GLP-1 receptor agonists are a class of peptide-based drugs that mimic a naturally occurring gut hormone. They have become widely prescribed for type 2 diabetes and, more recently, for weight management. Despite that growing use, most of what researchers know about their effects comes from trials designed to measure one or two specific outcomes at a time.
A recent analysis published in Nature Communications took a different approach. Instead of asking a narrow question, the researchers asked a broad one: across nearly one thousand distinct health conditions, how do outcomes differ between people prescribed a GLP-1 receptor agonist and people prescribed one of two other common diabetes drug classes? The answer involved some findings that researchers describe as surprising, particularly around infections and dental health.
The study drew on electronic health records from the All of Us Research Program, a large United States database that collects real-world clinical data from diverse populations. The final dataset covered 17,267 adults with type 2 diabetes, and the analysis spanned prescriptions written between January 2018 and October 2023.
Study design and comparisons
The researchers used a retrospective cohort design, meaning they looked backward through existing records rather than enrolling new participants in a controlled experiment. To make fair comparisons, they used propensity score matching, a statistical technique that pairs people who received different drugs but who looked similar in other important ways, such as age, other health conditions, and baseline lab values.
Two main comparisons were made. First, GLP-1 receptor agonists were compared against SGLT2 inhibitors, another class of diabetes drugs that work by causing the kidneys to excrete excess glucose. Second, GLP-1 receptor agonists were compared against DPP-4 inhibitors, a class that works by slowing the breakdown of natural GLP-1 in the body. Both are considered second-line treatments when a first medication is not enough on its own.
The analysis evaluated up to 974 different phenotypes, which is the technical word for measurable health conditions or traits. Researchers used two modeling approaches. One followed intention-to-treat logic, meaning participants were tracked from the time of prescription regardless of whether they kept taking the drug. The other followed per-protocol logic, meaning only time spent actually on the drug was counted. Both approaches pointed toward similar conclusions.
Genitourinary infections and SGLT2 inhibitors
One of the clearest findings involved genitourinary infections in women. When semaglutide, a specific GLP-1 receptor agonist included in the analysis, was compared to SGLT2 inhibitors, women taking semaglutide showed a notably lower rate of being diagnosed with vaginal yeast infections. The per-protocol hazard ratio was 0.31, with a 95 percent confidence interval of 0.17 to 0.55. In plain terms, that means the rate of that diagnosis was roughly 69 percent lower in the semaglutide group during the time they were on the drug.
This finding is not entirely surprising when you consider how SGLT2 inhibitors work. By pushing glucose into the urine, they can create conditions in the genital tract that favor the growth of fungi like Candida. GLP-1 receptor agonists do not share that mechanism, so the difference in infection rates is biologically plausible. Still, the size of the association in a real-world dataset of this scale adds meaningful weight to what had previously been noted mainly in smaller studies.
The researchers note this as one of the phenome-wide significant findings, meaning it cleared the statistical bar for being unlikely to be a chance result across hundreds of simultaneous comparisons.
Dental health and DPP-4 inhibitors
When GLP-1 receptor agonists were compared to DPP-4 inhibitors, a different pattern emerged. People in the GLP-1 group were less likely to receive diagnoses related to diseases of the hard tissues of teeth, a category that includes conditions like cavities and enamel breakdown. The hazard ratio was 0.45, with a 95 percent confidence interval of 0.33 to 0.61, suggesting roughly a 55 percent lower rate of those diagnoses in the GLP-1 group.
The biological path connecting GLP-1 receptor agonists to dental outcomes is less obvious than the genitourinary finding, and the researchers stop short of claiming a direct causal mechanism. One possibility the literature points toward is that GLP-1 receptor agonists tend to reduce intake of sugary foods and beverages, which is a major driver of dental decay. Another possibility is that improvements in blood sugar control itself may support better oral health. The study design cannot distinguish between these explanations.
The dental finding stood out partly because it was not an outcome that most people would have expected from a diabetes drug comparison. That is, in part, the value of a phenome-wide approach: it surfaces associations that would never have been tested in a more narrowly focused trial.
Cardiovascular and metabolic signals
The study also identified associations in cardiovascular and metabolic phenotypes, though the researchers describe some of these as suggestive rather than phenome-wide significant. That distinction matters. In a study evaluating nearly one thousand outcomes simultaneously, the statistical threshold for significance is set intentionally high to reduce false positives. Findings that fall just below that threshold are still worth noting, but should be interpreted more cautiously.
The cardiovascular signals are consistent with a growing body of literature suggesting that GLP-1 receptor agonists may influence heart and vessel health through multiple pathways, including effects on blood pressure, inflammation, and fat distribution. Earlier clinical trials focused on specific cardiovascular endpoints have shown benefits in high-risk populations, and this phenome-wide work adds a broader observational layer to that picture.
The researchers also examined time-to-event data, looking not just at whether a diagnosis occurred but at how long it took to appear. They found what they describe as modest delays for several key diagnoses in the GLP-1 group. This kind of analysis adds nuance, since a drug that does not prevent a condition entirely but delays its onset by months or years may still be clinically meaningful.
Limitations and what this means for research
Observational studies like this one cannot prove that a drug caused an outcome. Even with propensity score matching, there may be unmeasured differences between the groups that explain some findings. For example, people prescribed one drug class versus another may differ in ways that are not captured in electronic health records, such as dietary habits, socioeconomic factors, or how closely they follow up with their doctors.
The study also relies on diagnosis codes, which means it can only count what clinicians recorded. Some conditions may be underdiagnosed or coded differently across health systems. The researchers acknowledge these limitations directly.
Despite those caveats, the scale of the dataset and the breadth of outcomes examined make this one of the more comprehensive real-world comparisons of second-line diabetes therapies published to date. The authors conclude that GLP-1 receptor agonists, and semaglutide in particular, show a distinct downstream profile compared to SGLT2 inhibitors and DPP-4 inhibitors, with implications for how clinicians think about matching drugs to individual patients.
For researchers studying GLP-1 peptide biology more broadly, findings like these reinforce the idea that this signaling pathway touches multiple organ systems well beyond the pancreas and the gut. Early data points at dental tissue, the urogenital tract, and the cardiovascular system as areas where the downstream biology of GLP-1 receptor activation may be particularly active.
The peptide biology behind the findings
GLP-1, or glucagon-like peptide-1, is a short protein fragment produced naturally in the intestines after eating. It signals the pancreas to release insulin, slows the movement of food through the stomach, and acts on brain regions involved in appetite and satiety. The receptor agonist drugs studied here are synthetic peptides designed to bind and activate the same receptor that natural GLP-1 uses.
Semaglutide, which received its own subgroup analysis in this study, is a modified version of the natural GLP-1 peptide. Chemical modifications give it a longer half-life in the body compared to the natural hormone, which breaks down within minutes. That extended activity is part of what makes it effective when dosed weekly rather than daily.
The breadth of the associations found in this phenome-wide study reflects the fact that GLP-1 receptors are not found only in the pancreas. They are present in the heart, kidneys, brain, and other tissues, which helps explain why a drug targeting this receptor can produce effects across such a wide range of organ systems. Understanding those downstream effects, both beneficial and potentially adverse, is an active area of investigation in the research community.
