Retatrutide Trial Results: What the Headlines Leave Out

What follows is a living summary of retatrutide’s trial record, not a forecast of approval or availability. Retatrutide (LY3437943) is Eli Lilly’s investigational once-weekly GIP/GLP-1/glucagon “triple agonist”; it is not approved and legally obtainable only inside Lilly trials.

TL;DR — what the headlines leave out:

• The record numbers (24.2% in phase 2; 28.7% in TRIUMPH-4) are one dose at one timepoint on the efficacy estimand. Where trials are published, the as-randomized analysis is smaller: 25.0% vs the advertised 28.3% (TRIUMPH-1); 15.3% vs 16.8% (TRANSCEND-T2D-1).
• Type 2 diabetes costs roughly 6–8 weight-loss percentage points (−20.8% vs −28.3% in the 80-week trials).
• Discontinuation for adverse events reaches 18.2% at the top dose (TRIUMPH-4), and dysesthesia — altered skin sensation — recurs across trials.
• The cardiovascular signal is neutral, not proven: TRIUMPH-3’s hazard ratios cross 1; a 10,000-patient outcomes trial reports ~2029.
• Durability past two years, regain after stopping, and lean-mass effects in obesity are unmeasured.

Every retatrutide result so far, dated

Peer-reviewed:

• Phase 2 obesity (NCT04881760) — NEJM, June 26, 2023; 338 adults; 48 weeks [@jastreboff2023].
• Phase 2 type 2 diabetes (NCT04867785) — The Lancet, June 26, 2023; 281 adults; dulaglutide 1.5 mg comparator [@rosenstock2023].
• TRIUMPH-1 (NCT05929066) — phase 3 master trial with knee-osteoarthritis and sleep-apnea baskets; NEJM, September 29, 2026; 2,339 randomized; 80 weeks [@jastreboff2026].
• TRANSCEND-T2D-1 (NCT06354660) — phase 3, type 2 diabetes; The Lancet, June 6, 2026; 537 randomized; 40 weeks [@bajaj2026].

Company topline — details unpublished as of last check:

• TRIUMPH-4 (NCT05931367) — knee osteoarthritis; 445 randomized; 68 weeks; topline December 11, 2025 [@lilly2025].
• TRANSCEND-T2D-1 — first topline March 19, 2026, superseded by the June publication [@lilly2026a].
• TRIUMPH-1 — topline May 21, 2026; ADA detail June 6, 2026; NEJM September 2026 [@lilly2026b].
• TRIUMPH-2 (NCT05929079) — obesity/overweight plus type 2 diabetes; 1,152 randomized; 80 weeks; topline July 23, 2026 [@lilly2026c].
• TRIUMPH-3 (NCT05882045) — severe obesity (BMI ≥35) with established cardiovascular disease, ± diabetes; 1,949 randomized; 80 weeks; topline July 23, 2026 [@lilly2026c].

Still running (registry records, no readouts): a 10,000-participant cardiovascular-and-kidney outcomes trial (NCT06383390; primary completion listed February 2029); head-to-heads vs semaglutide in type 2 diabetes (NCT06260722; primary completion listed August 2026) and vs tirzepatide in obesity (NCT06662383; November 2026); a phase 3b maintenance trial (NCT06859268; April 2028); plus chronic-kidney-disease, chronic-low-back-pain, and MASLD trials [@ctgov].

Design and population

Phase 2 randomized 338 US adults (51.8% men; 4% with overweight — BMI 27–<30 plus a comorbidity — the rest with obesity) in a 2:1:1:1:1:2:2 split across placebo and six retatrutide arms from 1 mg to 12 mg, injected weekly for 48 weeks. The 4-mg and 8-mg levels each had two arms testing the starting dose (2 vs 4 mg); escalation every four weeks for up to 12 weeks reached the maintenance dose, which then held. Primary endpoint: percent weight change at 24 weeks; the 48-week change was secondary.

The phase 3 TRIUMPH program, started in 2023 with more than 5,800 people enrolled, standardized the ladder: 2 mg, stepping up every four weeks to targets of 4, 9, or 12 mg. TRIUMPH-1, in obesity/overweight without diabetes, is a master trial with two baskets — 574 with knee osteoarthritis pain, 243 with moderate-to-severe sleep apnea — plus a pre-specified 104-week extension for those with baseline BMI ≥35 who finished the 80-week main trial and tolerated their dose. TRIUMPH-2 covers type 2 diabetes with obesity or overweight; TRIUMPH-3, the earliest-started registrational trial, enrolls severe obesity (class 2–3) with established cardiovascular disease (1:1:2, extra placebo); TRANSCEND-T2D-1 is a 40-week monotherapy trial in diet/exercise-treated diabetes (baseline A1C 7.0–9.5%, BMI ≥23). Throughout, remember the populations aren’t interchangeable: diabetes consistently lowers achievable weight loss.

Weight and other endpoints

Phase 2 (48 weeks): least-squares mean weight change was −7.2% at 24 weeks (placebo −1.6%) and −24.2% at 48 weeks for 12 mg (placebo −2.1%); intermediate arms −8.7%, −17.1%, −22.8%. At 48 weeks, ≥10% and ≥15% loss occurred in 75% and 60% of the combined 4-mg arm, 91% and 75% of combined 8-mg, and 93% and 83% of 12-mg — vs 9% and 2% on placebo; 26% of the 12-mg arm lost ≥30%. Other endpoints: waist −19.6 cm (placebo −2.6); ~20% LDL-C reduction; 72% of those with baseline prediabetes reverting to normoglycemia (vs 22% placebo); 30–41% of the 8–12 mg arms dropping an antihypertensive drug [@jastreboff2023].

Phase 2, type 2 diabetes (36 weeks): A1C fell 2.02% at 24 weeks with 12 mg (placebo −0.01%; dulaglutide −1.41%); weight fell up to 16.94% at 36 weeks — significantly more than both comparators — still without a plateau; ADA-presentation figures put up to 82% of participants at A1C <6.5% [@rosenstock2023; @ada2023].

TRIUMPH-1 (80 weeks): Lilly’s efficacy-estimand table reads −19.0% (4 mg), −25.9% (9 mg), −28.3% (12 mg) vs −2.2% placebo from a baseline of 112.7 kg. The published NEJM paper — which made the as-randomized treatment-regimen analysis primary — reports −25.0% for 12 mg vs −3.9% placebo (a 21.0-point difference) and −23.7% for 9 mg, both P<0.001. By Lilly's table, 62.5%, 45.3%, and 27.2% of the 12-mg arm lost ≥25%, ≥30%, and ≥35% of body weight (placebo: 2.2%, 0.5%, 0.3%). The baskets met their endpoints in the peer-reviewed paper: WOMAC pain fell 3.5–3.6 points for 9–12 mg on the hybrid estimand (ITT 3.9–4.1) vs 1.9–2.5 on placebo, and apnea-hypopnea index fell ~32 events/hour at 12 mg (ITT −32.1; hybrid −31.7) vs 9.6 (ITT). In the 104-week extension (532 people: baseline BMI ≥35, dose-tolerant, all on maximum tolerated dose), 12 mg reached −30.3% and the placebo-assigned group — now on active drug — −19.2%: company topline, selected subgroup [@jastreboff2026; @lilly2026b].

TRANSCEND-T2D-1 (40 weeks): the Lancet’s treatment-regimen primary shows A1C −1.94% (12 mg) vs −0.81% placebo and weight −15.3% vs −2.6%; the topline efficacy-estimand figures were −2.0% and −16.8%. Weight had not plateaued by week 40; up to 90% reached A1C <7.0%, 85% ≤6.5%, and 46% normoglycemia <5.7% (company announcement; all those target proportions were significant in the paper) [@bajaj2026; @lilly2026a].

TRIUMPH-2 (80 weeks, diabetes; unpublished topline): −12.7% (4 mg), −19.1% (9 mg), −20.8% (12 mg) vs −4.0% placebo; best A1C reduction −1.6% from 7.7% [@lilly2026c].

TRIUMPH-3 (80 weeks, CVD; unpublished topline): −21.6% (9 mg), −22.6% (12 mg) vs −3.2% placebo; the 12-mg arm showed −37.0% triglycerides, −16.5% non-HDL cholesterol, −9.3 mmHg systolic blood pressure, −7.5 in (19 cm) waist circumference, and −51.2% hsCRP. In pre-specified analyses there were 44 MACE-5 events (all-cause death, heart attack, stroke, heart-failure event, or coronary revascularization) on retatrutide vs 52 on placebo, HR 0.82 (95% CI 0.55–1.22); the on-treatment variant, HR 0.73 (0.47–1.12), was not pre-specified; MACE-3 (cardiovascular death, heart attack, or stroke): 27 vs 23 events, HR 1.12 (0.64–1.96). Events accrued slowly in both arms — the study wasn’t powered to establish cardiovascular benefit [@lilly2026c].

TRIUMPH-4 (68 weeks, knee OA; unpublished topline): efficacy estimand −28.7% (12 mg) vs −2.1% placebo; treatment-regimen −23.7% vs −4.6%; WOMAC pain −4.5 points (75.8%) vs −40.3% placebo; ≥30% weight loss in 39.4% (12 mg) vs 0.8%; systolic blood pressure −14.0 mmHg at the top dose; post-hoc, 14.1%/12.0% of retatrutide-treated participants reported no knee pain vs 4.2% on placebo [@lilly2025].

The estimand fine print

Every headline depends on how intercurrent events — quitting, switching to other weight-loss treatment, intensifying diabetes therapy — are counted. In plain terms:

• Efficacy estimand: what would have happened had everyone stayed on their drug as intended, with no prohibited weight-loss therapy (estimates carried by modeling) — the larger figures Lilly’s releases lead with: −28.3%, −28.7%, −16.8%, up to −2.0% A1C.
• Treatment-regimen (as-randomized) estimand: everyone analyzed as randomized, dropouts and switchers included. This is the primary analysis in the published TRIUMPH-1 (NEJM: −25.0%) and TRANSCEND-T2D-1 (Lancet: −15.3%; A1C −1.94%) papers.
• Hybrid estimand (osteoarthritis and apnea baskets): certain intercurrent events counted as treatment failure, with ITT alongside — the reason pain and AHI figures shift modestly between company releases and the journal.

None of this makes the efficacy-estimand numbers wrong; it means “up to 28.7%” is the most favorable of two legitimate readings, and the as-randomized figure describes what actually happened to everyone enrolled. The gap also makes discontinuation material — see below.

Discontinuation and adverse events

The GI signature repeats everywhere: dose-related, mostly mild-to-moderate, concentrated during escalation, and partially mitigated by lower starting doses.

Percentages are company topline; TRIUMPH-2/-3 (and TRIUMPH-4’s details) await peer review. GI events were the most common reason for treatment discontinuation in phase 2 [@lilly2026b; @lilly2026c; @lilly2025; @jastreboff2023].

• Phase 2: overall adverse events in 70% (placebo) vs 73–94% (retatrutide); discontinuation for them in 6–16% of retatrutide participants, none on placebo; skin-sensitivity events 7% vs 1%; one acute pancreatitis (other enzyme rises asymptomatic); one severe arrhythmia (prolonged QT, in a participant also on ondansetron); no clinically significant hypoglycemia, medullary thyroid cancer, or C-cell hyperplasia; 15 serious events in 13 participants (4% in both arms), including one drowning death.
• Heart rate rose dose-dependently, peaked at 24 weeks, and declined thereafter, at magnitudes the authors compared to GLP-1 agonists — a class watch-item, not a resolved question.
• Dysesthesia is the novel signal vs earlier incretins: 12.5% (TRIUMPH-1 12 mg) vs 0.9% placebo; 20.9% (TRIUMPH-4 12 mg) vs 0.7%; ~6–7% in TRIUMPH-2/-3 — generally mild, mostly resolving, rarely leading to discontinuation.
• TRIUMPH-4’s 18.2% discontinuation was “highly correlated with baseline BMI and included discontinuations for perceived excessive weight loss”; at baseline BMI ≥35 it was 12.1% on 12 mg.
• TRANSCEND-T2D-1 (published): 2–5% discontinuation for adverse events vs 0% placebo; no severe hypoglycemia; two deaths, both 4-mg arm, judged unrelated to drug.

Durability and the questions that remain

Known: weight keeps falling through these treatment windows — no plateau in phase 2 at 48 weeks, none in TRANSCEND-T2D-1 at 40, and the TRIUMPH-1 extension added another 2–3 points between weeks 80 and 104 on the top dose. Off-drug follow-up is thin; TRIUMPH-2 kept four weeks.

Not known:

• Weight regain and stopping. No randomized stopping or switching data exist yet; the dedicated maintenance trial (NCT06859268) lists an April 2028 primary completion.
• Cardiovascular outcomes. TRIUMPH-3’s event counts are too small to establish benefit or harm; the 10,000-patient MACE-and-kidney trial (NCT06383390) doesn’t read out for years.
• Head-to-head position. No published trial has compared retatrutide with semaglutide or tirzepatide; both are in progress, with registry primary completions listed for late 2026.
• Body composition. A published analysis in the phase 2 diabetes cohort showed significantly greater fat-mass loss versus placebo and dulaglutide; equivalent lean-mass data in the obesity population aren’t published [@ldendio2025].
• Long-term safety. Two years at most, in a selected extension subset; dysesthesia, heart-rate, pancreatitis, and gallbladder signals need longer exposure and complete phase 3 tables.
• Pending publication. Detailed TRIUMPH-4, TRIUMPH-2, and TRIUMPH-3 results are promised at future meetings and journals. Lilly says it is completing its CMC package and plans a US biologics-license submission in Q1 2027 — a company statement, not a regulatory event.

The honest reading until then: efficacy unprecedented in published obesity pharmacotherapy, dose-dependent tolerance costs, a neutral (not negative) cardiovascular signal, and partly unpublished durability — every number subject to revision as the peer-reviewed record catches up with the press releases.

Want the next retatrutide update the moment the TRIUMPH-2/3 or head-to-head papers break? Join our research newsletter — trial record, not stock reaction.

Sources

1. Jastreboff AM, et al. N Engl J Med 2023;389:514–526 (phase 2, obesity). nejm.org
2. Rosenstock J, et al. Lancet 2023;402:529–544 (phase 2, T2D). thelancet.com01053-X/fulltext)
3. Jastreboff AM, et al. (TRIUMPH-1) N Engl J Med, published online September 29, 2026. nejm.org
4. Bajaj HS, et al. (TRANSCEND-T2D-1) Lancet 2026;407:2402–2413. thelancet.com00967-0/fulltext)
5. Lilly press releases — TRIUMPH-4, Dec 11, 2025; TRANSCEND-T2D-1, Mar 19, 2026; TRIUMPH-1, May 21, 2026; ADA data, Jun 6, 2026; TRIUMPH-2/-3, Jul 23, 2026. investor.lilly.com
6. Body-composition substudy (phase 2, T2D), Lancet Diabetes Endocrinol 2025. thelancet.com00092-0/abstract)
7. ClinicalTrials.gov: NCT04881760, NCT04867785, NCT05929066, NCT05929079, NCT05882045, NCT05931367, NCT06354660, NCT06383390, NCT06260722, NCT06662383, NCT06859268, NCT07035093, NCT07165028.

Suggest a correction