The Heart and Sleep Apnea Approvals: What Each One Actually Required

Answer first: each indication is a separate result, not a property of the drug

Reading the Indications and Usage section of four labels — tirzepatide under two brand names, semaglutide under two — the non-weight approvals are these. Semaglutide injection is approved to reduce the risk of major adverse cardiovascular events in adults with established cardiovascular disease and either obesity or overweight. Tirzepatide's weight-brand label adds an approval to treat moderate to severe obstructive sleep apnea in adults with obesity. Its diabetes-brand label carries a cardiovascular approval of its own, for adults with type 2 diabetes at high risk. And the semaglutide diabetes label carries both a cardiovascular approval and a separate kidney approval, each in adults with type 2 diabetes.

Every one of those is a distinct finding from a distinct trial. None of them transfers to the others, and none of them is a statement about what the molecule does in general. The trial that produced a cardiovascular result in people with established heart disease says nothing about people without it, because it did not enrol them.

What follows is what each trial required of the people in it, what it measured, and — this is the part that rarely travels — what it did not establish. The labels are unusually candid on that last point. They just say it in the footnotes.

The semaglutide cardiovascular trial: a large win on the composite, and two things it did not prove

This was a placebo-controlled trial in 17,604 adults, run on top of standard care rather than instead of it. Everyone was 45 or older with a body mass index of 27 or greater and established cardiovascular disease, which the trial defined narrowly as a prior heart attack, a prior stroke, or peripheral arterial disease. People with a history of type 1 or type 2 diabetes were excluded. Median follow-up was 41.8 months.

The primary endpoint was the time to a first occurrence of a three-part composite: cardiovascular death, non-fatal heart attack, or non-fatal stroke. That composite fell, with a hazard ratio of 0.80 and a confidence interval from 0.72 to 0.90. In plain counts, 8 percent of the placebo group and 6.5 percent of the treated group had one of those three events.

Now the part the label states and the advertising does not. Cardiovascular death on its own was the first confirmatory secondary endpoint in the testing hierarchy, and the label says superiority was not confirmed: the hazard ratio was 0.85 with an interval running from 0.71 to 1.01. All-cause death came out at 0.81, and the label marks it as not statistically significant under the prespecified hierarchy. Heart failure hospitalization gets a footnote of its own: effect on heart failure has not been established.

So the honest version is that the composite moved, driven most visibly by heart attacks, and that death from cardiovascular causes did not clear the bar the trial set for itself. Both facts are printed in the same table.

One number in that trial moves in the unwanted direction and is easy to miss. At two years, heart rate in the treated group was 3.1 beats per minute higher than placebo. The same label carries a Warnings and Precautions subsection on heart rate increase.

The tirzepatide cardiovascular trial was run against another drug, and did not beat it

This is the single most useful and least quoted result on any of these labels. The cardiovascular outcomes trial for tirzepatide in type 2 diabetes was not placebo-controlled. It randomized 13,299 adults with inadequately controlled type 2 diabetes and established cardiovascular disease to tirzepatide or to dulaglutide, an older once-weekly GLP-1 receptor agonist that already carried a cardiovascular indication of its own.

Median follow-up was 210.1 weeks — over four years, the longest follow-up on any of these labels. The result: tirzepatide was non-inferior to dulaglutide for the three-part composite, with a hazard ratio of 0.92 and an interval from 0.83 to 1.01. The label then states it plainly. Superiority to dulaglutide was not established.

All-cause death in that trial came out at 0.84, with an interval from 0.75 to 0.94, which looks like a win until you read the footnote attached to it: not controlled for family-wise type I error rate. That is the label telling you the number was not part of the plan that protects against a false positive, so it is a hypothesis rather than a conclusion.

The entry criteria were narrow in their own way. Everyone was 40 or older with type 2 diabetes, an HbA1c between 7 and 10.5 percent, and a body mass index of 25 or greater. Mean diabetes duration was 15 years. Almost half were already on insulin, 81 percent on metformin, 86 percent on statins and 83 percent on antiplatelet therapy. Whatever the trial found, it found on top of all of that.

The semaglutide diabetes cardiovascular trial was designed to show non-inferiority

This distinction matters and almost never survives into a summary. The cardiovascular outcomes trial on the semaglutide diabetes label enrolled 3,297 adults with inadequately controlled type 2 diabetes and atherosclerotic cardiovascular disease, and its primary analysis was set up to test non-inferiority to placebo using a risk margin of 1.3. In other words, the trial's stated job was to rule out a meaningful increase in cardiac risk, not to demonstrate a benefit.

It found a benefit anyway. The hazard ratio for the composite was 0.74, from 0.58 to 0.95, over a median observation of 2.1 years. Non-fatal stroke came in at 0.61 and non-fatal heart attack at 0.74. Cardiovascular death alone was 0.98, essentially unchanged.

The population is worth stating because it is broader than the weight-brand trial in one direction and narrower in another. Eligible patients were 50 or older with established cardiovascular, cerebrovascular or peripheral artery disease, chronic kidney disease, or class II or III heart failure — or they were 60 or older with specified risk factors and no established disease. That last group was 17 percent of the trial. Mean age was 65, mean diabetes duration 13.9 years.

The kidney indication is its own trial, with its own exclusions and one honest null

The semaglutide diabetes label carries a third approval: to reduce the risk of sustained decline in kidney function, end-stage kidney disease and cardiovascular death in adults with type 2 diabetes and chronic kidney disease. The supporting trial randomized 3,533 patients and followed them for a median of 41 months.

Entry required a measured kidney function in a defined band together with a urine albumin-to-creatinine ratio in a defined range, and required that patients already be on the maximum tolerated labeled dose of a renin-angiotensin-aldosterone blocking agent unless that was contraindicated or not tolerated. Ninety-five percent were on one at baseline. Congenital and hereditary kidney disease, autoimmune kidney disease and congenital urinary tract malformations were excluded outright.

The composite endpoint fell, at 0.76 from 0.66 to 0.88. All-cause death fell, at 0.80. And then this, in the label's own words: the treatment benefit on the primary composite endpoint was not evident in patients taking SGLT2 inhibitors at baseline, though there were few events in those patients. Sixteen percent of the trial was on one. That is a real and stated limit on who the result describes, and it exists nowhere in any advertisement.

The same label says the mechanism of kidney-related risk reduction has not been established.

The sleep apnea approval rests on two 52-week trials totaling 469 people

This is by far the smallest evidence base of the group, and it is the one most likely to be described loosely. Tirzepatide's obstructive sleep apnea indication comes from a master protocol containing two randomized, double-blind, placebo-controlled trials of 52 weeks, enrolling 234 and 235 adults respectively. Patients with type 2 diabetes were excluded from both, and everyone received diet and physical activity counseling throughout.

Entry required moderate to severe disease, defined as an apnea-hypopnea index of at least 15 events per hour, and obesity, defined as a body mass index of 30 or greater. The two trials differ in one deliberate way: the first enrolled people unable or unwilling to use positive airway pressure therapy, and the second enrolled people already on it.

The primary endpoint was the change in apnea-hypopnea index at week 52. In the first trial the index fell by 25.3 events per hour on treatment against 5.3 on placebo. In the second, 29.3 against 5.5. Expressed as proportions, 61.2 percent and 72.4 percent of treated patients achieved at least a halving of their index, against 19 and 23.3 percent on placebo. Remission or mild non-symptomatic disease was reached by 42.2 and 50.2 percent of treated patients against 15.9 and 14.3 percent.

Baseline severity in these trials was high — mean apnea-hypopnea index of 51.5 and 49.5, with roughly two-thirds of patients in the severe range. Mean baseline weight was between 112 and 117 kilograms, and mean body mass index around 39. These are not people at the edge of the criteria.

The sleep apnea trials contain a caveat that patients need and nobody repeats

In the second of the two trials, everyone was on positive airway pressure therapy and was instructed to suspend it for seven days before the endpoint was assessed. That is how a sleep study measures the drug rather than the machine. It also means the reported improvement describes what happened during a supervised week without the machine, inside a trial.

The label then adds a sentence that is worth reading twice. The clinical studies for obstructive sleep apnea did not evaluate the timing or appropriateness of discontinuing positive airway pressure therapy in patients who were previously compliant with it.

So the trial that produced the sleep apnea approval in people already using a machine did not study whether, when, or for whom the machine can be put away. Anyone reading a result about sleep apnea and hearing an implied answer to that question is hearing something the trial did not ask. It is a conversation for the clinician managing the sleep apnea.

What to carry into an appointment

Ask which trial the claim you have heard came from, and what its entry criteria were. Every one of these approvals names a population in a single sentence in Section 1 of the label, and every one of those sentences excludes far more people than it includes. Established cardiovascular disease means a prior heart attack, stroke or peripheral arterial disease in the trial that produced that indication. Moderate to severe sleep apnea means a measured index of at least 15 events an hour together with obesity.

Ask what the trial did not establish, because the labels answer it. Effect on heart failure has not been established. Superiority over the comparator drug was not established. The kidney benefit was not evident in patients already on an SGLT2 inhibitor. The mechanism of cardiovascular risk reduction has not been established.

And treat every one of these as separate from the weight result. The cardiovascular trials were run mostly in men with existing heart disease. The sleep apnea trials were run in 469 people with severe disease and a mean body mass index near 39. The weight percentages came from different trials with different entry criteria. A single drug can be the subject of all of them and none of the findings transfers between them.