🧬 Immunology teaches a tidy division of labor. CD4 T cells give the orders. CD8 T cells do the killing. In the blood of people who have lived past 110, that division has quietly come apart, and a Japanese team has now worked out roughly when it starts. The answer lands somewhere around a person's hundredth birthday.

The officer who picked up a rifle

CD4-positive killer T cells, usually shortened to CD4 CTLs, are the exception. They can attack target cells directly instead of only issuing instructions, and in most people they are a rounding error.

A study published in Cell Reports on August 20, 2026, Japan time, measured how common these cells are at different ages. It comes from researchers at the Institute for Protein Research at the University of Osaka, the Center for Supercentenarian Medical Research at Keio University School of Medicine and the RIKEN Center for Integrative Medical Sciences. Among people aged 70 to 99, CD4 CTLs sat at a median of 4 percent. In centenarians the figure was 9.6 percent. In people past 110 it reached 17.6 percent.

Worn-out T cells carry markers of exhaustion, and these did not: genes such as LAG3 and CTLA4 were expressed at low levels. They were also clonal. In each person a single lineage had multiplied out of all proportion to what surrounded it, with the biggest clone taking up 33.3 percent of that person's CD4 CTLs on average. In one centenarian, a single clone made up 53.8 percent.

The clone was not a monoculture either. When the team stimulated the cells in the lab, cells carrying identical receptors split into eight clusters with different cytokine profiles.

From 28 people to 1,500

The detailed work was done on 28 donors: eight aged 70 to 99, ten aged 100 to 109, and ten aged 110 or older. Their T cells were profiled one cell at a time, gene expression and surface proteins together.

Twenty-eight people is not a population, so the team took the molecular signature they had found, trained a model on it, and turned that model loose on public single-cell data from more than 1,500 people, spanning infancy to the 110s. The AI did not make the discovery; it checked whether the discovery held across a sample no single laboratory could recruit.

It held, and it produced the timing. The proportion stays low through the eighties and nineties, then climbs. Not as a rule, though. The team points out that the pattern was not confined to the very old, and that the single highest proportion in the whole study belonged to a participant who had not yet turned 100.

In November 2019, largely the same group reported in PNAS that supercentenarians carried unusual numbers of these cells, working from seven supercentenarians, five controls aged 50 to 80 and roughly 60,000 cells. That paper established the cells were there. This one establishes when they arrive.

A country that can find ten people over 110

The hard part of the 2026 study was not the sequencing. It was obtaining blood from ten people aged 110 or older, with their consent.

Keio's centenarian research dates to 1992, started by the physician Nobuyoshi Hirose, when Japan had 4,152 residents aged 100 or older. A nationwide study of people aged 105 and up has run since 2002. Hirose is a co-author on the 2026 paper. Yasumichi Arai, who directs the center, described its position in a January 2026 interview: "It is no exaggeration to say that ours is the only center in the world holding data on some 200 supercentenarians aged 110 and over."

As of September 1, 2026, Japan counted 107,677 residents aged 100 or older, the first time the figure has passed 100,000 and the 56th consecutive annual increase. When the count began in 1963, it listed 153 people. The United States, a far more populous country, had an estimated 101,000 centenarians in 2024.

Other countries run long-lived cohorts of their own. The New England Centenarian Study at Boston University, founded in 1995, describes itself as "the largest study of centenarians and their family members in the world," with more than 3,000 participants across its projects. The narrower the age band gets, the more the advantage moves: that study counts centenarians and their relatives, while Arai's claim is about people past 110.

Japan normally discusses these numbers as a bill coming due: pensions, care staffing, schools closing. But a country cannot decide to have supercentenarians. It either has them, has been counting them for decades, and has institutions old enough to have asked them for a blood sample, or it does not. Nobody would trade a pension crisis for a sample frame, but the sample frame is real, and it took thirty years of unglamorous fieldwork to build.

Scars are not strength

Two days after the announcement, a Japanese biology blogger set out the central objection. Picture examining the old soldiers who survived a long war, he wrote, finding them covered in scars, and concluding that scarring young soldiers would make them stronger. "The scars may not be the cause of the strength. They may be what is left after fighting for a long time and surviving." On a Japanese message board, someone made the same point in a line: the causality is the wrong way round.

They are describing survivor bias, and the researchers do not dispute it. The study does not show that having more CD4 CTLs prevents cancer or lengthens life. It cannot. Everyone in it had already lived past 100.

When the team compared the receptor sequences of the expanded clones against a public database, nearly three dozen of the hits came from patients with lung, breast and liver cancer, although no one in the study had been diagnosed with any of them. The team reads that as a possible trace of early immune activity, though Kosuke Hashimoto, the study's first author and an associate professor at the University of Osaka, adds a caveat: "Some CD4 CTLs may recognize cancer-related targets, although their exact targets remain unknown."

The authors set out their own limits. They did not establish what the cells do inside the body, they looked only at cells circulating in blood rather than resident in tissue, and the stimulation experiments were run on a small group. Hashimoto said in August that the next step is to look at tissue. The work ran on Japanese public research grants and domestic medical foundations.

"Immune aging is not simply a process of decline," Hashimoto said. "The selective expansion of certain T cells suggests that, even in extreme old age, the immune system may continue to adapt to age-related challenges."

That is as far as it goes. A system long assumed to be winding down turns out to be rearranging itself instead, in people old enough that almost nobody had ever looked.

Studies like this one get read as diet advice, and there is nothing here to act on. Japan found this because Japan had the people, the records and a head start of more than thirty years. If researchers in your country wanted ten volunteers over the age of 110 next month, could they find them? And is anyone there keeping count?

References