Separating mechanism from proof is the single most important discipline in reading this literature. The mechanistic case is strong and internally consistent. The clinical case is early, small, and deliberately modest in what it claims. Here it is worth being explicit about the hierarchy of evidence, from weakest to strongest: cell-culture experiments, animal models, uncontrolled human observations, small randomized safety trials, and finally large randomized efficacy trials with disease-relevant endpoints. NAD+ in Parkinson’s disease currently has a great deal at the lower rungs and very little at the top.1
At the preclinical level, NAD+ precursors have shown protective effects across multiple Parkinson’s models. Boosting NAD+ has improved mitochondrial function, reduced alpha-synuclein toxicity, and extended survival in fruit-fly and rodent systems, and the 2025 UPRmt/mitophagy work provided a specific mechanistic account of one way this protection might occur.1,5,6 These are meaningful signals, but animal models of Parkinson’s disease are notoriously imperfect predictors of human benefit; the graveyard of neuroprotective agents that worked in mice and failed in people is large. Preclinical success is a reason to run a human trial, not a substitute for one.
The most important human data come from a small set of Norwegian trials. The NADPARK study, published in Cell Metabolism in 2022, was a randomized, double-blind, placebo-controlled phase I trial in 30 newly diagnosed, treatment-naive patients who received 1,000 mg of oral nicotinamide riboside or placebo for 30 days.1 Its purpose was to establish safety and target engagement. It succeeded on both counts: NR was well tolerated and produced a significant, though variable, increase in cerebral NAD+ measured by phosphorus magnetic resonance spectroscopy, alongside changes in related metabolites in cerebrospinal fluid. In the subgroup whose brain NAD+ actually rose (the responders), the investigators observed altered cerebral metabolism on FDG-PET and reported an associated mild clinical improvement, and blood and muscle transcriptomics showed upregulation of mitochondrial, lysosomal, and proteasomal gene programs.1 These are encouraging exploratory findings. They are not proof of efficacy: the trial was not powered or designed to demonstrate a change in disease progression, the clinical signal was in a post-hoc responder subgroup, and 30 days is a fraction of the timescale over which Parkinson’s disease evolves.
The follow-up NR-SAFE trial, published in Nature Communications in 2023, tested a much higher dose, 3,000 mg of NR daily (1,500 mg twice daily), against placebo for four weeks in 20 patients, again primarily to assess safety.2 All 20 participants completed the study. There were 42 adverse events in total, 25 in the NR group and 17 in the placebo group, and critically all were graded mild, with no moderate or severe events and no statistically significant difference in adverse-event frequency between arms. No painful flushing was reported. The NR group showed a statistically significant improvement in total MDS-UPDRS score (from 51.0 to 40.3, p = 0.007) while placebo did not, but the authors themselves flagged this as preliminary and potentially confounded, including by differences in the timing of levodopa dosing relative to assessment.2 A responsible reading treats NR-SAFE as reassuring on high-dose safety and hypothesis-generating on efficacy, nothing more.
The decisive test is the NOPARK study (NCT03568968), a phase III randomized, double-blind, placebo-controlled trial of 1,000 mg oral NR daily over 52 weeks in roughly 400 patients with early Parkinson’s disease across multiple Norwegian centers, with the change in total MDS-UPDRS as its primary endpoint.3 This is the appropriately sized, appropriately long, efficacy-focused trial the field needs. As of this writing the trial has completed enrollment and follow-up, but its primary clinical results have not been published, so no conclusion about efficacy can be drawn. Until those results appear, the honest evidence level for “NAD+ precursors slow Parkinson’s progression” is: plausible mechanism, safe in the short term at the doses tested, and unproven in humans. Alongside these interventional data sit epidemiological signals that are hypothesis-supporting but causally weak, discussed in the next section.