There is real clinical evidence that cannabis helps some cancer patients feel better, no adequate human evidence that it treats the cancer itself, and confusing those two things has cost people their lives.
1. Two Questions That Get Blurred Into One
Question A — symptoms. Does cannabis help cancer symptoms and treatment side effects? There is real clinical evidence here, of varying strength.
Question B — the disease. Does cannabis act on the tumour itself? There is a large laboratory literature, some animal work, two very small early-phase human studies, and no adequate randomised evidence in people.
Answering A and letting readers carry the answer to B is how people get hurt. Nothing below claims cannabis cures, treats or shrinks cancer in a person; where a claim is widely repeated but unsourced, we say so.
2. How to Read Medical Evidence Without a Medical Degree
| Tier | What it tells you |
|---|---|
| Cell culture | A mechanism exists in a dish, not in a person. |
| Animal model | It can work in a living body, not that it works in humans. |
| Case report | Something happened once, not that it was caused. |
| Case series | Patterns to investigate, not cause and effect. |
| Observational study | Associations and safety signals, not causation. |
| Randomised controlled trial | Whether the treatment works. |
| Systematic review | The state of knowledge across trials. |
Why a dish result is the weakest basis for a decision
Bleach kills cancer cells in a dish. A dish has no liver, no bloodstream to dilute the compound, no immune system, no healthy tissue to damage. It opens a research question; it does not close one.
Six Questions for Any “Kills Cancer Cells” Headline
| Question | Why it matters |
|---|---|
| In what? Cells in a dish, animals or people? | Most compounds that work in a dish fail in animals, and most that work in animals fail in people |
| At what concentration, and can a human reach it? | Compare the study concentration with the blood levels people actually reach |
| Was the tested substance the product being sold? | A purified compound, a designed drug and an unrefined plant oil are different things |
| What did the paper conclude, in its own words? | “Holds therapeutic potential” is not “is effective”; the hedging is usually what gets cut |
| What were the negative findings? | Reviews also report unfavourable results, and those are rarely quoted |
| Where is the human trial? | If a mechanism has been known for decades and there is still no adequately powered randomised trial, that absence is information |
3. Sickness and Vomiting from Chemotherapy: The Strongest Case
Chemotherapy-induced nausea and vomiting is the one indication with a respectable evidence base. Two synthetic cannabinoid medicines are licensed: dronabinol, synthetic delta-9-tetrahydrocannabinol (THC), and nabilone, a THC analogue. Both carry United States Food and Drug Administration approval for this use; that agency has never approved cannabis itself for any condition.
A 2015 Cochrane review by Smith and colleagues pooled 23 randomised trials run from 1975 to 1991. Cannabinoids beat placebo for complete absence of vomiting, but Cochrane rated the evidence low to moderate quality, noted the trials predate modern antiemetics of the ondansetron class, and found dizziness, sedation and dysphoria far more common than with comparators. The National Academies rated the evidence for oral cannabinoids as antiemetics conclusive or substantial in 2017.
Where these medicines sit in guidelines
Second or third in line. In the National Comprehensive Cancer Network antiemesis guideline they sit in the breakthrough section, used after initial prophylaxis fails. The American Society of Clinical Oncology’s 2024 cannabis guideline says they “may improve refractory, chemotherapy-induced nausea and vomiting when added to guideline-concordant antiemetic regimens.” Refractory. Added to. Not instead of.
4. Cancer Pain: Mixed, and Largely Disappointing
An early phase 2 trial of a mouth spray combining THC and cannabidiol reported benefit in opioid-refractory cancer pain; larger later trials did not reproduce it. The 2023 Cochrane review by Häuser and colleagues pooled 14 trials and 1,823 participants and found moderate-certainty evidence that the nabiximols spray and THC are not effective here: 32 in 100 improved against 23 in 100 on placebo. The supportive care association’s 2023 guideline recommends against cannabinoids as an adjuvant analgesic in cancer pain.
5. Appetite, Weight Loss and Wasting
In a 2002 North Central Cancer Treatment Group trial (Jatoi and colleagues), 469 adults with advanced cancer and weight loss were randomised to megestrol acetate, oral THC or both. As the National Cancer Institute summarises it, megestrol produced a 75% rise in appetite and 11% in weight, against 49% and 3% on oral THC. A 2006 phase III trial from the Cannabis-In-Cachexia-Study-Group (Strasser and colleagues) randomised 243 patients to cannabis extract, THC or placebo and found no difference in appetite, weight or quality of life. The Oncology Nursing Society lists cannabis for cancer anorexia as not recommended for practice.
6. Sleep, Anxiety and Overall Symptom Burden
The Multinational Association of Supportive Care in Cancer reviewed cannabis for insomnia, anxiety and depression in cancer and concluded in 2023 that no high-quality evidence supports it: only 6 of 15 randomised trials suggested benefit, none with psychological symptoms as a primary outcome. Cannabis can also worsen anxiety in some people. In the MedCan1 trial (Hardy, Journal of Clinical Oncology, 2022), cannabidiol oil in 142 advanced cancer patients gave a symptom-distress response of 44.8% against 58.7% on placebo.
7. Evidence Strength Summarised by Indication
| Indication | Honest rating |
|---|---|
| Refractory sickness from chemotherapy | Moderate; add-on only |
| Cancer pain | Moderate certainty of no benefit |
| Appetite and weight loss | Weak to negative |
| Sleep | Insufficient |
| Anxiety and mood | Insufficient; can worsen anxiety |
| Treating the cancer itself | Insufficient; not a treatment |
8. What Cannabinoids Do to Cancer Cells in the Laboratory
Ramer and Hinz reviewed this in the British Journal of Cancer in 2022. In cell lines, cannabinoids induce apoptosis — programmed cell death — through ceramide accumulation and endoplasmic reticulum stress; are antiproliferative via Akt inhibition and cell-cycle arrest; and suppress angiogenesis, the blood supply a tumour needs.
The concentration problem, the crux of the argument
That review notes most studies showing pro-apoptotic effects used concentrations above 1 micromolar, while peak plasma concentrations of THC after inhaled or oral administration do not exceed 1 micromolar. The concentrations that kill cancer cells in a dish are ones a human body does not reach by consuming cannabis, and taking more does not help: absorption and tolerability impose ceilings. That is why researchers have tried delivering cannabinoids straight into a tumour. The review also records opposing effects: in human papillomavirus-positive head and neck models, submicromolar concentrations promoted proliferation.
Two More Gaps Between the Dish and the Patient
Serum and plastic. Many laboratory studies of cannabinoids on cancer cells are run in culture medium with little or no serum. A 2024 Penn State study tested this habit directly (Carkaci-Salli and colleagues, “Cannabinoids as Potential Cancer Therapeutics: The Concentration Conundrum”, Cannabis and Cannabinoid Research 9(4)). Four cannabinoids were tested at 10 micromolar on six cancer cell lines. Three of them, THC, cannabidiol and KM-233, worked better when serum was left out of the medium, and IC50 values were lower without serum in all cell lines. Without serum, THC and cannabidiol also bound heavily to the plastic of the dish. The authors note that cannabinoids are about 90 to 95% bound to proteins. In the body, only the small unbound fraction is available to act, so a common laboratory choice makes cannabinoids look more potent than they can be in a patient.
Immunity. The popular argument says cannabis helps the immune system fight cancer. At least one line of evidence points the other way: a 2022 study in Signal Transduction and Targeted Therapy reported that cannabis suppressed antitumour immunity by inhibiting JAK/STAT signalling in T cells through CNR2, the gene for the CB2 receptor. This matters more now that immunotherapy is central to treating many cancers, and it is one more reason to tell the oncology team about any cannabis use.
9. The Human Trial Record on the Cancer Itself
The brain tumour pilot study
In 2006, Guzmán and colleagues published a pilot study in the British Journal of Cancer in nine patients with recurrent glioblastoma. THC went straight into the tumour resection cavity through an implanted catheter, because the bloodstream cannot deliver the required concentrations. Median survival from relapse surgery was 24 weeks; the National Cancer Institute’s summary is blunt: no significant clinical benefit.
The later trial, and what it did and did not show
In 2021, Twelves and colleagues reported a phase 1b placebo-controlled trial of nabiximols spray added to temozolomide in recurrent glioblastoma, with safety the primary endpoint; part 2 randomised 21 patients, 12 to nabiximols, 9 to placebo.
The number that travels around the internet is one-year survival of 83% with nabiximols against 44% with placebo. The context rarely travels with it: 21 people, no power to detect a survival difference, two placebo patients dead within 40 days, and six-month progression-free survival identical at 33%. The authors concluded only that this justifies an adequately powered trial, and an editorial in the same journal warned against imprudent off-label use. That trial, ARISTOCRAT, is now running: temozolomide with or without cannabinoids in recurrent glioblastoma, protocol in BMC Cancer, 2024.
10. High-Dose Cannabis Oil Protocols
Much of the traffic here concerns concentrated whole-plant cannabis oil taken in escalating amounts over a fixed period, popularised under the name of the Canadian man who publicised it in the late 2000s. This article gives no amounts, no schedule and no method: Thai law prohibits it, and we will not print something a frightened person could follow instead of seeing an oncologist.
No randomised controlled trial of this protocol has been published, and no controlled case series. The claims attached to it — set quantities over a set number of days producing set outcomes — are, as far as we can establish, widely repeated and unsourced. Cancer Research UK states that the research so far has been laboratory work with mixed results, that it does not tell us whether cannabinoids can treat cancer in people, and that quality varies so much it is impossible to know what a preparation contains.
People reach for these protocols when the odds have been described as bad. That is human, not stupid. If you tried it, you are welcome here — but we will not tell the next person it works.
11. What Happens When Patients Substitute
Johnson and colleagues, in the Journal of the National Cancer Institute in 2018, used the National Cancer Database to find 281 patients with non-metastatic breast, prostate, lung or colorectal cancer who chose alternative medicine as their only treatment, matched to 560 who had conventional care.
| Cancer type | Five-year survival, alternative medicine | Conventional treatment | Hazard ratio for death |
|---|---|---|---|
| All four cancers | 54.7% | 78.3% | 2.50 |
| Breast | 58.1% | 86.6% | 5.68 |
| Colorectal | 32.7% | 79.4% | 4.57 |
| Lung | 19.9% | 41.3% | 2.17 |
| Prostate | — | — | 1.68 (not significant) |
A companion 2018 analysis by the same group in JAMA Oncology found that among people using complementary medicine alongside conventional treatment, the added risk came from refusing parts of conventional therapy, not from the complementary therapy. The danger is substitution and delay.
12. Drug Interactions That Matter in Oncology
THC is metabolised mainly by cytochrome P450 2C9 and 3A4, cannabidiol by 2C19 with help from 2C9 and 3A4. Both inhibit several of these enzymes in the laboratory: a 2023 review in Deutsches Ärzteblatt International lists THC inhibiting 1A2, 3A4, 2B6, 2C9 and 2D6, cannabidiol 3A4, 2C19 and 2C9 — the same enzymes many chemotherapy agents, antiemetics and opioids use.
| Interaction | Reported effect | Why it matters |
|---|---|---|
| Warfarin | The highest-risk documented cannabis interaction in a 2022 Journal of Clinical Medicine review; large rises in international normalised ratio | Bleeding risk if anticoagulated |
| Tacrolimus | Cannabidiol raised concentrations enough to force a large dose reduction | Transplant and graft-versus-host disease |
Cannabis also adds sedation to opioids and benzodiazepines and causes dizziness — not trivial in a thrombocytopenic patient, or one with bone metastases for whom a fall means a fracture.
Signals with immunotherapy
Taha and colleagues, in The Oncologist in 2019, compared 89 patients on nivolumab alone with 51 on nivolumab plus cannabis: response rate 37.5% against 15.9%, cannabis the only significant factor reducing response. Bar-Sela and colleagues, in Cancers in 2020, followed 102 patients starting immunotherapy, 34 of them cannabis users: median time to progression 3.4 against 13.1 months, overall survival 6.4 against 28.5 months. Both are observational and confounded, and neither proves harm — but two datasets pointing the same way warrant a conversation, and the supportive care association flags caution with checkpoint inhibitors.
13. Contamination and Quality Risk for Immunocompromised Patients
A 2020 study by Benedict and colleagues in Emerging Infectious Diseases compared 53,217 people carrying a cannabis-use diagnosis code with more than 21 million without. Fungal infections were recorded in 0.08% of users against 0.03% of non-users — odds ratio 3.5 adjusted for age and immunocompromised status — including aspergillosis and mucormycosis. The design cannot prove causation, but the authors flagged the potential for serious illness and death in immunocompromised people.
A 2023 review in Frontiers in Microbiology explains why: a cannabis mycobiome of over a hundred fungal species, Aspergillus, Penicillium and Mucor among them, able to produce aflatoxins and ochratoxin A, and heat from smoking or vaping may not kill all spores. Invasive aspergillosis in a neutropenic patient has very high mortality. Residual pesticides, heavy metals and solvents also concentrate in an oil with the cannabinoids.
Why unlabelled oil is a specific danger here
Bonn-Miller and colleagues tested 84 cannabidiol products sold online by 31 companies and published in JAMA in November 2017: only about 30% were within 10% of their label claim, and several contained undisclosed THC. An unlabelled, untested oil is worse on every axis: unknown potency, unknown contaminants, unknown solvent, no recall.
14. What a Patient or Carer Should Do
The highest-value action on this page is disclosure. The American Society of Clinical Oncology’s 2024 guideline calls for open, non-judgemental discussion of cannabis use between clinicians and patients, and recommends against cannabis as cancer-directed treatment outside a clinical trial. Your oncologist cannot check interactions with a drug they do not know you take. Questions for the appointment:
- Am I on a checkpoint inhibitor, and does that change your advice?
- Am I anticoagulated, thrombocytopenic, or at risk of falls?
- Is my white cell count low enough that inhaled plant material is an infection risk?
- Is a licensed cannabinoid medicine available instead?
The legal position in Thailand
Under the Notification of the Ministry of Public Health Re: Controlled Herbs (Cannabis), B.E. 2568 of June 2025, cannabis is prescription territory: supply requires a prescription from a licensed practitioner, and therapeutic advertising is prohibited. Any source telling you otherwise, particularly one attaching a cancer claim to a product, is outside the law as well as the evidence.
15. Where the Research Is Heading
A trial capable of answering the antitumour question would need a standardised, pharmaceutical-grade product; a delivery route reaching the concentrations the laboratory work implies are needed, the central unsolved problem; addition to standard treatment rather than replacement, since withholding effective therapy is not ethical; enough patients to detect a realistic survival difference; and pre-specified survival endpoints. On the symptom side, the useful work is head-to-head comparison against modern antiemetics.
Until then the position is not “cannabis does nothing.” It is a narrow, guideline-recognised role in refractory chemotherapy-induced sickness; a disappointing record in cancer pain; an insufficient record elsewhere; and no established role in treating the disease.
If you are a patient or a carer, this thread is open to you: tell us what you tried, what helped, what did not, and what you wish someone had told you earlier — those accounts matter to the next person who arrives here frightened at two in the morning. But be honest about what a forum is. Nothing posted here is a substitute for your oncology team, who know your staging, your blood counts and your drug list, as none of us do. Take what you read here to them, and do not stop, delay or substitute treatment because of a forum post, including this one.
This article is part of the Asiannabis Community educational series on medical cannabis. It is general information, not medical advice — treatment decisions belong with a licensed practitioner. Content is for educational purposes within jurisdictions where medical cannabis is legally permitted.