Back to Research Library
    Reviewed guideResearch Library / Retatrutide

    Retatrutide: what the evidence shows so far.

    A text-first guide to an investigational peptide drug candidate, the human studies behind the current discussion, and the questions those studies still leave open.

    How to read this guide

    The strongest-sounding number is not always the most useful fact. Keep the study population, endpoint, timepoint, comparator, source type, and limitation attached to every result.

    01 / Definition

    What is retatrutide?

    Retatrutide, also known as LY3437943, is an investigational peptide drug candidate developed by Eli Lilly. It is being studied as one molecule acting across GIP, GLP-1, and glucagon receptor systems.

    Human trials have reported average changes in body weight, HbA1c, and selected biomarkers. An MASLD substudy reported MRI-PDFF liver-fat changes. Those findings do not establish prevention of heart attack, stroke, cardiovascular death, cirrhosis, liver failure, or mortality. As of August 22, 2026, retatrutide is not an FDA-approved medicine or an established treatment.

    02 / Mechanism boundaries

    Mechanism and clinical proof are different questions.

    Coskun et al. described a single molecule with activity at glucagon, GIP, and GLP-1 receptors. In the reported in-vitro work, glucagon and GLP-1 activity was relatively balanced, while GIP activity was greater. This supports a receptor rationale, not a guarantee about human outcomes.

    Receptor rationale

    GIP and GLP-1 signaling may support glucose-dependent insulin secretion and lower caloric intake. Glucagon signaling may contribute to energy expenditure and broader metabolic effects. These are biological hypotheses.

    Preclinical work

    Obese-mouse findings are preclinical. They cannot establish the magnitude, durability, or safety of an effect in people.

    Human endpoints

    Human studies support observed intermediate endpoints. They do not by themselves prove a long-term mechanism, weight maintenance after stopping, or protection from uncommon harms.

    Source: Coskun et al., PubMed.

    03 / Human evidence

    The human record, study by study.

    These compact cards keep the population, endpoint, timeframe, comparator, finding, and limitation together. The source label identifies peer-reviewed evidence. The cards do not turn group averages into individual predictions.

    01Peer-reviewed / Cell Metabolism 2022
    View source

    Discovery and early clinical proof of concept

    Population / design
    Discovery work plus early exploratory human data.
    Timeframe
    Early program; not a long-term follow-up study.
    Endpoint
    Pharmacokinetic persistence and an exploratory weight signal.
    Comparator
    No long-term efficacy comparator.

    What the study reported: Coskun et al. described the molecule and early clinical observations that supported further study.

    Limit: This work does not establish long-term efficacy, weight maintenance, or safety.

    02Peer-reviewed / The Lancet 2022
    View source

    Phase 1b type 2 diabetes

    Population / design
    72 participants with type 2 diabetes.
    Timeframe
    12 weeks.
    Endpoint
    Glucose, HbA1c, body weight, treatment-emergent events, and discontinuations.
    Comparator
    Placebo-adjusted changes in higher-exposure groups.

    What the study reported: The study reported reductions in glucose, HbA1c, and weight in higher-exposure groups. GI disorders were the most frequent treatment-emergent events; 29 of 72 participants discontinued prematurely.

    Limit: Small, short, early-phase evidence cannot answer uncommon or delayed safety questions.

    03Peer-reviewed / The Lancet 2023
    View source

    Phase 2 type 2 diabetes

    Population / design
    281 randomized adults with type 2 diabetes.
    Timeframe
    HbA1c primary at week 24; outcomes through week 36.
    Endpoint
    HbA1c and body weight, with adverse events.
    Comparator
    Placebo and dulaglutide.

    What the study reported: HbA1c and weight declined more than placebo. GI events occurred in 67 of 190 retatrutide participants, 6 of 45 placebo participants, and 16 of 46 dulaglutide participants.

    Limit: The phase 2 design and follow-up limit conclusions about long-term outcomes.

    04Peer-reviewed / NEJM 2023
    View source

    Phase 2 obesity

    Population / design
    338 participants in a phase 2 obesity study.
    Timeframe
    Week 24 primary endpoint; week 48 secondary endpoint.
    Endpoint
    Body weight change from baseline.
    Comparator
    Placebo.

    What the study reported: Approximate least-squares mean weight changes at week 24 were 7%–18% across active groups versus about 2% with placebo. At week 48, changes were 9%–24% versus about 2% with placebo.

    Limit: These are group averages in a selected phase 2 population, with limited power for uncommon or long-term harms.

    05Peer-reviewed / Nature Medicine 2024
    View source

    MASLD MRI-PDFF substudy

    Population / design
    98 participants in a MASLD substudy.
    Timeframe
    48-week program; MRI-PDFF liver-fat change was primary at week 24.
    Endpoint
    Relative liver-fat change and the proportion below 5% liver fat on MRI-PDFF.
    Comparator
    Placebo.

    What the study reported: Approximate mean relative liver-fat reductions were 43%–82% versus about no change with placebo. The proportion below 5% liver fat was 27%–86% versus 0% with placebo.

    Limit: MRI-PDFF is a surrogate measure, not proof of histologic improvement or fewer liver events.

    06Peer-reviewed / The Lancet Diabetes & Endocrinology 2025
    View source

    Body-composition substudy

    Population / design
    189 people in the DXA substudy; 155 had baseline scans and 103 completed both scans.
    Timeframe
    Week 36.
    Endpoint
    Fat-mass change measured by DXA.
    Comparator
    Placebo and dulaglutide.

    What the study reported: The analysis observed greater fat-mass changes than placebo or dulaglutide at week 36.

    Limit: The selected substudy, missing scans, and DXA measure do not establish strength, function, bone health, or patient-centered outcomes.

    07Peer-reviewed / The Lancet 2026
    View source

    TRANSCEND-T2D-1 phase 3 diabetes

    Population / design
    537 randomized adults with type 2 diabetes inadequately controlled by diet and exercise.
    Timeframe
    40 weeks.
    Endpoint
    HbA1c and body weight.
    Comparator
    Placebo.

    What the study reported: Approximate HbA1c changes were 1.7–1.9 percentage points across retatrutide groups versus 0.8 with placebo. Weight change was about 12%–15% versus 3% with placebo.

    Limit: One selected 40-week setting cannot establish long-term safety or clinical outcomes.

    08Peer-reviewed / Diabetes, Obesity and Metabolism 2026
    View source

    Cardiovascular-biomarker analysis

    Population / design
    Post hoc analysis of phase 2 trial data.
    Timeframe
    Online ahead of print August 17, 2026.
    Endpoint
    Non-HDL cholesterol, ApoB, triglyceride-rich lipoprotein particles, LDL-particle measures, and selected inflammatory biomarkers.
    Comparator
    Surrogate and post hoc analysis, not a cardiovascular outcomes trial.

    What the study reported: The analysis reported changes across several cardiovascular-risk biomarkers.

    Limit: Biomarker changes do not establish prevention of heart attack, stroke, cardiovascular death, or mortality.

    04 / Interpretation

    Numbers need their setting.

    A reported percentage is an average result for a defined group at a defined timepoint. It is not a promise about what any one person will experience, and it cannot be lifted out of its comparator or endpoint.

    1. 01 / Population

      Who entered the study, and who remained for the analysis?

    2. 02 / Endpoint

      Was the outcome body weight, HbA1c, MRI-PDFF, a biomarker, or something else?

    3. 03 / Timepoint

      A week-24, week-40, or week-80 result answers a time-bounded question.

    4. 04 / Comparator

      The comparison may be placebo or another study arm, and the comparison must stay visible.

    5. 05 / Group average

      A group average does not establish durability after stopping, long-term safety, or clinical event prevention. MRI-PDFF changes are surrogate findings, not proof of fewer liver events.

    05 / Safety and uncertainty

    Signals deserve context, not certainty.

    Recurring gastrointestinal effects

    Across the phase 1b, phase 2, phase 3, and Lilly topline reports in this memo, recurring complaints included nausea, diarrhea, constipation, vomiting, decreased appetite, and related gastrointestinal effects. The frequency and context belong to each source and study population.

    Heart-rate signal

    The NEJM phase 2 obesity trial reported a graded heart-rate increase that peaked around week 24 and declined afterward. This is a signal requiring continued cardiovascular characterization. It is not proof of harm or harmlessness.

    Company-reported dysesthesia

    Preliminary signal

    Lilly reported dysesthesia, a symptom label whose interpretation remains unresolved. In TRIUMPH-1, Lilly reported approximately 5.1%–12.5% across active groups versus 0.9% with placebo. In TRIUMPH-2, it reported approximately 4.5%–7.3% versus 0.7%. In TRIUMPH-3, it reported 6.4% in each active group versus 1.3%. The supplied topline memo does not state a separate safety-analysis timepoint. These are company-reported topline figures. This guide does not rewrite dysesthesia as neuropathy or nerve damage.

    Discontinuations, deaths, and causality

    In the 12-week phase 1b diabetes study, 29 of 72 participants discontinued prematurely. In the 40-week TRANSCEND-T2D-1 peer-reviewed phase 3 report, adverse-event discontinuation was approximately 2%–5% in active groups versus 0% with placebo; two deaths in one retatrutide group were judged unrelated by investigators. These observations do not prove that serious harm is impossible, and a discontinuation alone does not prove adverse-event causality.

    Lilly’s preliminary toplines reported discontinuation ranges of 4.1%–11.3% active versus 4.9% placebo in TRIUMPH-1, 3.8%–11.6% versus 4.9% in TRIUMPH-2, and 9.8%–13.5% versus 4.8% in TRIUMPH-3; the supplied topline memo does not state a separate safety-analysis timepoint. Event counts and imbalances do not establish causation. Early and mid-stage studies may be underpowered for rare or delayed harms.

    06 / Regulatory context

    FDA and EMA records answer different questions.

    The FDA page says unapproved GLP-1 products are not reviewed like approved products. It specifically states that retatrutide and cagrilintide cannot be used in compounding under federal law, are not components of FDA-approved drugs, and have not been found safe or effective for any condition.

    At the August 22, 2026 cutoff, retatrutide is not FDA-approved.

    The EMA record documents a pediatric investigation-plan decision, P/0336/2024, dated September 13, 2024. A pediatric investigation plan is not a marketing authorization and does not give permission for personal use.

    Separate evidence tier

    Lilly-reported topline data

    Company-reported 2026 toplines.

    The following figures come from Lilly releases, not a peer-reviewed full report available in the supplied record at the August 22, 2026 cutoff. Lilly said full results were planned for future publication or presentation. Topline data can inform questions for review, but it should remain labeled as company-reported.

    Lilly reported 2,339 randomized participants. At 80 weeks, mean weight reductions were about 19.0%–28.3% across active groups versus 2.2% with placebo. In the highest tested group, 45.3% had at least 30% loss. A 104-week extension in 532 selected participants with baseline BMI at least 35 reported about 30.3% in the highest tested group.

    Lilly reported 1,152 randomized adults with obesity or overweight and type 2 diabetes. At 80 weeks, mean weight reductions were about 12.7%–20.8% versus 4.0% with placebo, and HbA1c reductions reached about 1.6 points versus 0.2 with placebo.

    Lilly reported 80-week mean weight reductions of 21.6% and 22.6% versus 3.2% with placebo in adults with severe obesity and established cardiovascular disease. MACE-5 was 44 pooled active versus 52 placebo, HR 0.82, 95% CI 0.55–1.22. MACE-3 was 27 versus 23, HR 1.12, 95% CI 0.64–1.96. Both intervals include 1, so cardiovascular benefit or harm is not established.

    Lilly’s stated plan to submit a BLA in Q1 2027 is a company plan, not an approval or regulatory decision.

    07 / Open questions

    The questions still open.

    01

    Durability after stopping and long-term weight maintenance

    02

    Long-term and rare harms, including the meaning of the heart-rate signal

    03

    Neurologic-symptom definitions, severity, and persistence

    04

    Whether MRI-PDFF changes translate to liver outcomes

    05

    Body composition, function, and bone health

    06

    Cardiovascular outcomes and comparative effectiveness

    07

    Subgroup generalizability and pediatric evidence

    08

    Registry transparency, enrollment discrepancies, and future regulatory requirements

    Editorial review checklist

    Confirm every result against the primary paper, registry, regulatory page, or company release.

    Keep population, endpoint, timepoint, comparator, and group-average limits beside each number.

    Separate peer-reviewed findings, registry records, regulatory statements, and Lilly-reported toplines.

    Recheck the FDA and EMA status language whenever this guide is materially updated.

    Keep the Axiolume review boundary visible: source coverage, evidence limits, disclosures, and general education.

    08 / Source trail

    Follow the record directly.

    The source trail separates peer-reviewed findings, registry records, regulatory statements, and company-reported material. Dates below follow the supplied evidence memo; no publication date is inferred where the memo did not provide one.

    Peer-reviewed

    Direct links / dated

    Coskun et al., Cell Metabolism, PMID 35985340

    Publication: 2022; accessed August 22, 2026.

    Urva et al., The Lancet, PMID 36354040

    Publication: 2022; accessed August 22, 2026.

    Rosenstock et al., The Lancet, PMID 37385280

    Publication: 2023; accessed August 22, 2026.

    Jastreboff et al., New England Journal of Medicine, PMID 37366315

    Publication: 2023; accessed August 22, 2026.

    Sanyal et al., Nature Medicine, PMID 38858523

    Publication: 2024; accessed August 22, 2026.

    Coskun et al., The Lancet Diabetes & Endocrinology, PMID 40609566

    Publication: 2025; accessed August 22, 2026.

    Bajaj et al., The Lancet, PMID 42250575

    Publication: 2026; accessed August 22, 2026.

    Ruotolo et al., Diabetes, Obesity and Metabolism, PMID 42608321

    Online ahead of print: August 17, 2026; accessed August 22, 2026.

    Registries

    Retrieved August 22, 2026

    ClinicalTrials.gov records are sponsor-submitted. “Completed” does not mean that results are posted or that a study proves efficacy. The two enrollment discrepancies below are disclosed rather than silently reconciled.

    Retatrutide registry records retrieved August 22, 2026
    RecordStatusEnrollmentResults record
    NCT03841630CompletedActual 45No posted results
    NCT04143802CompletedActual 72No posted results
    NCT04867785CompletedActual 281Results posted July 3, 2023
    NCT04881760CompletedActual 338Results posted September 13, 2023
    NCT06354660CompletedActual 537No posted results despite peer-reviewed phase 3 publication
    NCT05929066CompletedActual 2,335No posted results; Lilly May release says 2,339
    NCT05929079CompletedActual 1,152No posted results
    NCT05882045CompletedActual 1,946No posted results; Lilly July release says 1,949
    NCT05931367CompletedActual 445No posted results
    NCT06662383Active, not recruitingEstimated 800; actual unavailableNo posted results; completion estimated late 2026

    Regulatory

    Direct links / dated

    FDA, concerns about unapproved GLP-1 drugs used for weight loss

    Regulatory page; accessed August 22, 2026.

    EMA pediatric investigation plan P/0336/2024

    Decision: September 13, 2024; accessed August 22, 2026.

    Company-reported

    Direct links / dated

    Lilly TRIUMPH-1 release

    Company release: May 21, 2026; accessed August 22, 2026.

    Lilly TRIUMPH-2 and TRIUMPH-3 release

    Company release: July 23, 2026; accessed August 22, 2026.

    Closing disclosure

    Keep the boundary visible.

    Retatrutide remains investigational and not FDA-approved as of August 22, 2026. Axiolume reviewed this guide for source coverage, evidence boundaries, and disclosures. This is general health education, not medical advice, treatment guidance, a personal-use recommendation, medical or legal review, peer review, or regulatory approval. Recheck current regulatory and source records whenever the guide is materially updated.

    Axiolume provides general health education. It does not provide medical advice, diagnosis, dosing, protocols, treatment recommendations, individualized guidance, or personal-use instructions.