GLP-1 receptor agonists are a class of peptide-based compounds that have expanded rapidly in clinical use over the past decade, primarily for managing type 2 diabetes and obesity. Most people who have heard about this drug class know that stomach-related side effects, like nausea and vomiting, are the most commonly discussed. What the scientific literature has covered far less thoroughly is how these peptides interact with the respiratory system.
A systematic review published in 2026 set out to map exactly that. Researchers analyzed 19 studies drawn from randomized controlled trials, large observational cohorts, pharmacovigilance databases, and individual case reports, covering millions of patient records in total. Their goal was not to draw firm causal conclusions but to detect signals worth investigating further, a process researchers call signal detection and evidence mapping.
The findings paint a nuanced picture. Most respiratory events observed in controlled trials were mild and occurred at rates similar to people not taking these peptides. At the same time, rare but serious events appeared in case reports and safety databases, raising questions that the authors say will require dedicated prospective studies to answer properly.
What the review covered
The research team followed PRISMA 2020 guidelines, a rigorous standard for how systematic reviews should be conducted and reported. They searched three major academic databases from their earliest records through January 2026, looking for any study in adults that reported respiratory adverse events alongside GLP-1 receptor agonist use.
The final pool of 19 studies included four randomized controlled trials enrolling about 1,884 participants, five retrospective cohort studies covering more than 1.1 million participants, three pharmacovigilance analyses drawing on nearly 499,000 spontaneous adverse-event reports, and seven case reports or series describing 13 individual patients. Because the study designs were so different from one another, the researchers chose not to combine results into a single pooled statistic. Instead they used a method called narrative synthesis, which summarizes patterns across studies without collapsing them into one number.
To judge the quality of each study, the team used different validated tools depending on the study type. Randomized trials were evaluated with the Cochrane Risk of Bias 2 tool, cohort studies with the Newcastle-Ottawa Scale, and case reports and pharmacovigilance work with Joanna Briggs Institute checklists. They also applied the GRADE framework to rate overall certainty of the evidence at the level of each outcome category.
Upper respiratory infections: the most common finding
Upper respiratory tract infections, the kind of mild infections that affect the nose and throat, were the most frequently reported respiratory event across the literature. They appeared in 8 of the 12 studies that reported any respiratory adverse events at all, a proportion of about 67 percent.
Critically, the randomized controlled trial data showed that rates of these infections in people using GLP-1 receptor agonists were comparable to rates in control groups. The review assigned this finding a GRADE certainty rating of MODERATE, meaning the researchers were reasonably confident this is a real pattern and not a statistical artifact. In practical research terms, this suggests the peptides themselves are unlikely to be driving an unusual increase in common upper airway illnesses.
Rare but serious events in case reports
While common respiratory events looked reassuringly similar to control groups, the review also documented a set of rarer events that appeared in individual case reports and safety databases. These included anaphylaxis with bronchospasm in four cases, one case of acute eosinophilic pneumonia, multiple cases of perioperative aspiration pneumonitis or pneumonia drawn from two cohort studies and one case series, two cases of acute respiratory distress syndrome with one fatal outcome requiring a form of life support called extracorporeal membrane oxygenation, and one case of spontaneous pneumomediastinum.
The researchers were careful to note that these events were reported in temporal association with GLP-1 receptor agonist use, meaning they happened around the same time as the person was taking the peptide. Temporal association is not the same as causation. The certainty of evidence for all serious respiratory events was rated very low, because the data came primarily from case reports and high-risk pharmacovigilance sources. The authors frame these as hypothesis-generating signals, observations that suggest an area deserves closer study rather than confirmed risks.
Pharmacovigilance signals for breathing difficulty
Pharmacovigilance databases collect spontaneous reports of adverse events submitted by healthcare providers and patients around the world. These databases are useful for detecting patterns but are inherently limited because reports are unverified, duplicated, and not linked to a denominator of how many people took the drug without incident.
One analysis cited in the review found a disproportionate reporting signal for shortness of breath and asthma-like events, particularly for an exendin-4-based compound, with a reporting odds ratio of 2.14 and a 95 percent confidence interval of 1.88 to 2.43. A reporting odds ratio above 1 means that reports of that event were proportionally more common in association with this drug than with other drugs in the database. The review team stressed that this type of signal cannot establish causality or tell researchers how often the event actually occurs in a given population.
Why some peptide subtypes may carry different hypersensitivity profiles
The review noted that GLP-1 receptor agonists are not all structurally identical. Some are based on a molecule called exendin-4, which comes from a non-human biological source. The review found that exendin-4-based agents appear to carry the highest reported rate of hypersensitivity reactions in the literature, and the authors attributed this to the structural difference from human GLP-1. Peptides that more closely resemble the human version of GLP-1 appear to carry a lower hypersensitivity signal in the available data, though the certainty of this comparison remains limited.
This structural distinction is relevant to researchers studying how immune systems respond to different peptide classes and why members of the same drug class can have meaningfully different safety profiles.
A potential signal in the other direction
Not all of the respiratory findings pointed toward harm. One large retrospective cohort study of 331,863 matched patients found that people using GLP-1 receptor agonists had a lower rate of new pneumonia diagnoses compared with people using a different class of diabetes medication called DPP-4 inhibitors. The hazard ratio was 0.60 with a 95 percent confidence interval of 0.58 to 0.62, suggesting roughly a 40 percent lower rate in the GLP-1 group.
The review team assigned this finding a GRADE certainty of LOW and urged caution in interpretation. The study was retrospective, meaning data was collected after the fact rather than designed prospectively. The comparison group was not a true placebo but an active drug with its own pharmacological effects, and the study could not fully account for all differences between the groups. The researchers note this as an interesting signal that requires prospective replication before any firm conclusions can be drawn.
What the review says is still missing
The central conclusion of the systematic review is that the respiratory safety profile of GLP-1 receptor agonist peptides is neither uniformly safe nor uniformly dangerous, but that current evidence is too limited to be certain about much beyond the most common outcomes. Common upper airway symptoms appear mild and comparable to controls. Serious events have been documented but remain rare, low-certainty, and insufficiently characterized.
The authors call for prospective studies designed from the start with respiratory outcomes as primary endpoints, using standardized definitions and adequate statistical power. They also highlight perioperative aspiration as a specific area of clinical concern, suggesting that pre-procedural fasting protocols may need to account for the pharmacokinetic properties of the specific agent involved. For researchers and clinicians tracking safety signals in this space, the review serves as a structured map of what is currently known and where the evidence gaps remain largest.
