GLP-1 Resistance: Why 1 in 10 People Don't Respond.

What GLP-1 actually does — and how the receptor works

Glucagon-like peptide-1 (GLP-1) is a 30-amino-acid incretin hormone secreted from L-cells in the gut in response to food. Its jobs are precise: it amplifies insulin secretion from the pancreas in a glucose-dependent manner, suppresses glucagon, slows gastric emptying, and — critically for the weight-loss drugs — acts on GLP-1 receptors (GLP1R) in the hypothalamus and brainstem to reduce appetite and energy intake [5].

The GLP1R is a class B G-protein-coupled receptor. When a GLP-1 molecule — or a drug like semaglutide — binds to it, the receptor activates a G-protein (Gs), which triggers intracellular cyclic adenosine monophosphate (cAMP) production, activating protein kinase A (PKA). This cAMP/PKA cascade drives downstream effects: insulin secretion in the pancreas, satiety signaling in the brain, reduced gut motility [5]. There is a secondary arm mediated by beta-arrestin, which shapes the duration and trafficking of the receptor signal and may modulate side-effect profiles — but the primary weight-loss and glucose-control signal runs through cAMP [6].

The punchline: the entire drug class — semaglutide, liraglutide, tirzepatide's GLP-1 arm — depends on this one receptor firing properly. If the GLP1R signal is blunted at the receptor level, the drug cannot compensate by working harder. It has nowhere else to go.

The genetics are confirmed: the 23andMe Nature GWAS

On April 8, 2026, the 23andMe Research Institute published a genome-wide association study (GWAS) — a large-scale scan of the genome that identifies common genetic variants associated with a measurable trait — in Nature [1]. Lead author Adam Auton and colleagues analyzed data from 27,885 participants who had used a GLP-1 receptor agonist and had self-reported weight-loss outcomes. It is the largest pharmacogenetic GWAS of GLP-1 response conducted to date.

The key finding: a missense variant in the GLP1R gene, tagged by the single nucleotide polymorphism (SNP — a single-letter change in the DNA sequence) rs10305420, was significantly associated with blunted weight loss response. Carriers of each additional effect allele (the variant copy) lost approximately 0.76 kg less per copy of the variant — a modest per-allele effect, but one that reaches population significance at 27,885 subjects and translates into clinically meaningful differences when both copies are present [1].

This is a missense variant, meaning the DNA change alters a single amino acid in the GLP1R protein. The consequence is structural: the receptor is present, it binds the drug, but the downstream signaling cascade — the cAMP production, the PKA activation — fires at reduced efficiency. The receptor signal is blunted at its source.

This is the first large-scale GWAS to confirm genetic non-response at population scale. It does not explain all non-response — genetics rarely tells the whole story — but it establishes that a meaningful fraction of people who fail GLP-1 therapy are not failing because of poor adherence, incorrect dosing, or lifestyle factors. The biology is working against them.

It is worth being specific about what the 23andMe study is and is not. It is a consumer genetic cohort with self-reported outcomes — which introduces measurement noise. It does not prove causality in the strict clinical sense. What it does is provide the largest pharmacogenetic signal in this space to date, with a biologically plausible mechanism (missense variant in the receptor itself), published in the field's most rigorous venue. The signal is real. The clinical implications are directional, not yet quantitatively precise.

Related research adds texture. A 2025 study in Acta Diabetologica investigated two other GLP1R SNPs — rs6923761 and rs761387 — and found that these variants were not independently predictive of response to oral semaglutide in type 2 diabetes [9]. This underscores that not every genetic variant in GLP1R has the same downstream effect; the receptor's pharmacogenetics are layered, and rs10305420 appears to be the variant with the clearest weight-loss signal [9].

The receptor is present. The drug binds. But the signal it pulls — the cAMP cascade that drives satiety and insulin response — fires at a fraction of its intended strength. That is what GLP-1 resistance looks like at the molecular level.

How common is non-response? Sorting the numbers

"Roughly 1 in 10" is the conservative floor. The honest answer is that non-response rates depend heavily on how you define non-response and which population you are looking at.

In the STEP 1 trial — the pivotal phase 3 trial of semaglutide 2.4 mg weekly in adults with overweight or obesity (n=1,961), published in NEJM in 2021 — 7.6% of participants in the semaglutide arm lost less than 5% of their body weight at 68 weeks [2]. This is the trial non-responder floor: 7.6% of people who completed a well-run, well-monitored RCT with a confirmed-effective drug and strong adherence support still did not reach the minimum clinically meaningful weight-loss threshold [2].

Real-world numbers are higher. Squire and colleagues published a retrospective cohort analysis in BMJ Open in January 2025, examining 483 patients treated with GLP-1 analogues for obesity in a real-world clinical setting [3]. Mean total body weight loss (TBWL) across the cohort was 12.2%, but 17.8% of patients did not achieve the 5% TBWL threshold that defines a clinically meaningful response [3]. In a real clinic — with missed injections, dose adjustments, variable adherence, and the full messiness of actual care — closer to 1 in 6 patients are not responding adequately.

A 2025 expert review in Expert Opinion on Pharmacotherapy cited figures as high as 30% not achieving ≥5% weight loss across the broader clinical literature, using the most inclusive threshold definitions [4]. The range — 7.6% at best, up to 30% at broadest — reflects genuine methodological differences, not conflicting data.

For this article, "roughly 1 in 10" is the honest conservative floor that captures confirmed clinical non-responders across the best-controlled trials. In real-world practice, the number is likely higher.

Identifying non-responders clinically: the 5% threshold and the 12-week check

Clinically, distinguishing a true non-responder from an undertreated patient or a non-adherent patient requires careful sequencing. Non-response at week 4 on a starter dose means almost nothing. Non-response at week 20 on semaglutide 2.4 mg — with confirmed injection technique, no missed doses, and documented dietary context — is a meaningful signal.

The phenotypic markers that cluster with GLP-1 non-response in clinical observation include: minimal or absent nausea/appetite suppression (suggesting the satiety signal is simply not landing), persistently poor fasting glucose response despite adequate dosing in type 2 diabetes, and weight trajectories that plateau early and far short of the expected curve. None of these are diagnostic — they are signals worth flagging for clinical review.

Genetic testing for rs10305420 is not yet standard of care. The 23andMe data provides population-level confirmation; it does not yet offer a validated clinical decision tool. A clinician seeing a 12–16 week non-responder does not need the genetic result to act. The clinical picture is sufficient to prompt escalation.

The lean mass problem — worse for non-responders

Here is a fact about GLP-1 therapy that the promotional conversation consistently undersells: across the major trials, GLP-1 receptor agonists drive approximately 20–40% of total weight lost from lean mass — primarily skeletal muscle. In STEP 1, participants who lost ~15% of body weight lost roughly 4–5 kg of that as lean tissue. This is a consistent finding across the class.

For a successful responder who loses 15–20% of body weight, losing a fraction of that as muscle is a real concern — but one offset by significant fat mass reduction, metabolic improvements, and cardiovascular benefit. The tradeoff, for responders, is defensible with resistance training and adequate protein intake.

For a non-responder, the math is brutal. If you are losing 2–3% of body weight over 16 weeks — and 20–40% of that minimal loss is lean mass — you are losing negligible fat while actively sacrificing muscle. You are getting the worst of GLP-1 therapy with almost none of the benefit. The urgency to switch agents for confirmed non-responders is not just about missing weight loss targets. It is about stopping an ongoing lean-mass drain that is delivering no meaningful metabolic return. For a full breakdown of how to protect muscle during GLP-1 therapy, see our piece on lean mass preservation on GLP-1s.

The fallback options: dual and triple agonists

The single most important development for GLP-1 non-responders is the existence of receptor agonists that do not rely exclusively on GLP1R signaling. Two classes are now supported by phase 3 data.

Tirzepatide: GIP/GLP-1 dual agonist

Tirzepatide is a dual agonist: it activates both the GLP1R and the glucose-dependent insulinotropic polypeptide receptor (GIPR — GIP receptor). GIP (glucose-dependent insulinotropic polypeptide) is a second incretin hormone that works through a completely separate receptor pathway. In the SURMOUNT-1 trial (n=2,539), tirzepatide at 15 mg weekly produced mean weight loss of 22.5% at 72 weeks, versus 2.4% for placebo [7]. This is substantially greater than semaglutide's average in comparable trials.

For GLP1R non-responders, tirzepatide offers a parallel signal pathway. Even if the GLP1R signal is blunted, the GIPR arm is intact and functional. The GIPR pathway signals through a distinct cAMP cascade in adipose tissue and central appetite circuits. It is not a workaround to a broken GLP1R — it is an additional signal from a mechanistically different receptor [6].

The clinical implication is direct: a confirmed GLP1R-variant non-responder who switches to tirzepatide is not simply getting more of the same stimulus at higher dose. They are recruiting a second receptor system that the missense variant in GLP1R does not affect. For the head-to-head weight-loss data between the two drugs, see our tirzepatide vs semaglutide comparison.

Retatrutide: GLP-1/GIP/glucagon triple agonist

Retatrutide adds a third arm: glucagon receptor agonism. Glucagon, typically thought of as a counter-regulatory hormone that raises blood glucose, also drives thermogenesis and fat oxidation when signaled through the glucagon receptor in adipose tissue and the liver. The combination — GLP-1 satiety signaling, GIP insulin sensitization and fat storage modulation, glucagon-driven fat burning — appears to produce additive and in some contexts synergistic weight loss effects.

In a phase 2 trial published in NEJM in 2023 (n=338), retatrutide at 12 mg weekly produced mean weight loss of 24.2% at 48 weeks [8]. This is the largest weight-loss signal yet reported for any pharmacological agent in a controlled trial. Phase 3 trials are underway.

For GLP1R non-responders, retatrutide is the most aggressive pharmacological option currently in the pipeline. It is not yet approved (as of April 2026), which limits access to trial enrollment or compassionate use contexts.

A confirmed GLP-1 non-responder switching to tirzepatide is not chasing a higher dose of the same drug. They are recruiting a completely separate receptor signal that the GLP1R variant does not blunt.

Lifestyle augmentation: what the data actually says

GLP-1 therapy is not designed to replace lifestyle intervention — it is designed to make it more sustainable. For non-responders or partial responders, the evidence for lifestyle intensification alongside pharmacotherapy is real and should not be dismissed. In the STEP 3 trial, semaglutide combined with intensive behavioral therapy (including a low-calorie diet and increased exercise prescription) produced approximately 16% weight loss, compared to ~12% in the standard-lifestyle STEP 1 arm [10]. The behavioral component adds signal even when the receptor signal is partial.

However, lifestyle augmentation has limits. For a confirmed non-responder — defined by genetics, clinical non-response at 12–16 weeks, and absence of confounding factors — the ceiling of lifestyle intensification is real. It can add 2–4% TBWL in optimistic scenarios; it does not overcome a blunted receptor signal in any trial.

Tiered framework: what to do if the signal is blunted

We do not write protocols. We build frameworks that you take to a clinician. If you or a patient is at 12–16 weeks on a GLP-1 receptor agonist at target dose and the response is inadequate, here is how the decision tree looks.

What this actually changes

The 23andMe Nature GWAS matters for a reason beyond the scientific novelty. For years, patients who did not respond to semaglutide were implicitly — or explicitly — blamed for their non-response. They were eating too much. They weren't trying hard enough. The drug works; they must be doing something wrong.

The genetics are now confirmed. For roughly 1 in 10 GLP-1 users, the receptor signal is blunted before the drug even gets a fair shot. The missense variant in GLP1R is not a character flaw. It is a molecular fact. The appropriate clinical response is not to push higher doses of a drug against a broken signal, or to suggest that the patient exercise more willpower. It is to switch to a drug that uses a different signal pathway.

That switch is now supported by some of the most impressive phase 3 data in obesity pharmacotherapy history. Tirzepatide at 22.5% mean weight loss and retatrutide at 24.2% in phase 2 represent a fundamentally different therapeutic ceiling than GLP-1–only drugs. For non-responders, these are not second-line compromises. They are mechanistically superior options for a specific genetic profile.

The framework for the next five years of GLP-1 prescribing is coming into focus: baseline response assessment at 12 weeks, escalation to dual or triple agonism for non-responders, and — as pharmacogenetic testing matures — potentially pre-screening for rs10305420 before the first prescription is written. The era of one-size-fits-all incretin therapy is already ending. The data is making the case faster than the prescribing guidelines are catching up.

For more on the broader GLP-1 landscape and the next generation of incretin drugs, see The GLP-1 Era: What It Actually Changes and Next-Generation Incretins: Dual and Triple Agonists Explained.