Why GLP-1 Drugs Don’t Work the Same for Everyone: Two Studies Reveal the Genetic Divide

A Nature GWAS of 27,885 people and a Stanford enzyme study converge on the same conclusion: your DNA shapes how you respond to semaglutide, tirzepatide, and related drugs.

Two studies published within days of each other have reshaped the conversation around GLP-1 receptor agonists—and the implications extend well beyond the lab. A genome-wide association study of 27,885 people published in Nature on April 8 identified specific genetic variants that predict both how much weight someone will lose on GLP-1 drugs and how likely they are to experience nausea. Separately, Stanford researchers reported in Genome Medicine that roughly one in 10 people carry enzyme variants that make GLP-1 hormones less biologically active—even when blood levels of those hormones are elevated. Together, the two findings offer the strongest evidence to date that the wildly variable responses people report on semaglutide, tirzepatide, and related drugs are not random. They are, at least in part, written into DNA.

What the 23andMe/Nature Study Found

The larger of the two studies, conducted by the 23andMe Research Institute and published in Nature, analyzed genetic data from 27,885 people who had taken a GLP-1 receptor agonist. It is the largest pharmacogenomics study of GLP-1 drugs ever conducted.

Three findings stand out.

A GLP1R variant tied to weight loss. The study identified a missense variant in the GLP1R gene—the gene encoding the actual receptor these drugs target—that is associated with an additional 0.76 kg of weight loss per copy of the effect allele. In practical terms: two people on the same drug, same dose, same duration, will lose meaningfully different amounts of weight based in part on which version of this gene they carry.

A GIPR variant tied to tirzepatide-specific nausea. A variant in the GIPR gene was associated with increased nausea and vomiting—but only in people taking tirzepatide (Mounjaro/Zepbound), not those on semaglutide alone. This is the first genetic evidence that tirzepatide’s GIP receptor component drives a side effect profile distinct from pure GLP-1 agonists. For the millions of patients choosing between semaglutide and tirzepatide, this is directly actionable information—or it will be, once genetic testing catches up.

The range is enormous. Among study participants, weight loss ranged from 6% to 20% of starting body weight, and the probability of nausea or vomiting ranged from 5% to 78%. These ranges are driven by a combination of genetics and other factors, but the genetic contribution is now quantifiable for the first time.

What the Stanford PAM Study Found

The Stanford study, published March 29 in Genome Medicine, took a different approach. Rather than looking at the drug target, the researchers focused on an upstream enzyme called PAM (peptidyl-glycine alpha-amidating monooxygenase) that activates GLP-1 and many other hormones in the body.

The research was a decade in the making—an international collaboration involving experiments in humans and mice plus a meta-analysis of three clinical trials totaling 1,119 participants.

The key finding is counterintuitive. People carrying two specific PAM variants—roughly 10% of the general population—had higher circulating GLP-1 levels than non-carriers. The natural assumption would be that more GLP-1 means better response to GLP-1 drugs. The opposite was true. In the trial meta-analysis, PAM variant carriers were less responsive to GLP-1 medications, achieving smaller reductions in HbA1c (the standard measure of long-term blood sugar control).

The researchers’ explanation: the PAM enzyme is required to activate GLP-1 into its biologically functional form. When PAM is impaired, the body produces more GLP-1 to compensate—but that GLP-1 is less effective. It is the difference between having more currency and having currency that’s worth less.

The exact mechanism driving this resistance remains unresolved, and the researchers note it is likely multifactorial. But the clinical implication is clear: a subset of patients who appear to have adequate GLP-1 levels may still respond poorly to GLP-1 drugs, and genetic testing could help identify them before months of ineffective treatment.

Why This Matters

The GLP-1 story for the past three years has been dominated by a simple narrative: these drugs produce dramatic weight loss and blood sugar improvements, and the main question is access and cost. That narrative was always incomplete—clinicians have long observed wide variation in patient response—but until now, the “why” was largely speculative.

These two studies move the conversation from “some people respond better than others” to “here are specific genetic variants, in specific genes, with quantifiable effects on specific outcomes.” That is a fundamentally different evidentiary footing.

Three things this means for the compounds Peptidings covers:

Tirzepatide’s side effect profile has a genetic explanation. The GIPR nausea finding isn’t just academic. It suggests that patients experiencing severe GI side effects on tirzepatide—but not on semaglutide—may have a genetic predisposition tied to the GIP receptor component that makes tirzepatide unique. For readers weighing tirzepatide versus semaglutide, this is the kind of mechanistic evidence that informs the decision.

“GLP-1 drugs don’t work for me” may have a biological basis. The PAM finding means that the 10% of patients who respond poorly aren’t necessarily non-compliant or unusual—they may carry enzyme variants that reduce GLP-1 bioactivity. This has implications for retatrutide and other next-generation GLP-1 agonists as well: if the resistance is at the hormone-activation level, drugs targeting the same receptor may face the same limitation.

Precision medicine is coming for obesity pharmacology. These studies lay the foundation for a world in which a genetic test—before prescribing—helps clinicians select the right GLP-1 drug for the right patient. That world is not here yet. But the data now exists to start building it.

What These Studies Don’t Tell Us

The 23andMe study is observational and relies on self-reported data from a consented research population. The weight loss and nausea figures are associations—they do not establish that genetic testing will improve clinical outcomes. A randomized trial assigning treatment based on genotype would be needed to prove that.

The Stanford PAM study’s trial meta-analysis included 1,119 participants across three trials—a meaningful sample, but far smaller than what regulators typically require for pharmacogenomic labeling recommendations. The mechanism driving GLP-1 resistance in PAM variant carriers is still a hypothesis, not a confirmed pathway.

Neither study addresses long-term outcomes. Whether genetic predictors of initial response translate to differences in sustained weight maintenance, cardiovascular benefit, or other hard endpoints remains unknown.

These are important limitations. They do not diminish the significance of the findings—they frame it accurately.