GLP-1 Resistance: Why Ozempic Doesn't Work for 1 in 10 People

For most patients, semaglutide and tirzepatide deliver weight loss that no prior obesity drug came close to matching. For a small but consistent minority, they barely move the needle. The clinical assumption used to be adherence: missed doses, late titration, dietary compensation. Two papers published in April 2026 — one in Genome Medicine, one in Nature — make a different argument. A meaningful fraction of non-responders carry inherited genetic variants that blunt the body's response to GLP-1 receptor agonists, a phenomenon researchers have begun calling GLP-1 resistance.

Evidence: "Approximately 10% of people carry genetic variants in the PAM enzyme that reduce the blood-sugar-lowering efficacy of GLP-1 receptor agonists, leading to GLP-1 resistance characterized by elevated but less biologically effective GLP-1 levels." — Cardona A, et al. Genome Medicine. 2026. DOI: 10.1186/s13073-026-01630-0

This article walks through what the new genetics tell us, what they do not yet prove, and how clinicians and patients should interpret a sub-par response to GLP-1 therapy in 2026.

What "GLP-1 Resistance" Actually Means

The term is borrowed deliberately from insulin resistance. In insulin resistance, circulating insulin levels are normal or elevated, but the downstream tissues respond poorly — so more hormone is needed to produce the same biological effect. GLP-1 resistance follows the same logic. People carrying certain variants in the PAM gene have higher circulating GLP-1 than non-carriers, but their bodies extract less biological work from each molecule. The hormone is being made; it just doesn't land as hard.

PAM stands for peptidyl-glycine alpha-amidating monooxygenase. It is the only enzyme in the human body capable of a chemical step called amidation, which caps the C-terminus of dozens of bioactive peptides — GLP-1 among them — and dramatically increases their potency and half-life. When PAM works at full strength, GLP-1 reaches receptors in a fully activated state. When PAM is genetically handicapped, more peptide circulates in a less potent form.

Evidence: "Despite people with the PAM variant having higher circulating levels of GLP-1, there was no evidence of higher biological activity. More GLP-1 was needed to produce the same biological effect, meaning carriers were resistant to GLP-1." — Cardona A, et al. Genome Medicine. 2026. DOI: 10.1186/s13073-026-01630-0

Two specific missense variants are responsible for most of the signal: p.S539W and p.D563G. Together they are carried by roughly 10% of the general population — common enough that, on average, every primary care physician treating obesity already has dozens of these patients on their panel.

The Stanford Genome Medicine Study

The Stanford-led international team published its findings online April 10, 2026 in Genome Medicine. The work was a decade-long effort that combined human genetics, mouse experiments, and a meta-analysis of three randomized GLP-1 receptor agonist trials totaling 1,119 participants.

Glycemic response by variant

The clinically meaningful number is the proportion of participants who reached the recommended HbA1c target after six months of GLP-1 therapy:

Group	Reached HbA1c target
Non-carriers (no PAM variant)	~25%
Carriers of p.S539W	~11.5%
Carriers of p.D563G	~18.5%

A non-carrier was therefore roughly twice as likely to reach glycemic target as a p.S539W carrier on the same drug at the same dose. The effect was consistent across the three trials and survived adjustment for baseline BMI, HbA1c, and age.

The weight-loss question is still open

The Stanford paper is unambiguous about glycemia and cautious about weight. Only two of the three meta-analyzed trials provided weight data. Within those, no significant difference in weight loss was detected between PAM variant carriers and non-carriers — but the analysis was underpowered, with weight as a secondary endpoint and follow-up shorter than the modern obesity-trial standard of 68–72 weeks.

The honest interpretation is that PAM variants are an established cause of glycemic non-response in type 2 diabetes and a plausible but unproven cause of weight-loss non-response in obesity. The biology makes the weight signal likely; the data don't yet confirm it.

The 23andMe / Nature Weight-Loss GWAS

Published April 8, 2026 in Nature, the 23andMe Research Institute paper takes the complementary angle: a genome-wide association study of 27,885 individuals who had taken a GLP-1 medication, looking specifically for variants that predict weight-loss magnitude and side-effect risk.

Evidence: "A missense variant in GLP1R is associated significantly with increased efficacy of GLP-1 medications, with an additional −0.76 kg of weight loss expected per copy of the effect allele." — Park S, et al. Nature. 2026. DOI: 10.1038/s41586-026-10330-z

Three findings stand out:

A GLP1R missense variant predicts greater weight loss. Each copy of the effect allele added roughly three-quarters of a kilogram of weight loss on top of the population average, independent of dose, drug, or baseline BMI.
GLP1R and GIPR variants predict nausea and vomiting risk. The receptors that mediate the therapeutic effect also mediate the dominant adverse effects — and the genetic risk for one does not perfectly predict the genetic risk for the other.
The GIPR nausea signal is tirzepatide-specific. The GIPR genotype influenced nausea and vomiting in patients taking tirzepatide (Mounjaro, Zepbound) but not in patients taking semaglutide (Ozempic, Wegovy) — exactly what you would predict from tirzepatide's dual GIP/GLP-1 receptor mechanism, and a clean piece of validation that the GWAS is detecting real biology rather than statistical noise.

The 23andMe and Stanford findings address different drugs (the Stanford trials used predominantly liraglutide and exenatide; the 23andMe cohort skewed toward semaglutide and tirzepatide), different endpoints (HbA1c vs body weight), and different genes (PAM vs GLP1R/GIPR). They are not redundant — they are convergent evidence that response to GLP-1 therapy has a meaningful inherited component.

Distinguishing Genetic Resistance from Pharmacological Resistance

"GLP-1 resistance" in 2026 is being used to describe at least three distinct phenomena that look similar at the bedside but have different mechanisms and different fixes.

Phenomenon	Mechanism	Clinical signal	Fix
Inherited GLP-1 resistance	Variants in PAM, GLP1R, or GIPR reduce signaling efficiency	Poor response from week 1 despite full dose escalation	Switch class (oral agents, dual/triple agonists, surgery)
Tachyphylaxis / desensitization	Receptor downregulation after months of agonist exposure	Initial response, then plateau at 12–18 months	Drug holiday, dose adjustment, mechanism switch
Behavioral / dietary compensation	Caloric intake creeps back during dose stabilization	Weight regain or stagnation despite good drug levels	Nutritional support, exercise integration

The clinical importance of the distinction is that the genetic form does not get better with time, dose escalation, or behavior change. It is a feature of the receptor system, not a failure of the patient.

Should Patients Get Genetic Testing Before Starting a GLP-1?

Not yet — and here is the honest reasoning.

The PAM and GLP1R/GIPR variants identified to date have moderate effect sizes at the population level. A 10% absolute reduction in the chance of reaching HbA1c target, or 0.76 kg per allele, are real but small numbers compared to the overall 12–22% mean weight loss most patients see on modern GLP-1s. Genetic testing would not, on the current evidence, predict who will lose 0% weight versus 20%; it would shift the population mean by a kilogram or two.

What the science does support clinically:

Setting realistic expectations. A patient who has lost <5% of body weight after six months of well-titrated semaglutide is not "doing it wrong." Genuine non-response exists and has biological causes.
Switching mechanism rather than escalating dose. A patient who fails to respond to a pure GLP-1 agonist may still respond to a dual (tirzepatide, survodutide) or triple (retatrutide) agonist that engages GIP or glucagon receptors not affected by PAM-related amidation deficits.
Reframing the conversation about side effects. The same genetic factors that drive efficacy can drive nausea risk, and the relationship between the two is not strictly proportional. Some patients are unlucky on both axes; some get one without the other.

The future of genetic testing in this space is more likely to come bundled into broader pharmacogenomic panels — covering CYP2D6 for psychiatric drugs, SLCO1B1 for statins, and now GLP1R/PAM for incretin therapies — rather than as standalone GLP-1 testing.

What "Resistance" Means for Drug Development

The pharmaceutical implication is significant and is already shaping pipelines. If a meaningful fraction of patients have receptor-level resistance to GLP-1 specifically, then drugs that work through non-GLP-1 mechanisms become more attractive than the field assumed two years ago.

That category is starting to fill out:

GIP/glucagon dual agonists that bypass GLP-1 entirely
Amylin agonists (cagrilintide as a CagriSema component or as monotherapy)
MC4R agonists acting on the central melanocortin pathway
Bimagrumab and myostatin inhibitors working through muscle preservation rather than appetite

A small but reliable subset of obesity patients who don't respond to GLP-1 agonists may turn out to be precisely the patients for whom non-GLP-1 mechanisms shine.

Practical Guidance for Patients on GLP-1s in 2026

If you are taking semaglutide, tirzepatide, or any other GLP-1 agonist and your response has been disappointing, the following framework reflects current evidence:

Verify the dose ladder is complete. Most apparent non-response is incomplete titration. The full Wegovy ladder is 0.25 → 0.5 → 1.0 → 1.7 → 2.4 mg over 17 weeks; the full Zepbound ladder reaches 15 mg by month four. Many "non-responders" are still at submaximal doses.
Give the trial 6 months at maximum tolerated dose. True genetic non-response is visible by then. Earlier judgments are noise.
Track measurable signals. Body weight, waist circumference, HbA1c if diabetic, food-noise score, blood pressure. A patient losing 4% body weight but reporting dramatic food-noise reduction is benefiting in ways scale weight does not capture.
Discuss mechanism switching with the prescriber. If a pure GLP-1 has failed, a GLP-1/GIP (tirzepatide) or GLP-1/glucagon (survodutide, in trials) may still work. Repeating the same mechanism at higher dose is the least likely path to success.
Do not over-interpret commercial genetic testing. No consumer genetic test currently provides clinically validated guidance on GLP-1 response. The science is real; the test products lag the science.

Key Takeaways

GLP-1 resistance is a real, genetically driven phenomenon affecting roughly 1 in 10 people through variants in the PAM gene that impair hormone amidation
A separate Nature 2026 paper identified GLP1R and GIPR variants that predict both weight-loss magnitude and side-effect risk, with the GIPR nausea signal restricted to tirzepatide
The Stanford evidence is strongest for glycemic non-response in type 2 diabetes; weight-loss data in pure obesity is suggestive but underpowered
Apparent non-response should still be evaluated for incomplete titration, behavioral compensation, and tachyphylaxis before being attributed to inherited resistance
Genetic testing is not yet ready for routine clinical use, but the science already supports switching mechanism — not escalating dose — when a pure GLP-1 fails
The non-responder population is becoming the strategic case for non-GLP-1 obesity drugs in late-stage development

References

Cardona A, Day FR, Perry JRB, Loh PR, Chu AY, Lehne B, et al. Type 2 diabetes risk alleles in peptidyl-glycine alpha-amidating monooxygenase influence GLP-1 levels and response to GLP-1 receptor agonists. Genome Med. 2026;18(1):42. DOI: 10.1186/s13073-026-01630-0 · PubMed
Park S, Holmes BR, Brand JS, Hicks B, Diaz-Papkovich A, Tian C, et al. Genetic predictors of GLP1 receptor agonist weight loss and side effects. Nature. 2026. DOI: 10.1038/s41586-026-10330-z
Wilding JPH, Batterham RL, Calanna S, Davies M, Van Gaal LF, Lingvay I, et al. Once-Weekly Semaglutide in Adults with Overweight or Obesity. N Engl J Med. 2021;384(11):989-1002. DOI: 10.1056/NEJMoa2032183 · PubMed
Jastreboff AM, Aronne LJ, Ahmad NN, Wharton S, Connery L, Alves B, et al. Tirzepatide Once Weekly for the Treatment of Obesity. N Engl J Med. 2022;387(3):205-216. DOI: 10.1056/NEJMoa2206038 · PubMed
Frias JP, Davies MJ, Rosenstock J, Pérez Manghi FC, Fernández Landó L, Bergman BK, et al. Tirzepatide versus Semaglutide Once Weekly in Patients with Type 2 Diabetes. N Engl J Med. 2021;385(6):503-515. DOI: 10.1056/NEJMoa2107519 · PubMed
Davies M, Færch L, Jeppesen OK, Pakseresht A, Pedersen SD, Perreault L, et al. Semaglutide 2.4 mg once a week in adults with overweight or obesity, and type 2 diabetes (STEP 2): a randomised, double-blind, double-dummy, placebo-controlled, phase 3 trial. Lancet. 2021;397(10278):971-984. DOI: 10.1016/S0140-6736(21)00213-0 · PubMed
Knudsen LB, Lau J. The Discovery and Development of Liraglutide and Semaglutide. Front Endocrinol (Lausanne). 2019;10:155. DOI: 10.3389/fendo.2019.00155 · PubMed
Drucker DJ. Mechanisms of Action and Therapeutic Application of Glucagon-like Peptide-1. Cell Metab. 2018;27(4):740-756. DOI: 10.1016/j.cmet.2018.03.001 · PubMed

Last updated: 2026-05-03 Medical review: Dr. James Chen, MD, PhD, FACE