Roughly one in three cancer patients uses cannabis or cannabis-derived products during treatment. Ask them why, and about half will tell you they believe it is doing something to the tumour itself. That belief has been built over two decades of laboratory headlines, and it has never been tested in the one scenario that matters most: what happens when cannabinoids and chemotherapy are inside the same cell at the same time.
A team at the Harvard T.H. Chan School of Public Health has now run that test, and the answer is not the one the field expected. In a study published in Cell Death Discovery on 10 August 2026, Francesca Picucci, Kristopher Sarosiek and colleagues report that six common cannabinoids did not kill glioblastoma cells on their own — but did protect those cells from dying when chemotherapy drugs and radiation were applied. In some experiments, the number of cancer cells left alive after treatment roughly doubled.
Key Takeaways
The short version is this. Six cannabinoids — including CBD, cannabigerovarin (CBGV), cannabichromene (CBC) and a synthetic stand-in for THC — were tested against four different glioblastoma cell lines. On their own, at realistic doses, they mostly did nothing: the cancer cells lived. Combined with standard cancer drugs or with radiation, they repeatedly made those treatments less lethal, and the surviving cells went on to form new colonies weeks later. The researchers traced the effect to two mechanisms: the cannabinoids raised levels of a survival protein called BCL-XL, and they parked cells in a phase of the cell cycle where radiation does the least damage. Crucially, the effect did not run through CB1 or CB2, the cannabinoid receptors everyone assumes are responsible. And the one dose of CBD that reliably killed glioblastoma cells also wiped out 98% of healthy human neural stem cells in the same experiment. This is laboratory work in dishes, not a clinical trial, and it does not prove that cannabis harms patients — but it is the strongest signal yet that the question deserves a proper trial rather than a shrug.
What the researchers actually did
Glioblastoma is the most aggressive primary brain tumour in adults, and it is a deliberate choice of battlefield. The brain is dense with cannabinoid receptors, and glioblastoma is the cancer where claims about cannabis as a tumour-killer have been loudest.
The team took four glioblastoma cell lines — U251-MG, T98G, SNB75 and LN229 — and exposed them to six cannabinoids: CBD, CBGV, CBC, cannabicyclol (CBL), cannabielsoin (CBE), and CP-55,940, a synthetic molecule that behaves in cells much like THC does. Doses ran from 3 to 30 micromolar. For scale, smoking a cannabis cigarette containing 34 mg of THC produces a peak blood level of around 0.48 micromolar. That sounds like a mismatch, until you account for the fact that cannabinoids are fat-soluble and pile up in tissue: measured levels in organs can run ten to a thousand times higher than levels in blood. One post-mortem case in the literature found blood THC at 1.79 ng/mL and heart tissue at 472 ng/g.
Then came the part that had been skipped for twenty years. Instead of testing cannabinoids alone, the team added them alongside the drugs oncologists actually use — carboplatin, bleomycin, doxorubicin, etoposide, paclitaxel — and alongside the two treatments that define standard care for glioblastoma: temozolomide and ionising radiation.
The result nobody was looking for
On their own, the cannabinoids were unremarkable. Across the four cell lines, most doses left cancer-cell survival essentially unchanged. Only three exceptions appeared, all at the top of the dose range.
In combination, the picture inverted. Carboplatin alone left about 32% of glioblastoma cells alive after 72 hours. Carboplatin plus 30 micromolar CBGV left about 65% alive — roughly twice as many cancer cells surviving a drug designed to kill them. CBD blunted the effect of bleomycin by nearly 25 percentage points. CBL rescued cells from a class of experimental drugs built specifically to force cancer cells into suicide.
The protection extended to standard glioblastoma care. Against 10 and 20 Gray of radiation, CBGV, CBD and the THC-mimic all increased survival of the tumour cells. Against the clinical combination of radiation plus temozolomide, three of the six cannabinoids reduced cancer cell death. The effect even held when cannabinoids were given 24 hours before treatment rather than at the same time — which is closer to how a patient managing nausea or sleep would actually take them.
To be fair to the data, it was not uniformly one-directional. In two of the other cell lines, high concentrations of CP-55,940 and CBC did the opposite and made temozolomide plus radiation more lethal. Biology rarely gives a clean answer, and the authors do not pretend otherwise.
Why “delayed” and “prevented” are very different words
A sceptic’s first objection would be that the cannabinoids merely postponed the inevitable. The team tested this directly with a colony formation assay: treat the cells, wash the drugs out, and see what grows back over the following days.
More colonies grew back in the wells that had received chemotherapy plus CBGV than in the wells that had received chemotherapy alone. Time-lapse imaging over seven days told the same story from another angle. Carboplatin alone drove cell death up to a peak of 84%. With CBGV present, the killing started twenty hours later and topped out at 62%.
Those cells were not dying more slowly. Many of them were not dying at all — and they were still capable of regrowing a tumour.
How a plant molecule talks a cancer cell out of dying
Chemotherapy and radiation do not, for the most part, blow cells apart. They damage them badly enough that the cell triggers its own controlled shutdown, a process called apoptosis. Whether that trigger fires depends on a tug-of-war inside the mitochondria between proteins that push for death and proteins that push for survival.
A useful way to picture it: think of the survival proteins as sponges soaking up death signals. A cell that is “primed” has saturated sponges — one more drop of damage and it dies. A cell with dry sponges to spare can absorb a great deal of punishment and carry on.
Using a technique called BH3 profiling, the researchers found that cannabinoid-treated glioblastoma cells were measurably less primed. They needed more of a push to die. Protein analysis showed why: cannabinoid exposure raised levels of BCL-XL, one of the biggest sponges in the cell.
A second mechanism stacked on top. The cannabinoids increased a protein called p21, which halts cells in the G1 phase of their growth cycle. This matters enormously for radiotherapy, because cells are most vulnerable to radiation during the G2 and M phases and considerably tougher in G1. By nudging cells into the safer phase, the cannabinoids effectively moved the target before the shot was fired — buying the cell time to repair its damaged DNA and get back to work.
It wasn’t the receptors everyone assumed
Here the paper takes an unexpected turn. The obvious suspects are CB1 and CB2, the two classical cannabinoid receptors. The researchers silenced CB1, then CB2, then both at once. The protective effect survived all three.
Computer modelling of how each cannabinoid docks into CB1 and CB2 found no pattern that separated the protective molecules from the rest. Whatever is driving this runs through something else — perhaps GPR55, the PPAR receptors, or TRPV1 channels, or perhaps mechanisms with no receptor at all. Other groups have shown that cannabidiol can reduce how much platinum chemotherapy a cell takes up in the first place, and cannabinoids are well-established antioxidants, which could soften the oxidative damage that chemotherapy and radiation rely on.
This is the honest edge of the paper: the researchers have documented a robust effect and identified two mechanisms downstream, but not the switch at the top.
The finding that should give everyone pause
There is one more result, and it cuts against the anti-cancer story from a different direction.
Thirty micromolar CBD did kill glioblastoma cells — in every line tested. It is the closest thing in the paper to a win for cannabis as a tumour-killer. But the same experiment included ReNcells, an immortalised line of human neural stem cells used to model the developing brain. At that dose, 98% of them died. At 10 micromolar — a dose widely cited in the literature as evidence that CBD selectively kills cancer — the healthy neural cells were already dying in large numbers.
That is the crux. Many published studies concluding that CBD kills cancer cells never ran a healthy cell alongside for comparison. When you do run one, selectivity looks a lot less selective. It offers a plausible explanation for an odd earlier result in animals, where CBD injected directly into tumours shrank them while shortening overall survival.
Similar caution has been surfacing elsewhere in cannabinoid science. A systematic review that screened nearly 4,000 studies of CBD in neurological disease found only one condition where the evidence genuinely held up, and the recent landmark trial of a cannabis-based medicine in Alzheimer’s disease showed the same gap between headline and substance. Laboratory promise has a poor conversion rate.
What this means — and what it does not
It does not mean cannabis causes cancer to grow. It does not mean patients using cannabinoids for nausea, pain or sleep have harmed their treatment. These are human cells in plastic dishes, at concentrations that may or may not be reached in a real brain tumour, in a manuscript that is still in press and has not yet been through final editing.
What it does mean is that the reassuring assumption — that cannabinoids are at worst neutral during cancer treatment — has never been properly earned. Three of the six cannabinoids tested here are ones most patients have never heard of and cannot easily avoid, because they occur naturally in full-spectrum products sold across Europe under a patchwork of national CBD rules that vary wildly from country to country. Nobody is tracking what any given patient is actually taking.
The clinical world has started moving in the same direction. The American Society of Clinical Oncology’s 2024 guideline advises against cannabis use during immunotherapy, having found worse outcomes in patients who used it. Meanwhile, the ARISTOCRAT trial in the UK is actively testing temozolomide with and without cannabinoids in recurrent glioblastoma — which means the question in this paper is about to get an answer in humans, whether or not anyone likes it.
Until then, the practical advice is unglamorous and unchanged: tell your oncologist what you are taking. Not because cannabis is forbidden, but because your treatment team cannot account for a variable they don’t know about.
Frequently Asked Questions
No. It shows that in laboratory dishes, six cannabinoids reduced the amount of cancer cell death caused by chemotherapy and radiation in glioblastoma cell lines. There is no patient data here, no animal data, and no measurement of tumour growth in a living body. It is a warning flag for further research, not a finding about outcomes in people.
That is a decision for a patient and their oncology team, not for a laboratory paper. Two cannabinoid medicines are approved specifically for chemotherapy-induced nausea and vomiting when standard drugs fail, and those approvals stand. The clear takeaway is disclosure: any cannabis or CBD use should be on the record with the treating team, in the same way as a prescription drug or supplement.
Partly. The 3–30 micromolar range sits well above the blood levels produced by smoking cannabis, which peak below 1 micromolar. The researchers justified the range on the grounds that cannabinoids concentrate in tissue at levels far higher than in blood, and that medical users can take doses many times larger than recreational ones. Whether these concentrations are reached inside a human brain tumour is genuinely unknown, and it is the single biggest open question in the paper.
There is no single EU answer. CBD as an ingredient is treated as a novel food by the European Commission, meaning products require authorisation that has largely not been granted, while national rules on THC limits, permitted product formats and health claims differ substantially between member states. Making medical claims about a CBD product is prohibited across the bloc regardless of the product’s legal status. Anyone using CBD during cancer treatment should check the rules in their own country and, more importantly, tell their doctor.
Disclaimer
This article is written for general information and popular science purposes only. It is not medical advice, and it is not a recommendation to start, stop or change any treatment, medication or supplement. The research described here was conducted on cell lines in a laboratory setting and has not been tested in animals or in patients; the manuscript was published as an unedited article in press. Cannabis and cannabinoid products are regulated differently in different jurisdictions, and legal status varies. Anyone undergoing cancer treatment, or considering the use of cannabis or CBD products, should speak with a qualified oncologist or physician before making any decision.