17-DMAG dosage and evidence
Written by Aaron CuhaReviewed Sep 2026
The short answer
In people, it was only ever given into a vein to about 80 people with advanced cancer, in early dose-finding studies.
Evidence grade for 17-DMAG: Limited. The rule: people have taken it in a published study, but fewer than 100 of them. The sum on this page: 2 completed trials, 49 people given it. A grade counts how much has been done to people, not whether it works.
- People given it
- 49
- Human studies
- 2
- Found nothing
- 0
Dose
No human dose established
Amounts that were given in studies sit in Human evidence on this page. None of them is a personal dose.
2 completed trials, 49 people given it
3 effects reported, 3 unwanted, by people using it, not a trial
Goal pages: Ageing and longevity, Heart and circulation
In plain English
In plain English, from Aaron
This is how I would explain it to a friend. The evidence, with the numbers, is further down the page.
17-DMAG is a failed cancer drug. It was given into a vein, and there is no pill form. It is also studied as a drug that clears worn-out cells. It is not approved anywhere, and nobody is still working on it.
What people take it for
- Clearing worn-out cells, the idea behind senolytics.
- A longer life, after a mouse study.
What studies in people showed
Only about 80 people with advanced cancer have had it, all into a vein. At the top dose in one study, one person died from the treatment. Other harms included a heart attack, heart failure and eye swelling. In leukaemia, it cleared the disease in 3 of 17 people. No healthy person has taken it. Now mice. By the mouse program's main test, it did not help. A second look found fewer early deaths in males, fading with age. In fast-ageing mice, it kept them healthy for longer.
What people report
Reports, not study results
The good
- We have not collected user reports on this one yet.
The bad
- One death from the treatment at the top dose in one cancer study.
- Heart damage, including a heart attack and heart failure.
- Eye swelling, plus gut and liver problems.
Where these come from: The harms are from the cancer studies on this page. We have no user reports.
My bottom line
This one killed a patient in its first study and hurt hearts. Research chemical sellers still list it. I would never put it in my body.
An opinion, not a finding. I am a coach, not a doctor.
What it has been measured to do
Measured in people
Goals 17-DMAG has been tested for in people, biggest studies first. We count only what the studies on this page say, so the real numbers may be higher.
- Heart attacks and strokes24 people · 1 study
Animal and lab work, not yet tested in people
Real results, but from animals or cells, not people. What works in a mouse often does not work the same way in a person. Nobody can patent this, so nobody would ever earn back the cost of a study in people. That is usually why there is none. The record on this one is set out below.
- Living longer2 animal or lab studies
34 of the 36 indexed goals have no study of any kind behind 17-DMAG
No study in people, animals or cells on this page reports a result for these. That does not prove 17-DMAG does nothing for them. It does mean anything you have heard about 17-DMAG and one of these did not come from a study cited here.
Show full evidence
What is 17-DMAG?
In people, it was only ever given into a vein to about 80 people with advanced cancer, in early dose-finding studies. Heart damage, eye damage and one death from the drug set the dose limits, and work on it stopped around 2010. In mice, it failed the NIA lifespan program's standard test. A test that counts early deaths more found a 7 to 9 percent gain in typical male lifespan. It was also one of the first drugs found to clear old worn-out cells, and it kept fast-ageing mice healthier for longer.
17-DMAG (alvespimycin) is a water-soluble cousin of geldanamycin. It blocks HSP90, a helper protein that keeps many cancer-driving proteins stable. Kosan Biosciences and then Bristol-Myers Squibb developed it as a cancer drug. It reached early studies in solid tumours, acute myeloid leukaemia (AML) and HER2-positive breast cancer, and went no further. It is on this site for two reasons. A 2017 screen found that HSP90 blockers are a new kind of senolytic, a drug that clears senescent cells (old, worn-out cells that stop dividing but do not die). That study used 17-DMAG to keep a fast-ageing mouse healthier for longer. And the NIA mouse lifespan program tested it.
In people, three early studies make up the record. All gave the drug into a vein. In 25 patients with advanced solid tumours, it was given once a week, with the dose raised step by step from 2.5 to 106 mg/m2. At 106 mg/m2, two of four patients had harms severe enough to stop raising the dose, including one death caused by the treatment. Nobody had those harms at 80 mg/m2. Common problems hit the gut, the liver blood tests and the eyes. In 24 patients with advanced AML, it was given twice a week. The top dose they could stand was 24 mg/m2. At 32 mg/m2 the limits were heart damage, a raised troponin and a heart attack. The leukaemia cleared in 3 of 17 patients who could be assessed, though their blood counts did not fully recover. In 28 patients who also got trastuzumab, the limits were a severe drop in the heart's pumping and a severe but reversible eye inflammation. One patient's tumour shrank partly. A mid-stage study in HER2-positive breast cancer was stopped after 4 patients. No trial has been registered since 2010.
In mice, the NIA lifespan program fed 17-DMAG at 30 ppm from 6 months of age, in the run that began in 2015. In the main report, typical (median) male lifespan was 844 days against 772, 9 percent more, but the standard test said that could be chance. Females dipped 4 percent, also not clearly. The program reported no clear change in either sex. The 2024 second look with the Gehan test used slightly different mouse counts after some were left out. It put the male gain at 7 percent and said it held up. It found no effect in females. It also noted the effect on deaths seemed to fade with age. Separately, some mice lack a DNA repair gene (Ercc1) and so age very fast. In those mice, 17-DMAG given on and off kept them healthier for longer. It delayed several signs of ageing and lowered p16, a marker of worn-out cells.
17-DMAG was never approved anywhere and is no longer being developed.
How does 17-DMAG work?
17-DMAG plugs the fuel slot of HSP90 and stops it working. Without HSP90, the proteins it looks after, such as CDK4, HER2, AKT and faulty growth enzymes, get broken down by the cell. The cell makes more of another helper, HSP70, to make up for it. In the human studies, the helper HSP72 rose in blood cells from 20 mg/m2, and the looked-after proteins dropped at 80 mg/m2. So the drug did hit its target at doses people could stand. Its ability to clear worn-out cells was found by screening drugs in worn-out skin cells from fast-ageing mice, then checked in human cells. The idea is that worn-out cells lean on an active form of AKT to avoid dying, and 17-DMAG takes that prop away. That is dish and mouse work. The mouse program's papers do not explain why an HSP90 blocker would cut early deaths in male mice without helping the oldest ones. The second-look authors note that two other HSP90 blockers extend lifespan in worms (C. elegans).
What have studies in people found about 17-DMAG?
Early cancer studies only. About 80 patients with advanced cancer across three published trials and several registered ones. All got it into a vein, all in the mid 2000s. Nobody without cancer has taken 17-DMAG. There is no pill form. No study has looked at worn-out cells or ageing markers in a person.
- With trastuzumab, 28 patients: a drop in the heart's pumping and eye inflammation set the dose limit. One tumour shrank partly.
- Once a week into a vein, 25 solid tumour patients: harms severe enough to stop raising the dose at 106 mg/m2, including one death caused by the treatment. 80 mg/m2 was tolerated. Gut, liver and eye problems were common.
- Twice a week into a vein, 24 AML patients: top tolerated dose 24 mg/m2. A heart attack and a troponin rise at 32 mg/m2. The leukaemia cleared in 3 of 17.
- A mid-stage study in HER2-positive breast cancer was stopped after 4 patients. An early study in chronic lymphocytic leukaemia (CLL) was stopped too. No trial has been registered since 2010.
- Blood cell and tumour samples showed the drug hit its target at doses people could stand. So it does what it is meant to in people. The question was whether that was worth the harm, and the company's answer was no.
What this does not tell you: Cancer patients getting high doses into a vein tell you nothing about what low doses by mouth, over years, would do in a healthy older adult. That is the only case that matters for the mouse findings. The heart and eye harms are a real warning for a drug that would be given again and again to clear worn-out cells. No study in people has tested any HSP90 blocker for clearing worn-out cells.
The state of the research
The National Institute on Aging runs a mouse lifespan program called the Interventions Testing Program. It tests each compound at three labs at the same time: The Jackson Laboratory, the University of Michigan and UT Health San Antonio. The mice are a mixed breed, so a result is less likely to be a quirk of one strain. Males and females are both tested and judged separately. The three labs' results are then pooled. Every result gets published, whether or not the compound worked. It runs on public money from the National Institute on Aging, with no drug company money. 17-DMAG was fed at 30 ppm from 6 months in the 2015 run and reported as having no clear effect on lifespan in either sex.
Some results on these pages are called "Gehan-only". That is a weaker claim than "made mice live longer", so it needs explaining. The program's standard test counts every death the same. In 2024 a group at UT Health San Antonio went back over all the program's survival data from 2004 to 2022. That covered 132 compound and sex pairs. They used a second test, called the Gehan test. It gives more weight to deaths that happen young and in midlife. Five compounds that failed the standard test passed this one: metformin, enalapril and 17-DMAG in males, and CAPE and green tea extract in females. The authors say this is the pattern you would expect from something that stops mice dying young and then stops helping. In their words, the effects are "limited to midlife and may not be effective at older ages". They did not report any adjustment for running so many tests at once. So a Gehan-only result means fewer mice died young. By old age the groups caught up, and the oldest mice lived no longer.
For 17-DMAG the two reports give different typical (median) gains, 9 percent in the main paper and 7 percent in the second look. That is because a few mice were left out for fighting and deaths not linked to ageing, so the counts differ slightly. Both agree the standard test missed. The second look says the Gehan test passed, and that the survival curves fit the standard test's assumptions. Its authors flag that the effect on deaths seems to fade at older ages. The worn-out cell result is in a mouse that ages fast because it cannot repair its DNA. That models one kind of ageing damage, not normal ageing. And it measured years of good health, not lifespan.
Why nothing more will happen in people is simple. Bristol-Myers Squibb owned the drug and ran the trials. It saw heart and eye harm at doses that worked. The field was crowded with HSP90 blockers, and none has been approved for solid tumours. So the company stopped. The patents are old. A program to clear worn-out cells would need a new sponsor. It would have to take a dropped cancer drug, given into a vein, with a treatment death in its first study, into healthy older adults. Nobody has stepped up. The newer HSP90 blockers that did move forward are different molecules.
The evidence, in charts
Fig. 1a · how many people
77people, across 3 studies in people cited here
| Study | People | Share |
|---|---|---|
| Clinical Cancer Research 2012 | 28 | 36% |
| Clinical Cancer Research 2011 | 25 | 32% |
| Leukemia 2010 | 24 | 31% |
Counts are the number of people each study on this page says it signed up. That includes studies that only watched people and gave nobody the compound. 1 more study in people gives no headcount and is not counted here. Studies still signing people up are left out, and only studies cited on this page are counted.
Fig. 1b · evidence mix
4 of 7 sources here are studies in people. The rest are not.
- Study in people · done in people457%
- Animal study · done in animals343%
Counted from the list of sources on this page. It counts only what we cite, so there may be more studies out there. A review counts once, however many studies it sums up.
Fig. 2 · evidence over time
The studies cover 7 different years, 2008 to 2024, counted from the list of sources on this page.
Fig. 3 · legal status at a glance
US
Not approved
UK
Not approved
AU
Not approved
CA
Not approved
Not approved in any of the four countries shown. This is what the law says, not a verdict on the science. See Legal status below for the full text and any notes.
Fig. 4 · how much does what
Nobody has tested different amounts in people
This chart would show how the effect changes as the amount goes up. For this compound, nobody has measured that in people. So we leave it empty. We will not borrow an animal chart and pretend it works the same in people.
Fig. 5 · what the molecule is
- Kind of compound
- Small molecule
- Molecule weight
- Not on file
- Half-life
- None on file
- Chain length
- None on file
No amino acid sequence is on file for 17-DMAG, which is expected: a small molecule is not built from amino acids.
The studies we cite
- Study in people2011A phase I study of the heat shock protein 90 inhibitor alvespimycin (17-DMAG) given intravenously to patients with advanced solid tumors
This is a dose-finding cancer study, and at the top dose it killed someone. 25 patients were given alvespimycin into a vein once a week, at rising amounts. At the top dose, 2 of 4 patients had harms severe enough to stop raising the dose, including one death blamed on the treatment. Nobody did at the dose below. The usual problems hit the gut, the liver and the eyes. Target proteins in the blood did respond, which shows the drug reaches its target.
25 patients, weekly intravenous doses 2.5 to 106 mg/m2. At 106 mg/m2, 2 of 4 had dose-limiting toxicity including one treatment-related death; none at 80 mg/m2. Common adverse events gastrointestinal, liver function and ocular. HSP72 induction from 20 mg/m2, client protein depletion at 80 mg/m2. Academic UK trial with company drug supply.
- Study in people2010Phase I study of the heat shock protein 90 inhibitor alvespimycin (KOS-1022, 17-DMAG) administered intravenously twice weekly to patients with acute myeloid leukemia
In advanced leukaemia the drug helped a few people and hurt the heart at higher doses. 24 patients were given alvespimycin into a vein twice a week at rising amounts. At the step above the tolerated dose, the harms that set the limit were heart problems, including a troponin rise and a heart attack. Three of 17 patients who could be assessed had their leukaemia clear, though their blood counts did not fully recover. Kosan, the company behind the drug, paid for it.
24 evaluable patients with advanced AML, 8 to 32 mg/m2 twice weekly. Maximum tolerated dose 24 mg/m2; cardiac dose-limiting toxicities at 32 mg/m2 (troponin elevation, myocardial infarction). Complete remission with incomplete count recovery in 3 of 17 evaluable. Company sponsored (Kosan).
- Study in people2012A phase I dose-escalation trial of trastuzumab and alvespimycin hydrochloride (KOS-1022; 17 DMAG) in the treatment of advanced solid tumors
Combining the drug with trastuzumab caused heart and eye damage and produced very little benefit. 28 patients, 25 with breast cancer and 3 with ovarian, received weekly alvespimycin alongside trastuzumab. At the top dose one patient's heart pumping weakened enough to stop dosing, and two at the dose below developed serious but reversible inflammation of the cornea. One patient's tumour shrank partly and seven stayed stable. That is the whole benefit story, in 28 people with advanced disease.
28 patients (25 breast, 3 ovarian), 60 to 100 mg/m2 weekly with trastuzumab. Dose-limiting toxicities: grade 3 left ventricular systolic dysfunction (100 mg/m2) and reversible grade 3 keratitis in two patients (80 mg/m2). One partial response, seven stable disease in HER2-positive breast cancer.
- Study in people2008Clinical trial of intravenous alvespimycin in patients with HER2 positive breast cancer
The program was dropped. Bristol-Myers Squibb opened a mid-stage study in HER2-positive breast cancer, enrolled four people, and stopped it. No results were posted. The last alvespimycin study ever registered opened in 2010 in chronic lymphocytic leukaemia and was also stopped. Work on this drug has ended, and that is the most important fact about it for anyone reading about it today.
Phase 2, Bristol-Myers Squibb, TERMINATED after enrolling 4 participants, no results posted. The last alvespimycin trial to register started in 2010 (CLL, also terminated). Development has stopped.
- Animal study2020Rapamycin-mediated mouse lifespan extension: late-life dosage regimes with sex-specific effects
In mice. The NIA mouse lifespan program run that began in 2015. 17-DMAG at 30 ppm from 6 months: typical male lifespan 844 days against 772, 9 percent more, but the standard test said it could be chance. Survival to old age rose 4 percent, also not clear. Typical female lifespan fell 4 percent, not clearly. Reported as no clear change in either sex, alongside no results for minocycline, beta-GPA and MitoQ. Public NIA money paid for it.
ITP cohort begun 2015. 17-DMAG 30 ppm from 6 months: male median 844 versus 772 days (+9 percent), log-rank p = 0.12; 90th percentile +4 percent, p = 0.13. Female median -4 percent, p = 0.12. Reported as no significant change in either sex, alongside minocycline, beta-GPA and MitoQ nulls. NIA funded.
- Animal study2024The Gehan test identifies life-extending compounds overlooked by the log-rank test in the NIA Interventions Testing Program: metformin, enalapril, caffeic acid phenethyl ester, green tea extract, and 17-DMAG
In mice. A second look at the NIA mouse lifespan program. 17-DMAG at 30 ppm from 6 months: 152 treated males against 264 untreated ones. Typical male lifespan rose 7 percent. The standard test said that could be chance. The Gehan test, which counts early deaths more, said it held up. Females showed nothing. The effect on deaths seemed to fade at older ages. This was one of 132 comparisons, with no adjustment reported for running so many. Public NIA money paid for it.
17-DMAG 30 ppm from 6 months, 152 treated males against 264 controls: median +7 percent, log-rank p = 0.12, Gehan p = 0.032, proportional hazards assumption not violated. No female effect. Effect on mortality appears diminished at later ages. One of 132 comparisons reanalysed with no multiplicity correction reported. NIA funded.
- Animal study2017Identification of HSP90 inhibitors as a novel class of senolytics
In a dish and in fast-ageing mice. Researchers screened drugs on worn-out skin cells from mice missing a DNA repair gene (Ercc1). HSP90 blockers cleared worn-out mouse and human cells. In those fast-ageing mice, 17-DMAG kept them healthier for longer, delayed several signs of ageing and lowered p16, a marker of worn-out cells. These mice model fast ageing, not normal ageing, and nobody measured lifespan.
Screen of autophagy-regulating compounds in senescent Ercc1-deficient mouse fibroblasts identified HSP90 inhibitors as senolytic in mouse and human cells. In Ercc1-deficient progeroid mice, 17-DMAG extended healthspan, delayed onset of several age-related symptoms and reduced p16INK4a expression. Progeroid model, not normal ageing; no lifespan endpoint.
Common questions
Has 17-DMAG been tested in people?
Yes. 49 people given it across 2 completed trials cited here, 0 of which found nothing. In people, it was only ever given into a vein to about 80 people with advanced cancer, in early dose-finding studies.
What is the 17-DMAG dosage?
No human dose established. Amounts that were given in studies sit in Human evidence on this page. None of them is a personal dose. 2 completed trials, 49 people given it
Is 17-DMAG FDA-approved?
Investigational, never approved. Early safety studies in cancer finished between 2004 and 2012. A mid-stage study in HER2-positive breast cancer was stopped after 4 patients. No study registered since 2010. The lifespan hint in male mice comes only from a second look at the data using one statistical test (the Gehan test). The result on staying healthy for longer is in mice bred to age too fast.
Did 17-DMAG extend lifespan in the ITP?
Not by the program's standard test. At 30 ppm from 6 months, typical male lifespan was 7 to 9 percent higher, but that could have been chance. Females dipped 4 percent, also not clearly. A 2024 second look with the Gehan test, which counts early deaths more, said the male effect held up. It also said the effect faded with age. The program counts it as a no.
Is 17-DMAG a senolytic?
In a dish and in fast-ageing mice, yes. A 2017 screen found that HSP90 blockers clear worn-out cells. In mice missing a DNA repair gene (Ercc1), 17-DMAG kept them healthier for longer and lowered p16, a marker of worn-out cells. No person has taken it to clear worn-out cells. No HSP90 blocker has been tested for that in people.
Was it ever given to people?
Yes, into a vein, to about 80 patients with advanced cancer in early studies between 2004 and 2012. Harms that set the dose limits included a heart attack, heart failure, eye inflammation and one death caused by the treatment. A mid-stage study was stopped after 4 patients, and work on the drug stopped around 2010.
Can I buy 17-DMAG?
Research chemical sellers sell it for lab use. It has no approved form and no pill form. There is no human safety data outside cancer patients getting it into a vein, and it is known to harm the heart at doses that work. Nothing on this page supports taking it.