.
Thin or bounded evidence here means uncertainty remains visible. It is not evidence that an effect has been ruled out.
GLP-1 RA class MARKETED
Maintenance and regain Increase
Low evidence

Class-level synthesis: a narrative review of randomised-withdrawal trials, meta-analyses and real-world cohorts (>289,000 patients) concludes that GLP-1RA...

Shah E, AlShiab R, Abdo A et al., Diabetes Obes Metab, 2026 Source
Full findingClass-level synthesis: a narrative review of randomised-withdrawal trials, meta-analyses and real-world cohorts (>289,000 patients) concludes that GLP-1RA discontinuation is followed by substantial weight regain (60-90% within a year) and parallel reversal of cardiometabolic benefits, attributed to reactivation of orexigenic pathways and adaptive thermogenesis. It also notes the absence of validated tapering strategies.
PopulationNarrative review of randomised-withdrawal trials, systematic reviews/meta-analyses and observational cohorts (>289,000 patients).
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitspossible confounding by indication
Comparatorscontinued therapy
GLP-1 RA class MARKETED
Maintenance and regain Not directional
Low evidence

EMERGING CONCERN / GAP: no published study measured body composition (DXA/MRI fat vs lean) across a complete loss-then-regain cycle ON a GLP-1/incretin drug, so...

Gap synthesis across the confirmed D4 withdrawal records (no single source reports drug-s...
Full findingEMERGING CONCERN / GAP: no published study measured body composition (DXA/MRI fat vs lean) across a complete loss-then-regain cycle ON a GLP-1/incretin drug, so whether regained weight is disproportionately FAT with lean mass incompletely restored is untested. The withdrawal trials report TOTAL weight regain only; the human composition-across-cycle evidence (Biggest Loser) is diet/exercise, not drug. The worry is biologically plausible and reinforced by documented incretin-class lean-mass loss during the loss phase (D2), but direct drug-specific data are absent.
PopulationCross-study gap assessment across STEP-1 extension, STEP-4, SURMOUNT-4 (total weight only) and the metabolic-adaptation cohort.
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsDELIBERATE GAP-FLAG. Captures the 'a weight cycle worsens fat:lean ratio' concern as EMERGING with thin/absent direct human drug data. Anchors the open question; cross-ref D2 lean-mass loss (loss phase documented; regain-phase composition NOT). NOT a positive finding - records an absence. COUNCIL: rule whether a sourceless gap row belongs or should be an OQ only.
GLP-1 RA class MARKETED
Maintenance and regain Not directional
Very low evidence

EVIDENCE GAP: no randomised trial of a LOWER maintenance dose or an INTERMITTENT / reduced-frequency incretin dosing schedule for sustaining weight loss was located....

Gap synthesis; corroborated by Shah E et al., Diabetes Obes Metab, 2026 (no validated tap...
Full findingEVIDENCE GAP: no randomised trial of a LOWER maintenance dose or an INTERMITTENT / reduced-frequency incretin dosing schedule for sustaining weight loss was located. The maintenance trials prove FULL-dose continuation works but say nothing about whether a reduced or intermittent dose can hold weight off; the class discontinuation review explicitly notes no validated tapering strategies exist.
PopulationGap assessment across STEP-4, SURMOUNT-4 and the GLP-1RA discontinuation review.
Fundingnot-stated
Scope limitsno outcome data yet (ongoing)
Thin or bounded evidence here means uncertainty remains visible. It is not evidence that an effect has been ruled out.
GLP-1 RA class MARKETED
Obstructive sleep apnoea Not directional
High evidence

SURROGATE-vs-HARD-OUTCOME caveat: AHI is the registrational endpoint for incretin OSA trials, but lowering OSA severity does NOT guarantee hard-outcome benefit. The...

McEvoy RD et al. CPAP for Prevention of Cardiovascular Events in OSA (SAVE). N Engl J Med... Source
Full findingSURROGATE-vs-HARD-OUTCOME caveat: AHI is the registrational endpoint for incretin OSA trials, but lowering OSA severity does NOT guarantee hard-outcome benefit. The CPAP randomised precedent shows AHI/severity can be reduced without reducing major CV events - SAVE and ISAACC were both NULL for CV outcomes despite effective apnoea-severity reduction. The same caveat applies to drug-induced AHI reduction: the incretin OSA RCTs (SCALE, SURMOUNT-OSA) demonstrate AHI reduction (a surrogate), not reduced CV events, mortality or hard OSA outcomes. BOTH POLES (kept explicit): the CPAP precedent is WEAKENED by low device adherence (~3-4 hr/night) and selection of low-sleepiness populations, so it does not prove drug-induced AHI reduction is futile; but the narrow claim stands - AHI reduction alone is not established proof of hard-outcome benefit.
PopulationCross-cutting surrogate note. SAVE (N=2717, CV-disease population) and ISAACC (N=1264, post-ACS); both tested CPAP (not an incretin) and were academically led.
Fundingmixed-NHMRC (Australia) and Philips Respironics/device industry
Scope limitssurrogate/exploratory endpoint
Comparatorsusual care; sham/no CPAP
GLP-1 RA class MARKETED
Obstructive sleep apnoea Not directional
Low evidence

MECHANISM (the central question): incretin OSA benefit is WEIGHT-DOMINANT - AHI reduction tracks weight loss across SCALE and SURMOUNT-OSA, and the canonical mediator...

Wang SH, Keenan BT, Wiemken A, ... Schwab RJ. Effect of Weight Loss on Upper Airway Anato... Source
Full findingMECHANISM (the central question): incretin OSA benefit is WEIGHT-DOMINANT - AHI reduction tracks weight loss across SCALE and SURMOUNT-OSA, and the canonical mediator is reduction in upper-airway / parapharyngeal and especially TONGUE fat. A weight-INDEPENDENT/direct component (residual upper-airway-fat pathway, plus possible rostral fluid shift, lung-volume and ventilatory-control/loop-gain effects) is PLAUSIBLE but NOT PROVEN for incretins: no incretin OSA trial has isolated a weight-independent drug effect. Both poles stand.
PopulationMechanistic synthesis. Key imaging substrate: 67 adults with obesity + OSA, upper-airway + abdominal MRI and sleep study before/after weight loss (lifestyle or bariatric).
Fundingnot-stated
Scope limitsBalanced mechanism row (both poles). WEIGHT-DOMINANT: the simplest reading of SCALE + SURMOUNT-OSA is that AHI falls because weight (and parapharyngeal/tongue fat) falls. DIRECT-COMPONENT-PLAUSIBLE: tongue-fat reduction predicts AHI improvement even after adjusting for weight - but this is from lifestyle/bariatric weight loss, NOT demonstrated for any incretin drug; fluid-shift/loop-gain contributions are hypothesis-level. Attribution deliberately left open. Reta relevance: if weight-mediated, reta (largest weight loss) should lead on AHI; if a direct upper-airway pathway dominates, ranking could differ - untested.
Thin or bounded evidence here means uncertainty remains visible. It is not evidence that an effect has been ruled out.
GLP-1 RA class MARKETED
Insulin and beta-cell function Increase
Moderate evidence

NEGATIVE HEADLINE FIRST: no incretin, INCLUDING retatrutide, has demonstrated increased beta-cell MASS or disease modification in humans; all positive human data are...

Bunck MC, Diamant M, Corner A et al., Diabetes Care, 2009 Source
Full findingNEGATIVE HEADLINE FIRST: no incretin, INCLUDING retatrutide, has demonstrated increased beta-cell MASS or disease modification in humans; all positive human data are on-treatment FUNCTION surrogates (HOMA-B, C-peptide, proinsulin:insulin, disposition index, clamp ISR - none measures MASS), and the only off-drug test reverted at 1 year. POLE A (well-supported but confounded): every direct human beta-cell-function measure improves while ON drug (semaglutide, tirzepatide, exenatide). POLE B (the LEADING caution): the on-treatment gain is substantially confounded by large weight loss and relief of gluco-/lipotoxicity - the only direct human test (exenatide washout) reverted WITH weight regain, supporting mediation; the cited tirzepatide clamp with no weight correlation (Yamaguchi) measures insulin SENSITIVITY (glucose-infusion-rate), NOT beta-cell secretion, so a weight-INDEPENDENT beta-cell effect remains UNPROVEN. The positive on-treatment pole is also near-uniformly manufacturer-sponsored (a further confidence discount).
PopulationClass-level synthesis across human clamp / IVGTT / HOMA / off-drug studies (semaglutide, tirzepatide, exenatide).
Fundingindustry-Amylin Pharmaceuticals/Eli Lilly
Scope limitsopen-label (unblinded); small sample (N~69)
Comparatorsplacebo; insulin glargine; semaglutide
GLP-1 RA class MARKETED
Insulin and beta-cell function Mixed
Moderate evidence

GCGR antagonists lowered glucose in T2D but caused dose-dependent rises in hepatic fat, liver enzymes, LDL/serum lipids and body weight, and human GCGR...

Guzman CB et al. (GCGR antagonist trials review); Larger S et al. (Mahvash, PMID:30032256) Source
Full findingGCGR antagonists lowered glucose in T2D but caused dose-dependent rises in hepatic fat, liver enzymes, LDL/serum lipids and body weight, and human GCGR loss-of-function causes alpha-cell hyperplasia/hyperglucagonaemia (Mahvash disease) — establishing that blocking the glucagon axis has adverse hepatic/lipid consequences, the inverse-direction evidence framing why retatrutide AGONISES (not blocks) GCGR.
PopulationT2D trials (antagonists) + human GCGR-inactivating genetics
Fundingindustry - Eli Lilly and Company
Scope limitsReused from corpus L2-035 / T8-024. Inverse-direction (antagonist) evidence; bears on the glucagon-arm safety/metabolic logic. PMID not independently re-confirmed for this DOI by this collector — identifier carried as DOI per corpus.
Comparatorsplacebo
Thin or bounded evidence here means uncertainty remains visible. It is not evidence that an effect has been ruled out.
GLP-1 RA class MARKETED
Cardiovascular outcomes Decrease
Moderate evidence

Class-level meta-analysis of 8 GLP-1RA CVOTs in T2D: ~14% MACE reduction, with consistent reductions in each MACE component, all-cause mortality, HF hospitalisation...

Sattar N, Lee MMY, Kristensen SL et al., Lancet Diabetes Endocrinol, 2021 Source
Full findingClass-level meta-analysis of 8 GLP-1RA CVOTs in T2D: ~14% MACE reduction, with consistent reductions in each MACE component, all-cause mortality, HF hospitalisation and a kidney composite; no significant heterogeneity by structural homology or eight other subgroups.
PopulationSystematic review and random-effects meta-analysis of 8 placebo-controlled GLP-1RA CVOTs (ELIXA, LEADER, SUSTAIN-6, EXSCEL, HARMONY, REWIND, PIONEER-6, AMPLITUDE-O); PROSPERO CRD42021259711.
Fundingacademic/independent - no funding (Funding: None)
Scope limitsClass anchor: the ~14% MACE RRR headline. Graded MODERATE per the meta-analysis default rule even though constituents are pivotal CVOTs. STALE-ANCHOR caveat (Council): this is a 2021-vintage pool that pre-dates SELECT (2023), FLOW (2024) and SOUL (2025), so it excludes the non-diabetic-obesity trial most relevant to retatrutide; the 0.86 should NOT be cited as the current class figure. LUMP-NOT-SPLIT caveat: 'no significant heterogeneity by structural homology' is a power statement over sparse-event trials, NOT proof the two null constituents (ELIXA, EXSCEL) are noise; pooling is a value choice. TRANSFER caveat: homology homogeneity is WITHIN the GLP-1RA class and says nothing about molecules adding GIP or glucagon agonism; do not use the 14% RRR as a retatrutide expectation. Per-study COI not individually audited.
Comparatorsplacebo (pooled across constituent trials)
Thin or bounded evidence here means uncertainty remains visible. It is not evidence that an effect has been ruled out.
GLP-1 RA class MARKETED
comparative-efficacy Increase
Moderate evidence

EFFICACY LADDER (weight, as reported, with trial/dose/duration/population context). Semaglutide 2.4 mg ~15% (STEP-1, obesity, 68 wk); tirzepatide 15 mg ~20-22.5%...

Composite of SURMOUNT-5 (PMID 40353578), retatrutide ph2 (PMID 37366315), REDEFINE 1 (PMI... Source
Full findingEFFICACY LADDER (weight, as reported, with trial/dose/duration/population context). Semaglutide 2.4 mg ~15% (STEP-1, obesity, 68 wk); tirzepatide 15 mg ~20-22.5% (SURMOUNT-1, obesity, 72 wk); SURMOUNT-5 direct head-to-head tirz -20.2% vs sema -13.7% (72 wk); CagriSema ~20.4% (REDEFINE 1, 68 wk); retatrutide 12 mg ~24.2% phase 2 (48 wk), up to ~30% phase 3 TRIUMPH-1 (press, 80 wk); orforglipron ~7.6% (ACHIEVE-1 T2D, 40 wk).
PopulationMIXED across trials/doses/durations/populations - NOT a single comparison
Fundingindustry - multiple, each disclosed per component: Eli Lilly / Novo Nordisk
Scope limitsEXPLICIT LADDER CAVEAT: the only true APPLES-TO-APPLES rung is SURMOUNT-5 (tirz 20.2% vs sema 13.7%, same trial/population). Every other rung-to-rung gap is CROSS-TRIAL (different durations 40-80 wk, different populations T2D vs obesity-only, different placebo responses). The numbers are recorded as REPORTED, NOT as a validated ranking. CagriSema and tirzepatide overlap. Retatrutide top rung is phase-2/press, not yet head-to-head. 'Why' (receptor count, GCGR/GIP/amylin contributions) explicitly OPEN.
Comparatorscross-trial; placebo within each
Thin or bounded evidence here means uncertainty remains visible. It is not evidence that an effect has been ruled out.
GLP-1 RA class MARKETED
Energy expenditure and thermogenesis Mixed
Low evidence

Scoping review of human EE studies (23 studies): acute or chronic GLP-1RA MONOtherapy does not appear to impact energy expenditure independent of weight loss....

Vieira FT, Deng Z, Muller MJ, et al. Effects of GLP-1 Receptor Agonists (Mono and Combina... Source
Full findingScoping review of human EE studies (23 studies): acute or chronic GLP-1RA MONOtherapy does not appear to impact energy expenditure independent of weight loss. Combination with glucagon produced varied impacts on EE components (RQ, RMR, sleep metabolic rate); combination with GIP decreased RQ and increased fat utilisation. Eight studies (34.8%) concluded non-significant EE effects; 11 (47.8%) were inconclusive due to statistical analyses. Most assessed RMR; only 4 used 24-h whole-room calorimetry; none used doubly labelled water.
PopulationScoping review, humans, inception to Oct 2025; 23 studies (10 monotherapy: 4 exenatide, 4 liraglutide, 1 semaglutide, 1 beinaglutide; 13 combination: 11 dual, 2 triple agonists)
Fundingacademic/independent analysis (manufacturer not study sponsor; per-study COI not individually audited)
Scope limitsClass-level landmark for the 'whole-body-EE-null' pole in humans. Note the differential signal: +GIP lowers RQ/raises fat oxidation (matches the tirzepatide fat-oxidation finding, C-TIRZEPATIDE-EE-03), while +glucagon affects EE components variably. Co-author Heymsfield (Pennington). Per PubMed.
Comparatorsplacebo; across-study heterogeneity