The Organ Age Blood Test: Your Body Does Not Age at One Speed

Ten ages instead of one

For as long as medicine has existed it has worked with a single number for age, and everyone has always known the number is a lie. We have all met the sixty-year-old with the arteries of a forty-year-old and the liver of a seventy-five-year-old. What has been missing is the instrument.

That instrument now exists, and it is disarmingly simple. Draw one vial of blood. Measure thousands of the proteins floating in the plasma. Isolate the small subset whose genes are expressed far more strongly in one tissue than anywhere else — these are the proteins the heart, or the kidney, or the brain, sheds into circulation as it works and wears. Feed the pattern to a machine-learning model trained on tens of thousands of people, and it returns an age for that organ alone. Subtract your birthday. What remains is your organ age gap.

The proposition is seductive in exactly the way luxury medicine likes: not one crude verdict but ten precise ones, each supposedly pointing at what to fix. It is also, at this moment in August 2026, the fastest-moving idea in the biological-age field — three landmark cohorts inside eighteen months, a Stanford spin-out selling access through private clinics, and an imaging rival arriving from Columbia.

It deserves the attention. It also deserves a closer reading than it is getting, because the most striking result in the whole literature is the one that argues against the way the test is being marketed.

The number that started it: one in five

The founding paper came out of Tony Wyss-Coray's laboratory at Stanford and made the cover of Nature in December 2023.^[1] His group built clocks for eleven major organs and estimated organ age reproducibly across five independent cohorts totalling 5,676 adults spanning the adult lifespan.

Three findings from that paper have set the terms of everything since.

First, the distribution. Nearly 20% of the population showed strongly accelerated aging in a single organ. Only 1.7% were what the authors called multi-organ agers, aging quickly across the board. Aging is, for most people, focal rather than general — which is precisely why one whole-body number was never going to be enough.

Second, the consequence. Accelerated aging in an organ conferred a 20–50% higher mortality risk, and the organ-specific associations behaved as you would hope: people with accelerated heart aging carried a 250% increased risk of heart failure.

Third, the finding that made neurologists sit up. Accelerated brain and vascular aging predicted progression to Alzheimer's disease independently of, and as strongly as, plasma pTau-181 — the best blood-based Alzheimer's biomarker currently available.^[1] A general-purpose aging assay matching a purpose-built disease biomarker is not a small thing.

2026: the clocks leave the West

The obvious objection to any model trained on one population is that it may be measuring that population. In late 2025 a group led from Oxford, Harvard and Peking University answered it at scale, in a paper published in Nature Aging's 2026 volume.^[2]

They trained an organismal clock plus ten organ-specific clocks on the plasma proteomes of 43,616 UK Biobank participants, then tested them in a Chinese cohort of 3,977 and an American cohort of 800. Cross-cohort correlations came back at r = 0.98 and 0.93. Whatever the clocks are reading, they read it consistently across three continents.

Accelerated organ aging predicted disease onset, progression and mortality beyond clinical and genetic risk factors — the standard that separates a genuine biomarker from an expensive restatement of cholesterol and blood pressure. And the organ most strongly linked to mortality was not the heart. It was the brain.

The brain clock proved unusually rich. It tracked lifestyle, brain structure, and two genes, GABBR1 and ECM1. Alongside the artery clock it linked synaptic loss, vascular dysfunction and glial activation to cognitive decline. Most provocatively, it stratified Alzheimer's risk within APOE haplotypes: a super-youthful brain appeared to confer resilience even to carriers of APOE4, the field's most feared common variant.^[2] That is the first serious suggestion that a modifiable-looking measure might sit between a fixed genotype and its outcome. It is also, at present, an observation and not a mechanism.

Twenty years of follow-up, and a result nobody advertises

The longest look comes from Whitehall II, the British civil-service cohort. Mika Kivimäki's group at UCL took plasma collected from 6,235 middle-aged Londoners in 1997–99, computed nine organ age gaps using the openly published organage tool, and then followed those people through national health records for a mean of 19.8 years — 123,712 person-years in total.^[3]

Of 45 age-related diseases assessed, large organ age gaps predicted 30. On its face, a triumph.

Read the breakdown, and something else appears.

Only six of the thirty diseases were associated exclusively with accelerated aging of their own organ. The other twenty-four were linked to more than one organ age gap, or to organ age gaps with no anatomical relationship to where the disease appeared. An aged arterial clock raised the risk of developing two or more diseases in different organs by a hazard ratio of 2.03; the kidney clock, 1.78; the brain clock, 1.52.^[3]

This is the result that should govern how anyone reads their report. The organ clocks are excellent instruments of prognosis and mediocre instruments of localisation. They tell you, with real statistical force, that something systemic is running fast. They are much weaker at telling you where to point the intervention — which is exactly what the phrase "organ-specific" is doing in the sales copy.

The imaging rival

Proteomics is not the only claimant. In October 2025 a Columbia-led consortium published seven MRI-based organ age gaps — brain, heart, liver, adipose tissue, spleen, kidney and pancreas — derived from 313,645 individuals, then mapped them against 2,923 plasma proteins, 327 metabolites and nearly 6.5 million genetic variants.^[4] The analysis surfaced 53 genome-wide significant locus pairs and prioritised nine druggable genes as candidate anti-aging targets.

A third approach, published in Cell Reports in June 2026, drops below the organ to the cell type: circulating senescence signatures across fourteen human cell types, tested in 1,275 participants of the Baltimore Longitudinal Study of Aging and 997 from InCHIANTI. Pooled senescence proteins outperformed non-senescence proteins at predicting clinical parameters, and the immune-cell senescence signature was associated with mortality and future disease onset.^[5] It is the same intellectual move made one level finer — and it connects the organ clocks directly to the case for clearing senescent cells.

Three independent modalities — plasma proteins, MRI, cell-type senescence — converging on the same claim that aging is organ-heterogeneous is more persuasive than any one of them alone.

What it costs, and what you can actually buy

Vero Bioscience, co-founded by Wyss-Coray to commercialise the Stanford work, markets a proteomic OrganAge test built on more than five thousand measured proteins. Reporting through 2025 put the intended at-scale consumer price around $200. The reality in mid-2026 is narrower: access runs through Human Longevity's clinics rather than as an open purchase, effectively behind a concierge membership.

Two things follow. The first is that this is not an approved diagnostic, and no physician can order an organ age panel the way they order a liver function test. The second is more useful: the underlying mathematics is public. The organage package used by the Whitehall II investigators is openly available, which means the science is auditable even where the product is not — a distinction worth holding onto in a field where the two are routinely conflated.

The honesty clause

Four limitations deserve stating plainly, because none of them appears in a clinic brochure.

Every result here is observational

Not one trial has shown that acting on an organ age gap changes an outcome. The chain from measurement to intervention to benefit has never been closed — the same gap that runs through epigenetic age reversal and, most expensively, through plasma exchange.

Reverse causation is difficult to exclude

An organ shedding an aged protein signature may be an organ in the early, undiagnosed stages of disease. The clock would then be detecting subclinical pathology and calling it accelerated aging. Whitehall II excluded baseline cases and still found a twenty-year signal, which helps — but a decade of silent pathology is entirely possible.

Platform dependence is unresolved

The Stanford clocks were built on SomaScan aptamer measurements; the UK Biobank clocks on Olink's antibody-based panel. The two platforms do not measure identical protein sets and do not always agree. Cross-platform correlations within a study say nothing about whether your reading from one provider would survive at another.

Some organ clocks compete with cheap standards

A proteomic kidney age must justify itself against eGFR and albumin-to-creatinine, which cost almost nothing and are supported by decades of outcome data. The genuinely novel readouts are the brain, artery and immune clocks — the compartments where no cheap functional test exists. That is where the value is, and it is not evenly distributed across the ten numbers on the report.

The royal verdict

The organ clocks are the best biological-age instrument yet built, and they are being sold for the wrong reason.

They are sold as targeting: find the failing organ, direct the protocol, restore the balance. The twenty-year data does not support that. In Whitehall II, four-fifths of the diseases predicted did not respect organ boundaries at all. What the clocks appear to detect is a systemic process that surfaces first, and most legibly, in whichever tissue happens to be least resilient in a given body.

What they are genuinely excellent at is prognosis — and the honest version of that sentence is uncomfortable. The single most predictive number in the 2026 dataset is brain age, the organ against which we have the fewest credible interventions.^[2] The test's finest signal points at the place where medicine can currently do least. Whether the klotho line of research or anything else eventually changes that is an open question.

There is, all the same, a real case for knowing. A 20–50% mortality differential is not a wellness metric; it is the kind of number that reorganises a decade. If a brain or arterial gap moves cardiovascular risk management, sleep and training from intention to appointment, the two hundred dollars will have been the cheapest part of the exercise. Just do not mistake ten decimal places for ten instructions. The instrument has become extraordinarily precise about a question — how fast, and where — that we still cannot answer with a prescription. Precision and power are different things, and the gap between them is where most of the money in longevity is currently being spent. As with the inflammatory clock, the measurement is real; the promise attached to it is running ahead.

Common questions