CAR-T made inside the body: 15-month data on four myeloma patients

A team in Wuhan, China, has published 15-month follow-up on an experimental cancer therapy that does something conventional CAR-T cannot: it builds the engineered immune cells inside the patient's own body [s1]. The approach, called ESO-T01, was given to four people with relapsed or refractory multiple myeloma. All four responded — but the manufactured cells did not last, and the results read as a proof of concept with a clear ceiling rather than a breakthrough [s1].

CAR-T therapy re-programmes a patient's T cells to recognise and kill cancer. In its standard form it is a feat of logistics: doctors harvest the patient's T cells by apheresis, ship them to a factory to be genetically engineered over weeks, give chemotherapy to clear out the existing immune cells ("lymphodepletion"), then infuse the finished product back. That pipeline is slow, costly, and out of reach for much of the world. In vivo CAR-T aims to collapse it into a single injection that delivers the engineering instructions directly into the bloodstream, letting the body manufacture the cells itself — no apheresis, no factory, no lymphodepleting chemotherapy [s1].

What the study did

ESO-T01 targets B-cell maturation antigen (BCMA), a protein on myeloma cells and the same target as several approved conventional CAR-T products. The researchers had already reported preliminary safety and efficacy in the same four patients with up to three months of follow-up; the primary endpoints were safety and tolerability, with efficacy and the behaviour of the cells in the body as secondary measures [s1]. This paper extends the observation window to a maximum of 15 months to ask the questions that matter most for a brand-new platform: does it cause delayed harm, and does the benefit hold [s1].

The companion trial registry lists ESO-T01 as an early-phase-1, single-centre study sponsored by Union Hospital at Tongji Medical College, with a planned enrolment of 24 patients [s2]. As of the registry's latest entry the study is marked suspended — a reminder of how early this work sits [s2].

What it found

On efficacy, the early signal was striking: the objective response rate was 100%, with two patients reaching a stringent complete response and two a partial response [s1]. On safety, every patient developed cytokine release syndrome — the surge of inflammation that follows CAR-T activation — in a distinctive two-phase pattern, graded 1 to 3, along with grade 3 or worse drops in blood counts [s1]. One patient had grade 1 neurotoxicity. Reassuringly for a genetic therapy, the extended follow-up turned up no immune reactions against the construct and no toxicity linked to the DNA inserting itself in the wrong place, and no important new adverse events emerged beyond the first report [s1].

The durability is where the honesty lies. Of the four patients, only one maintained a stringent complete response out to 15 months [s1]. The median progression-free survival was just 4.0 months, with individual values ranging from 3.0 to 15.0 months [s1]. Three patients relapsed or progressed — all with disease that had spread outside the bone marrow — and all three eventually died during follow-up, though salvage treatments extended overall survival to between 5.5 and 10.9 months after the infusion [s1]. The authors attribute the short-lived benefit to limited persistence of the in vivo cells and say the platform needs further optimisation [s1].

How to read it

Four patients is not a basis for conclusions about whether a therapy works, and the authors do not claim otherwise. What this study establishes is narrower and still meaningful: that generating BCMA CAR-T cells inside the body is feasible, that it produced deep responses in every patient tried, and that over more than a year it did not throw off the kind of delayed genetic-safety signals that regulators watch for most closely [s1]. Those are the questions a first-in-human platform has to clear before efficacy is even worth testing at scale.

The ceiling is equally clear. The engineered cells faded, responses were mostly brief, and the one durable remission is hard to generalise from a single patient [s1]. The promise of in vivo CAR-T is access — a therapy that could skip the apheresis suites and cell-manufacturing plants that put conventional CAR-T out of reach for most patients worldwide. Whether it can be engineered to persist long enough to match that promise is unsettled. For context on the fast-moving myeloma field, see teclistamab moving to earlier lines of treatment, the approval of the oral CELMoD iberdomide, and an in-body CAR-T approach tested against glioblastoma.

What to watch

The decisive questions are persistence and scale: whether design changes can make in vivo cells expand and survive the way factory-made ones do, and whether larger, multi-centre trials reproduce the response rate without new safety surprises [s1]. Until then this is a landmark for what it demonstrates is possible, not for what it treats.

This article describes early-stage research and is not medical advice. Decisions about myeloma treatment belong with treating clinicians.

Sources

• Extended follow-up of in vivo BCMA CAR-T therapy in relapsed/refractory multiple myeloma — Nature Medicine, 6 October 2026
• ESO-T01 in relapsed/refractory multiple myeloma (NCT06691685) — ClinicalTrials.gov