Scientists Identify Biological Mechanism Behind Why Ozempic Fails for Some Patients

Semaglutide — sold as Ozempic for diabetes and Wegovy for weight loss — has become the most commercially successful drug class in a generation. But behind the dramatic before-and-after photos lies an uncomfortable statistic: up to 23% of patients in Novo Nordisk's own clinical trials qualified as non-responders, losing less than 5% of their body weight [1][2]. Two new studies published in April 2026 offer the first concrete genetic explanations for why, raising both hopes for precision prescribing and hard questions about who is paying for a drug that may never work for them.

The Scale of Non-Response

In clinical trials of semaglutide 2.4 mg (the STEP program), roughly 77% of participants lost at least 5% of their body weight — the threshold at which measurable health improvements begin. About half lost 10% or more, and a third achieved the striking 15%+ losses that dominate headlines [2][3]. That leaves approximately one in four patients who see minimal results despite weekly injections costing up to $1,349 per month at list price [4].

Source: STEP 1 Trial / Scientific American

Data as of Apr 11, 2026CSV

Real-world data paint a similar picture. As Scientific American reported in early 2026, "as many as one in four people on these drugs are nonresponders" [2]. The gap between trial results and real-world outcomes is a persistent theme in obesity medicine, where controlled trial populations tend to be more adherent and more motivated than patients in clinical practice.

Two New Genetic Studies, Two Different Angles

Two major studies published in April 2026 approach the genetics of GLP-1 non-response from different directions.

The PAM Enzyme Discovery (Stanford/ETH Zurich)

A team led by Anna Gloyn at Stanford and Markus Stoffel at ETH Zurich identified two variants in the gene encoding PAM (peptidyl-glycine alpha-amidating monooxygenase) — an enzyme that activates multiple hormones, including GLP-1. The variants, designated p.S539W and p.D563G, are carried by roughly 10% of the general population [5][6].

The finding was paradoxical. People carrying these PAM variants had higher circulating levels of GLP-1, not lower. But the hormone appeared biologically inert — their blood sugar levels did not drop as expected. "This was the opposite of what we imagined we would find," Gloyn said [5]. The researchers termed this phenomenon "GLP-1 resistance," analogous to insulin resistance in type 2 diabetes.

In a meta-analysis of three clinical trials totaling 1,119 participants, 25% of non-carriers achieved recommended HbA1c targets after six months of GLP-1 receptor agonist treatment. Among p.S539W carriers, only 11.5% hit those targets. Among p.D563G carriers, 18.5% did [5].

Source: Stanford/Genome Medicine

Data as of Apr 10, 2026CSV

Critically, the variants had no effect on response to other diabetes medications — sulfonylureas, metformin, or DPP-4 inhibitors — suggesting the mechanism is specific to the GLP-1 pathway [5]. The study was published in Genome Medicine on April 10, 2026.

The 23andMe GWAS (Nature)

Days earlier, on April 8, the 23andMe Research Institute published a genome-wide association study (GWAS) of 27,885 GLP-1 users in Nature [7]. This study took a broader approach, scanning the entire genome for variants associated with weight loss and side effects.

The headline finding: a missense variant in the GLP1R gene itself — the receptor that semaglutide binds to — was associated with an additional 0.76 kg of weight loss per copy of the effect allele [7][8]. Across participants, self-reported weight loss ranged from 6% to 20% of starting body weight, and nausea or vomiting risk ranged from 5% to 78%, depending on genetic profile [8].

Adam Auton, vice president of human genetics at the 23andMe Research Institute, called it "as clean a pharmacogenetic result as you can hope for" [9]. The study also identified a variant in the GIPR gene linked to nausea and vomiting specifically in patients taking tirzepatide (Mounjaro/Zepbound), but not semaglutide — a distinction that could eventually guide drug selection [7].

Demographic Patterns: Sex Matters, Race Appears Not To

A separate meta-analysis of 64 clinical trials published in JAMA Internal Medicine in March 2026 by researchers at Johns Hopkins found that GLP-1 receptor agonist effectiveness was "similar across" age groups, racial and ethnic populations, baseline BMI categories, and baseline HbA1c levels [10].

The one consistent demographic difference: sex. Women taking GLP-1 receptor agonists lost an average of 10.88% of their starting weight, compared to 6.78% for men — a statistically significant gap across 19,906 patients in six trials [10]. The biological reasons for this disparity are not fully understood and may overlap with, but are not explained by, the PAM or GLP1R variants identified in the two April studies.

"These results should give clinicians and their patients more confidence that GLP-1-RAs work similarly well across different racial and ethnic populations, and different ages and weights," said Hemal Mehta, the study's senior author [10]. The researchers cautioned, however, that clinical trials have historically enrolled disproportionately White and female participants, limiting confidence in subgroup analyses.

The Research Explosion Behind the Headlines

Scientific interest in GLP-1 has surged alongside commercial interest. According to OpenAlex data, 22,742 papers on GLP-1 were published in 2025 — up from just 2,681 in 2011, a nearly ninefold increase [11].

Source: OpenAlex

Data as of Jan 1, 2026CSV

That volume reflects not just the weight-loss boom but expanding research into GLP-1's effects on cardiovascular disease, neurodegeneration, addiction, and kidney function. The two April 2026 studies emerged from this broader acceleration.

The Obesity Context

The market for GLP-1 drugs is driven by a global obesity crisis. WHO data show adult obesity prevalence at 42% in the United States, 30.8% in South Africa, 30.2% in Australia, and 28.1% in Brazil [12]. Between 6% and 12% of American adults have used an injectable weight-loss medication, and more than 25,000 Americans per week were starting Wegovy as of May 2024 [13].

Source: WHO Global Health Observatory

Data as of Dec 31, 2022CSV

How Ozempic Compares to Alternatives

Semaglutide's non-response rate needs context. Other weight-loss interventions also have significant failure rates, often at comparable or higher costs.

Bariatric surgery produces dramatically more weight loss — an average 24% of total body weight at two years versus roughly 5% in real-world GLP-1 data, according to a study in JAMA Surgery [14]. The upfront cost ($15,000–$25,000) is lower over time than ongoing GLP-1 prescriptions, and a cohort study found surgery saved approximately $11,689 over two years compared to GLP-1 drugs [14]. However, 20–25% of surgery patients experience significant weight regain, and the procedure carries surgical risks including reflux (20% of patients) and blood clots (<1%) [14].

Liraglutide (Saxenda), the older GLP-1 drug, produces roughly half the weight loss of semaglutide in head-to-head comparisons, with higher non-response rates [15].

Lifestyle interventions alone — diet and exercise counseling — typically produce 3–5% body weight loss in clinical trials, with high rates of regain [3].

In short, "some patients" failing on semaglutide means roughly one in four — a figure in line with or better than most alternatives, but significant when millions of prescriptions are involved and monthly costs can exceed $1,000.

The Financial Toll on Non-Responders

At list price, Wegovy costs $1,349 per month [4]. Novo Nordisk has introduced reduced cash-pay options — $349 per month for the injectable and $149 per month for the new oral formulation at certain doses — but these prices apply only to specific programs with eligibility restrictions [4].

Insurance coverage remains patchy. Only about 20–25% of commercial insurance plans cover weight-loss medications, and Medicare is federally prohibited from covering them [4][16]. Blue Cross Blue Shield plans excluded Wegovy from member benefits in 2026 [4]. For patients paying out-of-pocket, three months of non-response at list price represents over $4,000 spent before a clinician would typically reassess.

With roughly 2.6 million semaglutide prescriptions filled at U.S. retail pharmacies by December 2023 — a figure that has only grown since — and a non-response rate of approximately 23%, hundreds of thousands of patients may be paying for a drug that provides minimal benefit [2][13]. The financial loss falls disproportionately on uninsured patients and those on high-deductible plans, while insured non-responders drive costs that employers and insurers ultimately pass along through higher premiums.

The Skeptics' Case: Is This Mechanism Overhyped?

Not everyone in the obesity medicine field is convinced that PAM variants or GLP1R polymorphisms are the primary explanation for non-response.

Adherence remains the elephant in the room. GLP-1 drugs cause nausea, vomiting, diarrhea, and constipation in roughly three-quarters of users [14]. Many patients discontinue or take doses inconsistently, and real-world adherence rates are substantially lower than in clinical trials. A patient who stops injecting after two months because of side effects will appear as a non-responder, regardless of their PAM genotype.

Gut microbiome variation is another confounding factor. A 2026 review found that gut bacterial diversity significantly differed between GLP-1 responders and non-responders, with short-chain fatty acid production and bile acid metabolism both influencing GLP-1 signaling through independent pathways [17]. Changes in diet and weight loss themselves alter the microbiome, making it difficult to disentangle cause from effect.

Phenotypic differences in appetite regulation may also explain much of the variance. Researchers at the Mayo Clinic have identified distinct subtypes — a "hungry brain" phenotype requiring abnormally high calorie intake to achieve satiety, and a "hungry gut" phenotype characterized by rapid return of hunger after eating [2]. Andres Acosta of the Mayo Clinic noted that patients "see themselves as a failure," though effectiveness "is likely outside of their direct control" [2].

A large Nature Medicine study of more than 10,000 GLP-1 users found no significant genome-wide associations with weight-loss outcomes, suggesting that non-genetic factors may predominate in explaining response variation [2]. Ruth Loos, a geneticist at the University of Copenhagen, was direct: "We cannot do that yet. We need better science" — referring to the prospect of using genetic scores to predict GLP-1 response [2].

Andrea Ganna, a health data scientist at the University of Helsinki, expressed skepticism that the 23andMe findings would change clinical practice in the near term [9]. The effect size — 0.76 kg per allele copy — is real but modest, and the PAM study's sample of 1,119 participants, while carefully analyzed, is small by modern genomics standards.

The Path to a Diagnostic Test

If the PAM and GLP1R findings hold up, the logical next step is a companion diagnostic — a genetic test administered before prescribing, designed to identify likely non-responders. The path from here to there is long and full of obstacles.

Scientifically, the PAM study's authors acknowledged that only two clinical trials in their meta-analysis included weight-loss data, which was insufficient for conclusive findings on that endpoint [5]. The exact mechanism by which PAM variants cause GLP-1 resistance remains unknown. "That is the million-dollar question," Gloyn said [5]. Replication in larger, more diverse cohorts is essential.

Regulatorily, companion diagnostics require FDA approval through the premarket approval (PMA) pathway, which typically takes 2–4 years and costs tens of millions of dollars. The FDA has approved companion diagnostics in oncology (e.g., for BRCA mutations before PARP inhibitors), but has never done so for an obesity medication [18].

Commercially, the incentives are misaligned. Novo Nordisk, which holds the patents on semaglutide, sells more drug when more patients try it — including non-responders. A test that screens out 10–23% of potential patients before prescribing would reduce revenue. Pharmaceutical companies possess genetic data from clinical trials that could accelerate validation, but as the Stanford researchers noted, they "have not widely shared it" [5]. Pharmacy benefit managers (PBMs) and insurers, by contrast, have strong incentives to require such a test as a prior authorization step — but only if it is validated and FDA-cleared.

Mayo Clinic has developed a genetic test that estimates an individual's "calories to satiation," which it says can predict GLP-1 response [18]. But this remains a research tool, not a clinically validated diagnostic. 23andMe has begun offering a GLP-1 response report to members of its Total Health subscription service [8], though this is a consumer product without FDA clearance for clinical decision-making.

A realistic timeline from the current state of evidence to a clinically validated, insurance-reimbursed companion diagnostic is 5–10 years — assuming continued research funding, successful replication, and regulatory cooperation.

What This Means for Patients Now

For the millions of Americans currently taking or considering semaglutide, the practical takeaway is straightforward: the drug works for most people, but not all, and science is beginning to understand why.

Patients who have taken a GLP-1 drug for three months without losing at least 5% of their body weight should discuss alternatives with their physician — including dose adjustments, switching to tirzepatide (which acts on both GLP-1 and GIP receptors), or considering bariatric surgery [2][14]. The emerging genetic findings do not yet support routine testing before prescribing.

The broader lesson is about the limits of blockbuster medicine. A drug that works for 77% of patients is a medical achievement. But when that drug costs over $1,000 per month and is prescribed to millions, the 23% for whom it does not work represent a massive clinical and financial problem — one that precision medicine may eventually solve, but has not solved yet.