Retatrutide Phase 2 Obesity Trial: NCT04881760

NCT04881760 was a randomized, double-blind, placebo-controlled phase 2 trial of the investigational compound retatrutide, also identified as LY3437943, in 338 adults with obesity or overweight plus a qualifying weight-related condition and without diabetes. At 24 and 48 weeks, the trial reported larger mean body-weight reductions in each retatrutide group than with placebo. The result supports a dose-ranging clinical signal for the sponsor's studied material under this protocol. It does not establish long-term health outcomes, FDA approval, equivalence of another material, or that an item described only as “GLP-3” is retatrutide.[1][2][5]

Study at a glance

The regimen descriptions report what the trial tested, not instructions for use. The registered protocol amendment (b), dated August 29, 2022, and final statistical analysis plan version 4.0, dated December 15, 2022, document the design and analysis framework.[3][4]

What the trial was designed to measure

The prespecified primary outcome was percentage change in body weight at week 24. The registry describes a mixed-effects repeated-measures analysis adjusted for treatment, time, baseline BMI category, sex, and baseline value.[2] Secondary outcomes included body-weight change at week 48 and the proportions reaching at least 5%, 10%, or 15% weight reduction.

The week-24 measure was the prespecified primary endpoint; week-48 measures were secondary. Prespecification does not mean that the comparisons were multiplicity-controlled: the paper and SAP section 2.1 state that no adjustment for multiple comparisons was made. Reported tests and confidence intervals are therefore nominal and unadjusted, not family-wise confirmatory evidence.[1][4]

The primary efficacy estimand asks what the mean treatment difference would have been if eligible randomized participants had continued their assigned treatment, allowing protocol-permitted dose reductions for gastrointestinal tolerability. This hypothetical continued-treatment question differs from the effect of assignment regardless of discontinuation.[1][3][4]

Primary outcome at week 24

For that efficacy estimand, the paper reported least-squares mean weight changes of:

The posted registry separately reports each of the six retatrutide arms. Its placebo-adjusted week-24 differences ranged from −5.64 percentage points for the 1 mg arm to −16.72 percentage points for one 8 mg arm. Each registry comparison reported a nominal, unadjusted two-sided 95% confidence interval that excluded zero and a nominal p-value below 0.001.[2][4]

The estimates support a weight-change signal under the stated model and estimand. Their nominal p-values and intervals do not control the combined false-positive risk across comparisons and should not be presented as definitive treatment-effect evidence. The paper combines starting-regimen groups for 4 mg and 8 mg and uses a dose-pooling model; its estimates and the registry's separately reported arm comparisons must remain identified by source.[1][2][4] The confidence-interval and multiplicity guide explains why effect magnitude, statistical uncertainty, and clinical importance are separate questions.

Secondary and later outcomes

At week 48, the paper reported least-squares mean changes of −8.7%, −17.1%, −22.8%, and −24.2% in the 1 mg, combined 4 mg, combined 8 mg, and 12 mg groups, respectively, compared with −2.1% with placebo.[1]

The proportions reported with weight reductions of at least 5%, 10%, and 15% at week 48 were 92%, 75%, and 60% in the combined 4 mg group; 100%, 91%, and 75% in the combined 8 mg group; and 100%, 93%, and 83% in the 12 mg group. The corresponding placebo proportions were 27%, 9%, and 2%.[1]

These secondary results use the paper's efficacy estimand; any associated tests or intervals remain nominal and unadjusted for multiplicity.[1][4] They are protocol-specific body-weight outcomes. The 48-week follow-up did not establish durability after the intervention ended, effects on mortality or major cardiovascular events, or a complete long-term safety profile.

Adverse events

The most common adverse events in the retatrutide groups were gastrointestinal. The paper described them as dose-related and mostly mild to moderate. It also reported dose-dependent heart-rate increases that peaked at week 24 and declined afterward.[1]

The safety population comprised 337 participants who received at least one dose: 70 placebo and 267 retatrutide. The registry reports serious adverse events in 3 of those 70 placebo participants and 10 of those 267 active-treatment participants.[2] These are participant counts, not numbers of separate events.

One death occurred in an active group. The paper identifies drowning and reports the site investigator's assessment that it was unrelated to retatrutide. Separately, an external physician committee classified whether the death was cardiovascular-related as undetermined; that was not a judgment of drug-relatedness.[1] This record reports those assessments without independently assigning causality. A trial of this size and duration cannot reliably characterize uncommon or delayed harms. The adverse-event and tolerability guide explains the distinction between an event, seriousness, and causal assessment.

Attrition and missing data

The paper reports that 81% of participants completed the 52-week trial and 78% completed the 48-week assigned treatment. Completion was 71% in the placebo group, compared with 76%–87% across the retatrutide groups for the overall trial.[1]

ClinicalTrials.gov lists 275 of 338 randomized participants as completing the overall study: 50 of 70 with placebo and 225 of 268 across the six assigned retatrutide groups. One randomized active-arm participant was never treated, leaving the safety population at 337.[2]

The registry's primary-outcome analysis denominator is 335: 68 placebo and 267 active participants. It required baseline and at least one post-baseline weight value, excluded participants who discontinued because of inadvertent enrollment, and excluded observations after study-drug discontinuation. The flow table records two placebo participants under inadvertent enrollment. The analysis count is not a count of 335 observed week-24 measurements.[2]

For the hypothetical efficacy estimand, the model uses pre-discontinuation observations to estimate outcomes that would have occurred with continued assigned treatment; it does not simply average treatment completers. Differential completion and the missing-data assumptions still matter.[1][4] See the estimands and missing-data guide for those distinctions.

How to interpret significance, effect size, and uncertainty

• Statistical significance: the week-24 registry comparisons reported nominal p-values below 0.001 under the stated model; no multiplicity adjustment was applied.[2][4]
• Effect size: the modeled placebo-adjusted differences were measured in percentage points of baseline body weight and varied by arm.
• Uncertainty: the nominal confidence intervals were unadjusted for multiple comparisons. They do not provide simultaneous coverage across the tested comparisons or cover every source of bias, missingness, or limited generalizability.[1][4]
• Clinical importance: weight change can be meaningful, but this phase 2 study did not directly establish long-term morbidity, mortality, or performance outside its protocol.

Important limitations

• The study was a phase 2 dose-ranging trial with 338 participants and 48 weeks of treatment, not a large long-term outcomes trial.
• All sites listed in the registry were in the United States; 298 of 338 participants were recorded as White, limiting some questions about broader representativeness.
• Adults with diabetes, recent large weight change, several medical conditions, and various medication exposures were excluded.
• The comparator was placebo rather than an active treatment.
• Tests and confidence intervals were not adjusted for multiple comparisons; the primary/secondary endpoint distinction does not supply family-wise error control.[1][4]
• Completion was lower with placebo, and the primary efficacy result targets hypothetical continued treatment using a restricted analysis set and model-dependent handling of missing observations.[1][2][4]

Eli Lilly sponsored and funded the trial. The paper states that Lilly designed and oversaw it and handled site monitoring, data collation, and analysis; investigators at the study sites collected the data. The academic first author and Lilly-employee last author prepared the first manuscript draft.[1] These roles are relevant context, not a standalone verdict on the validity of the result.

What this study does not prove

NCT04881760 does not prove that retatrutide is safe or effective for every person or condition, that its effects persist indefinitely, or that it improves outcomes the trial did not test. FDA currently states that retatrutide is not a component of an FDA-approved drug, has not been found safe and effective for any condition, and cannot be used in compounding under federal law.[5]

The registry and FDA status-check guide explains why trial registration, a publication, and product-specific approval must be checked separately.

The study also does not authenticate or validate a separately sourced material. Shared naming, a research-use label, a certificate, or a commercial shorthand such as “GLP-3” does not establish that another material is LY3437943, matches the trial formulation, or would reproduce the trial result.

Frequently asked questions

What was the primary endpoint in NCT04881760?

It was percentage change in body weight from baseline to week 24, estimated as a least-squares mean under a hypothetical continued-treatment efficacy estimand. Prespecification did not make the tests or intervals multiplicity-adjusted.[1][4]

How many people were randomized?

The trial randomized 338 adults across six retatrutide arms and one placebo arm. Of these, 337 received at least one dose and were included in safety analyses; the registry's primary-outcome analysis included 335, with its separate eligibility and available-data rules.[2]

Did every retatrutide group differ from placebo at week 24?

Each individual active-arm registry comparison reported a nominal p-value below 0.001 and an unadjusted 95% confidence interval excluding zero. Those model-based comparisons were not adjusted for multiplicity and are not family-wise confirmatory proof.[2][4]

Were the 48-week results the primary endpoint?

No. Week-48 weight change and threshold outcomes were secondary measures under the efficacy estimand. Their tests and intervals were also nominal and unadjusted; they do not replace the week-24 primary endpoint.[1][4]

Does this trial establish long-term safety?

No. Its sample size and duration cannot reliably characterize uncommon, delayed, or longer-term adverse outcomes.

Is retatrutide FDA approved?

No. FDA's current page states that retatrutide is not a component of an FDA-approved drug and has not been found safe and effective for any condition.[5]

Does the study validate a product called “GLP-3”?

No. The trial studied the sponsor's defined LY3437943 material. An informal label does not establish molecular identity, formulation, or equivalence.

Primary and official sources

1. Primary paper: Triple–Hormone-Receptor Agonist Retatrutide for Obesity, A Phase 2 Trial; PubMed record, PMID 37366315
2. ClinicalTrials.gov: NCT04881760 protocol and posted results
3. NCT04881760 study protocol
4. NCT04881760 statistical analysis plan
5. FDA: Concerns with unapproved GLP-1 drugs used for weight loss

Review and corrections

Sources checked September 2, 2026. This record describes NCT04881760 and its original phase 2 publication; later trials require their own evidence records. Serum Research is responsible for this editorial summary. No independent clinical, statistical, scientific, analytical, or regulatory review is claimed. This is not medical advice, dosing or administration guidance, or a finding that any separate material is authentic, equivalent, safe, or effective.

To report a source error or correction, contact Serum Research. Read how we review evidence, use the study-reading primer to put this record in context, or return to the Research Library.