evipedia.ai
← Back

17-DMAG for Health & Longevity

Evidence Review created on 09/10/2026 using AI4L / Opus 5

Also known as: Alvespimycin, Alvespimycin Hydrochloride, 17-dimethylaminoethylamino-17-demethoxygeldanamycin, KOS-1022, NSC 707545, BMS-826476

Motivation

17-DMAG (alvespimycin) is a laboratory-modified relative of geldanamycin, a compound first pulled out of a soil bacterium. It works by jamming a protein-folding helper that cells rely on to keep hundreds of other proteins in working shape. Blocking that helper strips fast-growing cells and worn-out cells of proteins they depend on, while pushing healthy cells to switch on their own repair machinery. The same single action can therefore look destructive or protective depending on which cell is watching.

The compound was built in the mid-2000s as a cancer drug that dissolved in water, unlike its predecessors, and it moved through several small human safety studies before its commercial development was stopped. Interest returned from an unexpected direction: work in ageing biology found that this class of compound selectively kills the worn-out cells that build up with age, and that 17-DMAG improved health measures and survival in short-lived mice.

This review examines what the human and animal record actually shows about 17-DMAG: how it acts, what benefits have been measured and at what level of evidence, what harms appeared in the dose-finding studies, and how far the ageing findings have travelled toward people.

Benefits - Risks - Protocol - Conclusion

This section collects high-level sources that treat 17-DMAG, or the HSP90 (heat shock protein 90, a molecular chaperone that folds and stabilises other proteins) inhibitor class it belongs to, in substantial depth, covering its proposed senolytic (senescent-cell-killing) action as well as the dose-limiting toxicity (the exposure at which side effects force treatment to stop) recorded in human dose-finding studies.

Coverage note for readers: of the priority platforms, only Lifespan.io carries directly relevant content. Rhonda Patrick, Peter Attia, Andrew Huberman, Chris Kresser and Life Extension have published nothing on 17-DMAG or on HSP90 inhibitors as senolytics, which is unsurprising for an intravenous investigational compound that never reached the consumer market. Conflict of interest, first citation: the human evidence base cited above and throughout this review was generated almost entirely by parties with a direct financial or institutional stake in the compound — Kosan Biosciences and its acquirer Bristol-Myers Squibb, which owned the asset, and the US National Cancer Institute, which co-developed and supplied it. No independently funded human trial of 17-DMAG exists.

Grokipedia

17-Dimethylaminoethylamino-17-demethoxygeldanamycin

A dedicated, chemistry-first article covering the compound’s structure, synthesis from geldanamycin, chaperone target and clinical trial history — useful for orientation before reading the primary literature.

Examine

No Examine article exists for 17-DMAG; direct searches for both “17-DMAG” and “alvespimycin” returned no results. As an unapproved investigational parenteral drug rather than a dietary ingredient, it falls outside Examine’s supplement database.

ConsumerLab

No ConsumerLab article exists for 17-DMAG. ConsumerLab tests and reviews commercially sold supplements; 17-DMAG is an investigational prescription-only compound that has never been marketed, so the site does not typically cover it.

Systematic Reviews

This section lists the systematic reviews and meta-analyses that evaluate 17-DMAG or the HSP90 inhibitor class to which it belongs.

The claimed effect (antitumour activity) and one principal risk (drug-induced pain) are each represented above. The senolytic and lifespan claims that motivate longevity interest in 17-DMAG are unrepresented: no systematic review or meta-analysis of HSP90 inhibitors as senolytic or geroprotective agents exists on PubMed as of 10 September 2026.

Mechanism of Action

17-DMAG occupies the N-terminal adenosine triphosphate (ATP, the cell’s energy currency) pocket of HSP90 and halts the chaperone cycle. It is a pan-HSP90 binder rather than an isoform-selective one, engaging both cytosolic forms and, at higher exposures, the related chaperones in the endoplasmic reticulum and mitochondrion. Stalled clients — among them AKT (a survival kinase), CDK4 (a cell-cycle driver), HER2 (a growth-factor receptor) and LCK (an immune-cell kinase) — are tagged and destroyed by the proteasome, the cell’s protein shredder. HSP90 simultaneously releases heat shock factor 1, switching on the heat shock response and raising HSP70/HSP72, a second chaperone family. That HSP72 rise is the target-engagement fingerprint measured in the weekly dose-escalation trial and every other human study.

Two competing readings of this same event exist. Under the senolytic account, senescent cells depend on an HSP90-stabilised AKT survival signal, and withdrawing it tips them into programmed cell death while normal cells are spared, as shown in the screen that identified the class. Under the cytoprotective account, the HSP72 surge is itself the therapeutic actor, buffering healthy tissue against septic and thermal injury — an effect that argues against sustained chaperone suppression.

Metabolism is oxidative and modest: human liver microsome work identifies CYP3A4 and CYP3A5 (drug-clearing liver enzymes) as the dominant route. At 80 mg/m², mean clearance was 18.9 L/h against a 385 L volume of distribution, implying an elimination half-life of roughly half a day; tissue concentrations exceed plasma, highest in liver, then spleen, lung and kidney.

Historical Context & Evolution

Geldanamycin was isolated in 1970 from Streptomyces hygroscopicus and pursued as an antibiotic and antitumour agent. Its target was unknown until 1994, when it was shown to bind HSP90; that discovery converted an unremarkable natural product into the founding molecule of a drug class. Geldanamycin proved too hepatotoxic for people. Its first semi-synthetic successor, 17-AAG (tanespimycin), was tolerable but barely water-soluble, required awkward solvent formulations, and was heavily metabolised.

17-DMAG was engineered by the US National Cancer Institute with Kosan Biosciences to remove exactly those liabilities: it dissolves in water, is more potent, is less extensively metabolised, and is orally bioavailable. Preclinical toxicology in rats and dogs defined gastrointestinal, bone marrow, hepatic, renal and gallbladder toxicity and set the phase I starting dose. Four phase I programmes ran between 2004 and 2012.

The compound was then shelved. It is worth stating what did and did not happen: 17-DMAG was never shown to be ineffective, and no phase III trial was conducted. Kosan was acquired by Bristol-Myers Squibb in 2008; its single-agent breast cancer study stopped after four patients, while the trastuzumab combination and the academic trials all completed and reported. What changed was the commercial and safety calculus, not a disproof.

Interest revived from the opposite direction in 2017, when a senescence screen identified HSP90 inhibitors as senolytics and 17-DMAG extended healthspan in progeroid mice, and again in 2024, when a reanalysis of mouse survival data surfaced a lifespan signal the original statistics had missed. Both readings remain open.

Expected Benefits

High 🟩 🟩 🟩

No benefit reaches High: the class of evidence required — a human clinical endpoint or a validated clinical surrogate demonstrated in more than one controlled trial — does not exist for 17-DMAG, because every human study conducted was an uncontrolled, single-arm dose-escalation cohort.

Medium 🟩 🟩

No benefit reaches Medium either: there is no single controlled human trial and no observational cohort of 17-DMAG users; the only human outcome data are the uncontrolled phase I response figures graded Low below.

Low 🟩

Tumour Response in Advanced, Treatment-Refractory Cancer

Durable objective responses occurred in heavily pretreated patients in the weekly solid-tumour trial, the acute myeloid leukaemia trial and the trastuzumab combination trial, plausibly because HSP90 stabilises the androgen receptor and HER2. All were open-label, single-arm and dose-escalating, so response cannot be separated from selection and natural history.

Magnitude: One complete response lasting 124 weeks and one partial response lasting 159 weeks among 20 evaluable solid-tumour patients; complete remission with incomplete count recovery in 3 of 17 evaluable acute myeloid leukaemia patients; one partial response plus seven stable diseases lasting 4–10 months in a 28-patient combination cohort.

Speculative 🟨

Senescent-Cell Clearance and Healthspan Extension

In progeroid mice, 17-DMAG delayed the onset of several age-related symptoms and lowered p16INK4a (a senescence marker), per the senolytic screen. No human senescence data exist.

Median Lifespan Extension in Male Mice ⚠️ Conflicted

A reanalysis of mouse survival data found roughly 7% longer median lifespan in male mice; the original analysis found none, and females showed nothing. Net reading: an unreplicated, statistic-dependent signal.

Confirmed Target Engagement in Human Tissue

Human trials showed sustained HSP72 induction and depletion of CDK4 and LCK in blood cells and in paired tumour biopsies. These are unvalidated pharmacodynamic biomarkers, not outcomes.

Protection Against Metabolic and Kidney Injury

Rodent work reports reduced ovariectomy-induced obesity, attenuated diabetic kidney disease and atherosclerosis and reduced kidney fibrosis. All are animal models with no human counterpart.

Vascular Protection in Animal Injury Models

17-DMAG attenuated abdominal aortic aneurysm formation in mice and reduced intimal thickening after balloon injury in rats. Both are animal injury models with no human counterpart.

Neuroprotection and Improved Protein Quality Control

In rodents, 17-DMAG preserved motor neurons in spinal and bulbar muscular atrophy, an inherited motor nerve disease, and reduced brain inflammation after a bleed into brain tissue. Basis is animal only.

Organ Protection During Acute Systemic Stress

Via HSP70 induction, 17-DMAG reduced multi-organ failure in rats given bacterial toxin and protected inner-ear hair cells from antibiotic damage. Both are animal or tissue experiments only.

Immune Modulation in Autoimmune Disease Models

17-DMAG reduced disease severity in a mouse model of lupus (an autoimmune disease attacking the body’s own tissues) and dampened macrophage inflammation in culture. Basis is animal and cell work only.

Activity Against Intracellular Parasites

17-DMAG suppressed several Babesia species in culture and in mice, a tick-borne infection relevant to some longevity-oriented adults. Evidence is in-vitro and rodent only, with no human report.

Benefit-Modifying Factors

  • NQO1 genotype: NAD(P)H:quinone oxidoreductase 1, an enzyme that converts quinone drugs to a more active form, varies by genotype. Per a review of quinone drug sensitivity, carriers of two NQO1*2 null alleles generate less active drug.
  • ABCB1 / P-glycoprotein expression: P-glycoprotein, a membrane pump encoded by ABCB1 that ejects drugs from cells, is induced during acquired 17-DMAG resistance. High baseline expression predicts blunted intracellular exposure.
  • Baseline HSP72 and client protein levels: Pre-treatment HSP72 in blood cells varied more than twofold between individuals, and the induction fold-change, not the absolute level, tracked target engagement in the dose-escalation cohort.
  • Senescent-cell burden: The senolytic rationale predicts benefit proportional to existing senescent-cell load. In progeroid mice with high burden the effect was large; effects in low-burden young tissue were minimal.
  • Sex: The only lifespan signal reported for 17-DMAG appeared in male mice alone, with no corresponding female effect. Whether this reflects biology, dosing per body weight, or statistical noise is unresolved.
  • Pre-existing hepatic and renal function: Liver holds the highest tissue concentrations, and preclinical work showed dose-disproportionate plasma accumulation in dogs. Impaired clearance raises exposure and shifts the risk-benefit balance unfavourably.
  • Age: Older adults carry more senescent cells, which should favour benefit, but the phase I cohorts enrolled participants to age 78 without reporting age-stratified response, so any age-related advantage remains untested in people.

Potential Risks & Side Effects

High 🟥 🟥 🟥

Gastrointestinal Toxicity

Nausea, vomiting, diarrhoea and anorexia were the most consistently reported drug-related events in the weekly, twice-weekly and leukaemia dose-escalation trials, matching the gut lining toxicity that was dose-limiting in rats. Events were low grade and reversible at tolerated doses but became dose-limiting above them, and gut epithelium is intrinsically chaperone-dependent, so this is on-target rather than idiosyncratic.

Magnitude: Present at every dose level tested and dose-limiting as grade 3 diarrhoea at 106 mg/m² weekly; the phase I reports describe incidence categorically and publish no pooled incidence figure for the compound.

Ocular Toxicity

Blurred vision, dry eye, keratitis (inflammation of the cornea) and conjunctivitis appeared in the weekly and trastuzumab combination trials, and reversible grade III keratitis was dose-limiting in the latter. Retinal photoreceptors are unusually HSP90-dependent, and the same lesion is reproducible in beagle dogs, so this is a class effect. Events were reversible, but they forced dose reductions and prophylactic lubricating drops.

Magnitude: Four of 25 patients experienced ten drug-related ocular events on the weekly schedule, nine of them at doses of 80 mg/m² or above, with two patients requiring dose reduction; grade III keratitis was dose-limiting in two of 13 patients at 80 mg/m² with trastuzumab.

Fatigue and Constitutional Symptoms

Fatigue was among the most frequent drug-related events across the weekly, twice-weekly and trastuzumab combination schedules and became dose-limiting at the top dose. It is the practical ceiling on tolerability for anyone taking the compound outside a cancer context, where a severe fatigue burden would not be accepted in exchange for a speculative longevity benefit.

Magnitude: Grade 3 fatigue was one of two dose-limiting toxicities recorded at 106 mg/m² weekly and appeared as grade 1–2 at most lower dose levels; grade III fatigue occurred in one of 28 patients in the trastuzumab combination cohort.

Cardiac Injury

Troponin elevation with myocardial infarction, and separately left ventricular systolic dysfunction presenting as congestive heart failure, were dose-limiting in the leukaemia and trastuzumab cohorts respectively. Every 17-DMAG protocol excluded participants with reduced ejection fraction, recent infarction or arrhythmia, which means the observed rate is an underestimate of what an unselected population would experience.

Magnitude: Cardiac dose-limiting toxicity defined the ceiling at 32 mg/m² twice weekly in acute myeloid leukaemia; grade III left ventricular systolic dysfunction occurred in one of six patients dosed at 100 mg/m² weekly alongside trastuzumab.

Acute Kidney Injury

Renal dysfunction was dose-limiting on the twice-weekly schedule, and one participant on the weekly schedule progressed from a liver enzyme rise and low blood pressure to anuric renal failure (kidney failure with no urine output) and died five days after dosing. Renal toxicity was also dose-limiting in dogs. This is the most severe documented harm and it appeared abruptly, within 24 hours of a single administration.

Magnitude: One treatment-related death from anuric renal failure among 25 patients (4%) on the weekly schedule, occurring at 106 mg/m²; renal dysfunction was one of two dose-limiting toxicities in the 31-patient twice-weekly cohort.

Musculoskeletal and Visceral Pain

Myalgia (muscle pain), arthralgia (joint pain), back pain and abdominal pain recurred across the twice-weekly and combination cohorts, and a class-level systematic review and meta-analysis found HSP90 inhibitors significantly increase abdominal pain risk with a weaker back pain signal. Preclinical work suggests HSP90 inhibition modulates pain sensitisation directly rather than only through tissue damage.

Magnitude: Directionally increased and among the most frequent events at 60–100 mg/m² weekly, with grade III myalgia and grade III back pain each recorded once in a 28-patient cohort; the meta-analysis reports a significant increase in abdominal pain risk but no incidence figure specific to 17-DMAG is published.

Medium 🟥 🟥

Hepatic Enzyme Elevation

Liver function disturbance was among the common adverse events on the weekly schedule and hepatic toxicity was dose-limiting in both preclinical species. 17-DMAG was designed to be less hepatotoxic than geldanamycin, and at tolerated doses the changes were low grade and reversible, but the abrupt severe rise seen at the top dose shows the margin is narrow.

Magnitude: Grade 4 aspartate aminotransferase (a liver enzyme) elevation in one of 25 patients (4%) at 106 mg/m² weekly, alongside grade 3 hypoalbuminaemia (low blood albumin); lower-grade, reversible liver enzyme changes were common at doses of 80 mg/m² and below.

Myelosuppression and Neutropenic Fever

Myelosuppression (a fall in the bone marrow’s output of blood cells) and the neutropenic fever it produces (fever arising when infection-fighting white cells fall dangerously low) were among the common toxicities in the acute myeloid leukaemia cohort, and bone marrow toxicity was dose-limiting in rats and dogs. Attribution in that trial is confounded because the participants had marrow failure from their disease, so the drug-attributable fraction in a person with normal marrow is unknown.

Magnitude: Directionally increased, with neutropenic fever listed among the most common toxicities at 8–32 mg/m² twice weekly in a marrow-compromised population; no incidence figure separating drug effect from disease effect is published.

Peripheral Neuropathy

Peripheral neuropathy (nerve damage in the hands and feet causing numbness, tingling or burning) was one of the two dose-limiting toxicities on the twice-weekly intravenous schedule. Long axons depend heavily on chaperone-supported protein transport, which offers a plausible on-target mechanism. Reversibility was not characterised in the published report, which matters because neuropathy that does not resolve is a poor trade for a speculative benefit.

Magnitude: Dose-limiting in the 31-patient twice-weekly cohort at doses above the 21 mg/m²/day maximum tolerated dose; the report gives no incidence rate or severity distribution.

Low 🟥

Unpredictable Exposure from Wide Pharmacokinetic Variability

A population pharmacokinetic analysis of 67 patients found large between-individual variability and no covariate explaining it. Two people given an identical dose can reach very different exposures, and the two participants with dose-limiting toxicity in the weekly study had the highest exposures recorded.

Magnitude: At the median 36 mg/m² dose the simulated 95% prediction interval (the range expected to cover 95 of every 100 individuals) for daily exposure spanned 1,059–9,007 mg/L·h, an approximately 8.5-fold range, versus 2,910–4,077 mg/L·h when only within-person occasion-to-occasion variation was modelled.

Electrolyte and Fluid Disturbance

Grade 3 hyponatraemia (low blood sodium) occurred at a dose as low as 5 mg/m² in the weekly cohort, and dehydration, acidosis (excess acid in the blood) and hypoalbuminaemia clustered with the fatal top-dose event. These uncontrolled single-cohort observations are partly secondary to gastrointestinal losses, not a distinct mechanism.

Magnitude: Grade 3 hyponatraemia in one of 25 patients (4%) at 5 mg/m² weekly; grade 3 dehydration, hyponatraemia and acidosis co-occurred in the single fatal case at 106 mg/m².

Speculative 🟨

Redox Stress from Benzoquinone Bioactivation

The molecule’s reactive benzoquinone ring is chemically reduced inside cells. In gastric cancer cells 17-DMAG raised reactive oxygen species and depleted antioxidant enzymes, per a cell-culture study. No human oxidative-stress outcome exists.

Loss of Cytoprotective Chaperone Function in Healthy Tissue

Because HSP90 chaperones hundreds of normal client proteins, sustained inhibition could plausibly impair stress tolerance in healthy tissue. The basis is mechanistic reasoning from the drug class overview only, with no controlled data.

Gallbladder and Biliary Toxicity

Dogs given 17-DMAG showed gallbladder toxicity that did not appear in rats and has no described human counterpart, per the preclinical toxicology package. The finding is species-limited and derived from isolated animal reports.

Risk-Modifying Factors

  • NQO1*2 polymorphism: The NQO1 enzyme activates quinone drugs such as this one. Individuals carrying two null copies produce less of the active form and may be relatively protected from redox-driven toxicity while gaining less effect, per the NQO1 review.
  • CYP3A4/CYP3A5 activity: Because these liver enzymes dominate oxidative clearance, poor metabolisers and anyone taking a strong inhibitor face raised exposure, and exposure tracked directly with dose-limiting toxicity in the weekly cohort.
  • Baseline liver enzymes, creatinine and albumin: Every trial protocol required near-normal liver and kidney chemistry for entry. Abnormal baseline values were an exclusion, so the published safety profile does not apply to anyone with pre-existing organ impairment.
  • Baseline cardiac status: Reduced ejection fraction, recent myocardial infarction, uncontrolled arrhythmia, a prolonged QT interval (slow electrical recovery) and left bundle branch block (a conduction delay) were protocol exclusions, so the observed event rate understates risk in unscreened adults.
  • Sex: No sex-stratified safety analysis was published for any 17-DMAG trial. Women were included throughout, and the female-specific QT threshold used at entry implies a recognised but unquantified sex difference in cardiac vulnerability.
  • Pre-existing dry eye or ocular surface disease: Ocular events concentrated at higher doses and were the most common reason for dose reduction in the weekly cohort; a compromised tear film plausibly lowers the threshold for symptomatic keratitis.
  • Age: Older adults more often carry the reduced kidney function, subclinical cardiac disease and polypharmacy that each amplify this compound’s specific toxicities, while trials enrolled a screened population up to age 78 with none of those features.

Key Interactions & Contraindications

  • Strong CYP3A4 inhibitors (ketoconazole, itraconazole, ritonavir, clarithromycin, grapefruit juice): Absolute caution — raises 17-DMAG exposure and risk of renal, hepatic and cardiac dose-limiting toxicity. Separation in time does not help; avoid co-administration or reduce dose under monitoring.
  • Strong CYP3A4 inducers (rifampicin, carbamazepine, phenytoin, St John’s wort): Caution — accelerates clearance and can drop exposure below the concentration needed for target engagement, producing toxicity risk without benefit.
  • P-glycoprotein inhibitors and substrates (verapamil, ciclosporin, digoxin): Monitor. Verapamil restored 17-DMAG sensitivity in resistant cell lines, implying co-administration raises intracellular drug and therefore intracellular toxicity.
  • QT-prolonging drugs (amiodarone, sotalol, methadone, ondansetron, macrolide antibiotics such as azithromycin): Absolute contraindication in the trial protocols, which barred any concurrent QT-prolonging agent; consequence is additive risk of ventricular arrhythmia.
  • Trastuzumab and other cardiotoxic agents (anthracyclines): Caution with mandatory cardiac monitoring — the combination trial recorded congestive heart failure as a dose-limiting toxicity. Baseline and serial ejection fraction measurement is the stated mitigation.
  • Over-the-counter analgesics and acid-suppressing agents (ibuprofen, naproxen, high-dose aspirin, cimetidine): Caution — non-steroidal anti-inflammatory drugs (ibuprofen-type painkillers) compound the renal injury that caused the single treatment-related death, and cimetidine inhibits CYP3A4, raising exposure. Paracetamol is the safer substitute.
  • Full-dose anticoagulants: Contraindicated in trial protocols. Consequence is bleeding risk compounded by drug-induced low platelet counts; prophylactic low-dose aspirin was permitted.
  • Other supplements with additive effects: Caution. Quercetin, fisetin and dasatinib act as senolytics through overlapping apoptosis pathways, so combining them risks additive cytotoxicity. Curcumin and resveratrol inhibit CYP3A4, raising exposure. High-dose antioxidants may blunt the drug’s redox component.
  • Combination with other investigational interventions: Trial protocols barred any concurrent investigational agent, growth factor or anticancer therapy; the consequence of ignoring this is uninterpretable attribution of any adverse event.

Populations who should avoid 17-DMAG:

  • Anyone not enrolled in a supervised clinical trial, since no approved source, dose or monitoring standard exists
  • Pregnant or breastfeeding women — teratogenic and abortifacient potential was assumed in every protocol
  • Left ventricular ejection fraction below 40%, New York Heart Association class III or IV heart failure, myocardial infarction within the past year, or active ischaemic heart disease within the past year
  • Corrected QT interval above 450 ms in men or 470 ms in women, congenital long QT syndrome (the inherited form), left bundle branch block, or any arrhythmia requiring antiarrhythmic medication
  • Transaminases above 2.5 times the upper limit of normal, bilirubin above 1.5 times normal, or Child-Pugh Class B or C liver impairment (a severity score for chronic liver disease)
  • Creatinine clearance below 50 mL/min, or serum sodium below 130 mmol/L
  • Absolute neutrophil count below 1,500/mm³, platelets below 100,000/mm³, or haemoglobin below 8 g/dL
  • Known hypersensitivity to geldanamycin or 17-AAG
  • People with active uncontrolled infection, symptomatic pulmonary disease, or on combination antiretroviral therapy (multi-drug HIV treatment)

Risk Mitigation Strategies

  • Ceiling the dose below the toxicity threshold: The fatal renal event and cardiac dose-limiting toxicities occurred at 106 mg/m² weekly and 32 mg/m² twice weekly; 80 mg/m² weekly or 21 mg/m²/day twice weekly avoided both, though ocular events persisted.
  • Prophylactic ocular lubrication and scheduled eye review: Preservative-free lubricating drops from the first dose, with slit-lamp examination at baseline and every 4 weeks, mitigate the keratitis and dry eye that forced dose reductions at 80 mg/m² and above.
  • Serial cardiac surveillance: Ejection fraction by echocardiogram at baseline and every 8–12 weeks, electrocardiogram before each dose, and high-sensitivity troponin if symptoms occur, mitigates the myocardial infarction and heart failure seen as dose-limiting toxicities.
  • Renal and electrolyte protection: Pre-dose and 48-hour creatinine, sodium and albumin, plus intravenous hydration around dosing, mitigate the anuric renal failure, hyponatraemia and dehydration that produced the single treatment-related death.
  • Therapeutic drug monitoring: Because exposure varied roughly 8.5-fold with no explanatory covariate, measuring plasma concentration and adjusting dose mitigates the unpredictable overexposure that preceded both dose-limiting events in the weekly study.
  • Blanket avoidance of CYP3A4 inhibitors and QT-prolonging drugs: A full medication and supplement reconciliation before each cycle, with a 5-half-life washout of any strong CYP3A4 inhibitor, mitigates exposure-driven hepatic, renal and arrhythmic toxicity.
  • Antiemetic and gastrointestinal support: Scheduled rather than as-needed antiemetics (anti-nausea drugs) and early loperamide for diarrhoea mitigate the volume depletion that links gastrointestinal toxicity to renal injury.
  • Complete blood count before every dose: Withholding treatment when the absolute neutrophil count falls below 1,500/mm³ or platelets below 100,000/mm³ mitigates neutropenic fever and bleeding.

Therapeutic Protocol

  • No established protocol outside trials: 17-DMAG has no approved indication, no marketed product and no compounding pathway. No clinic or practitioner administers it, so no consensus practice exists; every parameter below derives from published trial schedules.
  • Weekly intravenous schedule: 80 mg/m² by intravenous infusion once weekly was the recommended dose from the Institute of Cancer Research programme, established after a treatment-related death at 106 mg/m².
  • Twice-weekly intravenous schedule: The National Cancer Institute programme used 1–2 hour infusions twice weekly in 4-week cycles, with 21 mg/m²/day as the maximum tolerated dose; the leukaemia programme used 24 mg/m² twice weekly, 2 weeks of every 3.
  • Competing approaches presented without a default: The weekly high-dose approach maximises peak concentration and target engagement; the twice-weekly lower-dose approach maximises time above the effective concentration. Neither has been shown superior, and neither was tested for senolytic use.
  • Intermittent senolytic dosing as the untested alternative: The senolytic literature argues that brief, widely spaced exposures suffice to clear senescent cells, which would sidestep continuous chaperone suppression. No human study has tested this schedule with 17-DMAG.
  • Expert attribution: The weekly schedule came from Pacey, Workman and Judson at the Institute of Cancer Research; the twice-weekly schedules from Kummar at the National Cancer Institute and Lancet at Moffitt; the senolytic rationale from Robbins and Niedernhofer.
  • Best time of day: No chronopharmacology study exists for 17-DMAG. All trials infused during clinic hours, in the morning to early afternoon, which also allows same-day observation for the acute reactions that appeared within 24 hours.
  • Half-life: Clearance of 18.9 L/h against a 385 L volume of distribution at 80 mg/m² implies elimination over roughly half a day, and plasma concentration stayed above the cell-panel inhibitory concentration for more than 24 hours after a single dose.
  • Single versus split dosing: Both were tested. Splitting the weekly amount across two infusions lowered peak concentration and shifted the toxicity profile toward neuropathy and renal effects; the single weekly infusion shifted it toward ocular and hepatic effects.
  • Genetic factors in dose choice: NQO1*2 homozygosity predicts less bioactivation, and CYP3A5 expressor status alters clearance. Neither has been prospectively used to set a 17-DMAG dose, and no pharmacogenetic dosing algorithm exists.
  • Sex-based differences: No trial reported sex-stratified dosing or response. Body-surface-area dosing was used throughout, and the only lifespan signal in mice was male-specific, so extrapolation to women is unsupported.
  • Age-related considerations: Trials enrolled to age 78 with intact organ function. Age-related decline in kidney clearance and cardiac reserve argues for the lower twice-weekly schedule and closer monitoring in older adults, though no formal age adjustment was studied.
  • Baseline biomarkers influencing response: Pre-treatment HSP72 and client protein levels in blood cells confirmed engagement, and blood HSP72 rose highest in the two participants who suffered dose-limiting toxicity, making it a candidate toxicity marker too.
  • Pre-existing conditions influencing response: Every protocol required intact marrow, liver, kidney, cardiac and pulmonary function. Diabetes, chronic kidney disease and heart failure were either exclusions or unstudied, so response in those groups is unknown.

Discontinuation & Cycling

  • Not designed for continuous long-term use: All schedules were cyclical and finite, running to disease progression or unacceptable toxicity. No study administered 17-DMAG continuously, and no chronic-use safety data exist for any duration.
  • Cycling is intrinsic to every tested schedule: Twice-weekly protocols dosed 2 weeks in 3, or on days 1, 4, 8 and 11 of a 21-day cycle. Rest periods served marrow and gastrointestinal recovery, not efficacy.
  • Senolytic hit-and-run rationale: If the compound acts by clearing senescent cells, benefit should persist after withdrawal and repeat courses would be spaced by months. This is a hypothesis drawn from the senolytic literature and has never been tested with 17-DMAG.
  • No withdrawal syndrome described: No trial reported rebound, dependence or discontinuation symptoms. The heat shock response induced by dosing is transient, returning toward baseline within 96 hours, so an abrupt physiological rebound is not expected.
  • No tapering protocol: Because dosing is intermittent and intravenous, tapering does not apply. Trials stopped abruptly at progression or toxicity, and dose reductions rather than tapers were used for tolerable adverse events.
  • Reversibility of adverse effects on stopping: Ocular, hepatic and gastrointestinal events resolved after discontinuation in both human and animal studies, whereas the fatal renal event progressed after the final dose, so stopping does not guarantee reversal.

Sourcing and Quality

  • No pharmaceutical-grade source exists: 17-DMAG has never received marketing approval in any jurisdiction. There is no licensed product, no compounding monograph and no legitimate prescription route outside an investigational protocol.
  • Research-grade material is not a substitute: The compound sold by chemical suppliers is labelled for laboratory research only, typically at 95–99% purity, and is neither sterile, endotoxin-tested, nor manufactured to pharmaceutical quality standards.
  • What third-party testing would need to cover: For a parenteral benzoquinone, meaningful verification means certificate-of-analysis identity confirmation, high-performance liquid chromatography purity, residual solvent limits, bacterial endotoxin testing and sterility — a battery research suppliers do not perform.
  • Salt form and stability matter: Trials used alvespimycin hydrochloride, which is water-soluble and was the whole point of the molecule’s design. The benzoquinone core is light- and oxidation-sensitive, so stability depends on protection from light and frozen storage.
  • Trial supply was centrally controlled: Clinical material came from the National Cancer Institute’s Developmental Therapeutics Program and from Kosan Biosciences, both of which held a direct interest in the compound and neither of which supplies it outside a protocol.
  • Grey-market supply carries specific hazards: Because dose-limiting toxicity in trials tracked exposure, and exposure varied roughly 8.5-fold between individuals at a fixed dose, unverified concentration or purity compounds an already unpredictable exposure into an unmanageable one.

Practical Considerations

  • Time to effect: Target engagement is fast — HSP72 induction is measurable within 24 hours and sustained for 96 hours. Tumour responses in trials took weeks to months. Any senolytic or ageing effect has no measured human onset at all.
  • Common pitfall — treating mouse healthspan data as human evidence: The progeroid mouse and Interventions Testing Program reanalysis findings are the reason for longevity interest, but both are rodent survival endpoints, one sex-specific and one in an artificially accelerated ageing model.
  • Common pitfall — assuming shelving means safety was acceptable: Development stopped after a treatment-related death, cardiac dose-limiting toxicity and reversible keratitis. The narrow window between engagement and harm is the central practical fact about this compound.
  • Common pitfall — assuming an oral route is available: 17-DMAG is orally bioavailable in animals, but every human study used intravenous infusion. No oral human dose, schedule or bioavailability figure has been established.
  • Regulatory status: Unapproved worldwide. It is neither a dietary supplement nor a prescribable medicine; possession and use outside an authorised trial fall outside any regulatory framework, and no off-label pathway exists because there is no on-label use.
  • Cost and accessibility: Effectively inaccessible. No approved product can be bought at any price, and infusion plus the cardiac, ocular, renal and pharmacokinetic monitoring the safety record demands would require oncology-level infrastructure.
  • Payer incentives favour the cheap alternatives: Competing senolytic candidates such as fisetin, quercetin and generic dasatinib cost orders of magnitude less. Insurers and national health systems have a structural financial reason to fund trials of those instead, which plausibly shapes which senolytic questions get asked.

Interaction with Foundational Habits

  • Sleep: Direct and negative. Fatigue was among the most frequent drug-related events on every schedule, and gastrointestinal symptoms disrupt overnight rest in the 48 hours after infusion. No study measured sleep architecture. The practical implication is scheduling infusions to allow two recovery days rather than dosing before demanding periods.
  • Nutrition: Direct and bidirectional. Grapefruit juice inhibits CYP3A4 and raises exposure, and every trial protocol excluded it. Nausea and anorexia reduce intake exactly when volume status protects the kidney, making deliberate fluid and electrolyte replacement around dosing more important than any specific dietary pattern.
  • Exercise: Indirect and blunting. Exercise induces the same heat shock response the drug induces, so hard training on dosing days is mechanistically redundant and adds cardiac demand at the moment of highest cardiac risk. Light activity between cycles is unaffected; no trial measured performance outcomes.
  • Stress management: Indirect and potentiating in the mechanistic sense. Psychological and physiological stress both signal through the heat shock system that 17-DMAG manipulates. No human study measured cortisol or stress response on treatment, so this remains a mechanistic prediction rather than an observed interaction.

Monitoring Protocol & Defining Success

Baseline assessment matters here because every published trial used it as a gate rather than a record: participants with abnormal cardiac, renal, hepatic or haematological values were excluded, so the known safety profile does not extend to them. A complete pre-treatment panel covers liver enzymes, bilirubin, creatinine with estimated glomerular filtration rate (eGFR, a calculated measure of kidney filtering capacity), sodium, albumin, a complete blood count with differential, an electrocardiogram, an ejection fraction measurement and a formal ophthalmological examination including slit-lamp assessment.

Ongoing monitoring follows the tempo of the observed harms. Chemistry, blood count and electrocardiogram are checked before every dose and again at 48 hours, when the abrupt renal and hepatic events appeared. Ejection fraction is repeated every 8–12 weeks, and the ophthalmological examination every 4 weeks while dosing continues, then at 3 and 6 months after the last dose.

Biomarker Optimal Functional Range Why Measure It? Context/Notes
ALT and AST ALT 10–25 U/L (men), 10–20 U/L (women); AST similar Detects the hepatic injury that was dose-limiting in animals and rose to grade 4 in a patient ALT = alanine aminotransferase, AST = aspartate aminotransferase, both liver enzymes. Conventional labs flag only above 40 U/L, which is far too late here. Fasting not required
Total bilirubin 0.2–0.8 mg/dL Confirms hepatic excretory function, a trial entry gate Conventional range extends to 1.2 mg/dL. Fasting sample preferred; interpret with a Gilbert syndrome history (a common, harmless inherited cause of mildly raised bilirubin)
Creatinine and eGFR eGFR above 90 mL/min/1.73 m²; creatinine mid-reference Renal failure was the single fatal event and appeared within days Trials excluded creatinine clearance below 50–60 mL/min. Pair with cystatin C in people with high muscle mass. Hydrate before sampling
Serum sodium 138–142 mmol/L Grade 3 hyponatraemia occurred at a dose as low as 5 mg/m² Conventional range 135–145 mmol/L; trials excluded below 130 mmol/L. Best paired with serum osmolality when low
Serum albumin 4.2–5.0 g/dL Grade 3 hypoalbuminaemia accompanied dose-limiting toxicity and tracks fluid shifts Conventional lower limit is 3.5 g/dL. Interpret alongside C-reactive protein, since inflammation lowers albumin independently
Complete blood count with differential Neutrophils above 1,500/mm³; platelets above 150,000/mm³; haemoglobin above 13 g/dL (men), 12 g/dL (women) Marrow toxicity was dose-limiting preclinically and neutropenic fever was common clinically Trial floors were 1,500/mm³ and 100,000/mm³. Draw before each dose and at the expected nadir 7–10 days after
High-sensitivity cardiac troponin Below the assay’s 99th percentile; otherwise track change from the individual’s own baseline Troponin elevation with infarction was a dose-limiting toxicity No 17-DMAG-specific threshold exists, so serial change matters more than a single value. Avoid sampling immediately after strenuous exercise
Left ventricular ejection fraction 55% or above Heart failure was a dose-limiting toxicity in the combination trial Trials required 40% or above as a floor. Measured by echocardiogram or MUGA (multigated acquisition scan, a nuclear imaging method). Use the same modality each time
Corrected QT interval Below 440 ms Concurrent QT-prolonging drugs were barred and long QT was an exclusion Trial ceilings were 450 ms (men) and 470 ms (women). Correct for heart rate with the Fridericia formula; check potassium and magnesium alongside
Ocular surface and visual acuity No established numeric target; track change from the individual’s own baseline in acuity, tear break-up time and corneal staining Keratitis, dry eye and blurred vision forced dose reductions Requires slit-lamp examination, not a symptom questionnaire. Schedule before the day’s infusion, since acute blurring confounds acuity testing
HSP72 induction in blood mononuclear cells No established target; track fold-change from the individual’s own pre-dose baseline The only validated marker that the drug reached its target HSP72 = heat shock protein 72, the induced chaperone. Research assay only, not clinically available. Highest values occurred in those with dose-limiting toxicity
hs-CRP Below 1.0 mg/L Tracks the inflammatory burden that senolytic therapy is hypothesised to reduce hs-CRP = high-sensitivity C-reactive protein. Not validated as a senolytic response marker. Defer sampling 2 weeks after any infection

Qualitative markers worth tracking alongside the laboratory panel:

  • Visual clarity, glare sensitivity and eye comfort, recorded daily, since ocular symptoms preceded objective findings
  • Fatigue and exercise tolerance, since fatigue was the practical dose ceiling
  • Nausea, appetite and stool frequency in the 72 hours after each dose
  • Muscle and joint pain, and any new numbness, tingling or burning in hands or feet
  • Cognitive clarity and mood, which no trial measured formally but which fatigue of this magnitude typically degrades
  • Sleep quality and continuity in the two nights following each infusion

Defining success is unusually difficult here because no clinical endpoint has ever been validated for 17-DMAG outside oncology. In practice the only demonstrable success is confirmed target engagement without organ toxicity; any claim of a senolytic or longevity benefit currently has no measurable human endpoint attached to it.

Emerging Research

  • No active trials of the compound: None of the seven registered 17-DMAG studies is recruiting. The phase I chronic lymphocytic leukaemia study NCT01126502 was terminated in 2012 after 30 participants, and the phase II breast cancer study NCT00780000 was terminated with four.
  • The unfinished dose-finding record: The National Institutes of Health study NCT00088868 completed in December 2010 with a 40-participant target, pairing dose escalation with client protein assays in tumour tissue. Its pharmacodynamic dataset remains the reference for what target engagement looks like in people.
  • Senolytic repurposing could strengthen the case: Confirmation that intermittent, low-exposure dosing clears senescent cells in humans, as proposed in the senolytic drug class review by Fuhrmann-Stroissnigg and colleagues in 2018, would separate the ageing use from the toxic oncology dosing.
  • The lifespan signal could equally weaken: The Gehan reanalysis by Jiang and colleagues in 2024 was a secondary statistical test that the authors themselves note raises false-positive risk; the effect was male-only and absent under the original analysis. Independent replication would settle it.
  • Ocular toxicity may be designed out of the class: Retinal injury is reproducible in beagle dogs across HSP90 inhibitors, per Kanamaru et al., 2014. The selective successor pimitespib is in the active phase I imatinib-combination study NCT05245968, 78 participants, in gastrointestinal stromal tumours (a rare gut wall cancer).
  • Targeted delivery could widen the therapeutic window: Work on targeted exosome delivery of 17-DMAG by Park and colleagues in 2024 aims to concentrate the compound in target tissue. Success would address the exposure-driven toxicity that stopped development; failure would confirm the window is intrinsically narrow.
  • New indications keep surfacing from unbiased screens: A chemical screen by Fujimoto et al., 2025 identified alvespimycin for diabetic kidney disease. Such findings expand the mechanistic case but remain preclinical, and none has yet produced a registered human trial.

Conclusion

17-DMAG is a water-soluble derivative of a soil-bacterium natural product that blocks a protein-folding helper cells depend on. That single action explains both faces of the compound: it strips growing and worn-out cells of proteins they need, while triggering a stress-repair response in healthy tissue. The longevity interest rests on two rodent findings — clearance of worn-out cells with improved health measures in short-lived mice, and a modest survival gain in male mice that emerged only under a second statistical test.

Against that sits a human record that is small, uncontrolled and unforgiving. Four early-stage cancer studies produced some durable tumour shrinkage, but also a treatment-related death from kidney failure, heart injury, reversible eye damage, nerve injury and marked fatigue, with the gap between a working dose and a harmful one narrow and exposure varying several-fold between individuals for reasons nobody has identified.

The evidence base carries an unusual weight of interest: it was produced almost entirely by the compound’s owner, its acquirer, and the government body that co-developed it, and no independently funded human study exists. Development stopped without the compound ever being tested for the ageing use that now attracts attention, and without any human measurement of that use. Nothing about the current position is settled in either direction.

Top - Benefits - Risks - Protocol