Some patients on Ozempic drop 20 percent of their body weight. Others, taking the same drug at the same dose, barely lose five percent. Doctors have watched this play out across millions of prescriptions without a clear biological explanation. Now, a team of NIH researchers believes they have found one: deep in the hindbrain, individual appetite-regulating neurons respond to semaglutide in fundamentally different ways. Some cells sustain a strong chemical signal long after each dose. Others fire off a brief spike and go quiet. That cell-level split, described in a May 2026 paper in Nature Metabolism, may be a key reason the same prescription produces such wildly different results.
The signal that separates strong responders from weak ones
The research zeroes in on a brain structure called the area postrema, a small cluster of neurons in the hindbrain that sits outside the blood-brain barrier. Because it lacks that protective shield, the area postrema can directly sense drugs circulating in the bloodstream, making it one of the first places semaglutide reaches after injection.
When semaglutide binds to GLP-1 receptors on these neurons, it triggers a rise in cyclic AMP (cAMP), a molecule that acts as a relay, carrying the drug’s “reduce appetite” message inside each cell. The NIH team’s breakthrough was recording this process in individual neurons rather than measuring an average across the whole region. What they saw was striking: some neurons maintained elevated cAMP for an extended period, well beyond the initial drug exposure, while neighboring cells produced only a short-lived burst that faded quickly.
That difference is not trivial. When the researchers experimentally blocked sustained cAMP signaling in these hindbrain cells, semaglutide lost much of its power to reduce food intake and body weight in mice. The sustained signal, not just the initial activation, appears to be what drives the drug’s strongest appetite-suppressing effects, according to the NIH team’s summary of the work.
The paper also revealed that these GLP-1 receptor neurons do not rely on a single signaling cascade. They engage two distinct intracellular pathways, known as Gs-dependent and Gq-dependent signaling. Think of them as two separate volume knobs inside each cell. Depending on which pathway dominates in a given neuron, the drug’s signal gets amplified or dampened. That dual architecture helps explain how the same receptor, activated by the same molecule, can produce such different responses from one cell to the next.
Genetics add another layer
The cell-level findings do not exist in isolation. A separate analysis published in Nature drew on large genomic datasets to show that inherited differences in genes tied to the GLP-1 signaling pathway help explain why some patients respond strongly to these drugs and others see only modest results. The authors argued that genetic variation accounts for a measurable share of the patient-to-patient gap now visible in clinical practice, though the precise contribution of any single gene variant has not been quantified in large human trials.
Taken together, the picture is layered. Some people may carry genetic backgrounds that predispose their hindbrain neurons to sustain a strong cAMP response after each dose. Others may have variants that blunt the cascade. Neither study alone closes the case, but they point in the same direction: the variability patients experience is not random or purely behavioral. It has roots in biology.
The body’s own GLP-1 system complicates the picture
Semaglutide is not the only source of GLP-1 signaling in the body. The gut produces its own GLP-1 after meals, and a separate set of brainstem neurons, called PPG neurons, also releases the hormone. Research has shown that these central and peripheral systems can independently suppress eating, meaning the area postrema circuit the NIH team studied is not the only appetite brake in play.
Still, the new data suggest this hindbrain circuit carries a large share of the drug’s total weight-loss effect in mice, particularly through those neurons that maintain prolonged cAMP elevation. Understanding how the drug’s action in the area postrema interacts with the body’s native GLP-1 systems remains an open question, and one that matters for predicting who benefits most.
What about next-generation oral drugs?
A related line of NIH-supported research has examined oral GLP-1 receptor agonists such as orforglipron and danuglipron, small-molecule drugs still in clinical development. According to NIH, these compounds can penetrate deeper into the brain than injectable peptides and appear to alter reward circuitry in the central amygdala, reducing dopamine release during pleasure-driven eating in mice.
Whether these oral agents will show the same cell-to-cell cAMP variability seen with semaglutide in the area postrema is unknown. The current reports on orforglipron and danuglipron focus more on behavior and bulk circuit activity than on single-cell signaling dynamics. If future work finds that oral agents produce more uniform neuronal responses, they might, in theory, narrow the range of weight-loss outcomes. But no human data yet tie specific brain-circuit effects of these drugs to clinical weight loss, and the science remains early enough that even the framing of how they affect reward circuits is still being debated among researchers.
The large gaps that remain
Every mechanistic finding reported so far comes from mice. No team has recorded cAMP dynamics in living human area postrema neurons, and no published data link a specific cAMP response pattern in a person’s brain to kilograms lost over a treatment course. The leap from mouse hindbrain physiology to human clinical outcomes is plausible but unproven at the individual level.
The genetic findings face a parallel limitation. Genome-wide association studies can identify statistical links between gene variants and drug response across populations, but the effect sizes for individual variants tend to be small. No predictive genetic score currently exists that a clinician could hand a patient before the first injection to forecast whether semaglutide will produce a strong or weak result. Even if such a score emerges, it will have to be validated against real-world variables: adherence, diet, physical activity, coexisting conditions, all of which also shape weight change.
Clinical trial data from the STEP program illustrate the stakes. While average weight loss on semaglutide 2.4 mg hovers around 15 percent of body weight, the range is wide. Roughly 10 to 15 percent of trial participants lost less than five percent, a threshold often considered clinically insignificant. Understanding why those patients respond poorly, and whether their hindbrain neurons or genetic profiles differ in measurable ways, is the next critical question.
What this means for patients right now
For the millions of people currently taking semaglutide, tirzepatide, or similar injectable GLP-1 therapies, the new findings offer an explanation rather than a prescription change. They help clarify why two people on the same dose can see very different results and suggest that future diagnostic tools might one day predict response before treatment begins.
They do not show that the drugs are ineffective in any group already benefiting, and they do not yet justify switching medications or adjusting doses based on this brain-circuit research alone. Until human studies connect specific neuronal response patterns or genetic profiles to real-world outcomes, decisions about starting or continuing GLP-1 therapy will still rest on clinical trial evidence, side-effect profiles, and individual goals. The biology is finally catching up to what patients and doctors have observed for years. Turning that biology into personalized guidance is the work that comes next.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
