NR-aMCI Pilot
(2026)Objective
To evaluate the safety, tolerability, and preliminary efficacy of 12 weeks of the NAD+ precursor nicotinamide riboside (NR) 500 mg twice daily on cognitive function and cerebral blood flow (CBF) in older adults with amnestic mild cognitive impairment (aMCI).
Study Summary
• No improvement in cognitive function — primary composite (CVLT-III, WMS-IV, NIH Toolbox) all non-significant; largest effect on WMS-IV Logical Memory II (d=0.607, p=0.062)
• No improvement in total cerebral blood flow (48.5→55.7 vs 47.2→47.2 mL/min/100 g; p=0.135)
• No significant reduction in blood pressure, PWV, or carotid compliance
• Exploratory: regional CBF improved in the left hippocampus (p=0.033; d=0.736) and aortic augmentation index trended down (p=0.054; d=0.628); none survived multiple-comparison correction
• Well tolerated: no serious adverse events; adherence 85% NR vs 91% placebo (p=0.22); mild GI symptoms and headaches with NR
Intervention
Nicotinamide riboside 500 mg twice daily (1000 mg/day total) vs matching microcrystalline-cellulose placebo capsules, orally for 12 weeks. Capsules supplied by Niagen Biosciences (ChromaDex) under an FDA IND.
Inclusion Criteria
Community-dwelling adults ≥65 years with amnestic MCI (self-reported memory complaint + objective memory ≥1.5 SD below age norms + CDR global <1.0 per 2011 NIA-AA and Petersen criteria); TICS-m 21–34 or delayed recall ≤10.
Study Design
Arms: Nicotinamide riboside 1000 mg/day vs matching placebo (1:1) — stratified by sex and APOE genotype.
Patients per Arm: Randomized: 52 (NR=24, placebo=25 who started treatment; 3 withdrawn before dosing); completed: 42 (NR=22, placebo=20).
Outcome
• Secondary — total CBF: NR +7.2 mL/min/100 g vs no change with placebo, p=0.135
• Secondary — SBP, DBP, PWV, carotid compliance: all non-significant
• Exploratory — left hippocampal CBF significantly increased with NR (p=0.033; d=0.736); aortic AIx decreased (p=0.054; d=0.628)
• NAD+ metabolome ~2-fold increase with NR (p<0.001; d=1.88)
• Safety: no SAEs; NR tolerated similarly to placebo
Clinical Question
In older adults with amnestic mild cognitive impairment, does 12 weeks of oral nicotinamide riboside 500 mg twice daily improve cognitive function and cerebral blood flow compared with placebo, and is it safe and well tolerated?
Bottom Line
In this single-center phase-II pilot RCT of 42 completers, nicotinamide riboside safely doubled blood NAD+ but did not improve cognitive function, total cerebral blood flow, blood pressure, or arterial stiffness at 12 weeks; exploratory analyses suggested regional increases in hippocampal perfusion that require confirmation in larger trials.
Major Points
- Single-center, double-blind, placebo-controlled parallel-arm phase-II pilot trial at the University of Delaware (NCT03482167).
- Enrolled community-dwelling older adults (≥65 y) with amnestic MCI by NIA-AA/Petersen criteria; 52 randomized (stratified by sex and APOE), 42 completed (NR n=22, placebo n=20).
- Intervention: nicotinamide riboside 500 mg BID (1000 mg/day) vs matching microcrystalline-cellulose placebo × 12 weeks; conducted under an FDA IND.
- NAD+ pharmacodynamic engagement confirmed — blood NAD+ ~doubled with NR (24.1→48.5 µM) vs no change with placebo (p<0.001; d=1.88).
- Primary outcome (cognitive composite: CVLT-III, WMS-IV Logical Memory/Visual Reproduction, NIH Toolbox Fluid Cognition): no significant group×time interaction on any measure; largest effect on WMS-IV Logical Memory II (d=0.607, p=0.062).
- Secondary outcomes (total CBF, SBP, DBP, PWV, carotid compliance): all non-significant. Total CBF p=0.135; DBP p=0.052; PWV p=0.552.
- Exploratory: left hippocampal CBF increased significantly with NR (p=0.033; d=0.736); aortic augmentation index decreased (p=0.054; d=0.628); no correction for multiple comparisons.
- Safety: no serious adverse events in either group; adherence 85±5% NR vs 91±1% placebo (p=0.22); mild GI symptoms and headaches in NR arm; 4 discontinued due to AEs (3 placebo, 1 NR).
- Baseline plasma biomarkers: pTau-217 positivity only in ~32% (single-cutoff), suggesting many participants lacked biological AD — a source of heterogeneity that likely limited detection of cognitive effects.
Study Design
- Study Type
- Randomized Controlled Trial (Phase II, single-center, placebo-controlled, parallel-arm pilot)
- Randomization
- Yes
- Blinding
- Double-blind — participants, investigators, and biostatisticians blinded; coordinators had only blinded group labels (A/B) for capsule preparation. Allocation stratified by sex and APOE genotype using REDCap.
- Sample Size
- 52
- Follow-up
- 12 weeks
- Centers
- 1
- Countries
- USA
Primary Outcome
Definition: Change in cognitive function from baseline to week 12, measured by California Verbal Learning Test-III (CVLT-III), Wechsler Memory Scale-IV Logical Memory and Visual Reproduction subtests, and NIH Toolbox v2.0 Fluid Cognition Composite. Analyzed via generalized linear mixed model treatment×time interaction.
| Control | Intervention | HR/OR | P-value |
|---|---|---|---|
| - | - | - | - |
Limitations & Criticisms
- Small pilot sample (52 randomized, 42 completers) with wide confidence intervals — clearly underpowered to detect cognitive effects.
- Completer analysis rather than intention-to-treat; excluded 3 pre-randomization withdrawals and treats missing data as MAR.
- No adjustment for multiple comparisons across dozens of cognitive, CBF, biomarker, and vascular outcomes — inflates false-positive risk for exploratory left-hippocampal CBF and aortic AIx findings.
- Enrollment based on behavioral criteria for aMCI rather than biomarker-defined AD; only ~32% of participants were pTau-217 positive, so the majority may not have had underlying AD pathology, diluting any disease-modifying signal.
- Racial imbalance (Black/African American: 28% placebo vs 4% NR), which may confound cerebrovascular findings.
- 12-week treatment duration likely too short to detect meaningful cognitive change or effects mediated by slow brain NAD+ uptake.
- Single-center trial at one U.S. site; generalizability limited.
- COVID-19 pandemic caused protocol changes (BVMT-R replaced with WMS-IV Logical Memory for screening; some tests conducted virtually), producing measurement heterogeneity.
- Sample size was originally powered for vascular endpoints (BP, arterial stiffness) rather than for cognition, so the null cognitive result is expected rather than definitive.
Citation
Alzheimers Dement 2026;22(7):e71605