TRANSCEND-T2D-1 was a 40-week phase 3 trial of retatrutide in adults whose type 2 diabetes was inadequately controlled with diet and exercise alone. All three retatrutide groups had larger average reductions in hemoglobin A1c and body weight than placebo under the prespecified treatment-regimen estimand. Those findings are substantial randomized evidence for this exact study, but they are not long-term complication data, an individual prediction, or evidence that another retatrutide material is the intervention used in the trial. The live registry had no posted summary results when checked in August 2026, and FDA currently says retatrutide has not been found safe and effective for any condition.[1][2][3][4]
Study at a glance
| Field | Study record |
|---|---|
| Research question | At week 40, did retatrutide reduce HbA1c more than placebo in adults with relatively early type 2 diabetes managed with diet and exercise? |
| Design | Phase 3, multicenter, randomized, double-blind, placebo-controlled, parallel trial at 48 sites in the United States, Mexico, and India |
| Population | 537 randomized adults with HbA1c 7.0%–9.5% and BMI at least 23; most had not previously used antihyperglycemic medication |
| Intervention | Historical identifiers: three protocol-defined retatrutide escalation groups reaching nominal 4-, 9-, or 12-mg study exposures |
| Comparator | Matching placebo under the same 40-week protocol |
| Primary endpoint | Change from baseline in HbA1c at week 40 |
| Key secondary endpoint | Percentage change in body weight at week 40 |
| Registration | NCT06354660; completed; actual enrollment 537; no structured results posted at the August 27, 2026 check |
| Publication | The Lancet 2026; PMID 42250575; DOI 10.1016/S0140-6736(26)00967-0 |
| Sponsor | Eli Lilly; extensive sponsor roles and employee/shareholder authors disclosed |
The amounts, route, schedule, and escalation recorded in the protocol identify the experiment. They are not dosing, preparation, administration, self-use, or treatment guidance.[1]
Design and population
Investigators screened 930 people at 48 sites and randomly assigned 537: 134 to the nominal 4-mg group, 133 to 9 mg, 136 to 12 mg, and 134 to placebo. Eligible adults were at least 18 years old, had type 2 diabetes with HbA1c from 7.0% to 9.5%, and had not used a diabetes medication during the preceding 90 days. The question was therefore retatrutide as studied against placebo on a diet-and-exercise background, not an add-on comparison across common multidrug regimens.[1][2]
At baseline, mean age was 48.8 years, HbA1c 7.9%, body weight 96.9 kg, BMI 35.8, and diabetes duration 2.5 years. There were 296 women and 241 men. Eighty-five percent had never used an antihyperglycemic medicine. This relatively early and largely medication-naive population matters: results should not be assumed for people with long-duration diabetes, extensive complications, different medication backgrounds, or populations not represented at the three participating countries.[1]
Four participant counts answer four different questions
Of 930 screened people, 537 were randomized and included in efficacy analyses. The paper says 504 completed the study, while 490 completed 40 weeks on assigned study intervention. A person could complete follow-up without remaining on intervention, so those counts are not contradictions and should not be collapsed. Likewise, neither completion count replaces the randomized denominator for the treatment-regimen analysis.[1]
The live ClinicalTrials.gov record independently reports actual enrollment of 537 and completed status. It lists the actual primary-completion date as January 22, 2026, and study completion as February 20, 2026. The record was first posted April 9, 2024, before study enrollment began, and last updated March 5, 2026. Its results tab still said “No Results Posted” on August 27, 2026. The paper, rather than a registry result table, is therefore the available result source.[2]
Primary endpoint and two prespecified estimands
The primary endpoint was week-40 change in HbA1c. The headline analysis used a treatment-regimen estimand: in plain language, it asked about the average effect of assignment regardless of whether participants remained fully adherent or later received additional glucose-lowering or rescue therapy. Missing outcomes were handled through the prespecified imputation and ANCOVA procedure.[1]
The paper also reports an efficacy estimand that asks a different hypothetical question, what the effect would be if participants stayed on assigned intervention and did not use rescue therapy, using a mixed model for repeated measures. Both were prespecified, but they should not be blended with each other or with a completer-only comparison. This page uses the treatment-regimen estimand for the main numeric claims because that is the designated primary framework.[1]
The testing hierarchy controlled the family-wise type I error rate at .05 across primary and key secondary hypotheses. The paper also warns that ordinary confidence intervals were not adjusted for multiplicity. A small p value addresses compatibility with a no-difference hypothesis under the analysis model; it does not determine clinical importance or guarantee the same response for an individual.[1]
HbA1c result at week 40
Mean HbA1c changes under the treatment-regimen estimand were −1.69 percentage points in the 4-mg group, −1.86 in the 9-mg group, −1.94 in the 12-mg group, and −0.81 with placebo.[1] The active-versus-placebo differences were:
- 4 mg: −0.88 percentage points (95% CI −1.18 to −0.59)
- 9 mg: −1.04 (95% CI −1.32 to −0.76)
- 12 mg: −1.12 (95% CI −1.39 to −0.85)[1]
All three comparisons had reported p values below .0001.[1] These are randomized-group average differences at 40 weeks, not promises that a particular person’s HbA1c will change by one of those amounts. The confidence intervals quantify sampling uncertainty under the model, while population selection, missing-data assumptions, adherence, and follow-up length create additional uncertainty.
Body-weight result was key secondary, not primary
Mean body-weight changes were −11.5%, −13.9%, and −15.3% in the three active groups, compared with −2.6% under placebo. Differences were −8.9 percentage points (95% CI −10.5 to −7.2), −11.3 (−13.1 to −9.5), and −12.7 (−14.4 to −11.0), all p<.0001.[1]
Body weight had not reached a plateau at week 40. That makes the time point especially important: the study does not show where weight would eventually stabilize, how much would persist after exposure ended, or whether longer follow-up would change the benefit–harm balance. Weight change also cannot substitute for a tested reduction in cardiovascular, kidney, eye, nerve, hospitalization, or mortality outcomes; those were not established here.[1]
Adverse events and discontinuations
Treatment-emergent adverse events occurred in 84/134 participants (63%) at 4 mg, 85/133 (64%) at 9 mg, 85/136 (63%) at 12 mg, and 77/134 (57%) with placebo. Adverse events led to intervention discontinuation in 3/134 (2%), 6/133 (5%), 7/136 (5%), and none of 134 placebo participants. Serious adverse events occurred in 6/134 (4%), 1/133 (1%), 5/136 (4%), and 2/134 (1%). A 537-person, 40-week trial still has limited ability to detect rare or delayed harm.[1]
Gastrointestinal events were common and occurred mainly during escalation. Diarrhea affected 25/134, 35/133, 31/136, and 6/134; nausea 22/134, 26/133, 36/136, and 5/134; vomiting 21/134, 20/133, 24/136, and 3/134. Dysesthesia was reported in 6/134, 3/133, 6/136, and no placebo participants. Three dysesthesia events were severe, one in each active group, and two led to discontinuation.[1]
No severe hypoglycemia was reported. Level-2 hypoglycemia, defined in the paper by glucose below 54 mg/dL, occurred in zero, two, one, and zero participants across the four groups. Two participants died in the 4-mg group. Investigators described both deaths as unrelated to study drug. That is their causality assessment, not proof supplied by this page that the intervention did or did not contribute.[1]
Sponsor role, disclosures, and regulatory status
Eli Lilly funded the trial and was involved in conception, design, supervision, monitoring, data collection, and analysis. Six authors were employees and shareholders; investigator relationships with industry were also disclosed. Sponsor involvement does not make the findings unusable, but it raises the importance of transparent methods, complete reporting, and independent replication.[1]
As of the evidence review, FDA states that retatrutide is not a component of an FDA-approved drug and has not been found safe and effective for any condition.[4] A clinical-trial publication can report efficacy signals without creating marketing approval. An NDC/SPL entry, export listing, supplier name, or matching ingredient name would not establish an approved application or show that another material matches the clinical study intervention.
Statistical significance, effect, uncertainty, and importance
- Statistical significance: all three active groups differed from placebo for primary HbA1c and key secondary weight outcomes under the controlled hierarchy.
- Effect size: HbA1c differences ranged from −0.88 to −1.12 percentage points; weight differences ranged from −8.9 to −12.7 percentage points.
- Uncertainty: intervals were reasonably bounded, but estimand assumptions, intervention discontinuation, population selection, and 40-week follow-up remain relevant.
- Clinical importance: intermediate metabolic outcomes improved; long-term diabetic complications, durability, rare harms, and survival were not established.
Important limitations
The study lasted 40 weeks, included only three countries, and enrolled mostly medication-naive adults with relatively early diabetes. It did not test long-term clinical complications, provide a mature weight plateau, or resolve uncommon and delayed safety outcomes. Thirty-three randomized participants did not complete the study and 47 did not complete the treatment period on assigned intervention. The treatment-regimen estimand addresses that reality analytically, but no missing-data method removes every assumption.[1]
What this study does not prove
TRANSCEND-T2D-1 does not prove prevention of heart attack, stroke, kidney failure, neuropathy, retinopathy, hospitalization, or death. It does not establish long-term durability or safety, an individual result, or effects in every diabetes population. It also does not authenticate a retatrutide vial, powder, blend, compounded preparation, catalog material, supplier, or lot. Only the named study intervention under this protocol was tested.[1][2][4]
Frequently asked questions
What was the trial’s primary outcome?
The prespecified primary outcome was change from baseline in HbA1c at week 40. Body-weight change was a key secondary outcome, not the primary endpoint.[1]
Were 537 or 490 people analyzed?
All 537 randomized participants entered efficacy analyses. The smaller number, 490, completed 40 weeks on assigned study intervention; 504 completed the study follow-up.[1]
What does the treatment-regimen estimand mean?
It estimates the average effect of assignment regardless of later adherence or additional glucose-lowering/rescue therapy, using prespecified methods for missing outcomes.[1]
How much did HbA1c differ from placebo?
Average active-versus-placebo differences were −0.88, −1.04, and −1.12 percentage points across the 4-, 9-, and 12-mg study groups.[1]
Did the trial show long-term diabetes protection?
No. It measured 40-week metabolic outcomes and safety observations, not long-term vascular, kidney, eye, nerve, hospitalization, or survival outcomes.[1]
Were severe hypoglycemia or deaths reported?
No severe hypoglycemia was reported. Two deaths occurred in the 4-mg group and were assessed by investigators as unrelated; this page does not independently adjudicate causality.[1]
Is retatrutide FDA approved or does this validate another material?
No. FDA currently says retatrutide has not been found safe and effective for any condition, and this trial cannot establish the identity or equivalence of another material.[4]