There is no human study of how a vaccine performs in someone taking a GLP-1 receptor agonist. Not a trial, not a cohort, not a case series. The word vaccine does not appear anywhere in any of the six current labels in this class. What exists instead is one review arguing that the question has been neglected, a body of work about obesity that points the opposite way from that review, and a single result that reverses the obesity findings by changing what gets counted. That combination is the article.
The label census, and the control that makes it mean something
Read on September 15, 2026 against the current prescribing documents for Wegovy, Ozempic, Zepbound, Mounjaro, Saxenda and the semaglutide tablet, the string vaccin occurs zero times in all six.[1][2][3] There is no drug interaction entry, no warning, no counseling line and no postmarketing report.
An absence claim needs proof that the instrument was reading. In those same six documents the phrase “gastric emptying” appears between six and eight times each, and “oral medication” between three and nine times each.[1][2][3] These labels have a great deal to say about interfering with other drugs. Everything they say is about the oral route and the rate at which the stomach empties into the small intestine, which is the subject of the oral medication article. An injected or intranasal vaccine does not use that route at all, which is the likeliest reason the interaction sections are silent: the only mechanism these labels describe cannot reach it.
The literature census, with three controls
On the same date, PubMed indexes 32 records in which one of semaglutide, tirzepatide, liraglutide, dulaglutide or exenatide appears in the title or abstract alongside vaccine, vaccination or immunization. Adding any of antibody, titer, seroconversion or immunogenicity to that requirement reduces it to three, of which exactly one is about the subject; the other two are a circular-RNA engineering paper and a review of peptide-antibody fusions. Restricting the first query to influenza vaccine returns three records, none of them on the question — a cardiology conference summary and two serial trial bibliographies. Restricting it to COVID-19, SARS-CoV-2 or mRNA vaccines returns one.
Those numbers only mean something next to a query of the same shape that finds a literature. The identical five drug names paired with oral contraceptives or contraception return 26 records, and paired with levothyroxine, warfarin or digoxin return 35. Methotrexate, a drug with a genuine and well-characterized effect on vaccine response, paired with vaccine and antibody or immunogenicity returns 220. The query shape finds interaction literature when interaction literature exists. Here it does not exist.
The one paper on the subject is a hypothesis, and says so
A 2024 review in a cytokine biology journal is the entire dedicated literature. Its argument begins from a pharmacological asymmetry: native GLP-1 has a half-life of under five minutes and circulates in the picomolar range, while the agonists prescribed for weight reduction have half-lives of several days and are given at supraphysiological doses. GLP-1 receptor signaling reaches tissues beyond the pancreas and gut, including immune cells.[4]
From there it assembles evidence that these drugs may attenuate vaccine responses through direct effects on immune cells and through modulation of other tissues, and discusses how GLP-1 receptor signaling might create a tolerogenic environment that reduces vaccine immunogenicity. Its own framing is that, given how widely these drugs are used, understanding the effect on immune responses is important.[4] That is a case for doing the study. It is not the study, and the review contains no vaccine response measured in any person taking any of these drugs.
The premise underneath it does not hold up
The intuition that makes the hypothesis feel obvious is that obesity blunts vaccine responses, so a drug acting on that physiology might change them. The first half of that has been tested directly, and it came out the other way.
A 2025 systematic review and meta-analysis screened 865 studies and took 140 to full-text review, finding eleven that reported seroprotection or seroconversion for participants with and without obesity one month after licensed influenza vaccination, across samples of 44 to 1,132 participants. Obesity was associated with a marginally higher likelihood of response: for A/H1N1, seroconversion at a relative risk of 1.04 (95% CI, 1.01 to 1.08) and seroprotection at 1.08 (1.02 to 1.14); for A/H3N2, seroconversion at 1.11 (1.02 to 1.21). Influenza B strains showed no significant difference. The stated conclusion is that obesity does not impair influenza vaccine immunity at one month and may enhance antibody responses, possibly through a proinflammatory profile.[5]
A prospective cohort in children reached the same place from a different direction. Comparing normal-weight children with those whose obesity was metabolically healthy and those whose obesity was metabolically unhealthy, influenza antibody responses were statistically indistinguishable across all three groups, while one cytokine differed.[6]
Then the endpoint changes, and the answer inverts
A prospective observational study followed 1,022 vaccinated adults through the 2013–2014 and 2014–2015 influenza seasons, with serum drawn before vaccination and 26 to 35 days after. Among participants with obesity, 9.8% developed laboratory-confirmed influenza or influenza-like illness, against 5.1% of healthy-weight participants — a relative risk of 2.01 (95% CI, 1.12 to 3.60; P = 0.020). Seroconversion and seroprotection rates were not different between the two groups.[7]
The authors drew the conclusion that matters here: despite robust serological responses, vaccinated adults with obesity were twice as likely to get sick, which challenges the standard correlate of protection and suggests that using antibody titers to judge vaccine effectiveness in this population may be misleading.[7] The meta-analysis that reads as reassurance is pooling exactly the endpoint that paper says cannot be trusted in exactly this group. Neither result is wrong. They measure different things, and the more comforting one measures the thing with the weaker claim to matter.
Would losing weight change any of it
One experiment has asked, in mice. Diet-induced obese mice were switched from a high-fat diet to a control diet at different times around influenza vaccination. High-fat exposure limited T cell maturation into the memory compartment at the time of vaccination, which impaired recall of effector memory T cells on viral challenge. A diet switch after vaccination did not improve it. Metabolic dysfunction of T cells was reversed, and a functional recall response restored, only when weight loss occurred four weeks beforevaccination.[8]
That is a mouse, a diet and a timing question, not a person, a drug or a clinical outcome. It is worth reporting only because it is the closest thing in the literature to the question a reader on this drug is actually asking, and because its direction — weight loss before vaccination helping rather than hurting — is the opposite of the concern in the review above.
The word immunogenicity does appear, and it means something else
Anyone searching these drug names and that term will find a large, well-powered answer to a completely different question. An analysis of seven phase 3 trials examined 5,025 evaluable tirzepatide-treated patients for antibodies against the drug itself. Treatment-emergent antidrug antibodies developed in 51.1% of them, in similar proportions across doses, with maximum titers ranging from 1:20 to 1:81,920.[9]
The consequential fractions are much smaller. Neutralizing antibodies against tirzepatide activity at the GIP and GLP-1 receptors appeared in 1.9% and 2.1% of patients, and fewer than 1.0% had neutralizing antibodies cross-reactive against native GIP or GLP-1. Antibody status, titer and neutralizing status had no effect on the pharmacokinetics or efficacy of the drug. More antibody-positive patients experienced hypersensitivity or injection-site reactions, most of them nonserious and nonsevere.[9] That is the immune system responding to a peptide injection, which is expected, and it is not evidence about how it responds to anything else. Injection-site effects are covered in what the trials recorded.
What the absence does and does not license
No evidence of an interaction is not evidence that there is none, and the review above is a real argument that somebody should look. Equally, a mechanistic hypothesis with no human data behind it is not a reason to change anything, and deferring a vaccine is not a neutral act — it is a decision with its own risk, taken on the strength of something nobody has measured. This site does not make that call in either direction, and a page that resolved it would be inventing a finding.
Two adjacent situations do have documented answers and are worth keeping distinct. Drugs that genuinely suppress vaccine responses are a different category with a real literature, taken up in the immunosuppressants article. And being acutely unwell while on this class raises a separate and better-evidenced set of questions about dosing and fluid, covered in the sick-day article. Vaccination timing, eligibility and any interaction with a specific regimen belong to the clinician holding the chart.
What a compounded prescription changes
Nothing about the biology, and everything about where a finding would land if one were ever established. Compounded semaglutide and tirzepatide are not FDA-approved, and the FDA does not review them for safety, efficacy or quality before they are dispensed. The drug interaction section of an approved label is the place a confirmed vaccine interaction would be recorded and distributed; a compounded vial ships without one, so the mechanism that would carry such a finding to patients does not exist for most of this market. What that regulatory gap does and does not mean is set out in the article on what not FDA-approved means.
In the meantime the only honest statement is the census. Six labels, no mention. One review, no data. Three searches, one relevant paper. The underlying population evidence pointing two ways depending on whether you count antibodies or illnesses. Whether a seller can tell a patient that much, rather than improvising an answer, is the sort of thing recorded in the provider reviews.