THC, CBD, and Cytokine Storms: A Strong Laboratory Signal and One Negative Human Trial

The laboratory evidence that cannabinoids dampen cytokine production is genuine and predates the pandemic by more than a decade. It comes from splenocytes, human peripheral blood mononuclear cells, and mouse models of systemic and pulmonary inflammation.

According to PubMed, the one randomized, double-blind, placebo-controlled trial of cannabidiol in COVID-19 patients found no effect. Daily administration of 300 mg for 14 days failed to alter clinical evolution, and by day 28 symptom resolution was 83.3 percent with cannabidiol and 90.2 percent with placebo.

Cytokine storm became a household phrase in 2020, and a large volume of cannabinoid commentary was written in the months that followed, most of it extrapolating from preclinical immunology to a disease nobody had yet tested it in. That commentary is still circulating, largely unrevised.

Patients with autoimmune disease, chronic inflammatory conditions, and post-viral syndromes read it and reasonably ask whether cannabinoids will calm an overactive immune system. The answer requires distinguishing a measurable effect on cytokine output in a laboratory from a demonstrated benefit to a sick person, and those are not close to the same thing.

Cytokine release syndrome, commonly called cytokine storm, is a dysregulated surge of inflammatory mediators that can cause organ injury and death. It occurs in severe infection, in sepsis, in some autoimmune flares, and as a toxicity of certain immunotherapies. The mediators most often named are tumor necrosis factor alpha, interleukin-6, and interleukin-1 beta.

According to PubMed, the case that cannabinoids modulate this system rests on mechanistic work that long predates COVID-19. A 2009 review in Future Medicinal Chemistry by Nagarkatti and colleagues at the University of South Carolina summarized evidence that CB1 and CB2 receptors are present on immune cells, that administering THC to mice triggered apoptosis in T cells and dendritic cells producing immunosuppression, and that cannabinoids downregulate cytokine and chemokine production and in some models upregulate regulatory T cells.

A mechanism specific to cannabidiol was described the year before. According to PubMed, Kaplan, Springs, and Kaminski reported in Biochemical Pharmacology in 2008 that cannabidiol suppressed interleukin-2 and interferon gamma production in activated murine splenocytes and purified T cells, and suppressed the transcriptional activity of activator protein 1 and nuclear factor of activated T cells. Using splenocytes from mice lacking both CB1 and CB2, the authors showed this suppression occurred independently of either receptor.

Two studies published in 2022 represent the strongest preclinical work on cannabinoids in cytokine storm specifically.

According to PubMed, Maayah and colleagues at the University of Alberta and Qatar University used a lipopolysaccharide-induced sepsis model, giving mice cannabidiol at 15 mg/kg or vehicle one hour before a single 10 mg/kg lipopolysaccharide injection, then assessing them at 24 hours. Cannabidiol improved cardiac function and reduced renal injury, and reduced both systemic and renal inflammation. The authors proposed that cannabidiol could be repurposed to reduce morbidity in cytokine storm, particularly in sepsis.

According to PubMed, Aswad and colleagues at the Rambam Health Care Campus in Haifa compared high-THC and high-CBD extracts and found significant differences between them, then focused on one high-CBD extract. It reduced proinflammatory cytokines in human peripheral blood mononuclear cells, neutrophils, and T cells; in systemically inflamed mice it reduced tumor necrosis factor alpha and interleukin-1 beta while raising the anti-inflammatory cytokine interleukin-10; it impaired T-cell migration toward the chemoattractant SDF1 and reduced phosphorylation of the T-cell receptor signaling proteins Lck and Zap70; and in a lung-inflamed mouse model it reduced leukocyte migration into the lungs along with interleukin-1 beta, MCP-1, interleukin-6, and tumor necrosis factor alpha.

That is a coherent and well executed body of work. Every result in it comes from cells in culture or from mice.

According to PubMed, the CANDIDATE study is the randomized test the preclinical literature called for. Crippa and colleagues at the Ribeirao Preto Medical School of the University of Sao Paulo ran a randomized, parallel-group, double-blind, placebo-controlled trial at two Brazilian sites between July 7 and October 16, 2020, recruiting patients from an emergency room. Raphael Mechoulam was among the authors.

Of 321 patients assessed for eligibility, 105 were randomized, 49 to cannabidiol and 42 to placebo as reported, with 91 participants included in the efficacy analysis. Patients received 300 mg of cannabidiol or placebo added to standard symptomatic care for 14 days. The primary outcome was prevention of deterioration from mild or moderate to severe or critical illness, or the natural course of symptom resolution, assessed at days 14, 21, and 28.

The result was null. Median time to symptom resolution was 12 days with cannabidiol and 9 days with placebo, a difference that did not reach significance by log-rank test. By day 28, 83.3 percent in the cannabidiol group and 90.2 percent in the placebo group had resolved symptoms. There were no between-group differences on secondary measures. Cannabidiol was well tolerated, with mostly mild and transient side effects. The authors’ own conclusion was that daily administration of 300 mg for 14 days failed to alter the clinical evolution of COVID-19, and they suggested future trials explore severe disease at possibly higher doses.

A related line of research is frequently merged into the cytokine storm claim and should be kept separate. According to PubMed, Nguyen and colleagues at the University of Chicago reported in Science Advances in 2022 that cannabidiol and its metabolite 7-OH-CBD, but not THC or other cannabinoids tested, blocked SARS-CoV-2 replication in lung epithelial cells and in mice, acting after viral entry, in part by upregulating the host IRE1-alpha endoplasmic reticulum stress response and interferon signaling. In matched patient groups from the National COVID Cohort Collaborative, prescribed cannabidiol oral solution showed a negative association with positive SARS-CoV-2 tests.

That is an antiviral mechanism, not an anti-cytokine one, and the authors were unusually direct about the limits of their own work. They described cannabidiol as a potential preventative agent meriting future clinical trials, and explicitly cautioned against the use of non-medical formulations including edibles, inhalants, and topicals as prevention or treatment.

Note also the dose gap. The cannabidiol oral solution associated with the epidemiologic signal is a pharmaceutical product dosed by weight for seizure disorders. Retail cannabidiol products deliver a small fraction of that, with content that varies between batches.

The field did not stall after the negative trial, but it did shift away from COVID-19. According to PubMed, a 2025 study in Biomedicine and Pharmacotherapy from the Universidade Federal de Minas Gerais stimulated human peripheral blood mononuclear cells with inactivated chikungunya virus and treated them with cannabidiol, reporting that it altered natural killer cell and monocyte subsets and attenuated inflammatory pathways, particularly the interleukin-17A and interferon gamma axis. The authors framed cannabidiol as a candidate for further immunomodulatory investigation in alphavirus disease.

That is the appropriate place for this research to sit: identifying which inflammatory pathways cannabidiol touches, in which cells, at which concentrations, so that a future trial can be designed around a specific mechanism and a specific population rather than a general hope that an anti-inflammatory compound will help a severe illness.

It also illustrates the correction this page exists to make. The interesting question was never whether cannabinoids affect cytokine production. They do. The question is whether that effect is large enough, fast enough, and targeted enough to change what happens to a patient, and the only randomized answer so far is no.

The preclinical evidence is strong for what it measures. Multiple independent groups, across mouse and human cell systems and two decades, report that cannabinoids reduce proinflammatory cytokine output, and at least one receptor-independent mechanism for cannabidiol has been identified using CB1 and CB2 double knockout cells. That is more mechanistic consistency than many repurposing candidates ever accumulate.

The clinical evidence is one randomized trial and it was negative. A single trial of 300 mg daily in mild to moderate COVID-19 does not close the question for severe disease, for other causes of cytokine storm, or for higher doses. It does mean that anyone asserting a clinical benefit is asserting something that has been tested once and not found.

The CANDIDATE trial has real limitations that should not be used to dismiss it and should not be ignored either. It enrolled 105 patients against a target informed by a much larger assumed effect, studied mild to moderate rather than severe illness, and used 300 mg daily, a dose the investigators themselves suggested may have been too low. Patients were recruited from an emergency room in 2020 under pandemic conditions.

On the preclinical side, the mouse sepsis model used a single lipopolysaccharide injection with cannabidiol given an hour beforehand, which tests prevention of an induced inflammatory surge rather than treatment of an established one. Cannabis extract studies also introduce composition variability, and the Israeli group found meaningful differences between high-THC and high-CBD extracts, which means results attach to specific preparations rather than to cannabinoids as a class.

None of this work shows that cannabis, cannabidiol, or THC prevents, treats, or reduces mortality from cytokine storm in humans. It does not identify a dose, a route, a timing window, or a patient population, and it does not establish that reducing measured cytokine levels improves outcomes even when it occurs.

It also does not show that cannabinoid immunosuppression is desirable. Suppressing T-cell function during an active infection is a double-edged intervention, and the same mechanistic literature that supports an anti-inflammatory effect also documents immunosuppression as a potential harm in people who need an intact immune response.

This topic is a clean case study in how an early-pandemic hypothesis hardens into folk knowledge. The mechanistic groundwork was legitimate and predated 2020. What followed was a wave of review articles and commentary proposing cannabinoids for COVID-19, written before any trial existed, and that layer of secondary literature is what most people encounter.

The trial that tested the hypothesis was published in October 2021 and is cited far less often than the speculative reviews that preceded it. When a page claims cannabinoids calm cytokine storms without mentioning CANDIDATE, that is usually the reason.

I remember the enthusiasm in 2020, and I understand where it came from. The immunology was real, the need was urgent, and cannabidiol was already sitting on pharmacy shelves as an approved drug. Proposing it was not unreasonable. Continuing to promote it after a randomized trial found nothing is a different matter.

In practice the patients who bring this to me are not worried about COVID-19 anymore. They have rheumatoid arthritis, or long COVID, or an inflammatory bowel condition, and they have read that cannabinoids quiet an overactive immune system. What I tell them is that cannabinoids demonstrably change cytokine output in a laboratory, that this has not yet translated into a demonstrated clinical benefit in any inflammatory condition tested with a control group, and that immunosuppression is not automatically a good thing when you are fighting something.

That is not a no. It is an accurate description of where the science sits, and most patients find it more useful than a promise. If someone is using cannabis for pain or sleep alongside a rheumatologic regimen, that is a reasonable conversation. If they are hoping it will replace a biologic, it will not.

Cannabinoids reduce inflammatory cytokine production in cells and in animals, through mechanisms including receptor-independent suppression of NFAT and AP-1 by cannabidiol. The single randomized placebo-controlled trial of cannabidiol in COVID-19 patients found no effect on clinical course. Anyone using cannabinoids for an inflammatory condition should treat this as an open research question rather than an established therapy, and should not substitute them for disease-modifying treatment.

The sentence worth carrying forward is that measurable cytokine suppression and clinical benefit are separate claims requiring separate evidence, and only the first has been demonstrated. The sentence worth discarding is any version of cannabinoids treat cytokine storms, which describes a hypothesis that was tested once in a randomized trial and did not hold.