Dead ends
Every landmark marked as a dead end: 52 in all, of which 43 are trials in people. The rest are failed gene claims, ideas that did not hold up, and lab work that went nowhere.
The public version, grouped by approach with what was learned, is What has failed.
| 2027 | ION464 / BIIB101 (synuclein antisense) | Clear the bad protein | Synuclein lowering | Trial in people | 1 | 40 | The first synuclein-lowering drug given to patients, an antisense injected into the spinal fluid, was safe in the first 22 people, but it is being tested in multiple system atrophy (MSA), not Parkinson's, and Biogen dropped it in 2025. | NCT04165486 | 22 Oct 2026 |
| 2026 | LRRK2 antisense BIIB094 | Treat the inherited causes | LRRK2 | Trial in people | 1 | 82 | A spinal-fluid antisense drug cut LRRK2 protein by up to 59% in an 82-person safety trial, but Biogen halted its development in early 2025 and it was never tested for benefit. | NCT03976349 | 19 Oct 2026 |
| 2026 | ProSavin and OXB-102 | Replace or repair what's lost | Gene therapy & delivery | Trial in people | – | – | A lentiviral gene therapy delivering three dopamine-making enzymes looked safe in 15 people in 2014; its successor was stopped in 2022 after a sponsor collapse, with a small 2026 report showing mixed, unpowered results. | NCT00627588 | 19 Oct 2026 |
| 2026 | BIIB122 / DNL151 (LUMA) | Treat the inherited causes | LRRK2 | Trial in people | 2 | 648 | An oral LRRK2 inhibitor cut its blood target by over 90% in 648 people with early Parkinson's yet did not slow progression, so Biogen and Denali dropped it in general Parkinson's and kept only a mutation-carrier study. | NCT05418673 | 19 Oct 2026 |
| 2026 | AZA-PD azathioprine | Look beyond the brain | Inflammation & immunity | Trial in people | 2 | 66 | The first trial of an immune-suppressing drug in early Parkinson's (66 people, 12 months) did not meet its primary measure (off-state gait and axial score at 12 months); exploratory analyses hinted at immune-marker effects. | – | 19 Oct 2026 |
| 2026 | NOPARK: nicotinamide riboside Phase 3 | Help cells clean up | Mitochondria & energy | Trial in people | 3 | 410 | Boosting the cell's energy molecule NAD with a daily supplement did not help 410 people with early Parkinson's and left them slightly worse than placebo after a year. | NCT05589766 | 19 Oct 2026 |
| 2026 | BIA 28-6156 (ACTIVATE) in GBA1 carriers | Help cells clean up | Lysosome & GBA | Trial in people | 2 | 273 | An oral pill that raises activity of the recycling enzyme GCase, tested for 18 months in 273 people with GBA1-linked Parkinson's, left progression no different from placebo, and Bial stopped the drug for this use. | NCT05819359 | 19 Oct 2026 |
| 2026 | EFFACE-PD faecal transplant (Warsaw) | Look beyond the brain | Gut–brain origin | Trial in people | 2 | 59 | In a 59-person Polish placebo-controlled trial, one colonoscopic faecal transplant did not change the Parkinson's motor score over 12 months. | NCT05204641 | 18 Oct 2026 |
| 2026 | CAP-003 intravenous GBA1 gene therapy (Capsida) | Replace or repair what's lost | Gene therapy & delivery | Trial in people | 2 | – | Capsida's planned 31-person trial of an intravenous gene therapy for GBA1-linked Parkinson's was withdrawn with no patients dosed, after a child died in the company's trial of a drug using the same engineered virus shell. | NCT07011771 | 19 Oct 2026 |
| 2026 | TRANSEURO fetal grafts | Replace or repair what's lost | Cell replacement | Trial in people | – | 11 | Only 11 people could be grafted because fetal tissue was scarce, and 3 years later the group showed no overall clinical gain, which argued for a stem-cell source of dopamine neurons. | NCT01898390 | 22 Oct 2026 |
| 2026 | AMBITIOUS: ambroxol in GBA carriers | Help cells clean up | Lysosome & GBA | Trial in people | 2 | 65 | The first placebo-controlled ambroxol trial in people with Parkinson's and a GBA1 mutation found no slowing of cognitive decline over 52 weeks, in a preprint that has not yet been peer reviewed. | NCT05287503 | 19 Oct 2026 |
| 2026 | Vodobatinib: PROSEEK Phase 2 | Keep vulnerable cells alive | Cell survival & anti-apoptosis | Trial in people | 2 | 513 | A c-Abl blocker built to enter the brain, with spinal-fluid levels shown to be high enough, did no better than placebo on motor decline at 40 weeks and was stopped in April 2024. | NCT03655236 | 19 Oct 2026 |
| 2025 | Exenatide Phase 3 (Exenatide-PD3) | Look beyond the brain | Metabolic drug repurposing | Trial in people | 3 | 194 | The first Phase 3 of a GLP-1 drug in Parkinson's, 194 people for two years, found no benefit from weekly exenatide, and in June 2026 the journal added an expression of concern about one trial site. | NCT04232969 | 19 Oct 2026 |
| 2025 | Stockholm exenatide trial | Look beyond the brain | Metabolic drug repurposing | Trial in people | 2 | 59 | A 59-person Swedish Phase 2 found exenatide shifted brain metabolic networks on PET scans but gave no movement benefit, matching the UK Phase 3. | NCT04305002 | 19 Oct 2026 |
| 2024 | ORCHESTRA: minzasolmin | Clear the bad protein | Synuclein misfolding & spread | Trial in people | 2 | 496 | The first pill designed to stop alpha-synuclein misfolding inside cells, minzasolmin (UCB0599), showed no clinical benefit in a Phase 2 trial of about 500 people. | NCT04658186 | 22 Oct 2026 |
| 2024 | Buntanetap Phase 3 | Clear the bad protein | Synuclein lowering | Trial in people | 3 | 523 | Annovis Bio's oral drug buntanetap, said to lower several toxic proteins including alpha-synuclein, was tested in 523 people for six months; the registry results show no overall change on the primary measure, and the company's reported benefit is in subgroups. | NCT05357989 | 22 Oct 2026 |
| 2024 | Finnish faecal transplant trial | Look beyond the brain | Gut–brain origin | Trial in people | 2 | 48 | In a placebo-controlled trial of 48 Finnish patients whose gut bacteria were out of balance, one colonoscopic faecal transplant did not change the combined movement and daily-life score at six months, and caused more stomach side effects. | NCT04854291 | 19 Oct 2026 |
| 2023 | Venglustat (MOVES-PD) in GBA1 carriers | Help cells clean up | Lysosome & GBA | Trial in people | 2 | 221 | Venglustat, a pill that blocks the enzyme making GCase's raw material, did not slow early Parkinson's in 221 people with a GBA1 variant; the score trended slightly worse than on placebo. | NCT02906020 | 19 Oct 2026 |
| 2023 | Smoking and coffee: protective associations | Look beyond the brain | Environment & lifestyle | Discovery | – | – | One of the most replicated findings in Parkinson's epidemiology is lower risk in smokers and coffee drinkers; genetic analysis supports the smoking link, but a nicotine-patch trial found no benefit. | NCT01560754 | 19 Oct 2026 |
| 2023 | NLY01 Phase 2 | Look beyond the brain | Metabolic drug repurposing | Trial in people | 2 | 255 | NLY01, a long-acting GLP-1-type drug designed to calm brain immune cells, did not beat placebo on its main measure in 255 people with early Parkinson's, according to the company's 2023 announcement. | NCT04154072 | 19 Oct 2026 |
| 2023 | Fosgonimeton (SHAPE) | Replace or repair what's lost | Growth factors | Trial in people | 2 | 28 | An injected drug meant to boost a growth-factor pathway (HGF) was stopped at 28 of a planned 75 people with Parkinson's disease dementia or dementia with Lewy bodies, and missed its main target. | NCT04831281 | 19 Oct 2026 |
| 2022 | SPARK: cinpanemab | Clear the bad protein | Synuclein misfolding & spread | Trial in people | 2 | 357 | Biogen's anti-synuclein antibody cinpanemab did nothing on any dose in 357 people with early Parkinson's, so Biogen ended its development in February 2021. | NCT03318523 | 22 Oct 2026 |
| 2022 | FAIRPARK-II: deferiprone | Help cells clean up | Oxidative stress | Trial in people | 2 | 372 | Removing iron from the substantia nigra with deferiprone worked on scans but made symptoms worse in newly diagnosed people, a reminder that hitting the target can backfire. | NCT02655315 | 19 Oct 2026 |
| 2022 | Simvastatin (PD STAT) | Look beyond the brain | Metabolic drug repurposing | Trial in people | – | 235 | A UK trial of 235 people found the statin simvastatin, tested because statin use had been linked to lower Parkinson's risk, did not slow progression and, on the main measure, trended slightly worse. | – | 19 Oct 2026 |
| 2021 | SURE-PD3: inosine | Help cells clean up | Oxidative stress | Trial in people | 3 | 298 | A Phase 3 trial that deliberately raised blood urate, an antioxidant linked to slower Parkinson's in observational studies, found no slowing and more kidney stones. | NCT02642393 | 19 Oct 2026 |
| 2021 | Mendelian randomisation: urate and BMI | Help cells clean up | Oxidative stress | Discovery | – | – | Using inherited gene variants as a natural randomised experiment, researchers found that the apparent protection of high urate was probably not causal, helping explain why raising urate failed, while higher body mass index may weakly protect. | – | 19 Oct 2026 |
| 2021 | VY-AADC Phase 1b and RESTORE-1 | Replace or repair what's lost | Gene therapy & delivery | Trial in people | 2 | 14 | After a 15-person Phase 1, Voyager and Neurocrine ran a 16-person Phase 1b and then RESTORE-1, a sham-surgery-controlled trial of an AADC gene therapy; MRI abnormalities led to an FDA hold in December 2020 and Neurocrine ended the Parkinson's programme in 2021. | NCT01973543 | 19 Oct 2026 |
| 2021 | AKST4290 (TEAL) | Look beyond the brain | Inflammation & immunity | Trial in people | 2 | 110 | A 110-person, 12-week Phase 2 trial of a pill that blocks the immune receptor CCR3 found no advantage over placebo on its main motor measure. | NCT04369430 | 19 Oct 2026 |
| 2021 | Molecular hydrogen trials | Help cells clean up | Oxidative stress | Trial in people | 2 | 178 | Drinking hydrogen-rich water improved scores in a 17-person Japanese pilot, but the 178-person follow-up found no difference from placebo, and a US trial stopped after enrolling two people. | NCT03971617 | 19 Oct 2026 |
| 2021 | Nilotinib: Georgetown and NILO-PD | Keep vulnerable cells alive | Cell survival & anti-apoptosis | Trial in people | 2 | 151 | Two placebo-controlled Phase 2 trials of the leukaemia drug nilotinib ran in parallel; the first reported shifts in spinal-fluid markers, the second found almost no drug in the brain, no marker change and slightly worse motor scores, and advised against going further. | NCT02954978 | 19 Oct 2026 |
| 2021 | HTRA2/Omi (PARK13) did not hold up | Treat the inherited causes | Genetic architecture | Discovery | – | – | A 2005 screen of German patients linked a mitochondrial protease gene, *HTRA2*, to Parkinson's; a 15,000-person association study in 2011 and a 5,800-person sequencing study in 2021 found no effect, so the claim is treated as unproven. | – | 20 Oct 2026 |
| 2021 | DNAJC13 and TMEM230: disputed genes | Treat the inherited causes | Genetic architecture | Discovery | – | – | A 2014 paper named *DNAJC13* as a Parkinson's gene and a 2016 *Nature Genetics* paper named *TMEM230* from the same family, but replication failed for both and the second was challenged directly. | – | 20 Oct 2026 |
| 2020 | Isradipine (STEADY-PD III) | Keep vulnerable cells alive | Selective vulnerability | Trial in people | 3 | 336 | The blood-pressure drug isradipine, chosen to ease calcium stress on dopamine neurons, did not slow Parkinson's in 336 people over three years. | NCT02168842 | 19 Oct 2026 |
| 2019 | GDNF infusion trials (Amgen, Bristol) | Replace or repair what's lost | Growth factors | Trial in people | 2 | 69 | Two placebo-controlled trials (34 and 35 people) pumped the growth factor GDNF into the putamen; neither improved symptoms, though the Bristol trial showed more dopamine activity on PET scans. | – | 19 Oct 2026 |
| 2019 | LEAP: early vs delayed levodopa | Behind the scenes | Trial methods & regulation | Trial in people | 3 | 445 | In 445 people, levodopa begun at once did no better after 80 weeks than levodopa begun after 40 weeks of placebo, so it does not slow the disease. | ISRCTN30518857 | 19 Oct 2026 |
| 2018 | Parkin/PINK1–STING paper retracted | Look beyond the brain | Inflammation & immunity | Discovery | – | – | A high-profile *Nature* paper linked failed mitochondrial clean-up to inflammation through the STING pathway, but its authors retracted it in 2025 after finding unreliable and irreproducible data. | – | 19 Oct 2026 |
| 2015 | NET-PD LS-1: creatine | Help cells clean up | Mitochondria & energy | Trial in people | 3 | 1,741 | After creatine cleared a government "screen then confirm" futility screen, a 1,741-person trial found no benefit and stopped early. | NCT00449865 | 19 Oct 2026 |
| 2015 | Pioglitazone (FS-ZONE) | Look beyond the brain | Metabolic drug repurposing | Trial in people | 2 | 210 | A 210-person NINDS "futility" trial found pioglitazone, a diabetes drug that protects neurons in animals, was unlikely to slow early Parkinson's at either dose tested. | NCT01280123 | 19 Oct 2026 |
| 2015 | CERE-120 (AAV2-neurturin) | Replace or repair what's lost | Growth factors | Trial in people | 2 | 109 | A gene therapy that made neurons produce neurturin, a GDNF relative, failed to beat sham surgery in 58 people (putamen only) and then in 51 people (putamen plus substantia nigra). | NCT00400634 | 19 Oct 2026 |
| 2015 | GIGYF2 and EIF4G1: disputed genes | Treat the inherited causes | Genetic architecture | Discovery | – | – | Two papers named GIGYF2 (2008) and EIF4G1 (2011) as causes of inherited Parkinson's, but larger screens found the same changes in healthy people and the claims are now treated as unproven or refuted. | – | 19 Oct 2026 |
| 2014 | QE3: coenzyme Q10 fails in Phase 3 | Help cells clean up | Mitochondria & energy | Trial in people | 3 | 600 | A mitochondria-support supplement that looked promising in an earlier trial failed at scale, the cautionary tale that pushed the field to demand stronger evidence before Phase 3. | NCT00740714 | 19 Oct 2026 |
| 2013 | PROUD: pramipexole delayed start | Behind the scenes | Trial methods & regulation | Trial in people | 3 | 535 | In 535 people, starting pramipexole at once rather than 6–9 months later made no lasting difference to symptoms or to dopamine-transporter scans. | NCT00321854 | 19 Oct 2026 |
| 2011 | Spheramine (STEPS) | Replace or repair what's lost | Cell replacement | Trial in people | 2 | 71 | Retinal pigment cells stuck to tiny gelatin beads and placed in the putamen did not improve movement more than a fake operation in 71 people with advanced Parkinson's. | NCT00206687 | 19 Oct 2026 |
| 2011 | 9E4 antibody in mice (paper retracted 2024) | Clear the bad protein | Synuclein misfolding & spread | Lab study (cells or animals) | – | – | A mouse antibody against the tail of alpha-synuclein, given to synuclein-overproducing mice, was reported to reduce protein build-up and memory deficits and became prasinezumab, but the paper was retracted in November 2024 over image concerns. | – | 19 Oct 2026 |
| 2010 | MitoQ (PROTECT) | Help cells clean up | Mitochondria & energy | Trial in people | 2 | 128 | A pill designed to sit inside mitochondria and mop up damaging oxygen molecules did nothing to slow progression in 128 newly diagnosed people over 12 months. | NCT00329056 | 19 Oct 2026 |
| 2009 | ADAGIO: rasagiline | Help cells clean up | Oxidative stress | Trial in people | 3 | 1,176 | A delayed-start trial of rasagiline hinted at slower progression on the 1 mg dose, but the 2 mg dose showed nothing and an FDA panel voted 17-0 that slowing was not proven. | NCT00256204 | 19 Oct 2026 |
| 2008 | Urate epidemiology | Help cells clean up | Oxidative stress | Discovery | – | – | Two studies found that people with higher blood urate (a natural antioxidant) were less likely to get Parkinson's and declined more slowly, which launched the urate-raising trials that later failed. | NCT00040404 | 19 Oct 2026 |
| 2007 | PRECEPT: CEP-1347 | Keep vulnerable cells alive | Cell survival & anti-apoptosis | Trial in people | 2 | 806 | An anti-apoptotic drug that protected nerve cells in mouse and cell models did nothing in 806 people with early Parkinson's; the trial was stopped when an interim look showed it could not succeed. | NCT00040404 | 19 Oct 2026 |
| 2006 | TCH346 (GAPDH binder) | Keep vulnerable cells alive | Cell survival & anti-apoptosis | Trial in people | 2 | 301 | TCH346 (CGP 3466), a drug that binds the enzyme GAPDH to stop a cell-death pathway, did not differ from placebo in 301 people with early Parkinson's. | NCT00407212 | 19 Oct 2026 |
| 2003 | Olanow: sham-controlled fetal grafts | Replace or repair what's lost | Cell replacement | Trial in people | 2 | 34 | In 34 people with advanced Parkinson's, fetal grafts did not significantly beat a placebo operation at 24 months, and 56% of grafted patients had dyskinesias that continued off medication. | – | 19 Oct 2026 |
| 2001 | Freed: sham-controlled fetal grafts | Replace or repair what's lost | Cell replacement | Trial in people | 2 | 40 | In 40 people with severe Parkinson's, fetal dopamine-cell grafts did not beat sham surgery on the main measure, helped only patients aged 60 or under, and left 15% with dystonia and dyskinesias. | – | 22 Oct 2026 |
| 1998 | UCHL1 I93M (PARK5) | Treat the inherited causes | Genetic architecture | Discovery | – | – | A single German sibling pair carrying a change in UCHL1, an enzyme of the cell's ubiquitin protein-disposal system, was reported as a Parkinson's gene, but no other family has confirmed it. | – | 19 Oct 2026 |
Why slowing Parkinson's is hard
- Most neurons are gone by diagnosis. At the earliest point studied, 50–90% of substantia nigra dopamine neurons were already lost (Kordower, Brain 2013).
- Slow, variable progression means long, expensive trials. Symptom drugs and strong placebo effects blur the results (Foltynie et al., J Parkinsons Dis 2026). A LRRK2 Phase 3 was cancelled partly because it would not have finished until 2031.
- No measure of progression speed. The seed test shows whether the disease is present, not how fast it moves, and LUMA showed that hitting the target does not guarantee benefit.
- "Parkinson's" is many diseases. Drugs may have been tested in the wrong people. Known common genetic variants explain only about 16–36% of heritable risk ([Nalls, Lancet Neurol 2019](<https://doi.org/10.1016/S1474-4422(19)30320-5>)), and the two 2024 biological definitions disagree.
- Getting drugs into the brain. The blood-brain barrier blocks large molecules, so growth factors need surgery or gene therapy.
An editorial note from the project's dead-ends.md, as of 22 Oct 2026. It is the project's view, not generated from the data.