If you’ve been following this newsletter, you know I opened this whole series with retatrutide back in June. The triple agonist. 28.3% weight loss at 80 weeks. The number that made me write the sentence “a decade ago I would have called it fantasy.”

At the end of that piece I listed what we still needed. Full Phase 3 safety data across all TRIUMPH trials. Cardiovascular data. And a proper understanding of the dysesthesia signal and the higher discontinuation rate at the top dose.

On 23 July, Eli Lilly and Company delivered two of those three.

TRIUMPH-2 and TRIUMPH-3 read out on the same day. Both met their primary endpoints. Neither reproduced TRIUMPH-1.

And I want to be clear that this is not a disappointment story. It’s the opposite. TRIUMPH-1 studied people with obesity and no diabetes. These two trials studied the patients who are genuinely hard to treat — people with type 2 diabetes, and people with Class 2/3 obesity and established cardiovascular disease. The numbers came down. They came down in exactly the way the class has taught us to expect. And that makes them far more useful for actual clinical conversations than the headline figure ever was.

Let me walk you through the data.

What Was Studied

TRIUMPH-2 (NCT05929079)

Design: Phase 3, 80-week, randomised, double-blind, placebo-controlled basket trial in adults with type 2 diabetes and obesity or overweight. 1,152 participants randomised 1:1:1:1 to retatrutide 4 mg, 9 mg, 12 mg, or placebo. Retatrutide arms started at 2 mg and escalated stepwise every four weeks to the assigned target dose.

Baseline: mean weight 106.4 kg (234.6 lb), mean BMI 38.2, mean A1C 7.7%.

TRIUMPH-3 (NCT05882045)

Design: Phase 3, 80-week, randomised, double-blind, placebo-controlled trial in adults with severe obesity (Class 2 or 3, BMI ≥35) and established cardiovascular disease, with or without type 2 diabetes. 1,949 participants randomised 1:1:2 to retatrutide 9 mg, 12 mg, or placebo.

Baseline: mean weight 111.4 kg (245.6 lb), mean BMI 40.4.

One technical note before the numbers, and long-time readers will recognise it from my Foundayo piece. Everything below is the efficacy estimand — the treatment effect if everyone had stayed on drug and not started rescue therapy. It’s the more flattering of the two standard analyses. Lilly has released no treatment-regimen figures. With orforglipron, the gap between the two was more than a full percentage point. Assume something similar here, and hold these numbers loosely until the publications land.

The Efficacy Numbers

TRIUMPH-2 — percent body weight change at 80 weeks:

Placebo: −4.0% (−9.3 lb)
Retatrutide 4 mg: −12.7% (−29.8 lb)
Retatrutide 9 mg: −19.1% (−45.4 lb)
Retatrutide 12 mg: −20.8% (−49.6 lb)

TRIUMPH-2 — A1C change from a baseline of 7.7%:

Placebo: −0.2%
4 mg: −1.4%
9 mg: −1.6%
12 mg: −1.5%

TRIUMPH-3 — percent body weight change at 80 weeks:

Placebo: −3.2% (−7.7 lb)
Retatrutide 9 mg: −21.6% (−52.7 lb)
Retatrutide 12 mg: −22.6% (−55.8 lb)

The T2D Gap Now Has a Third Data Point

I’ve been circling this question for three articles now, so let me put it all in one place.

In the Wegovy HD piece I flagged the persistent gap between what semaglutide does in obesity and what it does in obesity plus type 2 diabetes: 20.7% versus 14.1%. In the Foundayo piece, orforglipron showed the same shape: 12.4% versus 10.5%. I called it one of the most interesting open questions in metabolic medicine, and said the counter-regulatory mechanisms that defend body weight in the setting of insulin resistance are still not fully understood.

Retatrutide now joins the list.

Weight loss in obesity vs. obesity + T2D, highest dose:

Retatrutide: 28.3% → 20.8%
Semaglutide 7.2 mg: 20.7% → 14.1%
Orforglipron: 12.4% → 10.5%

Placebo-adjusted, retatrutide 12 mg delivered about two-thirds of the effect in type 2 diabetes that it delivered in the non-diabetic TRIUMPH-1 cohort — 16.8 points versus 26.1 points.

I have to add the caveat properly, because I would want someone to add it for me. These are separate trials, run in different populations over different periods. TRIUMPH-1’s cohort started heavier — 112.7 kg and BMI 40.0, against 106.4 and 38.2 in TRIUMPH-2. No trial has randomised diabetic and non-diabetic participants to retatrutide inside the same protocol. Cross-trial subtraction generates hypotheses; it doesn’t prove them.

But three molecules, three mechanisms, same direction. That’s a pattern.

And here’s the part I find most interesting. I had quietly assumed the glucagon arm might buy retatrutide an exemption — that if the problem in type 2 diabetes is partly about defended energy expenditure, a molecule that directly increases energy expenditure might punch through it.

It didn’t. Whatever is defending body weight in insulin-resistant physiology, adding a third receptor doesn’t overcome it. That’s a genuinely informative negative result, and it tells us the mechanism we’re up against is not simply an energy-expenditure deficit.

The Glucagon Question, Finally Answered

This was the single most-asked question I got after the June piece, and it deserves a direct answer. Glucagon receptor agonism promotes hepatic glucose output. Does the third receptor cost you glycaemic control?

No.

The strong evidence is TRIUMPH-2’s A1C data. You do not get 1.4 to 1.6 percentage point reductions from a baseline of 7.7% out of a molecule that is meaningfully impairing glucose handling. That is the answer, and it’s a clean one.

The weaker evidence is TRIUMPH-3, where hyperglycaemia was reported as an adverse event in 3.9% (9 mg) and 3.1% (12 mg) of retatrutide participants versus 13.4% on placebo. I’ve seen that gap being passed around as though retatrutide is actively protecting against hyperglycaemia. I’d be careful. These are spontaneously reported adverse events in a mixed diabetic and non-diabetic cohort, not measured glycaemia, and most of that placebo excess is almost certainly untreated disease progression rather than anything the drug did.

Either way, the mechanistic worry I raised in June is resolved. The energy expenditure benefit is not being purchased with glycaemic deterioration.

One more thing in the A1C data worth flagging. The curve is essentially flat from 9 mg upward — 1.6% at 9 mg, 1.5% at 12 mg. I’m not going to make anything of a 0.1-point difference without confidence intervals. But if glycaemic benefit really does plateau around 9 mg while weight loss and adverse events keep climbing, then dose selection in the patient whose primary problem is glycaemic becomes a real clinical decision rather than a titrate-to-max reflex. Watch for that when the full data is presented.

TRIUMPH-3 and the Cardiovascular Data (Where I Need to Correct Myself)

In June I wrote that TRIUMPH-3 was critical because “cardiovascular outcomes data will likely be required for full regulatory approval.”

I was half right, and I want to fix that in public.

TRIUMPH-3 was never powered as a cardiovascular outcomes trial. It is a weight management trial conducted in a cardiovascular population. Those are different studies, and the readout makes the difference painfully obvious.

Cardiovascular risk factors at 12 mg — genuinely excellent:

Triglycerides: −37.0%
Non-HDL cholesterol: −16.5%
Systolic blood pressure: −9.3 mmHg
Waist circumference: −19.0 cm (−7.5 in)
hsCRP: −51.2%

MACE — this is where you slow down.

Lilly reported two analyses: the pre-specified in-study analysis, and a non-pre-specified on-treatment analysis excluding events occurring more than 35 days after discontinuation.

MACE-5 (all-cause death, MI, stroke, heart failure event, coronary revascularisation) — 44 events on retatrutide, 52 on placebo:

In-study: HR 0.82 (95% CI 0.55–1.22)
On-treatment: HR 0.73 (95% CI 0.47–1.12)

MACE-3 (CV death, MI, stroke) — 27 events on retatrutide, 23 on placebo:

In-study: HR 1.12 (95% CI 0.64–1.96)
On-treatment: HR 0.92 (95% CI 0.51–1.65)

Lilly states plainly that MACE occurred less frequently than anticipated in both arms.

Read the confidence intervals, not the point estimates. All four cross 1.0, comfortably. In-study MACE-5 spans 0.55 to 1.22 — an interval compatible with a 45% risk reduction and with a 22% increase.

Then look at how much the point estimates move between the two analyses. MACE-3 swings from 1.12 to 0.92 depending on which defensible analytic choice you make. When two reasonable analyses land on opposite sides of unity, you are looking at noise. Not signal.

So: anyone circulating the 0.82 as proof of cardioprotection is reading a hazard ratio as though it were a p-value. Anyone circulating the 1.12 as evidence of harm is making the identical mistake in the other direction. Both takes are already in my feed, and both are wrong.

The cardiovascular and renal outcomes trial is running separately. That is the readout that answers this question. Not this one.

Safety: The Number I Flagged in June

In June I called TRIUMPH-1’s 11.3% discontinuation rate at 12 mg “a number worth watching,” and compared it to roughly 6–7% for tirzepatide 15 mg in SURMOUNT.

Here’s what these two trials added.

Discontinuation due to adverse events:

TRIUMPH-2 — placebo 4.9%, 4 mg 3.8%, 9 mg 11.6%, 12 mg 7.7%
TRIUMPH-3 — placebo 4.8%, 9 mg 9.8%, 12 mg 13.5%

TRIUMPH-3’s 13.5% is the highest figure we have for this molecule. In a Class 2/3 population with established cardiovascular disease, roughly one in seven patients on the top dose came off drug because of adverse events. Put that next to orforglipron’s 10.3% and tirzepatide’s 6–7%, and retatrutide is sitting at the rough end of the class. That is a real number for a risk–benefit conversation, not a footnote.

TRIUMPH-2’s pattern is odd — 11.6% at 9 mg but only 7.7% at 12 mg — and I want to resist the temptation to explain it. This was a parallel-group design. Participants were randomised to a target dose at baseline, and anyone discontinuing during escalation is still counted in the arm they were randomised to. There is no mechanism that could enrich the 12 mg arm for tolerant patients. With roughly 288 people per arm, the most probable explanation is chance. TRIUMPH-3, with a larger sample and the expected dose ordering, is the more trustworthy read.

Most common adverse events, TRIUMPH-2 (4 mg / 9 mg / 12 mg vs. placebo):

Diarrhoea: 27.4% / 33.5% / 33.6% vs. 13.2%
Nausea: 13.7% / 20.8% / 28.0% vs. 8.0%
Constipation: 14.0% / 16.2% / 16.8% vs. 9.4%
Decreased appetite: 5.8% / 12.3% / 17.1% vs. 4.5%
Vomiting: 5.5% / 10.2% / 15.7% vs. 4.2%

Most common adverse events, TRIUMPH-3 (9 mg / 12 mg vs. placebo):

Diarrhoea: 30.1% / 24.4% vs. 8.7%
Nausea: 21.7% / 22.4% vs. 5.8%
Constipation: 18.0% / 15.7% vs. 7.1%
Decreased appetite: 13.5% / 14.5% vs. 3.0%

The Dysesthesia Signal Has Not Gone Away

This is the one I asked about in June, and it’s still here.

TRIUMPH-2: 4.5% / 5.6% / 7.3% at 4, 9, 12 mg vs. 0.7% on placebo
TRIUMPH-3: 6.4% / 6.4% at 9, 12 mg vs. 1.3% on placebo

Roughly a tenfold excess over placebo in TRIUMPH-2, fivefold in TRIUMPH-3, and dose-dependent in the diabetes trial.

Lilly describes these events as generally mild to moderate with the majority resolving during treatment. I’ll take that at face value, while noting that the sentence sits immediately after the dysesthesia and UTI figures in the release, so it isn’t entirely clear how far back it reaches. I wouldn’t assume it was written to cover the GI events.

Dysesthesia is not a typical incretin class effect. Semaglutide doesn’t do this. Tirzepatide doesn’t do this. Orforglipron doesn’t do this. The most plausible explanation is the glucagon arm, which is the one thing retatrutide has that the others don’t — and if that’s right, it may be an unavoidable cost of the mechanism rather than something you can engineer away.

Before this molecule reaches a broad population I want time to onset, persistence, and reversibility properly characterised. A tingling sensation that resolves is a footnote. One that doesn’t is a labelling problem.

What Excites Me — And What I’m Watching

Here’s my honest take.

What excites me is that 22.6% in Class 2/3 obesity with established cardiovascular disease is a remarkable number for a population that has historically had one route to that magnitude of loss, and that route involved an operating theatre. And 20.8% in type 2 diabetes still comfortably exceeds anything else available in that group. Retatrutide got less spectacular between June and July. It also got a lot more real.

What I’m watching:

The cardiovascular outcomes trial. TRIUMPH-3 didn’t answer this and was never going to. Until the dedicated CVOT reads out, retatrutide has no cardioprotection claim — and given that semaglutide has SELECT, that gap will matter commercially as much as clinically.

Dysesthesia characterisation. Time to onset, persistence, reversibility. This is my single biggest open question on this molecule and has been since June.

The treatment-regimen estimand. Every number in this article is the efficacy estimand. The real-world figure will be lower. I want to see by how much before I quote any of this to a patient.

Whether 12 mg survives. Between the flat A1C curve above 9 mg, the 13.5% discontinuation rate, and dose-dependent dysesthesia, there’s a reasonable case that 9 mg is the better clinical dose for many patients. Watch what the label does.

The T2D gap itself. Three molecules, three mechanisms, same attenuation. Somebody needs to run the mechanistic study rather than us all noting the pattern article after article.

The Bigger Picture

I’ve now written about four molecules in this series, and a framing has emerged that I keep returning to.

Retatrutide is chasing the ceiling on efficacy. Foundayo is chasing access. Wegovy HD is squeezing another gear out of the molecule we trust most. CagriSema is combining proven pathways. Amycretin is engineering dual action into a single molecule.

TRIUMPH-2 and TRIUMPH-3 add something none of those headline stories do. They tell us what happens when the most potent molecule in the field meets the patients whose physiology fights back hardest — and the answer is that it does very well, just not as well, and at a tolerability cost that goes up in exactly the population least able to absorb it.

Efficacy in easy populations tells you what a drug can do.

Efficacy in hard populations tells you what it will do.

In June the clinical researcher in me was excited by the numbers, the pharmacist in me was cautiously optimistic about the safety profile, and the evidence nerd in me was watching every TRIUMPH readout.

Six weeks later: the researcher has revised downward and is completely fine with it. The pharmacist has one specific question and it’s about dysesthesia. And the evidence nerd is quietly relieved that the cardiovascular data came in underpowered rather than overinterpreted — because it means we still have to wait for the trial that was actually designed to answer the question.

Five positive Phase 3 trials. BLA targeted for Q1 2027, pending the CMC package.

This is metabolic medicine’s moment. And it’s just getting started. ✌️

Key References

Eli Lilly Press Release. Lilly’s triple agonist, retatrutide, successful in two additional Phase 3 obesity trials, delivering significant improvements in weight and A1C. 23 July 2026. Eli Lilly Investor Relations
Eli Lilly Press Release. Lilly’s triple agonist, retatrutide, delivered powerful weight loss in pivotal Phase 3 obesity trial (TRIUMPH-1). May 2026. Eli Lilly Investor Relations
Giblin K, Kaplan LM, Somers VK, et al. Retatrutide for the treatment of obesity, obstructive sleep apnea and knee osteoarthritis: Rationale and design of the TRIUMPH registrational clinical trials. Diabetes Obes Metab. 2026;28(1):83-93. PubMed
Jastreboff AM, Kaplan LM, Frías JP, et al. Triple-Hormone-Receptor Agonist Retatrutide for Obesity — A Phase 2 Trial. N Engl J Med. 2023;389(6):514-526. PubMed
Rosenstock J, Frías JP, Jastreboff AM, et al. Retatrutide, a GIP, GLP-1 and glucagon receptor agonist, for people with type 2 diabetes. The Lancet. 2023;402:529-544. The Lancet
ClinicalTrials.gov: TRIUMPH-2 (NCT05929079), TRIUMPH-3 (NCT05882045).
Leteyski T. Retatrutide: The Triple Agonist That Could Redefine Metabolic Medicine. The CRA Wizard, June 2026.
Leteyski T. Wegovy HD (Semaglutide 7.2 mg): The Workhorse Just Found Another Gear. The CRA Wizard, July 2026.
Leteyski T. Foundayo (Orforglipron): The First GLP-1 Pill You Can Take With Your Morning Coffee. The CRA Wizard, July 2026.

Disclaimer: Retatrutide is an investigational compound. It has not been approved by the FDA, EMA, or any other regulatory agency. The TRIUMPH-2 and TRIUMPH-3 results discussed here are topline data from a company press release and have not yet been peer reviewed or presented in full. This article is for educational and informational purposes only and does not constitute medical advice. Always consult a qualified healthcare professional before starting, stopping, or changing any medication.

The CRA Wizard

Retatrutide Part Two: What TRIUMPH-2 and…