What the Latest Physiology Research Reveals About Cannabis Formulation and Diabetes Risk

In diet-induced obese mice, both purified delta-9-tetrahydrocannabinol and whole cannabis extracts reduced body weight and fat mass, and both reversed obesity-associated changes in the expression of adipokines that regulate signaling between fat tissue and the pancreas.

The extracts did two things THC alone did not. They normalized adipokine expression more effectively, and they restored glucose clearance in obese mice to the level seen in lean animals. Losing fat and regaining glucose control turned out to be separable outcomes.

The most common metabolic question patients ask about cannabis is whether it makes you gain or lose weight. That framing assumes weight is the outcome that matters. In metabolic disease, glucose handling is frequently the outcome that determines risk, and this study shows the two can move independently under the same drug.

It also puts a testable mechanism behind an epidemiological oddity that has sat unexplained for over a decade. Population surveys have repeatedly found lower obesity prevalence among frequent cannabis users despite cannabis being famous for stimulating appetite. A mouse study cannot resolve that, but it can propose where to look.

Bryant Avalos, Martin Olmos, Courtney Wood, Camila Alvarez, Haley Read, Parima Udompholkul, Theodore Garland, and Nicholas DiPatrizio of the University of California, Riverside published the study in The Journal of Physiology on May 11, 2026. Garland and DiPatrizio are affiliated with the UCR Center for Cannabinoid Research.

Male mice were fed either a high-fat, high-sucrose diet or a low-fat, no-sucrose diet for 60 days. At day 30, animals received either THC at 5 mg per kilogram or cannabis extracts matched for THC content, given daily for the following 30 days. The team measured adipocyte biology, glucose tolerance, insulin sensitivity, endocannabinoid system expression, body weight, food intake, and motor activity.

A parallel set of experiments used 3T3-L1 cells, a standard cultured preadipocyte line, to examine what the compounds do to fat cell development and to cellular energy metabolism directly, without the confounds of appetite, activity, and whole-animal physiology.

THC and the extracts both reduced body weight and fat mass in the diet-induced obese animals. On that measure the two interventions were not meaningfully different, which is itself worth noting given how much attention THC receives as the plant’s active molecule.

Both interventions also reversed obesity-associated changes in the expression of adipokines, the signaling proteins fat tissue releases to communicate with the pancreas and other organs. The extracts did this more effectively than THC alone, a difference of degree rather than of kind.

The categorical difference appeared in glucose clearance. Extracts, but not THC, normalized glucose clearance in obese mice to levels found in lean mice. An animal can therefore lose fat on THC and still handle sugar like an obese animal. That is the result that would change how a clinician thinks about formulation if it holds up in people.

The adipoinsular axis describes the two-way signaling relationship between adipose tissue and the insulin-producing beta cells of the pancreas. Fat tissue releases adipokines, including leptin and adiponectin, that influence how much insulin the pancreas secretes. The pancreas, through insulin, influences how fat tissue stores energy.

In diet-induced obesity this loop degrades. Adipokine output shifts, the pancreas receives distorted information, and insulin secretion and sensitivity drift out of alignment. That drift precedes and contributes to type 2 diabetes, which makes it a mechanistically sensible place to intervene rather than a convenient label applied after the fact.

The endocannabinoid system is embedded in both ends of that loop. CB1 receptors are expressed in adipose tissue and in pancreatic islets as well as in liver, gut, and brain. That is why cannabinoid effects on this axis are biologically plausible rather than surprising, and it is also why a single receptor-targeting molecule may behave differently from a mixture acting at several points.

In 3T3-L1 cells, both THC and the extracts produced anti-adipogenic effects and altered cellular energy metabolism in a concentration-dependent manner. Anti-adipogenic means the compounds interfered with the process by which preadipocytes mature into fat-storing cells.

That matters because it removes some obvious alternative explanations. A whole animal that eats less or moves more will lose fat for reasons having nothing to do with the fat cell itself. A cultured cell line has no appetite and no exercise wheel, so an effect observed there points toward direct action on adipocyte biology.

Concentration dependence is also a useful signal. An effect that scales with dose in a controlled system behaves more like pharmacology and less like an artifact, though cultured cells are exposed to concentrations that do not map cleanly onto what reaches human adipose tissue after an oral or inhaled dose.

The epidemiological backdrop is real but limited. Yann Le Strat and Bernard Le Foll, writing in the American Journal of Epidemiology in 2011, analyzed two nationally representative US surveys and found adjusted obesity prevalence of 22.0% and 25.3% among adults reporting no cannabis use in the past year, compared with 14.3% and 17.2% among those using at least three days per week. Those are cross-sectional associations, not evidence that cannabis causes lower body weight.

The closest human interventional data come from a different cannabinoid. Khalid Jadoon and colleagues, in Diabetes Care in 2016, randomized 62 adults with non-insulin-treated type 2 diabetes across five arms for 13 weeks. Tetrahydrocannabivarin at 5 mg twice daily lowered fasting plasma glucose by an estimated treatment difference of 1.2 mmol/L and improved a measure of pancreatic beta cell function. The trial’s own primary endpoint, a change in HDL cholesterol, was not met, and the combination arms showed no significant effect.

Neither of those supports a clinical recommendation. Together they establish that the question is worth asking properly in humans, which is a lower bar than it sounds and one this field has rarely cleared.

As preclinical work this is well constructed. Matching the extracts to THC content is the design decision that makes the comparison meaningful, because it isolates the contribution of everything in the plant other than THC rather than comparing two different THC doses. Pairing whole-animal physiology with a cultured adipocyte line adds a mechanistic layer that a mouse study alone would lack.

As evidence about people it carries no weight yet, and that is not a criticism of the authors. Mice and humans differ in cannabinoid pharmacokinetics, metabolite profile, and the physiology of diet-induced obesity. The animals were male only, which leaves half the relevant biology unexamined in a field where sex differences in adipose tissue and endocannabinoid signaling are well documented.

The extracts are not chemically specified in a way that lets anyone reproduce the active ingredient. Cannabis extract is a category, not a compound. Until the responsible constituents are identified, the finding cannot be translated into a product, a dose, or a recommendation, and the authors present the identification of those constituents as work still to be done.

Male mice only. Adipose tissue distribution, adipokine secretion, and endocannabinoid tone differ by sex in both rodents and humans, and metabolic interventions have a long record of behaving differently in females. A single-sex design in metabolic research is a real limitation rather than a formality.

The 3T3-L1 concentration-response work uses exposures applied directly to cultured cells. Those concentrations do not translate simply to what circulates in a person after an oral or inhaled dose, and a concentration-dependent effect in a dish is not a dose-response curve for a patient.

This study does not show that cannabis improves blood sugar in humans. No human was studied. Extrapolating a normalized glucose clearance curve in mice to a hemoglobin A1c target in a patient is not supported by anything in this paper.

It does not identify which non-THC compounds are responsible for the difference between the extracts and purified THC. Without that, no product on any shelf can be matched to the finding, and a label reading full spectrum carries no assurance that it contains whatever did the work here.

It does not test cannabis as a treatment for obesity or type 2 diabetes, and it does not address interactions with the medications most patients with those conditions are taking. Nothing here belongs in a treatment plan.

Targeting the endocannabinoid system for metabolic disease is not a new idea, and its history counsels humility. Rimonabant, a CB1 receptor blocker, produced meaningful weight loss and metabolic improvement in humans, was approved in Europe, and was withdrawn there in 2008 after psychiatric adverse effects including depression and suicidality emerged. It was never approved in the United States. The mechanism was real and the drug was unusable.

Research since has moved toward approaches that avoid central nervous system exposure or work through endogenous cannabinoid tone rather than blunt receptor blockade. Plant compounds that modulate the system without producing intoxication sit in that category, which is part of why a finding that non-THC constituents carry metabolic activity is interesting beyond the cannabis field.

The practical reading for a clinician is narrow and worth stating plainly. This work does not change what anyone should do on Monday. It does argue that the whole-plant versus isolate question, often treated as a marketing debate, has a legitimate pharmacological version that is testable and has not been tested in humans.

The part of this study I find genuinely useful is the separation of weight loss from glucose control. Patients and clinicians both tend to treat those as the same outcome, and they are not. An intervention that reduces fat mass while leaving glucose handling impaired is doing less than it appears to be doing, and that is a distinction worth carrying into any conversation about metabolic treatment.

I want to be careful about the second part. It is tempting to read this as vindication for whole-plant preparations, and I have argued for taking formulation seriously for a long time. But this is a mouse study with an unspecified extract, in male animals only, with no human data behind it. That is a hypothesis worth funding, not a reason to choose a product.

What I tell patients in the metabolic program has not changed. If you are using cannabis, tell me, because it interacts with medications and it is relevant to how I interpret your labs. If you are asking whether to start cannabis for blood sugar, the answer is that the human evidence does not exist. This study is a reason to run the trial, not a substitute for having run it.

In obese mice, purified THC and whole cannabis extract both reduced fat mass, but only the extract restored normal glucose clearance. Treat that as a mechanistic hypothesis about formulation, not as guidance. No human trial has tested a whole-plant cannabis preparation for glucose control, and nothing here should change a metabolic treatment plan.

The design feature to notice is that the extracts were matched to THC content, which is what makes the comparison interpretable. The finding to carry forward is that fat loss and glucose control came apart under the same class of drug. The findings not to carry forward are any claim about humans, any claim about a specific product, and any implication that the responsible compounds have been identified. They have not.