Today's Parkinson's medicines replace missing dopamine but cannot bring back lost brain cells. Cell therapy tries to replace those cells. Gene therapy tries to change how brain cells work. Both have moved forward quickly since 2023. One stem-cell product is now approved in Japan, and the first large controlled trials are under way. But neither approach is a cure, and most results still come from small studies.
Key points
- In March 2026, Japan gave the world's first approval to a stem-cell treatment for Parkinson's (AMCHEPRY). It is a conditional, time-limited approval, based on a study of only seven people. Its benefit still has to be confirmed.
- Bemdaneprocel, made from embryonic stem cells, is in a phase 3 trial with a sham (fake) surgery comparison. About 102 people are taking part. Results are not expected before 2027.
- Small early studies show the transplanted cells can survive, make dopamine and avoid forming tumours. Some patients improved, but none of these studies had a comparison group.
- Gene therapies (GDNF, GAD, GBA1) are in phase 1 or 2 trials. Earlier gene and cell therapies looked good in small open studies and then failed or stalled. This is a real reason for caution.
- All of these treatments need brain surgery or a spinal injection. Cell transplants from donors usually need about a year of drugs that weaken the immune system.
Why replace cells?
Parkinson's symptoms come mainly from the loss of brain cells that make dopamine. These cells live deep in the brain and send dopamine to an area called the putamen. By the time symptoms start, many of them are already gone.
The idea behind cell therapy is simple. Grow new dopamine-making cells in a lab, then place them in the putamen. If they settle in, they could release dopamine more steadily than pills. Cell therapy does not stop the disease itself, and it is not expected to help symptoms that do not come from dopamine loss, such as balance or memory problems.
Lessons from fetal tissue
Doctors first tried this in the 1980s and 1990s using tissue from aborted fetuses. Two US trials compared transplants with sham surgery. In the 2001 NEJM trial of 40 patients, the transplant group did no better overall after one year. In the 2003 trial of 34 patients, there was again no overall benefit, and 56% of those transplanted developed extra movements (dyskinesia) even without medicine. These are called graft-induced dyskinesias.
The European TRANSEURO trial tried again with better methods. It was published in 2025. Tissue was so hard to get that only 11 people were treated. There was no overall benefit at three years, although major graft-induced dyskinesias were not seen. The authors concluded that a stem-cell source is needed. That is where the field has now moved.
Stem-cell transplants
Two kinds of stem cells are used. Embryonic stem cells come from early embryos left over from IVF. Induced pluripotent stem cells (iPS cells) are made by reprogramming adult cells, such as skin or blood. Either kind can be turned into young dopamine cells in large, standard batches.
| Therapy | Approach | Results so far | Stage |
| AMCHEPRY / raguneprocel (2026) Sumitomo Pharma; Nature 2025 | Donor iPS cells (Kyoto University), placed in both sides of the brain | 7 patients. No serious side effects. Of 6 assessed, 4 improved off-medicine and 5 on-medicine. Dopamine scans up about 45%. Dyskinesia rose in some. | Late-stage Conditional approval, Japan, March 2026 |
| Bemdaneprocel (2025) BlueRock/Bayer; Nature 2025 | Embryonic stem cells, "off-the-shelf", one year of immune suppression | 12 patients. No side effects linked to the cells, no graft-induced dyskinesia. High-dose group improved about 23 points off-medicine at 18 months, about 18 points at 3 years. | Late-stage Phase 3 recruiting |
| STEM-PD (2026) Lund/Cambridge; Nature Medicine 2026 | Embryonic stem cells, one year of immune suppression | 8 patients. No serious side effects from the cells, no graft-induced dyskinesia, no tumours. One died of a lung infection. Scans suggest cells survived. | Early trials |
| Sasineprocel / ANPD001 (2026) Aspen Neuroscience press release | The patient's own skin cells turned into iPS cells; no immune suppression | 8 patients at 12 months: about 2 more hours a day of good "on" time and 13–16 points better off-medicine. Company data, not yet peer-reviewed. | Early trials |
| TRANSEURO (2025) Nature Biotechnology 2025 | Fetal tissue transplants | 11 patients. No overall benefit at 3 years. | Stopped / setback |
What Japan's approval means. Japan has a fast-track system for regenerative medicine. It lets a product reach patients early if it looks safe and likely to help. Sumitomo Pharma must now run a further study and track patients to win full approval within a set period. If benefit is not confirmed, the approval can lapse. The product is approved only for people whose symptoms are not well controlled by medicines, including levodopa. A small US trial of the same cells is under way in San Diego.
Why the phase 3 trial matters. Every stem-cell study so far was "open-label": patients and doctors knew who was treated. Brain surgery itself can create a strong placebo effect, as the fetal trials showed. The bemdaneprocel phase 3 trial, exPDite-2, compares real surgery with sham surgery, and neither patients nor raters know who got which. Its main measure is extra good "on" time without troublesome dyskinesia after 78 weeks. The lead investigator of the earlier study has urged "caution in interpreting the positive trends."
Gene therapy
Gene therapy uses a harmless virus (called AAV) to carry a gene into brain cells. It is usually given once, during surgery. The gene then keeps working, possibly for years. That also means it cannot easily be switched off if problems arise.
| Therapy | Approach | Results so far | Stage |
| AB-1005 / AAV2-GDNF (2025) AskBio (Bayer); Movement Disorders 2025 | Delivers GDNF, a growth factor meant to protect and restore dopamine cells | 11 patients, no comparison group. Mild group stable. Moderate group about 20 points better off-medicine at 18 months. No serious side effects linked to the gene. | Early trials Phase 2 under way |
| AAV-GAD (2011; 2024) Lancet Neurology 2011; MeiraGTx | Calms an overactive brain circuit (subthalamic nucleus) | Sham-controlled 2011 trial: 8.1 vs 4.7 points better at 6 months, a modest edge. Small 2024 study: high dose about 18 points better (company data, 14 people). | Early trials Phase 3 planned |
| VY-AADC (2021) Voyager/Neurocrine; Neurology 2022 | Helps brain cells turn levodopa into dopamine | Small open study: lower medicine needs at 3 years. Phase 2 put on FDA hold in 2020 after abnormal MRI findings. Partner ended the program in 2021. | Stopped / setback |
| PR001 / LY3884961 (PROPEL) Prevail (Lilly); ClinicalTrials.gov | Adds a working GBA1 gene, for people with a GBA1 mutation | Safety study of about 32 people. No results published. | Early trials |
| BIIB094 / ION859 (2026) Nature Medicine 2026; Biogen | Antisense drug, by spinal injection, that lowers the LRRK2 protein | 82 people. Spinal fluid LRRK2 lowered by up to 59%. No serious drug-related side effects. Biogen stopped development in February 2025. | Stopped / setback |
| BIIB101 / ION464 Ionis; Biogen Feb 2025 | Antisense drug that lowers alpha-synuclein | Biogen dropped it in February 2025. Ionis continues a safety trial in multiple system atrophy, not Parkinson's. | Stopped / setback |
GDNF has a long history. Earlier GDNF and similar trials gave mixed results, partly because the treatment reached only part of the target. A US government study reported in 2026 covered only about a quarter of the putamen and found no clear change over five years. The newer AB-1005 method reached about 63%. The ongoing REGENERATE-PD trial compares it with a partial sham surgery, which is the real test.
Antisense drugs are a different idea. They are not permanent. They are given by repeated spinal injections and lower a harmful protein. The LRRK2 study showed the drug hit its target, but it was a safety study and could not show whether symptoms slowed.
Risks to know about
- Brain surgery. Nearly all these treatments are placed through small holes in the skull. Risks include bleeding, infection, seizures and headache. Most side effects in the trials were tied to the surgery, not the treatment.
- Immune suppression. Donor cells usually need about 12 months of anti-rejection drugs. These raise the risk of infections. The STEM-PD team said the main risks in their study came from these drugs. Using a patient's own cells avoids this, but each batch must be made one person at a time.
- Dyskinesia. Fetal grafts caused uncontrolled movements in some people. New stem-cell trials have not seen this so far, but in the Kyoto study dyskinesia rose in people who stayed on full medicine doses.
- Tumours. Stem cells could in theory overgrow. No tumours have been seen on scans so far, but follow-up is short.
- Permanence. Transplanted cells and gene therapy cannot be removed if something goes wrong.
- Unknown long-term effects. The new cells are not expected to stop the disease itself, and they will not fix non-dopamine symptoms.
How strong is the evidence?
Honestly, it is still early. Every positive stem-cell result so far comes from fewer than 15 people per study, without a comparison group. Brain-scan evidence that the cells survive and make dopamine is fairly consistent across studies. That is real progress. Whether people feel meaningfully better than they would after sham surgery is not yet known. In the past, VY-AADC, fetal tissue and several GDNF approaches all looked promising early on and then disappointed.
What's coming next
- Bemdaneprocel phase 3 (NCT06944522). Recruiting in the US, Canada and Australia. The registry lists primary completion as March 2027 (estimate). With a 78-week main measure, results may well come later.
- AMCHEPRY in Japan. Sumitomo Pharma plans a post-marketing study to confirm benefit. US trial NCT06482268 runs to about December 2027.
- AB-1005 REGENERATE-PD (NCT06285643). Now listed as no longer recruiting. Main results estimated August 2028.
- Sasineprocel (NCT06344026). Aspen says it is moving toward phase 3. No phase 3 trial was registered as of October 2026.
- STEM-PD (NCT05635409). Patients will be followed to 3 years.
- AAV-GAD. MeiraGTx said in March 2026 it expected to start phase 3 "in the coming months." No phase 3 trial was registered as of October 2026.
- PR001 (NCT04127578). No longer recruiting. Five-year safety follow-up runs to 2031.
Realistically, outside Japan, no cell or gene therapy for Parkinson's is likely to be widely available before the late 2020s, and only if the controlled trials succeed.
Summary as of October 2026. This is general information, not medical advice. Talk to your neurologist before considering a clinical trial.