Senolytics in humans: what the trials show

The practical question is whether a particular senolytic improves a health problem enough to justify its risks. A change in a cell marker, a possible improvement on an eye chart and an over-the-counter supplement claim are different kinds of evidence. The findings below reflect research through September 22, 2026. [1] [2] [3] [4] [5] [6] [7] [8] [10] [11]

What the evidence supports

• Shown: early studies suggest that senescence-related measurements can change. Controlled studies have also identified negative primary results and uncertain secondary or subgroup signals, giving more precise questions for further research. [1] [2] [3] [4] [5] [10] [11]

• Not established: a favorable benefit–risk balance for routine dasatinib plus quercetin or fisetin use by otherwise healthy people. [2] [3] [4] [5] [6]

• What would change the assessment: a clinical benefit chosen in advance—such as better walking, fewer fractures or preserved vision—confirmed against an appropriate comparison group, with enough follow-up to judge durability and harm. [2] [3] [4] [5] [8] [11]

What a senolytic is intended to do

A senescent cell has stopped dividing but remains alive and chemically active. Some release inflammatory signals and other substances that affect surrounding tissue. Senolytics aim to disrupt these cells' survival mechanisms so that targeted cells die. Dasatinib plus quercetin, often shortened to D+Q, is one approach; fisetin and locally injected experimental medicines are others. [1] [4] [6]

This mechanism does not mean every senescent cell is harmful or should be removed. A drug might change a laboratory marker without removing the relevant cells from the relevant organ. Even successful removal would still need to improve an outcome that matters to the patient.

The route and molecule matter, too. Oral D+Q, oral fisetin, a knee injection and an eye injection deliver different compounds to different tissues. Success with one would not validate every product called a senolytic. A failure would not disprove every way of targeting senescence.

Six controlled studies, six distinct questions

These studies illustrate why the compound, comparison and outcome must be considered together. Randomization assigns treatment by chance. Masking, or blinding, means participants or assessors do not know the assignment. A sham procedure mimics aspects of treatment without giving the study medicine.

The primary outcome is the main measurement chosen in advance; secondary outcomes are additional measurements. Exploratory findings suggest questions to test in future studies. Adverse events are health problems reported during a study; they are not necessarily caused by treatment.

In idiopathic pulmonary fibrosis, lung tissue becomes scarred. Diabetic macular edema is swelling in the central retina, the light-sensitive tissue at the back of the eye. Both UBX1325 trials enrolled people who had responded inadequately to earlier eye treatment. They tested an injection into the eye, not an oral supplement for general aging. [3] [8] [11]

Why the main outcome comes first

The bone study asked whether D+Q reduced a marker of bone breakdown after 20 weeks. It did not. A bone-building marker improved relative to control at weeks 2 and 4, but the difference was no longer significant at week 20. Women with higher baseline senescence-related measurements showed an exploratory signal. That finding offers a hypothesis about whom to study next; it does not make the overall trial a positive treatment result or demonstrate fewer fractures. [2]

The fisetin knee study is particularly relevant to joint-pain supplement claims. Investigators randomly assigned 34 participants to fisetin and 40 to placebo and found no significant benefit across pain, function or cartilage measures. This was a conference report, with less methodological detail than a full trial paper, but the null result deserves attention. It does not establish that every other formulation would behave identically. [4]

BEHOLD illustrates a different problem: a favorable-looking estimate with substantial uncertainty. Its 5.6-letter vision difference was a secondary outcome in a small study primarily focused on safety. The confidence interval ranged from a small disadvantage to a larger advantage. The published finding therefore supports further investigation, not a conclusive claim that vision improved because of the medicine. [11]

ASPIRE compared UBX1325 with aflibercept, an established eye treatment. Its noninferiority test asked whether UBX1325 was no worse by more than a predefined margin. That is different from testing whether it was better. The main analysis, averaging weeks 20 and 24, missed this goal. The sponsor reported noninferiority at week 36; both the later favorable finding and the unsuccessful main analysis belong in the account. [8]

What early pilot studies can tell us

A 2019 study in nine people with diabetic kidney disease compared tissue and blood samples before D+Q treatment with samples taken 11 days after a short course. Its original report described decreases in several senescence-related measurements. A 2020 correction states that reanalysis changed some conclusions. The original report must therefore be read with that correction; its initial list of marker reductions cannot be accepted unchanged. The uncontrolled study did not establish better kidney outcomes. [1] [10]

STAMINA, published in 2025, included 12 older adults with mild cognitive impairment and slow walking who completed the study. It was an open-label pilot: participants and investigators knew the treatment, and there was no randomized placebo group for cognition or mobility. Over 12 treatment weeks, with final assessment around week 14, the overall change on the MoCA screening test of thinking and memory was 1.0 point, with a 95% confidence interval of −0.7 to 2.7. The result was statistically uncertain. [5]

A subgroup with lower starting scores looked more promising. A separately recruited biomarker comparison group did not provide a randomized control for the cognitive results. Without that control, ordinary variation, familiarity with repeated tests and movement back toward a person's usual score remain possible explanations. A favorable exploratory subgroup needs independent confirmation. [5]

The 2026 TROFFi paper describes the rationale and design of a placebo-controlled fisetin study in postmenopausal breast cancer survivors previously treated with chemotherapy. It sets out planned physical-function measurements; it does not report a successful treatment result or establish current trial availability. [6]

Counting “human senolytic trials” misses these distinctions. A protocol describes a plan, a pilot may establish feasibility, and a controlled trial can test a specific benefit. The completed study, its outcome and its comparator tell us more than the count.

What small safety studies leave unanswered

Small studies can identify frequent, immediate problems. They are less able to reveal uncommon events, delayed harm or risks of repeated treatment over years. Participants may also be selected to exclude medical conditions or interacting medicines.

The lung pilot makes the limit concrete. There were six participants in each group and 65 nonserious adverse events with D+Q versus 22 with placebo. These are event counts, not numbers of affected people. No serious event was attributed to D+Q, but six treated participants cannot provide broad reassurance about safety. [3]

Dasatinib first received US approval in 2006. Its August 2026 US prescribing information covers specified Philadelphia chromosome-positive leukemias—particular forms of blood cancer—not a senolytic regimen for longevity. The earlier July 2024 US label also concerns leukemia treatment. Warnings include low blood-cell counts, bleeding, fluid retention, cardiovascular problems, pulmonary arterial hypertension—high pressure in the arteries supplying the lungs—and important medication interactions. [9]

Much of this safety experience comes from cancer treatment. Its event rates cannot simply be assigned to intermittent experimental use; a shorter schedule also cannot make the risks disappear by assumption.

Quercetin and fisetin should not inherit a safety guarantee from being present in food. Concentrated products, formulations and patterns of use are different exposures. In the fisetin knee trial, reported adverse-event measures did not significantly differ between groups. That does not establish long-term safety for every retail product or repeated regimen. [4]

A useful safety discussion asks who was excluded, what monitoring occurred, how adverse events were collected and how long follow-up lasted. “Well tolerated in this selected group during this study” has a narrower meaning than “safe.”

How Healthy Longevity Clinic experts evaluate the evidence

For a reader hoping to preserve mobility, protect bone or maintain cognition, the decisive question is which patient outcome improved. The bone study's transient formation-marker change did not become a positive primary result or evidence of fracture prevention. The fisetin joint trial did not improve pain or function. The eye studies address a different disease, molecule and route, so they cannot validate an oral longevity regimen. [2] [4] [8] [11]

These findings support a specific practical conclusion: the evidence does not establish a favorable benefit–risk balance for routine D+Q or fisetin use solely for longevity. That conclusion leaves room to investigate better-defined populations and targeted approaches. It also keeps the discussion focused on the person's actual health problem and the relevant alternatives, rather than on “senolytic” as a product category.

Evidence that would change the assessment should identify the population and main clinical outcome in advance, use a comparator that can distinguish a treatment effect from ordinary change, and follow participants long enough to assess durability and harm. Better senescence measurements might help select patients, but a test used to choose treatment must itself be validated. A marker shift is not a reason on its own to intensify an experimental medicine. [2] [3] [4] [5] [8] [11]

A clinical-trial discussion can address these uncertainties, oversight and eligibility. It should not promise rejuvenation or assume access from the existence of a registration.

Three questions for a clinician or trial team

1. What exact problem would this intervention treat, and which controlled human study supports that outcome in people like me?

2. Does the proposed molecule, formulation and route match the study, and what interactions, exclusions and monitoring matter?

3. What alternative is the intervention being compared with, how long would benefit and harm be followed, and what happens if an adverse effect occurs?

Common questions

If a senescence marker falls, have harmful cells been removed?

A marker change can support a biological hypothesis, but it does not by itself establish removal of the relevant cells from the relevant organ or a patient benefit. The corrected kidney pilot also shows why later corrections matter to the original claim. [1] [10]

Does a failed primary outcome mean all senolytics have failed?

No. It limits the claim for the particular compound, population and outcome tested. It does, however, prevent a favorable secondary measurement or subgroup from being presented as though the main clinical question had been answered positively. [2] [7] [8]

Is fisetin safe because it occurs in foods?

Food exposure does not establish the safety of concentrated products or repeated experimental use. The small knee trial's similar adverse-event measures cannot settle uncommon or long-term risks. [4]

Does dasatinib's approval cover anti-aging use?

Its US prescribing information is for specified leukemias. That approval does not establish a favorable benefit–risk balance for senolytic longevity use, and its safety and interaction warnings remain relevant. [9]