What does it mean to be a GLP-1 non-responder?
Clinical trials of incretin drugs report averages. When a study says participants lost a given percentage of body weight over 72 weeks, that number sits at the center of a wide distribution, and the tails of that distribution are real people. At one end are people whose results outstrip the headline figure. At the other are people who take the medication as prescribed, tolerate it, complete the full titration their prescriber directs, and see comparatively little change on the scale or in their glucose measures.
That second group is what the literature calls non-responders, and the label is less tidy than it sounds. Whether someone counts as a non-responder depends entirely on the threshold chosen — a five percent weight reduction, a ten percent reduction, a specific drop in HbA1c — and on the time window over which it is measured. Different papers use different cutoffs, which is one reason the reported proportion of non-responders varies so much between studies.
The new review in Diabetes, Obesity & Metabolism
The paper prompting this piece is a review article by A. Kurylowicz and L. Czupryniak, indexed in Diabetes, Obesity & Metabolism and titled "Responders Vs. Non-Responders or How to Predict the Response to GLP-1 or GLP-1/GIP Receptor Agonist Therapy." PubMed classifies it as a journal article and a review, which matters for how much weight to give it: a review gathers and interprets published findings rather than generating new ones.
That distinction is worth spelling out, because reviews and randomized trials often get quoted in the same breath online. A randomized controlled trial produces new outcome data under controlled conditions. A review assembles what already exists, weighs it, and identifies gaps. Reviews are how a field takes stock, and they frequently set the agenda for the trials that follow — but a review cannot establish that a given marker predicts response. It can only report how strong the existing case is.
The framing in the title is itself the news for this audience. The question is not whether variation in response exists — anyone who has compared notes with other people on the same medication already knows it does. The question is whether that variation can be anticipated before therapy begins, from characteristics measurable at baseline, rather than discovered months later from a scale and a lab panel.
The bibliographic record indexed on PubMed for the responders review lists it as a review in Diabetes, Obesity & Metabolism by Kurylowicz and Czupryniak, without a public abstract summary of its conclusions.
If you are reconciling your own records against a prescriber-directed schedule, the semaglutide dosage calculator handles the unit arithmetic without changing anything about the plan itself.
The tirzepatide PCOS analysis, published in the Journal of the Endocrine Society by Clift, Reisel, Johnson and colleagues, compared weight loss outcomes in women with and without self-reported polycystic ovary syndrome. the tirzepatide PCOS analysis.
Why subgroup studies feed the response question
Predicting response usually starts with subgroups. If a defined population reliably does better or worse than the overall trial average, that population becomes a candidate signal. A recent analysis in the Journal of the Endocrine Society by Clift, Reisel, Johnson and colleagues did exactly that kind of comparison, examining weight loss outcomes with tirzepatide in women with and without self-reported polycystic ovary syndrome.
PCOS is a plausible place to look. It is a common endocrine condition associated with insulin resistance and with difficulty losing weight through diet and activity alone, so whether a dual GLP-1/GIP receptor agonist performs differently in this group is a question with practical consequences for a lot of people. The methodological caveat is in the study's own title: the PCOS status was self-reported rather than confirmed against diagnostic criteria or clinical records, which introduces the possibility of misclassification in both directions.
Subgroup analyses of this kind are hypothesis-generating. They can point toward a predictor without proving one, particularly when the grouping variable depends on what participants said about themselves. Readers should treat the comparison as a piece of a developing picture rather than a verdict on how any individual with PCOS will respond.
For the dual GLP-1/GIP agonist discussed in the PCOS subgroup analysis, the tirzepatide dosage calculator covers the same conversion work.
The Obesity Pillars paper on muscle mass, by Mendias and Awan, addresses increasing skeletal muscle mass and strength during incretin-based weight loss. the Obesity Pillars paper on muscle mass.
Is weight even the right measure of response?
A paper by Mendias and Awan in Obesity Pillars, titled "Increasing skeletal muscle mass and strength during incretin-based weight loss," points at a complication underneath the entire responder debate. Total body weight does not distinguish fat mass from lean tissue, and concern about lean mass loss during rapid weight reduction has become one of the most active discussion topics in obesity medicine.
If two people lose the same amount of weight but one loses proportionally more muscle, calling them equal responders is misleading. Conversely, someone whose scale weight barely moves while body composition shifts might be classed as a non-responder by a weight-based threshold. Body composition measurement is far less available in routine care than a scale and a tape measure, which is part of why weight-based cutoffs persist in the literature.
For anyone tracking a protocol, the takeaway is about record-keeping rather than intervention: the measures you happen to capture determine the conclusions you can draw later, and weight alone is a narrow lens.
Separating a genuine limited response from an interrupted course depends on adherence tracking, which is why missed doses and pauses belong in the log alongside the numbers.
How many people does non-response actually affect?
The size of the affected group is growing, which is what turns a clinical curiosity into a public health question. An analysis by Wartko, Zhang, Johnson and colleagues in Diabetes Research and Clinical Practice examined trends in the use of GLP-1 receptor agonists, SGLT2 inhibitors, and metabolic bariatric surgery among U.S. adults with diabetes and obesity — the kind of utilization work that establishes the denominator.
As prescribing expands well beyond the populations enrolled in the original registration trials, the range of real-world outcomes widens with it. Trial participants are screened, monitored, and supported in ways that everyday patients often are not. A proportion of people who see limited benefit is unremarkable in a trial appendix and consequential when the treated population runs into the millions.
That is the practical reason the responder question is drawing review-level attention now: the decision of what to do when a medication is not working has become a common clinical conversation rather than an edge case.
Consistent side-effect logging gives a prescriber context for judging whether tolerability, rather than biology, is shaping the outcome.
What is still unresolved
No validated pre-treatment test for predicting individual response to incretin therapy is part of standard prescribing. Response is still determined retrospectively, from measurements collected over weeks and months, and the definition of adequate response is a judgment made between a patient and their prescriber rather than something fixed by a threshold in a paper.
The related literature indexed alongside these papers shows how broadly the class is being studied — from a matched cohort study in Inflammatory Bowel Diseases on symptomatic remission in ulcerative colitis to preclinical work in iScience comparing semaglutide, tirzepatide and retatrutide in mouse models of renal fibrosis. Those are early and separate lines of inquiry, and none of them speaks to how a given person will respond.
What the responders review signals is that the field is treating variation in outcome as something to be explained rather than tolerated. Whether that produces a usable predictor, or simply a better vocabulary for describing what clinicians already observe, is not settled by a review article.
The review indexed in Diabetes, Obesity & Metabolism by Kurylowicz and Czupryniak is listed on PubMed as a journal article and review addressing how to predict response to GLP-1 or GLP-1/GIP receptor agonist therapy. the review indexed in Diabetes, Obesity & Metabolism.