Tirzepatide Long-Term Side Effects: Where the Evidence Stops

The longest published randomized experience with tirzepatide in obesity is 176 weeks — a little under three and a half years — and that figure comes from one subgroup of one trial.^[4] In type 2 diabetes there is a cardiovascular outcome trial with a median of 46.9 months of treatment.^[7] Past those horizons, nobody has data; they have an opinion. So the useful version of “long-term side effects” is two separate questions: what the approved label warns about, and how much of that warning is backed by a counted number.

How far the record reaches

Tirzepatide is not an old drug. FDA cleared the molecule first for type 2 diabetes, under the name Mounjaro, on May 13, 2022, and then for weight reduction, as Zepbound, on November 8, 2023.^[9][10] Everything anyone knows about taking it for years comes from a small number of trials that were designed to end.

The three-year report is the one people mean when they say the drug has long-term data. It found substantial weight reduction sustained to week 176 and markedly fewer diabetes diagnoses than placebo, 1.3 percent against 13.3 percent, and its safety conclusion was a single sentence: no new safety signals were identified.^[4] That sentence is worth reading precisely. It says nothing new appeared over three years. It does not say nothing new exists at five.

A trial built to answer the outcome question in obesity is running. SURMOUNT-MMO enrolled 15,374 participants, is listed as active and no longer recruiting, and has a primary completion date of October 2027.^[11] Until it reports, the morbidity and mortality question for people taking tirzepatide without diabetes has no trial answer at all.

A warning is not a rate

This is the distinction that decides how worried to be, and almost nothing written about GLP-1 drugs preserves it. A warning in an approved label is a disclosure FDA required, based on a plausible association, an animal finding, or reports gathered after approval. A rate is the result of counting events among people who were randomized. The two look identical on a page of bullet points and mean very different things.

The table below reads the Zepbound label, because that is the presentation approved for weight reduction. Mounjaro, the same molecule approved for type 2 diabetes, carries the same set of warnings, with the retinopathy one written for patients who already have a history of diabetic retinopathy.^[2] If you are taking tirzepatide for diabetes, that is the label your prescription is written against.

Read the pancreatitis row twice. Pancreatitis is the risk people fear most on these drugs, it carries a full warning section, and in the pooled weight-reduction trials the adjudicated rate on tirzepatide was the same as the rate on placebo. That does not make the warning wrong or removable. It means the warning is doing what a warning does — telling a clinician what to watch for — and it is not evidence that the drug caused pancreatitis at a measurable rate in those trials.

The boxed warning, read carefully

The strongest warning on the label is the one at the top, and it is entirely an animal finding.

In a two-year rat study, thyroid C-cell tumors rose with the size of the dose and with how long the animals were exposed, at exposures the label calls clinically relevant.^[1] Whether the same thing happens in people has, in the label's own phrasing, not been determined.^[1] Nothing stronger than that is claimed, and nothing weaker is permitted either — the warning sits in a black box at the top of the document.

Two things follow from it, and only two. The first is a hard contraindication covering two groups: anyone carrying type 2 multiple endocrine neoplasia syndrome, and anyone whose own medical record or family record includes medullary thyroid carcinoma.^[1] That is absolute, and it is a question to answer honestly with a prescriber before starting rather than a risk to weigh. The second is a short list of symptoms worth knowing: a mass in the neck, difficulty swallowing, shortness of breath, persistent hoarseness.^[1]

What does not follow is a screening plan. Routine calcitonin testing, and thyroid ultrasound too, are described in the label as being of uncertain value when it comes to catching MTC early — and as carrying their own risk of setting off procedures nobody needed.^[1] Anyone selling you a thyroid panel as GLP-1 safety monitoring is going past the label.

What the cardiovascular trial does and does not cover

SURPASS-CVOT is the longest controlled safety experience with tirzepatide that exists. It randomized 13,299 adults who had type 2 diabetes alongside established atherosclerotic disease, assigning them either tirzepatide or dulaglutide, a drug already shown to reduce cardiovascular events, and ran to a median treatment duration of 46.9 months.^[6][7] The primary composite of cardiovascular death, myocardial infarction or stroke occurred in 12.2 percent on tirzepatide and 13.1 percent on dulaglutide: noninferior, and not statistically superior.^[6]

For a long-term side effect question, the value of that trial is the four years of blinded follow-up rather than the efficacy verdict. Adverse events were broadly similar between the two groups, with more gastrointestinal events on tirzepatide — 42.5 percent against 35.9 percent in the post hoc analysis of the same population.^[7] Note the ceiling on what it transfers to, though. Everyone in it had diabetes and established heart disease, and they were compared against another incretin drug rather than against nothing. Zepbound's approved indications are weight reduction and obstructive sleep apnea; the label carries no cardiovascular risk-reduction claim.^[1]

Muscle: the part that is measured, and the part that is not

Lean-mass loss is the long-term worry that has traveled furthest ahead of its evidence, so here is the actual measurement. A substudy of SURMOUNT-1 scanned 160 of the 2,539 participants by DXA at baseline and week 72. On tirzepatide, body weight fell 21.3 percent, fat mass 33.9 percent and lean mass 10.9 percent; on placebo the corresponding figures were 5.3, 8.2 and 2.6 percent.^[8]

The finding that matters is the ratio rather than any single number. About 75 percent of what came off was fat and about 25 percent was lean tissue — and that split was about the same in the placebo group.^[8] In other words, the composition of the loss looked like weight loss, not like a specific effect of the drug on muscle. Losing a quarter of your lost weight as lean mass is still worth knowing about, particularly if you are older, and it is a reason to raise resistance training and protein intake with a clinician. It is not evidence that tirzepatide dissolves muscle.

Stopping: the best-documented long-term effect there is

SURMOUNT-4 was built to answer exactly this. Participants took tirzepatide open-label for 36 weeks, at which point 670 of them were randomized either to continue or to switch to placebo for another 52 weeks. Between week 36 and week 88, weight changed by -5.5 percent in the group that continued and by +14.0 percent in the group that stopped.^[5] Of those who continued, 89.5 percent still held at least 80 percent of their run-in weight loss at week 88, against 16.6 percent of those who stopped.^[5]

The three-year trial adds a second angle. After 176 weeks of treatment and then 17 weeks with no drug at all, 2.4 percent of the tirzepatide participants had type 2 diabetes against 13.7 percent of the placebo participants — so some of the metabolic benefit survived the washout, even as weight did not.^[4]

The practical reading is unglamorous: this behaves like ongoing therapy for a chronic condition, not like a course of something. That has consequences for supply, for cost, and for what happens when a program ends for reasons unconnected to anything you did — which is the situation we walk through in what to do when a compounded GLP-1 program ends. If cost is the reason you are considering stopping, say that to your prescriber before you do, and compare the dated figures on our price tracker rather than a seller's headline.

Where the record simply stops

• Beyond about three and a half years in obesity. 176 weeks is the end of the published randomized road, and that report covers the prediabetes subgroup.^[4]
• Cancer risk in humans. The boxed warning is a rat study. A three-and-a-half-year trial is not long enough to settle a question about tumors that take decades to appear, and no trial of this length could be.^[1][4]
• Outcomes in people without diabetes. The cardiovascular evidence comes from a diabetes population with established heart disease. SURMOUNT-MMO does not complete until October 2027.^[6][11]
• Function, as opposed to tissue mass. The body-composition data describe what was lost, not what changed about strength.^[8]
• Anything about compounded tirzepatide over years. There is no trial of any compounded formulation, at any duration, which is a separate problem we take apart in compounded tirzepatide side effects.

None of that is a reason for alarm and none of it is reassurance. It is the shape of a young drug's evidence base, and a seller who answers a long-term safety question with more confidence than the trials support is telling you something about the seller. Which pharmacy stands behind a given program is recorded in our pharmacy register, and the programs themselves sit on the compounded tirzepatide board.