Beyond GLP-1

A long-standing puzzle in cannabis science has resurfaced, this time with real mechanistic weight behind it. People who use cannabis chronically tend to have lower body weight and a lower prevalence of type 2 diabetes than non-users, a pattern that seems to sit awkwardly against the plant’s best-known acute effect, which is stimulating appetite. In May 2026, a research team at the University of California, Riverside published a preclinical study in The Journal of Physiology that begins to explain the contradiction, and the answer points away from THC as the sole actor and toward the broader chemical profile of the whole plant[1].

The timing is notable. The finding arrives just as the limitations of the dominant pharmacological approach to obesity, the GLP-1 receptor agonists, are coming into sharper focus even while their commercial success accelerates. This article examines that study, situates it against the GLP-1 landscape and its documented downsides, and argues that the extract-versus-isolate distinction it surfaces is a hypothesis worth carrying into human investigation. It is not an endorsement of cannabis as a metabolic therapy, and the researchers themselves are explicit on that point.

Fast Facts

• A 2026 preclinical study found that whole-plant cannabis extract, but not isolated delta-9 THC, normalized glucose clearance in diet-induced obese mice, even though both treatments reduced body weight [1].
• GLP-1 receptor agonists such as semaglutide and tirzepatide produce average weight reductions of roughly 15 to 22 percent in clinical trials, yet about half of patients discontinue within one year [8][19].
• Chronic cannabis users show lower rates of obesity and type 2 diabetes across multiple epidemiological studies, despite the well-known appetite-stimulating effect of THC [1].
• The UCR study used a matched-THC design, so the only variable between the two active treatments was the presence or absence of the plant’s other compounds [1].
• No cannabinoid product is currently approved or recommended for the treatment of obesity or type 2 diabetes.

The Scale of the Problem

The case for pursuing new metabolic interventions rests on numbers that have barely moved in a decade. Roughly 40.3 percent of American adults have obesity, and 9.4 percent meet the threshold for severe obesity, according to National Health and Nutrition Examination Survey data covering 2021 through 2023 [2]. When overweight is included, the figure climbs to 72.4 percent of adults, meaning a clear majority of the adult population carries elevated metabolic risk [3]. Obesity is estimated to cost the United States healthcare system approximately 173 billion dollars annually [4].

The diabetes picture is similarly stark. More than 40 million Americans were living with diabetes as of 2023, about 12 percent of the population, with type 2 diabetes accounting for an estimated 90 to 95 percent of diagnosed cases [5]. Upstream of that sits prediabetes, which affects an estimated 115.2 million adults, more than two in five, roughly eight in ten of whom do not know they have it [6]. The total estimated cost of diagnosed diabetes reached 412.9 billion dollars in 2022 [7]. These are not problems at the margins of public health. They are central, expensive, and largely unmoved by lifestyle messaging alone, which is precisely why pharmacological appetite and glucose control has become such a dominant strategy.

The GLP-1 Era: Efficacy, Economics, and What Gets Left Out

Glucagon-like peptide-1 receptor agonists work by mimicking an endogenous gut hormone that stimulates insulin secretion, suppresses glucagon, slows gastric emptying, and promotes satiety [8]. The clinical results have been genuinely impressive relative to anything that came before. In the STEP trial program, semaglutide produced average weight reductions of roughly 15 percent over 68 weeks, and in the SURMOUNT program, the dual GIP/GLP-1 agonist tirzepatide reached reductions of up to 21 to 22 percent over 72 weeks [8]. For conditions where previous medications delivered 5 to 10 percent at best, this was a step change.

The market has responded accordingly. Tirzepatide generated 24.8 billion dollars in revenue through the first nine months of 2025, making it, by that measure, the world’s best-selling drug despite a relatively recent approval [9]. One industry analysis projects that semaglutide and tirzepatide products combined could reach roughly 470 billion dollars in cumulative United States revenue by the end of 2030 [10]. Figures of this magnitude signal two things at once, enormous unmet demand and a powerful commercial incentive to keep patients on therapy indefinitely. Both are relevant context for why alternative and adjunctive approaches deserve research attention, not as competitors to effective medicine, but because no single drug class should be the only well-funded avenue for a problem this large.

GLP-1 receptor agonists such as semaglutide and tirzepatide have transformed obesity pharmacotherapy, but their mechanism, cost, and side-effect profile leave room for complementary research directions.

The Origin Question: Where the Active Ingredients Come From

A detail often lost in popular coverage is where these molecules actually come from, and it offers an instructive contrast with plant-derived compounds. The GLP-1 drug class traces its lineage to an unlikely source. In the early 1990s, researchers studying the saliva of the Gila monster identified a peptide, exendin-4, that closely resembled human GLP-1 but resisted the rapid enzymatic breakdown that makes the native human hormone impractical as a drug [11]. That discovery produced exenatide, the first GLP-1 receptor agonist, approved in 2005 as a synthetic version of the lizard peptide [11].

Modern agents have moved well beyond that template. Semaglutide is a synthetic analogue of human GLP-1, manufactured using recombinant DNA technology in yeast cells and then chemically modified, including a fatty acid side chain that extends its half-life to roughly one week [12]. It is, in other words, a bioengineered peptide with no remaining biological material from any natural source, designed for consistency, purity, and large-scale production [13]. Phytocannabinoids sit at the opposite end of the spectrum. They are produced by the cannabis plant itself, dozens of structurally related compounds appearing together in ratios that vary by cultivar. That difference, a single engineered molecule versus a naturally co-occurring chemical ensemble, turns out to be the crux of the UCR findings, and it raises a standardization challenge worth returning to.

The Documented Downsides of GLP-1 Therapy

GLP-1 therapy is effective and, for most patients, reasonably well tolerated. But an honest accounting of its limitations is necessary to understand why researchers keep looking for other levers, and these limitations are well documented rather than speculative.

Gastrointestinal effects are the most common. Nausea, vomiting, diarrhea, and constipation affect an estimated 40 to 50 percent of users, driven directly by the slowing of gastric emptying that also produces satiety [14]. The large majority of these events are mild to moderate and tend to ease after the dose-escalation period, but they are frequent enough that discontinuation due to adverse effects ranged from essentially zero up to 26 percent across trials, compared with up to 9 percent for placebo [15]. Rarer but more serious complications, including gastroparesis and, in 2023, an added warning for ileus, become less reassuring at population scale when millions of people take these drugs on an open-ended basis [16].

Loss of lean tissue is a second concern, and it calls for careful framing rather than alarm. Across randomized and controlled studies, GLP-1-induced weight loss reduces fat mass more than lean mass, but absolute lean body mass still declines, generally representing about 20 to 40 percent of total weight lost [17]. The clinical significance of this is genuinely contested. A 2025 review concluded that functional measures such as strength generally appear preserved, and that some markers of muscle quality may even improve, while stressing that longitudinal data integrating body composition with standardized strength testing remain limited [17]. Bone mineral density loss has likewise been flagged as a consequence of rapid weight reduction, with clinical guidance recommending resistance training, adequate protein, and monitoring of calcium and vitamin D status to mitigate both muscle and bone effects [18]. The substrate of the concern is real; the magnitude of its long-term consequences is still being worked out.

A third limitation is durability. These medications manage weight while they are taken, but they do not appear to reset the underlying physiology. Roughly half of patients discontinue within one year and about 70 percent within two, citing gastrointestinal effects, lean mass loss, injection burden, or cost [19]. On stopping, patients regain approximately two-thirds of lost weight within a year as the hormonal regulation of appetite and energy expenditure reasserts the prior weight set point [16]. This is less a failure of adherence than a feature of the mechanism, but it means the commercial projections above describe, in effect, a chronic-use model.

A fourth and more emerging signal sits at the intersection of the drugs’ central mechanism and the brain’s reward system, and it connects directly to appetite suppression itself. The same dampening of reward-related signaling that reduces what patients describe as “food noise” appears, in some users, to extend to other sources of pleasure. Clinicians and patients have begun describing a blunting of enjoyment, informally labeled “Ozempic personality,” that aligns with the clinical concept of anhedonia, a reduced capacity to experience pleasure or motivation [20]. Reporting in 2025 and 2026 documented case-level accounts of patients losing interest not only in food but in hobbies, socializing, and other previously rewarding activities [21]. The proposed mechanism is that chronic GLP-1 receptor activation blunts phasic dopamine signaling in reward pathways, an effect that may be functional rather than structural and that some clinicians distinguish from clinical depression [22]. Precision matters here. This is anecdotal and emerging, not an established adverse effect, and nobody yet knows whether it is dose-related, temporary, or limited to a subset of users [23]. But the mechanistic plausibility, the idea that you cannot fully separate the suppression of food reward from the suppression of reward more broadly, is exactly why it warrants mention alongside the better-established effects. Many patients, it should be said, report improved well-being on these drugs. The concern is heterogeneity of response, not a uniform outcome.

The Cannabis Metabolic Paradox

Against that backdrop, the cannabis observation is intriguing precisely because it runs in the opposite direction from what the plant’s acute pharmacology would predict. Cannabis at low doses reliably stimulates appetite through a CB1 receptor mechanism, the familiar “munchies” effect [1]. Yet a substantial body of human epidemiological work has found that chronic cannabis users develop type 2 diabetes and obesity at lower rates than non-users. The UCR authors anchor this claim in more than a dozen human studies, noting that cannabis users are roughly half as likely to develop type 2 diabetes and that the inverse association with obesity is most pronounced among daily users [1].

The most rigorous human data point in that literature is a randomized, double-blind, placebo-controlled trial published in Diabetes Care in 2016, in which 62 patients with non-insulin-treated type 2 diabetes received one of several cannabinoid regimens or placebo over 13 weeks [24]. That trial is worth citing carefully, because its primary endpoint was a change in HDL cholesterol, with glycemic measures as secondary and tertiary endpoints. The improvements in glucose control and pancreatic beta-cell function attributed to tetrahydrocannabivarin were secondary findings from a pilot study, not the headline result [24]. Read with that caveat, the human literature establishes an association and some preliminary interventional signal, but it does not establish mechanism or causation. That is the gap the UCR study set out to address in a controlled preclinical model.

The UCR Study: What It Actually Found

The study, led by Avalos and colleagues, used a mouse model designed to approximate human dietary patterns [1]. Male mice were placed on a high-fat, high-sucrose Western-style diet or a low-fat control diet for 60 days. Beginning at day 30, the diet-induced obesity animals received either pure delta-9 THC at 5 milligrams per kilogram or a whole-cannabis oil extract matched for the same THC content, administered daily for 30 days. That matched-THC design is the methodological heart of the work, because it isolates the contribution of everything in the extract other than THC.

The central result is a dissociation. Both THC alone and the whole extract significantly reduced body weight and fat mass in the obese mice relative to vehicle-treated controls, and they did so without significant changes in daily motor activity and despite largely sustained food intake after an initial transient dip [1]. On weight, in other words, THC and extract performed similarly. But on glucose regulation they diverged sharply. The whole extract, but not THC alone, normalized glucose clearance in the obese mice to levels found in lean animals, as measured by glucose tolerance testing, and improved insulin sensitivity [1]. Mice given THC alone lost weight yet continued to show impaired glucose homeostasis, a hallmark of the diabetic state.

The proposed mechanism centers on what the authors call the adipoinsular axis, the signaling network through which fat tissue and the pancreas communicate to regulate insulin and glucose [1]. Diet-induced obesity dysregulates this axis. The study found that both treatments reversed obesity-associated changes in the expression of key adipokines, the signaling molecules secreted by fat, but that the extract normalized this adipokine expression more effectively than THC alone [1]. Both treatments also appeared to restore elements of endocannabinoid system tone that are disrupted in obesity, lowering the elevated tissue levels of the endocannabinoids 2-AG and anandamide seen in the obese state [1].

One finding complicates any tidy mechanistic story and deserves to sit alongside the headline result. In cultured fat cells, blocking the CB1 and CB2 cannabinoid receptors with antagonists did not reverse the cannabinoids’ effect on lipid accumulation; if anything, it further reduced it [1]. The authors interpret this as evidence that chronic cannabinoid exposure downregulates receptor signaling over time, producing outcomes that resemble those of receptor blockade, an effect consistent with the reduced CB1 receptor expression they observed both in tissue and in culture [1]. The practical implication is that the mechanism is not a simple matter of receptor activation, and that the relevant pathways are not yet fully mapped.

The study centers on the adipoinsular axis, the signaling network between fat tissue and the pancreas. Whole-plant extract normalized adipokine expression more effectively than isolated THC.

Why the Extract-Versus-Isolate Distinction Matters

The most consequential element of the study for the broader field is not the weight loss, which THC produced on its own, but the glucose effect that only the whole extract delivered. The authors are measured about why. They suggest the difference may reflect the contribution of non-THC cannabinoids and other bioactive compounds in the extract, noting that minor phytocannabinoids such as CBD and tetrahydrocannabivarin have been reported to improve glucose homeostasis without affecting body weight, and that the enhanced effect of extract over THC alone “may therefore reflect combinatorial or synergistic interactions among multiple cannabinoids” [1]. They explicitly frame dissecting the relative contribution of individual phytocannabinoids as a direction for future work, rather than claiming the interaction as established.

This is the appropriate register, and it is worth preserving. The idea that the constituents of whole-plant cannabis act together to produce effects that isolated compounds do not, often called the entourage effect, remains a hypothesis with preliminary support rather than a settled principle. The UCR data add a clean, controlled mechanistic data point to that hypothesis, because the only variable distinguishing the two active treatments was the presence or absence of the plant’s other compounds. For a fuller treatment of how cannabis constituents interact with the body’s signaling systems, and what is and is not established about those interactions, see Hytiva’s overview, How Does Cannabis Affect the Body and Brain?.

There is a hard practical corollary. A finding that depends on the composition of a whole-plant extract is only translatable if that composition can be reproduced. The cannabis used in this study was a single, chemically characterized batch quantified by mass spectrometry, whereas real-world cannabis varies widely in cannabinoid ratios from cultivar to cultivar and even batch to batch. The authors themselves stress that future studies must “consider and accurately report the specific phytocannabinoid composition” because different strains could have different metabolic effects [1]. In other words, the extract-versus-isolate result makes standardization a precondition for any serious clinical translation, not an afterthought. Hytiva, whose research program follows developments in cannabis therapeutics closely, has examined why this matters for research integrity in Cannabis Product Standardization and Quality: Why Consistency Matters for Research, and the present findings are a concrete illustration of the principle. Without reproducible composition, an extract effect cannot become a therapy.

The Honest Gap: What Would Have to Be True for Clinical Translation

Given the GLP-1 limitations catalogued above, it is tempting to frame cannabinoids as a gentler alternative. The recent findings do offer a genuine reason to investigate that possibility, but the responsible reading treats the safety comparison as an open question rather than a settled advantage, and several gaps stand between this study and any clinical claim.

The study used only male mice. The authors are direct that sex differences in both endocannabinoid signaling and metabolic regulation are well documented, that cannabinoid receptor expression and signaling can differ substantially between males and females, and that female cohorts will be necessary to determine whether these effects generalize [1]. No direct measurement of whole-body energy expenditure was performed, which led the authors to label their interpretations regarding energy balance as “hypothesis-generating” and to call for metabolic-cage studies [1]. The animals were housed below their thermoneutral zone, a condition that alters baseline energy expenditure and may have influenced the magnitude of the responses observed [1]. The highest concentration used in the cell-culture experiments exceeded the blood levels reported in mice given the same dose, another limitation the authors flag [1]. And the receptor-antagonist result described earlier means the mechanism itself is not fully resolved.

Beyond the study’s internal limitations lies the larger question its popular framing tends to skip, which is that cannabis is not free of downsides. Chronic use carries documented risks, including the potential for dependence and cognitive and respiratory effects depending on the route of administration, alongside a regulatory status that complicates both research and clinical use. Delta-9 THC in particular is the intoxicating constituent of the plant, and any therapeutic application would have to contend with its psychoactivity, a consideration the UCR team addresses directly by stating that their forward plan is to identify non-intoxicating compounds that might deliver metabolic benefit without THC’s effects [1]. Readers seeking background on that specific compound can consult Hytiva’s primer, What Is Delta-9 THC?. The point is not that cannabis is dangerous where GLP-1 drugs are safe, nor the reverse. It is that the comparative safety and tolerability of a hypothetical cannabinoid-based metabolic intervention against an established drug class is simply unknown, and presenting it as a foregone conclusion would repeat exactly the kind of overclaiming that good cannabis science needs to avoid.

Conclusion: A Research Agenda, Not a Recommendation

What the UCR study earns is a place on the research agenda, not a place in the medicine cabinet. It converts a long-observed epidemiological curiosity into a controlled mechanistic observation, and it does so in a way that sharpens the most interesting question in the field, namely whether the whole plant does something its most famous single molecule cannot. The dissociation between weight loss, which THC achieved alone, and glucose normalization, which required the full extract, is a clean result that points toward the minor cannabinoids and their interactions as the place to look next.

A credible path forward would mean isolating and testing individual non-intoxicating compounds to identify which carry the glucose benefit, replicating the work in female animals and at thermoneutral conditions, adding direct energy-expenditure measurement, and, critically, building all of it on standardized, compositionally reproducible material. Only after that preclinical groundwork would carefully designed human trials be warranted. The lead investigator’s own framing is the right note to end on. The appropriate stance is not to use cannabis to manage weight or diabetes, but for clinicians, researchers, and policymakers to pay attention to this space and pursue evidence-based approaches to understanding both the risks and the potential benefits of cannabis and its components [1]. The science here is a beginning, and it should be described as one.

---