In the first randomized, placebo-controlled trial of cannabinoids as an anticancer add-on, patients with recurrent glioblastoma who received THC+CBD spray plus chemotherapy had 83% one-year survival versus 44% for placebo — but with only 21 randomized patients, the results are preliminary and require confirmation in larger trials.
Read this if you want to know where the evidence stands for cannabis-based drugs actually treating brain cancer — not just managing symptoms.
83% vs 44% one-year survival — but only 21 patients were randomized
The Backstory
For twenty years after Manuel Guzman's team showed that THC could shrink brain tumors in rats, the question hanging over cannabinoid cancer research was simple and unanswered: does this work in people?
In 2021, Chris Twelves — Professor of Clinical Cancer Pharmacology and Oncology at the University of Leeds — published the first answer. It wasn't definitive. It was 27 patients. But the numbers that came out of that small trial were striking enough to make oncologists pay attention and to launch the largest clinical trial of cannabinoids as an anticancer treatment ever attempted.
The Disease
Recurrent glioblastoma is among the most lethal diagnoses in medicine. Glioblastoma multiforme is the most aggressive primary brain tumor, accounting for roughly half of all malignant brain cancers. Standard first-line treatment — maximal surgical resection, radiotherapy, and temozolomide chemotherapy — extends median survival to about 15 months. When the cancer recurs — and it almost always does — options narrow dramatically. Median survival after recurrence is typically 6-9 months.
There is no standard second-line treatment that reliably extends life. Patients and their families know this. So do their oncologists.
Into this landscape, Twelves and colleagues introduced a question: could pharmaceutical-grade cannabinoids, added to standard chemotherapy, change the trajectory?
The Trial
The study was a Phase 1b trial — designed primarily for safety, with survival as a secondary measure. It had two parts.
Part 1 was open-label: six patients received nabiximols (Sativex) plus dose-intense temozolomide. The goal was to establish that the combination was tolerable and to find a workable dosing scheme.
Part 2 was what matters: a randomized, double-blind, placebo-controlled trial. Twelve patients received nabiximols; nine received placebo. Both groups took dose-intense temozolomide. Neither patients nor doctors knew who was getting the cannabinoid spray and who was getting an identical-looking placebo.
12 vs 9
The number of patients randomized to nabiximols versus placebo in Part 2 — the controlled portion of the trial. By oncology standards, this is a very small trial. Phase III trials typically require hundreds to thousands of patients.
For context, the follow-up ARISTOCRAT trial is recruiting 230+ patients across 14 UK hospitals — more than ten times larger — to determine whether the survival signal seen here is real.
Twelves et al. (2021), Br J Cancer 124(8):1379-1387
The Drug
Nabiximols (brand name Sativex) is a pharmaceutical-grade oromucosal spray manufactured by GW Pharmaceuticals (now Jazz Pharmaceuticals). Each spray delivers 2.7 mg of THC and 2.5 mg of CBD — roughly equal parts of each major cannabinoid, extracted from cannabis plants under pharmaceutical conditions. It is approved in multiple countries for MS spasticity but has never been approved for cancer.
Patients could take up to 12 sprays per day (32.4 mg THC + 30 mg CBD), with individualized dose escalation. The mean dose in the randomized portion was 7.5 sprays per day. The spray is applied inside the cheek or under the tongue — not smoked, not eaten, not vaped.
The Results
Safety
The primary finding: nabiximols plus temozolomide was safe and tolerable. The most common adverse events — vomiting, dizziness, fatigue, nausea, headache — were consistent with what you'd expect from both cannabinoids and chemotherapy. No new drug-drug interactions were identified. No unexpected toxicity emerged.
This mattered. Before this trial, the theoretical concern was that cannabinoids might interfere with chemotherapy metabolism or add intolerable central nervous system effects on top of a brain tumor. Neither happened.
Survival
83% vs 44%
One-year survival: 83% of patients receiving nabiximols plus temozolomide were alive at one year, compared to 44% of patients receiving placebo plus the same chemotherapy (p = 0.042).
For a disease with median post-recurrence survival of 6-9 months, these numbers are extraordinary. But the sample size — 12 versus 9 patients — means they must be interpreted with extreme caution. A single patient dying earlier or later in either group could substantially change the percentages.
Twelves et al. (2021), Br J Cancer 124(8):1379-1387
The one-year survival difference was statistically significant (p = 0.042). In a larger trial, this p-value would be persuasive. In a trial of 21 patients, it's a signal — an important one, but not proof.
The Puzzle
One finding complicates the story. Progression-free survival at six months — the percentage of patients whose tumors hadn't grown — was 33% in both groups. Identical. The tumors progressed at the same rate regardless of whether patients received cannabinoids.
This raises an uncomfortable question: if cannabinoids didn't slow tumor growth, why did patients in the cannabinoid group live longer? Several hypotheses exist:
Process
Why Might Survival Differ Without a PFS Difference?
Post-progression survival
Delayed tumor escape
Statistical noise
True anticancer effect
Twelves et al. (2021); editorial commentary, Br J Cancer
This ambiguity is why larger trials are essential. With 21 patients, you can't distinguish between these explanations.
The Researchers
Chris Twelves has spent his career in clinical cancer pharmacology — specifically, understanding how anticancer drugs behave in the body and how to use them more effectively. He held the Chair of Clinical Cancer Pharmacology and Oncology at the University of Leeds and was a consultant medical oncologist at St James's University Hospital. This was not a cannabis enthusiast testing a pet theory; it was a mainstream oncology pharmacologist evaluating a hypothesis generated by 20 years of preclinical research.
Susan Short, the trial's senior author, is Professor of Clinical Oncology and Neuro-Oncology at Leeds. The trial was funded by GW Pharmaceuticals, which provided the pharmaceutical-grade nabiximols — appropriate for a drug manufacturer testing its product, and declared as a conflict of interest in the publication.
What Comes Next
Research Timeline
From Pilot to Proof: The Path Forward
Galve-Roperh/Guzman show THC shrinks gliomas in rats
Preclinical proof of concept — cannabinoids can kill brain cancer cells.
Guzman conducts first human pilot: 9 patients, intracranial THC
Safety demonstrated. Modest signal. Too small for conclusions.
Twelves publishes Phase 1b: 83% vs 44% one-year survival (21 patients)
First randomized controlled data. Encouraging survival signal, but sample too small.
ARISTOCRAT Phase II trial begins recruiting at UK hospitals
230+ patients. Randomized, double-blind, placebo-controlled. Powered to answer the survival question.
ARISTOCRAT results expected
If positive, would likely trigger Phase III trials and a potential path to approval. If negative, would close this chapter.
Multiple publications and trial registries
The ARISTOCRAT trial — "A Randomised phase II trial of temozolomide with or without cannabinoids in patients with recurrent glioblastoma" — is the direct consequence of Twelves' findings. Led by Susan Short at Leeds and coordinated by Cancer Research UK's Clinical Trials Unit at the University of Birmingham, it will randomize more than 230 patients at 14 NHS hospitals across England, Scotland, and Wales. Patients will receive temozolomide plus either nabiximols or placebo in a 2:1 ratio.
The primary outcome is overall survival. Secondary outcomes include PFS, quality of life, and adverse events. If the survival benefit seen in the Phase 1b trial replicates at this scale, ARISTOCRAT could provide the evidence needed to consider Phase III trials — and potentially the first approved cannabinoid anticancer therapy.
If it doesn't replicate — and the history of small encouraging oncology trials that fail at larger scale is long — it will establish that the Phase 1b survival difference was statistical noise rather than a biological signal. Either answer would be valuable.
The Honest Assessment
This study occupies an uncomfortable middle ground that is easy to misrepresent in either direction.
The optimistic misrepresentation: "Cannabis doubles brain cancer survival!" This extrapolates a 21-patient Phase 1b safety trial into a treatment recommendation. It ignores the identical PFS, the tiny sample, the lack of dose-response data, and the long history of small positive trials that fail at scale.
The dismissive misrepresentation: "It's only 21 patients, it means nothing." This ignores that the survival difference was statistically significant, that the safety data enabled a much larger trial, and that the preclinical mechanism supporting cannabinoid anticancer effects is reproduced across dozens of independent studies over 25 years.
The honest position: a pharmaceutical-grade THC+CBD combination can be safely added to standard chemotherapy for recurrent glioblastoma, and the first controlled survival data are encouraging. Whether the survival benefit is real will be determined by the ARISTOCRAT trial. Until then, nobody — not patients, not oncologists, not cannabis advocates — should treat this as proven.
Key Takeaways
A phase 1b randomised, placebo-controlled trial of nabiximols cannabinoid oromucosal spray with temozolomide in patients with recurrent glioblastoma
Twelves C, Sabel M, Checketts D, Miller S, Tayo B, Jove M, Brazil L, Short SC (2021) · British Journal of Cancer
Related Research
Key studies in this area
()
()
()
()
What the researchers found
One-year survival was 83% for nabiximols-treated patients versus 44% for placebo (p=0.042). PFS at 6 months was 33% in both groups. Nabiximols had acceptable safety and tolerability with no drug-drug interactions with temozolomide. The most common adverse events were vomiting, dizziness, fatigue, nausea, and headache.
Why it matters
This was the first randomized, placebo-controlled trial testing cannabinoids as an anticancer add-on — not just for symptom relief, but with survival as the outcome measure. It bridged 20 years of preclinical research (beginning with Guzman 2000) and the clinical question: can pharmaceutical-grade cannabinoids actually help people with brain cancer live longer? The survival difference was large enough to justify the ARISTOCRAT Phase II trial (230+ patients, actively recruiting), which could provide the definitive answer.
How the study worked
Phase 1b trial with two parts. Part 1: open-label dose escalation (n=6). Part 2: randomized, double-blind, placebo-controlled (n=12 nabiximols, n=9 placebo). All patients received dose-intense temozolomide. Nabiximols (Sativex): oromucosal spray delivering THC:CBD 1:1 (2.7mg THC + 2.5mg CBD per spray), up to 12 sprays/day, individualized dose escalation.
Who was studied
Adults with first recurrence of glioblastoma after standard first-line treatment (surgery + radiotherapy + temozolomide). Part 1: n=6 open-label. Part 2: n=12 nabiximols, n=9 placebo.
What this study cannot tell us
Very small sample size (21 randomized patients). Phase 1b design — primarily a safety study, not powered for efficacy. Glioblastoma has highly variable outcomes; 21 patients cannot account for this variability. PFS at 6 months was identical between groups (33%), raising questions about the survival mechanism. The p=0.042 is nominally significant but would not survive multiple testing correction.
How to read the evidence
This is a Phase 1b randomized controlled trial — a legitimate clinical trial with randomization and blinding, but designed primarily for safety, not efficacy. The sample size is too small for definitive efficacy conclusions.
When this study was published
Published in 2021. The follow-up ARISTOCRAT Phase II trial (230+ patients) was actively recruiting at UK hospitals as of 2025.
The bigger picture
The ARISTOCRAT trial — a Phase II, multicenter study of 230+ patients at 14 UK hospitals — is the direct follow-up and was actively recruiting as of 2025. If ARISTOCRAT confirms the survival signal, it could lead to Phase III trials and potentially the first approved cannabinoid anticancer therapy. But the history of small encouraging trials that fail at larger scale is long, and expectations should be calibrated accordingly.
Questions still open
- Will the survival benefit replicate in the larger ARISTOCRAT trial? Why did PFS not differ between groups despite the survival difference? What is the optimal dose and schedule of nabiximols with chemotherapy? Could the effect be driven by palliative benefits (better appetite, less distress) rather than direct anticancer activity?
Common questions
Does this study prove THC+CBD treats brain cancer?
What is nabiximols/Sativex?
Should brain cancer patients use cannabis?
When will we know if this really works?
Read the original research
A phase 1b randomised, placebo-controlled trial of nabiximols cannabinoid oromucosal spray with temozolomide in patients with recurrent glioblastoma
British Journal of Cancer, 124(8), 1379-1387
Citation
Twelves, Chris; Sabel, Michael; Checketts, Daniel; Miller, Sharon; Tayo, Bola; Jove, Maria; Brazil, Lucy; Short, Susan C. (2021). A phase 1b randomised, placebo-controlled trial of nabiximols cannabinoid oromucosal spray with temozolomide in patients with recurrent glioblastoma. British Journal of Cancer, 124(8), 1379-1387. https://doi.org/10.1038/s41416-021-01259-3