The source here is not a self-published pamphlet. It is a real molecular biology department at a Spanish university, and its finding is accurately reported: in laboratory studies, cannabinoids killed glioma cells and left normal brain cells alone. The problem is the sentence that other writers attach right after it, and who is likely to be reading.
If you or someone you love has been diagnosed with a malignant brain tumour, please read section 5 before anything else. Nothing here is medical advice, nothing here is a treatment, and nothing here should delay or replace neuro-oncological care. In Thailand, cannabis flower is a controlled herb. It may be dispensed only against a prescription or certificate from an authorised practitioner (PT 33). See topic 1358 for the current rules.
1. What the Research Actually Says
The work comes from the Department of Biochemistry and Molecular Biology at Complutense University of Madrid. The passage that cannabis literature quotes is a fair paraphrase of the group’s own summary (Velasco et al., Mol Neurobiol 2007;36:60–67; see also Guzmán, Nat Rev Cancer 2003;3:745–755):
Cannabinoids, the active components of Cannabis sativa L., act in the body by mimicking endocannabinoids, which activate specific cell-surface receptors. They have several palliative effects in cancer patients. In addition, cannabinoids inhibit the growth of different types of tumour cells, including glioma cells, in laboratory animals. They do this by modulating key cell-signalling pathways, mainly the endoplasmic reticulum stress response, which leads to apoptosis of tumour cells and inhibition of tumour angiogenesis. Of particular interest, cannabinoids appear to be selectively antitumour: they kill glioma cells but not the corresponding non-transformed astroglial cells.
Read carefully, that passage is scrupulous. It speaks of tumour cells and laboratory animals. For patients it claims palliative effects, meaning symptom relief. It makes no claim about treating human brain tumours.
The mechanism behind it is covered in topic 1308. THC causes glioma cell death through cannabinoid-receptor-dependent ceramide synthesis, which triggers endoplasmic reticulum stress. It also kills glioma cells by stimulating autophagy through inhibition of the Akt/mTORC1 axis. Glioma is the tumour type with the deepest preclinical cannabinoid literature.
The selectivity claim is the interesting one. Many substances kill cancer cells in a dish. A compound becomes a candidate drug when it kills cancer cells and spares the equivalent healthy cells. That claim is what brought this work attention, and it is legitimate.
2. The Sentence That Does Not Belong
Popular cannabis books that quote this passage sometimes add, right next to it, that THC has been shown to kill glioma cancer cells and is an effective cure, and that many patients owe their lives to cannabis. Those sentences come from the secondary book, not from the Complutense group. The researchers never wrote them.
Neither statement is supported by the passage beside it, and neither is supported by any other evidence. This is the failure examined in topic 1314 (why testimony is not evidence) and topic 1308: a carefully hedged scientific statement, quoted correctly, with an unhedged conclusion attached in the next line. The change happens so smoothly that a frightened reader will not notice that the paragraph break is doing all the work.
3. What Exists in Humans
Very little, and all of it concerns pharmaceutical cannabinoids, not home preparations.
Guzmán and colleagues, 2006 (Br J Cancer 2006;95:197–203). This was a pilot study in which delta-9-THC was given directly into the tumour to nine patients with recurrent glioblastoma. It was non-randomised and had no control group. The treatment was reported as safe. Establishing safety is what a phase 1 study is designed to do; it is not an efficacy result. Median survival was 24 weeks from the start of THC administration (n=9, non-randomised). Without a control group, that figure cannot show any effect of THC.
Twelves and colleagues, 2021 (Br J Cancer 2021;124:1379–87). This was a phase 1b trial that added nabiximols spray to temozolomide in patients at their first recurrence of glioblastoma. In the randomised, placebo-controlled part, twelve patients received nabiximols and nine received placebo (n=21). One-year survival was 83% versus 44% (p = 0.042). The trial was built to test safety and tolerability, so this survival comparison is effectively exploratory, with a nominal p≈0.04. The abstract also notes that two placebo patients died within the first 40 days of enrolment. The authors called for an adequately powered randomised trial.
That second result is quoted often, and it should be read with its denominator visible: twelve patients against nine. In a trial that small, with early deaths in one arm, a difference like this could easily reflect chance or imbalance between the groups. It is a signal worth pursuing, which is exactly what the authors said. It is not a basis for treatment.
The field’s own reviews describe the anticancer evidence as essentially preclinical (for example, Hinz & Ramer, “Cannabinoids as anticancer drugs: current status of preclinical research”, Br J Cancer 2022). Adequately powered randomised, placebo-controlled trials in cancer patients have not been done.
4. The Two Findings Nobody Quotes
Laboratory studies usually use cannabinoid concentrations well above anything human dosing reaches in the blood, let alone in tumour tissue behind the blood–brain barrier. Some findings also point the other way. At low concentrations, cannabinoids have stimulated growth in certain cancer cells. A 2022 study in mice found that cannabinoids suppressed antitumour T-cell immunity through CNR2 and JAK/STAT signalling (Xiong et al., Signal Transduct Target Ther 2022). Topic 1308 covers both the concentration and the immunity arguments in detail.
5. What Glioblastoma Actually Involves
“Malignant glioma” is not one disease. Under the current WHO classification of brain tumours (CNS5, 2021), glioblastoma (IDH-wildtype, grade 4) is the most aggressive form. IDH-mutant astrocytomas and oligodendrogliomas (grades 2–3) have a much better prognosis and different treatment. For grade 2 IDH-mutant tumours, that treatment now includes vorasidenib, which the US FDA approved in 2024. The figures below apply to glioblastoma only.
The prognosis figure in older cannabis literature, a median survival of 40 to 50 weeks from diagnosis, reflects an earlier treatment era. Current standard of care for newly diagnosed glioblastoma is maximal safe surgical resection, then radiotherapy with concurrent and maintenance temozolomide. That means temozolomide 75 mg/m²/day for six weeks alongside 60 Gy of radiotherapy in 30 fractions, followed by six maintenance cycles.
In the EORTC/NCIC trial that set this standard (Stupp et al., N Engl J Med 2005;352:987–96), the regimen gave a median overall survival of 14.6 months, against 12.1 months for radiotherapy alone, and 2-year survival of 26.5%, against 10.4%. The five-year analysis (Stupp et al., Lancet Oncol 2009;10:459–66) reported 9.8% alive at five years, against 1.9%. Trial patients are selected, so survival across the whole population is generally lower. Adding tumour-treating fields to maintenance temozolomide extended median survival further in the EF-14 trial: 20.9 months against 16.0 (Stupp et al., JAMA 2017). At recurrence, re-operation, re-irradiation or bevacizumab can control the disease or its symptoms in selected patients. Almost all patients still experience tumour progression.
Those numbers are grim, and stating them plainly is the point. This is exactly the situation in which a “cure” claim does the most damage. The patient is not credulous; the alternative on offer is so poor that any chance looks worth taking.
That is why the survival data in topic 1309 belong here. Among patients with curable cancers, using alternative medicine instead of conventional treatment carried a hazard ratio for death of 2.50. Glioblastoma is not curable, which makes the calculation different and more finely balanced. But surgery, temozolomide with radiotherapy, and tumour-treating fields buy months and sometimes years, and they are the treatments with randomised evidence behind them.
6. What Is Reasonable
Symptom relief is a real and legitimate role. The Complutense group says so in its own words: cannabinoids have several palliative effects in cancer patients. Nausea, appetite, sleep and pain are genuine targets. Topic 1312 covers chemotherapy-induced nausea and vomiting, the indication with the strongest evidence of all.
Telling the treating team is essential. Temozolomide becomes active by spontaneous chemical breakdown, with little CYP enzyme involvement, so it is not the main interaction concern. The Twelves trial reported no drug–drug interaction between nabiximols and temozolomide. The real risks lie elsewhere:
- Dexamethasone, the steroid widely used for brain swelling, is handled by CYP3A4, an enzyme pathway that cannabinoids also use.
- Enzyme-inducing anticonvulsants such as carbamazepine and phenytoin can change cannabinoid levels.
- CBD raises levels of clobazam’s active metabolite, which increases sedation.
- Levetiracetam, the anticonvulsant most often used in brain tumours, has little known interaction potential.
None of this can be judged at home: tell the treating team everything you are taking.
Trials exist. A patient interested in cannabinoids for glioma should ask their neuro-oncologist about clinical trials. That is how the twelve-versus-nine result becomes a real answer instead of a permanent rumour.
The evidence here is the same as for every “cannabis cures cancer” claim: preclinical, with no adequately powered human trial. The only difference is that the Complutense group never made the claim; someone else attached it. If you have a brain tumour diagnosis and someone has sent you material about cannabis oil, please bring it to your neuro-oncology team rather than acting on it. If you post it here, members can discuss it against the sources.
This is general educational information, not medical advice. Cannabis is not a treatment for glioma, and no regulator has approved any cannabinoid as a cancer treatment.