🧬 Japan approved the world's first medicine grown from iPS cells back in March. On July 31 the company behind it finally attached a date to the part that decides whether the approval meant anything: October, Kyoto University Hospital, one patient. Thirty-four more follow over roughly three and a half years, and only then will anyone be able to say the regulators were right.

The date that turns an approval into a treatment

At its quarterly results briefing on July 31, 2026, Sumitomo Pharma said the first transplant of AMCHEPRY since the product went on sale is scheduled for October at Kyoto University Hospital. One or two patients will be treated before the Japanese fiscal year ends in March 2027. The full group of 35 is due to be transplanted by the first half of fiscal 2029.

Seven hospitals will share the work: Kyoto University Hospital, two more in Kansai, the region around Osaka and Kyoto; two in Kanto, the region around Tokyo; one in Tohoku in the north; and one in Chubu, the central belt that includes Nagoya.

AMCHEPRY, whose international non-proprietary name is raguneprocel, received conditional and time-limited approval on March 6, 2026. It was not alone. Cuorips' iPS-derived cardiac muscle sheet, ReHeart, cleared the same panel and was approved the same day, so the world's first two products of this kind arrived together rather than in sequence. Japanese regulatory coverage at the time described the pair as the first products anywhere derived from pluripotent stem cells, whether iPS or embryonic. Both licences run seven years and expire on March 5, 2033. The decision rested on an investigator-initiated trial at Kyoto University Hospital involving seven patients aged 50 to 69, published in Nature in April 2025. Among the six assessed for efficacy, four improved on the standard motor rating scale measured during off periods, and the grafts produced dopamine without forming tumours. The authors were candid about the limits of their own study: it was open-label with no control group, which leaves placebo effect and observer bias in play.

Seven patients is a thin basis for licensing a medicine, which is precisely why the approval came with conditions attached. The 35 are the price of the licence.

Why 35 patients takes three and a half years

The product is a suspension of dopaminergic neural progenitor cells, the immature cells that grow into the dopamine-producing neurons Parkinson's destroys. They are grown from a healthy donor's iPS cells, then delivered by stereotactic surgery through small holes in the skull into the putamen, a deep brain structure involved in movement.

None of that scales like a pill. The cells ship unfrozen, which means they cannot be stockpiled; manufacturing has to be timed to the operating theatre rather than to a warehouse. Enrolling a patient and getting them to surgery takes up to six months. The study fills in sequence: 30 patients aged 18 to 65 first, and only once all 30 have been transplanted do the remaining five, all over 65, begin. Each patient is then followed for two years before the routine surveillance phase starts.

Sumitomo Pharma expects to file for full approval in fiscal 2032, six years after the conditional one.

A provisional licence, and what it costs

The route Japan used was created in the 2014 revision of the Pharmaceuticals and Medical Devices Act. If safety is confirmed and efficacy can be reasonably estimated, a regenerative product can be approved for a period of up to seven years, during which the developer gathers the evidence that would normally come first. A 2025 review in a Japanese regulatory science journal counted more than 20 regenerative products approved in Japan as of May that year, six of them by this route.

Approval brings public insurance coverage immediately, which is both the appeal and the exposure. AMCHEPRY was listed on May 20, 2026 at 55,306,737 yen per patient, roughly $348,000 at the rate of 159 yen to the dollar on July 31, 2026. The national health insurance system pays while the evidence is still being assembled.

That exposure is not hypothetical. Two products that took this route before AMCHEPRY were struck from the reimbursement lists in 2024 after failing to prove themselves. Terumo's HeartSheet, cleared in 2015 on a single-arm study at three hospitals, could not show any advantage over an external comparison group on its primary endpoint, and its approval was wound up in July 2024. The number of patients behind that 2015 decision was also seven. Anges's Collategene lapsed in June of the same year, after an open-label post-marketing survey failed to reproduce the double-blind trial result that had won it approval in the first place. Both cases pushed Japan's central insurance council to open a review, in October 2025, of how conditionally approved regenerative products should be priced and covered at all. The drug-safety watchdog group Yakugai Ombudsperson went further in April 2026, formally asking the health ministry to revoke both iPS approvals and refuse them insurance coverage.

ReHeart, developed by the Osaka University spinout Cuorips, is a few weeks ahead of AMCHEPRY on the calendar. It joins the insurance list on September 1, 2026 at 53.2 million yen, about $335,000, with the first insured transplant targeted for late September as of the company's July 22 comments. Its post-marketing study needs 75 transplanted patients plus 150 untreated ones for comparison.

Washington and Brussels took a different road

That reading of March 2026 as a global first is itself the clearest statement of where the United States and Europe stand. Neither the FDA nor the EMA has licensed anything grown from pluripotent stem cells. What they have approved in this field is overwhelmingly something else: gene-modified versions of a patient's own cells, gene therapies delivered by virus, or products built from adult tissue. The FDA cleared its first mesenchymal stromal cell therapy, Ryoncil, only in December 2024, and the newest advanced therapies cleared in Europe run to umbilical cord cells and genetically modified blood stem cells.

It is not that the West lacks fast lanes, and it would be wrong to paint them as strict. The United States has accelerated approval and RMAT designation; the EU has conditional marketing authorisation and the PRIME scheme. The EMA is explicit that a conditional authorisation rests on less comprehensive clinical data than normally required, granted where immediate availability outweighs the risk of the remaining gaps. The distance from Japan is a matter of degree. American and European regulators still want an effect measured against something, a control arm or at least a surrogate endpoint. Japan's route can approve when efficacy is only estimated.

That margin of degree is enough to explain why Japan has an approved iPS cell product and nobody else does, and why the rest of the field is watching the 35 rather than the approval.

Two answers, arriving out of step

Bayer's subsidiary BlueRock Therapeutics is running the closest competing programme. Bemdaneprocel is an allogeneic pluripotent stem cell-derived cell therapy for Parkinson's, and its Phase 3 trial, exPDite-2, dosed its first patient on September 22, 2025. The design is everything Japan's is not: about 102 participants, randomised, double-blind, controlled against sham surgery. Data are expected in 2027. BlueRock also picked up Japan's own pioneering designation for the product in December 2025.

So somewhere around 2027 the field should get a controlled read on whether transplanted dopamine cells genuinely help. Japan's answer, drawn from 35 patients with no randomised comparison, is not due until that 2032 filing.

The stakes are not abstract. The health ministry's triennial patient survey counted about 250,000 people being treated for Parkinson's in Japan in 2023.

In October a patient in Kyoto will receive a treatment the world has never commercially delivered, paid for by public insurance, while the proof is still being collected. Somewhere else, the same patient would be waiting for a trial to read out. Which of those two bargains would your country's health system take?

参照