Hundreds of testimonials of success. Dates, doses, photographs, tears in the comments. There is a version of the internet where fenbendazole is the cure your oncologist won’t tell you about. There is another where you’ve exhausted every protocol, every supplement, every promise, and the cancer never noticed.
This is arithmetic.
Give an inert pill to 100,000 people with advanced cancer. Depending on which symptom you ask about, somewhere between 6,000 and 27,000 of them will report feeling better. One or two will have a documented tumor regression with no treatment at all, and if the cancer in question is kidney cancer, closer to a thousand will. If 1 in 100 of the people who felt better writes about it, you have a couple of hundred stories, complete with dates, doses, and photographs, for a substance that did not affect the tumor, at all.
Those figures come from the placebo arms of 82 randomized cancer trials and a database of 21,000 drug development programs.. All linked as we go.
My goal in writing this, isn’t to suggest that people are foolish to be moved by the story. I want the story to be true, too. And, wanting it to be true is the most honest human response to being told you have a terminal disease, or that there is not much left to try. Learning from other people’s experience is how our species figured out nearly everything. Someone ate the root, the fever broke, the village remembered. That instinct is beautiful, and rooted deeply. Through, I think, cancer might be the one place it reliably betrays all of us, for reasons that have nothing to do with how carefully you are thinking.
My goal isn’t to take hope away. It's to make sure hope is informed. Hope doesn't harm people. Misinformation, misrepresentation, and an incomplete understanding of the evidence can.
Placebo does not shrink tumors, and that is exactly the problem
A 2022 analysis in eClinicalMedicine pooled the placebo arms of 45 phase 3 trials in advanced solid tumors, covering 5,684 patients. The objective response rate on placebo was 1%, with a confidence interval running from 0 to 2. Complete response was 0%. Sugar pills do not dissolve tumors.
Now let’s take a look at what placebo pills did do. Chvetzoff and Tannock reviewed 37 placebo-controlled cancer trials and found that among individual patients on placebo, 0 to 21% reported reduced pain or cut their analgesic use, 8 to 27% had appetite improvement, 7 to 17% gained weight, and 6 to 14% improved in performance status.
Their own conclusion is worth quoting, because it cuts against the lazy version of this argument: substantial improvements in symptoms and quality of life are unlikely to be due to placebo. What they mean is that those improvements are probably not caused by the pill. Pain and appetite fluctuate on their own, symptom scores drift back toward average after people enroll at their worst, and being asked attentively how you feel changes how you answer. For our purposes the cause is beside the point. A meaningful fraction of people taking something inert will report feeling better, and that report is the raw material of every testimonial you have ever read. Because testimonials are not written by tumors. They are written by a person describing how they feel, and feeling somewhat better is the one thing placebo does produce.
Spontaneous regression is real, and happens in some cancers more than others
Cancers sometimes regress with no treatment at all. Everson and Cole collected 176 documented cases and estimated the rate across all cancers at roughly 1 in 60,000 to 1 in 100,000. That estimate is from 1964 and is almost certainly low, since it depends on cases being noticed, confirmed, and published.
It also hides enormous variation. For example, in clear cell renal cell carcinoma the figure is closer to 1%, particularly after the primary tumor is removed. Melanoma and B cell malignancies also regress more often than average, which is unsurprising, since these are the tumors the immune system is most capable of recognizing.
These regressions happen on their own, on no treatment, on nobody's protocol. What matters is who happens to be taking something when one occurs, because whatever they were taking gets the credit. Give a sugar pill to 100,000 people with mixed cancers and one or two of them will have a documented regression. Give it to 100,000 people with kidney cancer and the number is closer to 1,000. I imagine, every one of those people will write about it in some way, because a thing like that demands to be written about.
You cannot tell that world apart from a world where the drug works by reading the stories. The stories look identical from the inside, and no amount of reading more stories resolves the issue.
Dishes and mice are not small people
Dare I say, that most every protocol circulating online rests on preclinical data, meaning based on cells in a dish or tumors in mice. Those experiments are usually real, but very few of them survive the complexities of a live human body.
Turns out, killing an isolated cancer cell outside its environment is easy.
A drug that kills cancer cells in a dish has to kill them at a concentration a human body can actually reach. Let’s take, ivermectin, for example. In culture, ivermectin kills cancer cells when the fluid around them holds about 3µM of drug. A person taking a standard dose peaks at roughly 1/60th of that in their blood. Push to the very high doses the protocols call for and you might get to 1/10th. Then subtract further, because ~93% of circulating ivermectin is stuck to blood proteins and unavailable to act on anything. The free drug actually reaching a tumor cell is more than a hundred times below what the dish experiment required for killing.
Mouse studies are better, but not by a lot. The tumor is usually a single cell line, genetically uniform, implanted under the skin of a mouse bred to have no immune system, and treated when it is a few days old and a few millimeters wide. A human cancer has evolved inside a living immune system for years, is genetically heterogeneous from one region to the next, and sits in tissue with its own blood supply and stroma. Doses are given in milligrams per kilogram, and a mouse dose does not convert one for one into a human dose. Do the conversion properly and the protocols circulating online usually land well below the dose that worked in the mouse. Additionally, the endpoint is usually tumor volume over a few weeks and where survival is reported, it is a mouse surviving a few extra weeks.
Mak, Evaniew, and Ghert put the rate of successful translation from animal models to human cancer trials at under 8%. Begley and Ellis, at Amgen, tried to reproduce 53 landmark preclinical cancer papers and were able to confirm 6. The Reproducibility Project: Cancer Biology spent eight years repeating 50 preclinical experiments from 23 high-profile papers, and where replication succeeded at all, the effect sizes came in 85% smaller than the original data suggested.
So when a post tells you a drug “kills cancer cells” and links a paper, the paper is usually real... and the gap between that paper and a person is real too.
The people who are no longer with us, are no longer posting
I don’t mean to sound crass. But testimonials have a trajectory. Survivors write. The ones who took the identical protocol and progressed drift offline, and their families do not come back six months later to post the correction.
Then add the “attribution problem.” Most people start an off-label drug alongside a new line of standard treatment, or while on immunotherapy, which can produce responses months after the starting of it. The response gets credited to the thing they chose rather than the thing they were given. The drug you sought out, paid for, and swallowed on a schedule feels like something you did. The infusion you sat through feels like bad weather.
A possible example of this and, a story known by many is this. In 2019 an average Joe with metastatic small cell lung cancer took veterinary fenbendazole, had a dramatic response, and wrote about it. He was also on an experimental immunotherapy at the time. His blog reached millions of people, and the immunotherapy part mostly did not travel with the story.
Why certainty should make you pause
Of every cancer drug that enters phase 1 testing, 3.4% eventually reach approval. Oncology is the worst performing field in medicine (though it’s marginally getting better). Everything outside oncology runs at about 21%.
Even after a cancer drug has survived phase 1 and phase 2 and entered a large randomized phase 3, it reaches approval only 35.5% of the time. Every one of these drugs had a plausible mechanism, real money behind it, a trial designed by statisticians, and a control group to compare against. Each was tested in the cancer its developers believed was most likely to respond. 2 out of 3 still fail at the last step. The median oncology program spends ~13 years in the clinic to find that out.
There is a name for part of what is happening: regression to the mean. A phase 2 result that looks spectacular tends to be followed by a phase 3 result closer to nothing, because extreme results are extreme partly by luck, and, alas, luck does not repeat. The stories that reach you are selected the same way. They are the tail of a distribution, and tails regress.
So when someone hands you an exact protocol for an off-label drug, mandatory additional off-label therapeutics, mandatory pulsing strategy with precise dosing strategy - a confident claim about what happens in stage 4 disease for all cancer types - they are claiming a level of certainty that no one has. Cancer is not one disease. It is thousands, behaving differently in different bodies, evolving continuously under treatment and without. Certainty in that setting means someone has stopped looking, or has something to sell.
The clinicians I trust most say “I think,” “we do not know yet,” and “let’s watch this closely and change course if we need to.” That is what expertise sounds like in this field.
There is no one protocol that works for all cancers.
But I object!
There is an elephant in the room with a real argument.
Nobody is going to fund a phase 3 trial of an off-patent animal dewormer. There is no patent, no return, and no sponsor. Demanding a randomized trial for a drug that will never receive funding for one can function as a permanent no, and that is a legitimate grievance.
It is also largely true. It explains why the evidence is missing, but t does not fill the gap. A drug without a comparison group has neither taken the test, not failed a lower bar.
You might notice that by my own numbers, cancer drugs fail 96.6% of the time. Why should the bar be higher for something cheap and widely available? It isn’t higher. The difference is that we know those drugs failed, because somebody measured. A paid newsletter with a mandatory dosing schedule is not a solution to the funding problem. If anything it relieves the pressure to solve it, because it lets everyone feel the question has already been answered.
Same arithmetic, different bottle
Ivermectin, fenbendazole, high-dose vitamin C, mistletoe, low-dose naltrexone, metformin, ketogenic protocols, hyperthermia, ozone… I could go on. They are not equivalent, some integrative/alternative cancer therapies have more to show for their name. Metformin, mistletoe, high-dose vitamin C, and hyperthermia have been studied in humans, some in randomized trials, with results ranging from encouraging to not. Fenbendazole, ozone, and low-dose naltrexone have almost no human oncology data at all. Ivermectin has a phase 1/2 trial hopefully being read out this year.
I use the ones with human data in my practice. I use others in cancer patients for treatment on conditions other than cancer. I advise on risk stratification for the therapies that don't have a lot of data behind them.
What integrative oncology shares is a thick bias layer because most of these things you are actively choosing, paying for, and believing in. The test is not whether something is conventional vs. off-label vs. investigational. Plenty of standard care started as somebody’s odd idea. The test is whether anyone has ever compared people who took “it” to people who did not.
The cost is not only money
The usual defense is that these drugs are cheap and mostly safe, so there isn’t a downfall. Let’s look at what has actually been published about humans taking fenbendazole.
Six case reports of drug-induced liver injury. A woman with lung cancer on pembrolizumab. A woman with colon cancer on nivolumab. A man with cirrhosis and liver cancer. A patient on immunotherapy whose cholestatic markers rose tenfold. A woman who took it for a year for a precancerous skin lesion. A man with prostate cancer taking fenbendazole and ivermectin together, whose ALT reached 1764. All resolved after stopping.
Notice what these cases share. Many were on immunotherapy, and the liver injury looked exactly like immune-related hepatitis, which is treated with high-dose steroids and stopping immunotherapy. So the risk is not only an overloaded liver. It is also losing the treatment that might have been working.
What the positive reports actually say
There are two of them (that I could find), describing four patients.
A 2020 report describes an 83-year-old man with stage IV diffuse large B-cell lymphoma who declined chemotherapy, took fenbendazole, and had improving lymphadenopathy over twelve months. He also developed peripheral neuropathy and had to cut the dose, which is what tubulin poisons do. And, from our earlier discussion, B cell lymphomas are among the tumors more likely to regress on their own.
A 2021 series describes three patients with genitourinary cancers. The first received three doses of nivolumab and a course of steroids the same month he started fenbendazole, and his own authors write that he may well have responded to nivolumab alone. The second had prior pembrolizumab and chemotherapy, and took fenbendazole alongside vitamin E, curcumin, and CBD oil. The third received six cycles of standard combination chemotherapy with fenbendazole running concurrently.
A third case series was published in 2025 and retracted.
That is the complete human record for tumor response. Four patients. Three of them receiving standard cancer treatment at the same time.
What to do with the next story you find
Read it. Save it. Evaluate 3 factors:
What else was on board? Very few take an off-label drug alone. If this person was also on chemotherapy, immunotherapy, radiation, a clinical trial, or 7 other off-label therapies, it’s likely that several things changed at once and only one of them is being credited.
Who is missing? You cannot count them, and that is the problem. Everyone who took this and got worse stopped posting.
Compared to whom? Someone has to have looked at people who took it alongside people who did not. Without that comparison there is no way to know what would have happened anyway.
If the answer to the third is no, what you have is a hypothesis. That is a real thing to have. Hypotheses are where every treatment I use started, and some of them are worth exploring. It is just not evidence yet.
Bring the protocol to your provider anyway. I would rather read it with you and tell you honestly what it is built on than have you run it alone.
I have seen off-label therapies work:
And I have seen them not:
There is a middle ground between a story and a phase 3 trial, and it is where most of my work happens. If we decide something is worth trying, we agree in advance on what we are watching, how long we are giving it, and what would make us stop. Markers, imaging intervals, liver enzymes, counts, symptoms, whatever is relevant to you and the drug. That is a structured trial of one. It will not tell the world anything, but it will tell you something, which is more than a protocol run on faith can offer.
Either way, you were right to look. Hope and evidence are not opponents. I would rather help you hold both than watch you pick one.
Ksenia Malarkey, ND is a naturopathic physician specializing in integrative oncology, hormones, and metabolic health.
She provides medical care for Washington residents and offers global health coaching.
To schedule a free 15-minute consult, visit drkseniamalarkey.com