An Oral GLP-1 Pill Just Passed Its First Cardiovascular Test. Read the Death Numbers Carefully.

For years the deal was simple. If you wanted the cardiovascular benefit of a GLP-1 receptor agonist, you accepted a subcutaneous injection. ACHIEVE-4, published in The Lancet on 30 September 2026, is the first cardiovascular outcomes trial to put a once-daily oral pill through that test, and it changes what we can honestly claim about non-peptide incretins.

The compound is orforglipron, a small-molecule GLP-1 receptor agonist discovered by Chugai and licensed by Eli Lilly in 2018. It is not a peptide. That detail is the entire point of the study, and it is where most of the argument about this dataset belongs.

What ACHIEVE-4 actually did

ACHIEVE-4 (NCT05803421) was an event-driven, randomised, open-label, active-comparator phase 3 trial run at 317 sites across 16 countries and territories. Between 1 May 2023 and 5 September 2024, 2,749 adults with type 2 diabetes and a BMI of 25 or higher were randomised 1:1. Every participant had established cardiovascular disease or chronic kidney disease, and 85.9% had established cardiovascular disease specifically. Median follow-up was 104 weeks.

The comparator was titrated insulin glargine, not placebo, and that choice matters more than it first appears. Insulin glargine is a glucose-lowering drug with a known safety profile and a well-understood propensity for hypoglycaemia and weight gain. Beating it on safety is a much easier bar than beating placebo on safety.

Non-inferiority was declared if the upper bound of the two-sided 95% confidence interval for the hazard ratio sat below 1.8. That is a deliberately generous margin, adopted because it allows for the excess risk of a glucose-lowering agent in a high-risk population.

The number Lilly put on the front of the press release

Beyond the primary endpoint, pre-planned analyses showed a 53% lower risk of cardiovascular death (HR 0.47, 95% CI 0.25 to 0.90, p=0.019) and a 57% lower risk of all-cause death (HR 0.43, 95% CI 0.25 to 0.75, p=0.002). Hospitalisation for heart failure was 0.7% versus 1.5% (HR 0.44, 95% CI 0.20 to 0.97). In absolute terms, 62 deaths occurred during the study: 19 on orforglipron and 43 on glargine.

Those analyses were not controlled for family-wise type 1 error. Lilly states this plainly in its own release footnotes. With no multiplicity control, a nominally significant result in a subgroup or secondary endpoint is a hypothesis, not a finding. Treating the 57% all-cause mortality reduction as an established effect would be reading the sponsor's own caveat backwards.

There is a second, less comfortable fact in the same dataset. All-cause mortality is exactly the kind of endpoint where differential dropout matters, and differential dropout was substantial. Of participants assigned to orforglipron, 948 of 1,371 (69.1%) completed the study on treatment, against 1,096 of 1,378 (79.5%) on glargine. That 10-point gap sits underneath a mortality claim.

The tolerability ledger does not flatter the pill

This is where the non-inferiority framing starts to strain. Gastrointestinal adverse events occurred in 62.1% of orforglipron participants versus 14.2% on glargine. Nausea hit 31.9% versus 1.4%, vomiting 21.0% versus 1.6%, diarrhoea 20.4% versus 4.9%. Treatment-emergent adverse events of any kind: 82.3% versus 68.0%.

Discontinuation due to adverse events ran at 15.4% on orforglipron against 5.4% on glargine. The mitigating detail, from the paper, is that most nausea, vomiting and diarrhoea were mild to moderate, clustered early during dose escalation, and tended to diminish over time. That is the standard incretin profile and it is genuinely reassuring for the individual patient who tolerates month three. It does not change what happened in this trial's attrition accounting.

Set against that, hypoglycaemia ran the other way and strongly. Clinically significant or severe hypoglycaemia, defined as glucose below 3 mmol/L, occurred in 6.8% on orforglipron versus 19.2% on glargine. Severe level 3 events: 0.1% versus 0.9%. If you are one of the 37.6% of participants with chronic kidney disease, or you take a sulfonylurea alongside, that is a meaningful advantage.

Why this matters beyond one pill

The reason this trial was run at all is that cardiovascular benefit had been demonstrated for peptide GLP-1 receptor agonists, and it was genuinely unknown whether a non-peptide molecule acting on the same receptor would reproduce it. Mechanistically that is not obvious. Small molecules and peptides reach the receptor differently, dock differently, and clear differently. Receptors do not care about the drug's chemical family, but the drug's shape and pharmacokinetics can change how much, and for how long, the receptor is engaged.

ACHIEVE-4's answer, within its limits, is that cardiovascular safety is a property of GLP-1 receptor agonism itself rather than of the peptide scaffold. That is a bigger claim than it sounds, because it suggests the oral non-peptide class will keep the safety record of the injectable one. The word doing the work there is suggests.

Weight and glycaemic control moved in the expected direction. At 52 weeks from a mean baseline HbA1c of 8.22%, orforglipron lowered HbA1c by 1.6 percentage points versus 1.0 on glargine, a treatment difference of -0.66 (95% CI -0.74 to -0.58, p<0.001). Body weight from a mean baseline of 90.9 kg fell 8.8% on orforglipron and rose 1.7% on glargine, a difference of -10.42 percentage points (95% CI -10.92 to -9.93, p<0.001). At 104 weeks, 54.6% of orforglipron participants reached HbA1c of 6.5% or lower versus 23.8% on glargine, and 35.9% achieved at least 10% weight reduction versus 3.4%.

Lilly's release also reports cardiometabolic risk-factor movement at week 52: waist circumference down 7.0 cm, systolic blood pressure down 4.4 mmHg, triglycerides down 18.9%, non-HDL cholesterol down 6.6%, and high-sensitivity CRP down 38.9%. Kidney signals ran in the same direction, with a 2.8 mL/min/1.73 m2 smaller reduction in eGFR and a 27.1% greater reduction in albuminuria versus glargine. None of these were controlled for multiplicity either.

What we still do not know

The trial was open-label, which matters for a subjective endpoint family and for how side effects get reported. It enrolled people with type 2 diabetes, not the obesity-only population where orforglipron is already approved. Median follow-up of two years is short for cardiovascular outcomes; the SUSTAIN and SELECT trials that established benefit for injectable GLP-1s ran considerably longer. And the ongoing ATTAIN-OUTCOMES study, which targets adults with obesity or overweight with or without diabetes, is where the obesity-specific outcomes question gets answered.

The comparator gap deserves emphasis. Nobody should read ACHIEVE-4 as evidence that oral orforglipron beats injectable semaglutide or tirzepatide for cardiovascular protection. It was not tested against them. Insulin glargine is a real and useful baseline, and beating it on hypoglycaemia while tying it on events is a solid result. It is simply not the same claim.

Evidence Grade: B — randomised, active-controlled, event-driven phase 3 in 2,749 high-risk participants, published in The Lancet and presented at EASD 2026, with prespecified non-inferiority met. Held back from A by open-label design, a 2-year median follow-up, no superiority on the key secondary endpoint, roughly 10 points of differential on-treatment completion, and mortality analyses explicitly not corrected for multiplicity. Robust for safety and non-inferiority. Preliminary for any mortality claim.

The Oria take

There is a temptation in this dataset to reach for the mortality number, because 57% is the kind of figure that travels well on social media. The honest read is narrower. ACHIEVE-4 established that a daily oral GLP-1 pill does not carry excess cardiovascular risk against basal insulin in a high-risk diabetes population, which is a genuinely useful and underappreciated result. It also produced a lower-discontinuation, better-weight, far-more-hypoglycaemia profile than insulin in that setting. The mortality signal is interesting and worth watching in ATTAIN-OUTCOMES, where it will be tested with a proper control group and no open-label bias.

For anyone weighing an oral incretin against a daily basal insulin, the practical takeaways are already usable: expect meaningful gastrointestinal effects concentrated in the first months, expect to be one of the roughly one in seven who stops because of them, and expect markedly fewer low-blood-sugar events unless you are also on a sulfonylurea. That trade is probably worth making for a lot of people. It just is not the trade headline numbers imply.

Sources

Klein KR, Wysham C, Tuttle KR, et al. Cardiovascular safety of orforglipron versus insulin glargine in adults with type 2 diabetes at increased cardiovascular risk (ACHIEVE-4): a phase 3, event-driven, randomised, open-label, non-inferiority, active comparator trial. The Lancet, published online 30 September 2026. DOI 10.1016/S0140-6736(26)01865-9. PMID 42815506.

Eli Lilly and Company. Lilly's oral GLP-1, Foundayo (orforglipron), demonstrated cardiovascular safety alongside sustained A1C reduction and weight loss in its largest and longest type 2 diabetes study. News release, 30 September 2026.

ClinicalTrials.gov. ACHIEVE-4 (NCT05803421). Phase 3 event-driven cardiovascular outcomes trial of orforglipron versus insulin glargine.

Medical disclaimer: This article is for general education only and is not medical advice, diagnosis, or treatment. It summarises published clinical research and cannot account for your individual medical history, medications, or conditions. Never start, stop, or change any prescription medication without a licensed clinician who can evaluate you directly. If you have side effects from a medication, contact your provider promptly.