Abstract
Nicotinamide mononucleotide (NMN) and nicotinamide riboside (NR) can increase circulating NAD+ metabolites in humans, but the functional results are inconsistent. A 2025 meta-analysis found no convincing preservation of muscle mass or function in adults whose mean ages were above 60. That gap is the starting point for this research note: perhaps supplying an NAD+ precursor is not enough when mitochondrial quality control is impaired.
The hypothesis is biologically plausible, not clinically proven. A 28-day randomized trial published in Nature Aging in October 2025 found that urolithin A altered two prespecified features of CD8+ T-cell phenotype and metabolism. It did not show fewer infections, a better vaccine response, slower aging, or direct mitophagy flux in people. This reviewed edition separates those human data from animal and cell findings, reports the null trials alongside the positive signals, corrects a spermidine dose claim, and attributes the KiSel-10 result to the tested combination of selenium plus coenzyme Q10 rather than to coenzyme Q10 alone.
01 · The paradox
NAD+ rises more consistently than function
NAD+ is both a redox cofactor and a consumed substrate for enzymes that include sirtuins and poly(ADP-ribose) polymerases. Human measurements do not support one uniform age curve across every tissue, cellular compartment, and assay. The precise statement is therefore narrower than the familiar slogan: NAD+ homeostasis can change with age, and oral precursors can alter parts of the circulating NAD+ metabolome.
What follows is less settled. A 2024 systematic review of 10 NMN randomized trials, 437 participants, doses from 150 to 1,200 mg/day, and a mean follow-up of 9.6 weeks described the aggregate physical-performance changes as non-significant [1]. A more focused 2025 systematic review and meta-analysis included adults with mean ages from 60.9 to 83 years and found no significant effect of NMN on skeletal-muscle index, left or right grip strength, gait speed, or the five-times chair-stand test. Its narrative synthesis found no consistent improvement in knee-extension strength, short physical performance battery, or thigh muscle mass. The authors concluded that the available evidence did not support NMN or NR for preserving muscle mass and function in this age group [2].
The null is not limited to healthy volunteers. In a randomized, double-blind, placebo-controlled crossover study of 25 people with moderate-to-severe chronic kidney disease, six weeks of NR at 1,000 mg/day changed several metabolic and lipidomic measures but did not improve peak oxygen consumption, total work, or work efficiency compared with placebo [3]. A separate 2024 meta-analysis of eight short NMN trials in 342 mostly non-diabetic adults found no significant pooled benefit for fasting glucose, fasting insulin, HbA1c, HOMA-IR, or the measured lipid profile [4].
There are positive signals, and they should not be erased. In a 10-week randomized trial of 25 postmenopausal women with prediabetes who were overweight or obese, 250 mg/day of NMN increased insulin-stimulated glucose disposal per kilogram of fat-free mass by 25 ± 7% from baseline, while placebo did not. The effect was specific: hepatic and adipose-tissue insulin sensitivity, fasting glucose, body composition, muscle mitochondrial oxidative capacity, strength, and fatigability did not improve. Muscle NAD+ concentration itself did not rise, although NMN metabolites suggested altered turnover [5]. This small, narrowly selected study is evidence for one physiological effect in one population, not a general result for aging, diabetes prevention, or physical performance.
NAD+ is a substrate, not an outcome. But the stronger claim — that damaged mitochondria are the reason NAD+ trials disappoint, or that adding a mitophagy-active compound will solve the problem — remains a hypothesis until the combination is tested.
The “fuel without a furnace” metaphor is useful if it prompts a systems question. It becomes misleading if it is treated as a result. NAD+ participates in many cellular compartments, and mitochondrial function is constrained by far more than mitophagy alone: substrate delivery, electron transport, redox balance, biogenesis, dynamics, proteostasis, inflammation, vascular supply, training status, and disease can all matter. A null precursor trial does not identify which one was rate-limiting.
02 · The 2025 trial
What MitoImmune found — and what it did not
MitoImmune was a single-centre, randomized, double-blind, placebo-controlled proof-of-concept trial in 50 generally healthy adults aged 45 to 70. Participants received 1,000 mg/day of urolithin A or placebo for 28 days. The primary outcomes concerned peripheral T-cell populations and immune-cell metabolism, not infections, vaccine response, hospitalization, or aging itself. The study is registered as NCT05735886, and the peer-reviewed report was published on 31 October 2025 [6].
Two between-group estimates deserve attention. First, CD8+ T-cell fatty-acid and amino-acid oxidation capacity increased by 14.72 percentage points relative to placebo (95% confidence interval 6.46 to 22.99; P = 0.0061). Second, a naive-like, less terminally exhausted CD8+ population increased by 0.50 percentage points relative to placebo (95% confidence interval 0.16 to 0.83; P = 0.0437). The investigators also reported exploratory or secondary changes in PGC-1α expression, natural-killer-cell and monocyte populations, stimulated T-cell cytokine output, monocyte uptake of bacterial particles, and single-cell transcriptional pathways.
These are laboratory measurements from blood cells after four weeks. The trial did not establish that participants had a “younger immune system,” resisted infection, or mounted a better vaccine response. Nor did it directly quantify Pink1/Parkin-dependent mitophagy flux inside participants. The paper used ex-vivo experiments and prior mechanistic work to support that interpretation. The metabolic assay also changed from the planned real-time profiling method to SCENITH; the authors reported this protocol deviation and explained that SCENITH allowed subset-specific analysis with fewer cells. Importantly, many comparisons were made without adjustment for multiple testing, which increases the chance of nominally significant findings when numerous endpoints are examined.
The mechanistic foundation is stronger in models. In 2022, the same research programme reported that urolithin A induced mitophagy-related signalling and expanded T-memory-stem-like cells in mice, improving antitumour T-cell responses; human CAR-T cells were also studied ex vivo [7]. That paper supports a mechanism and a translational hypothesis. It is not evidence that oral urolithin A prevents or treats cancer in people.
03 · Human function
The urolithin-A muscle trials are promising in places and null in others
If mitochondrial quality control is the missing variable, whole-body function is where the hypothesis should eventually pay rent. The current record is mixed.
ATLAS: secondary strength findings, missed primary endpoint
ATLAS randomized 88 healthy, overweight, middle-aged adults to 500 mg/day, 1,000 mg/day, or placebo for four months. The prespecified primary endpoint, peak power output during exercise, was not significantly different between either urolithin-A group and placebo. Hamstring average peak torque rose from baseline by about 12% in the 500 mg group and 9.8% in the 1,000 mg group; each change differed nominally from placebo. The placebo group declined from baseline on those measurements, which contributed to the between-group contrast.
At 1,000 mg, peak oxygen consumption, predicted maximal oxygen consumption, and six-minute walk distance improved within the treatment group, but the comparisons with placebo were trends rather than conventionally significant results (P = 0.058 for peak oxygen consumption and P = 0.098 for six-minute walk distance). The study was exploratory, involved many outcomes, and was funded by Amazentis; most authors were employees, board members, or scientific advisers of the sponsor [9]. The correct summary is not “no effect,” but neither is it “a 12% global strength gain.” It is a missed primary endpoint with encouraging secondary hamstring-strength signals that require confirmation. Trial record: NCT03464500.
ENERGIZE: endurance moved; primary outcomes did not
ENERGIZE randomized 66 adults aged 65 to 90 to 1,000 mg/day or placebo for four months. Urolithin A improved the number of repeated contractions until fatigue in hand and leg muscles at two months, a secondary outcome. However, neither primary result — six-minute walk distance nor maximal ATP production in hand muscle — improved significantly versus placebo. The original primary endpoints were hand-muscle ATP production and endurance; after the study began, endurance was replaced by walking distance because the investigators considered it more functionally important. The statistical plan was prepared before unblinding. Both groups increased walking distance substantially, and no between-group benefit was shown. The trial was sponsored by Amazentis, which had a role in design, conduct, analysis, manuscript preparation, and the decision to publish [10]. Trial record: NCT03283462.
ENDURO: recovery signals without a performance advantage
ENDURO studied 42 highly trained male distance runners during a four-week altitude camp. Urolithin A at 1,000 mg/day did not improve 3,000-metre performance versus placebo, and the treatment-by-time interaction for aerobic capacity was not significant. Ratings of perceived exertion and post-race creatine-kinase exposure were lower, and muscle proteomics shifted, but mitochondrial function did not change. Only 22 runners completed the time trial and 20 contributed muscle biopsies, making those analyses small. The sponsor was again Amazentis, and multiple authors were affiliated with Amazentis or Nestlé Health Science [11]. Trial record: NCT04783207.
Taken together, urolithin A has reproducible biological activity and several human functional signals. It does not yet have a clean, independently replicated demonstration that mitophagy activation improves a prespecified clinical aging outcome. Population also matters: an older adult with low reserve, an overweight middle-aged adult, and an elite runner are different tests of the same molecule.
04 · The autophagy arm
Spermidine: strong biology, an association, and a randomized null
Spermidine can induce autophagy in experimental systems, and genetic disruption of autophagy can abolish lifespan effects in model organisms. That biology makes it a serious research molecule. Human evidence, however, must be separated into observational and interventional layers.
In the Bruneck cohort, 829 adults aged 45 to 84 completed repeated dietary assessments over two decades. After adjustment for age, sex, energy intake, lifestyle, established mortality predictors, and other dietary features, each one-standard-deviation higher estimated spermidine intake was associated with a hazard ratio of 0.76 for all-cause mortality (95% confidence interval 0.67 to 0.86). Mortality rates declined across thirds of intake from 40.5 to 15.1 deaths per 1,000 person-years [12].
That is a substantial and carefully adjusted association, not proof that spermidine caused the difference. Higher intake can accompany diets rich in whole grains, legumes, vegetables, and other factors that are difficult to measure completely. Residual confounding and dietary-estimation error remain possible even when sensitivity analyses are favourable.
The randomized SmartAge trial is more sobering. One hundred adults aged 60 to 90 with subjective cognitive decline were assigned to a wheat-germ extract providing 0.9 mg/day of spermidine or placebo for 12 months. The primary memory endpoint did not improve, and neither did the secondary outcomes overall. Exploratory verbal-memory and inflammation signals were explicitly hypothesis-generating [13]. Trial record: NCT03094546.
A later report is often used to argue that 3.3 mg/day succeeds where 0.9 mg/day failed. That comparison is not valid. The higher-dose report followed 45 nursing-home residents with dementia who all received spermidine for one year; their Mini-Mental State Examination scores improved from baseline, but there was no randomized placebo group [14]. Practice effects, regression to the mean, changes in care, attrition, and natural variability cannot be separated from treatment. The two studies therefore do not establish a dose threshold, an effective cognitive dose, or efficacy in dementia.
The honest spermidine conclusion is “mechanistically interesting, observationally associated, and not yet confirmed for cognitive benefit in a substantial randomized trial.”
05 · Biogenesis and electron transport
PQQ and coenzyme Q10 require two different evidence labels
PQQ: mostly a mechanistic case
Pyrroloquinoline quinone (PQQ) increased PGC-1α expression and measures of mitochondrial biogenesis through CREB signalling in cultured mouse liver cells [15]. A separate mouse-fibroblast experiment found higher cellular NAD+, SIRT1-dependent PGC-1α activation, and increased mitochondrial markers after PQQ exposure [16]. These are cell experiments, not evidence that oral PQQ creates new, better-functioning mitochondria in people.
One small human study randomized 23 untrained men to 20 mg/day of PQQ or placebo while both groups completed six weeks of supervised endurance training. PQQ did not add an aerobic-performance or body-composition benefit. PGC-1α protein increased more in the PQQ group, but a signalling-protein change is not the same as demonstrated mitochondrial biogenesis or improved health [17]. The earlier source note misattributed the cellular NAD+ paper to Hwang; the correct authors are Saihara and colleagues.
KiSel-10: a combined selenium–coenzyme Q10 intervention
Coenzyme Q10 is an electron carrier in the mitochondrial respiratory chain. KiSel-10 nevertheless cannot tell us the effect of coenzyme Q10 alone. The trial randomized 443 community-living Swedish adults aged 70 to 88 to a combination of 200 mg/day coenzyme Q10 plus 200 micrograms/day selenium yeast, or matching placebos, for 48 months. At a median 5.2 years of follow-up, cardiovascular mortality was 5.9% in the combined-treatment group and 12.6% in the placebo group; NT-proBNP and an echocardiographic cardiac-function score also favoured treatment [18]. Trial record: NCT01443780.
Follow-up analyses reported a cardiovascular-mortality hazard ratio of 0.51 at ten years and persistent group separation at 12 years [19] [20]. The registry record was first posted in 2011, after the 2003–2010 intervention period, so it cannot provide the assurance of prospective registration. The investigators themselves described the long-term finding as hypothesis-generating because it came from a small, geographically homogeneous cohort with low baseline selenium status, subgroup analyses, deaths and withdrawals during the intervention, and no measurement of supplement use after the intervention ended. Part of the analysis costs came from the product manufacturer.
A 2023 sub-study found higher serum SIRT1 concentrations after four years in participants with available samples, and lower SIRT1 was associated with later cardiovascular death [21]. That exploratory analysis may suggest a pathway; it does not establish mediation, and it still concerns the combined intervention. The defensible wording is therefore “selenium plus coenzyme Q10 was associated with lower cardiovascular mortality in KiSel-10,” not “coenzyme Q10 reduced mortality by 54%.” The result should not be generalized to younger, selenium-replete populations or used to guide individual selenium intake without clinical context.
06 · The proposed system
Mitochondrial quality control is a loop, but the stack remains a bet
The design logic is coherent. NAD+ precursors can expand precursor availability. Mitophagy removes selected damaged mitochondria. Biogenesis supplies replacements. Electron carriers support the respiratory chain. Autophagy handles a wider set of cellular material. Those processes interact, so single-target thinking can miss a bottleneck.
Urolithin A also shows overlap with NAD+ biology. A 2020 study reported increased skeletal-muscle NAD+, ATP, SIRT1, and angiogenic signalling after long-term oral urolithin A — in mice, with supporting work in mouse muscle cells [22]. It should not be presented as a human result. MitoImmune reported PGC-1α-related remodelling in human immune cells, but did not show that urolithin A raised tissue NAD+ or mitophagy flux in vivo.
This distinction changes the conclusion. It is reasonable to hypothesize that an NAD+ precursor and a mitophagy-active compound could be complementary. It is equally plausible that they are redundant in a particular tissue, that one changes the pharmacology of the other, that a pathway is already saturated, or that neither pathway limits the chosen outcome. “Complementary mechanisms” is not a synonym for synergy.
07 · What could falsify the idea
The next useful trial is a combination trial, not another pathway diagram
A rigorous test would randomize an appropriate population to four arms: placebo, an NAD+ precursor alone, urolithin A alone, and the combination. It would prespecify one or two functional primary endpoints, measure blood and relevant tissue NAD+ separately, include a defensible measure of mitochondrial turnover rather than assuming mitophagy from downstream markers, and adjust for multiple comparisons. Duration and sample size would have to match the outcome. An independent replication would matter because much of the urolithin-A clinical literature comes from the ingredient developer.
The combination hypothesis would be weakened if the joint arm failed to outperform the better single ingredient, if tissue exposure did not change, or if biomarker movement failed to translate into function. Those are not hostile tests; they are how a mechanistic story becomes — or fails to become — useful evidence.
Two further boundaries are non-negotiable. First, the National Institute on Aging Interventions Testing Program found that NR at 1,000 mg per kilogram of diet, started at eight months of age, did not extend lifespan in genetically heterogeneous mice [23]. Mouse lifespan is not human lifespan, but this null argues against casually treating NAD+ elevation as a longevity intervention. Second, no compound or combination in this article has been shown to extend human lifespan. A shorter trial with a favourable biomarker cannot answer that question.
There are practical reasons human lifespan trials are rare: they are long, expensive, vulnerable to changes in background care and adherence, and difficult to interpret. Those barriers explain why researchers use intermediate endpoints. They do not convert those endpoints into evidence of longer life.
Questions
Frequently asked questions
Do NMN and NR raise NAD+ in humans?
They can raise circulating NAD+ metabolites, but the response depends on compound, dose, assay, tissue, and population. Blood changes do not prove that NAD+ increased in the target tissue or that function improved.
What did the 2025 MitoImmune trial prove?
It showed that 1,000 mg/day of urolithin A for 28 days changed two prespecified CD8+ T-cell measures versus placebo in 50 healthy middle-aged adults. It did not prove fewer infections, better vaccine response, slower immune aging, or direct in-vivo mitophagy.
Is urolithin A the same as pomegranate extract?
No. Urolithin A is a metabolite that gut microbes can produce from ellagitannins in foods such as pomegranate and walnuts. Production and exposure vary among people. A sponsor-funded crossover study found that direct 500 mg urolithin A produced more consistent exposure than pomegranate juice, but bioavailability does not establish a clinical benefit [24].
Did a higher spermidine dose improve dementia?
The 3.3 mg/day report had no placebo group, so improvement from baseline cannot be attributed to spermidine. The 12-month randomized SmartAge trial at 0.9 mg/day was null for its primary memory outcome. Together, these studies do not establish an effective cognitive dose.
Did coenzyme Q10 reduce cardiovascular mortality by itself?
KiSel-10 tested coenzyme Q10 together with selenium yeast. The contribution of either component cannot be isolated, and the result came from an older Swedish population with low selenium status. It should be cited as a combined-intervention finding.
Should I combine an NAD+ precursor with urolithin A or spermidine?
This article does not make an individual treatment recommendation. The combination has not been tested against its components in a randomized human trial. Supplements can have adverse effects and interactions; decisions should account for health status, medicines, pregnancy, planned procedures, and qualified clinical advice.
Sources
References
- Wen J, et al. Improved Physical Performance Parameters in Patients Taking Nicotinamide Mononucleotide: A Systematic Review of Randomized Control Trials. Cureus. 2024;16:e65961. PMID 39221308; PMCID PMC11365583.
- Prokopidis K, et al. The Effect of Nicotinamide Mononucleotide and Riboside on Skeletal Muscle Mass and Function: A Systematic Review and Meta-Analysis. Journal of Cachexia, Sarcopenia and Muscle. 2025;16:e13799. PMID 40275690; PMCID PMC12022230.
- Ahmadi A, et al. Randomized crossover clinical trial of coenzyme Q10 and nicotinamide riboside in chronic kidney disease. JCI Insight. 2023;8:e167274. PMID 37159264; PMCID PMC10393227; NCT03579693.
- Chen F, et al. Effects of Nicotinamide Mononucleotide on Glucose and Lipid Metabolism in Adults: A Systematic Review and Meta-analysis of Randomised Controlled Trials. Current Diabetes Reports. 2025;25:4. PMID 39531138; PMCID PMC11557618.
- Yoshino M, et al. Nicotinamide mononucleotide increases muscle insulin sensitivity in prediabetic women. Science. 2021;372:1224–1229. PMID 33888596; PMCID PMC8550608; NCT03151239.
- Denk D, et al. Effect of the mitophagy inducer urolithin A on age-related immune decline: a randomized, placebo-controlled trial. Nature Aging. 2025;5:2309–2322. PMID 41174221; NCT05735886.
- Denk D, et al. Expansion of T memory stem cells with superior anti-tumor immunity by urolithin A-induced mitophagy. Immunity. 2022;55:2059–2073.e8. PMID 36351375.
- Denk D, et al. Author Correction: Effect of the mitophagy inducer urolithin A on age-related immune decline. Nature Aging. 2026;6:463. PMID 41491872; PMCID PMC12920124.
- Singh A, et al. Urolithin A improves muscle strength, exercise performance, and biomarkers of mitochondrial health in a randomized trial in middle-aged adults. Cell Reports Medicine. 2022;3:100633. PMID 35584623; PMCID PMC9133463; NCT03464500.
- Liu S, et al. Effect of Urolithin A Supplementation on Muscle Endurance and Mitochondrial Health in Older Adults: A Randomized Clinical Trial. JAMA Network Open. 2022;5:e2144279. PMID 35050355; PMCID PMC8777576; NCT03283462.
- Whitfield J, et al. Evaluating the Impact of Urolithin A Supplementation on Running Performance, Recovery, and Mitochondrial Biomarkers in Highly Trained Male Distance Runners. Sports Medicine. 2025;55:3183–3200. PMID 40839339; PMCID PMC12628386; NCT04783207.
- Kiechl S, et al. Higher spermidine intake is linked to lower mortality: a prospective population-based study. American Journal of Clinical Nutrition. 2018;108:371–380. PMID 29955838; NCT03378843.
- Schwarz C, et al. Effects of Spermidine Supplementation on Cognition and Biomarkers in Older Adults With Subjective Cognitive Decline: A Randomized Clinical Trial. JAMA Network Open. 2022;5:e2213875. PMID 35616942; NCT03094546.
- Pekar T, et al. The positive effect of spermidine in older adults suffering from dementia after 1 year. Wiener Klinische Wochenschrift. 2024;136:64–66. PMID 37284840; PMCID PMC10776733.
- Chowanadisai W, et al. Pyrroloquinoline quinone stimulates mitochondrial biogenesis through cAMP response element-binding protein phosphorylation and increased PGC-1α expression. Journal of Biological Chemistry. 2010;285:142–152. PMID 19861415; PMCID PMC2804159.
- Saihara K, et al. Pyrroloquinoline Quinone, a Redox-Active o-Quinone, Stimulates Mitochondrial Biogenesis by Activating the SIRT1/PGC-1α Signaling Pathway. Biochemistry. 2017;56:6615–6625. PMID 29185343.
- Hwang PS, et al. Effects of PQQ Supplementation on Aerobic Exercise Performance and Indices of Mitochondrial Biogenesis in Untrained Men. Journal of the American College of Nutrition. 2020;39:547–556. PMID 31860387.
- Alehagen U, et al. Cardiovascular mortality and N-terminal-proBNP reduced after combined selenium and coenzyme Q10 supplementation: a five-year prospective randomized double-blind placebo-controlled trial among elderly Swedish citizens. International Journal of Cardiology. 2013;167:1860–1866. PMID 22626835; NCT01443780.
- Alehagen U, et al. Reduced Cardiovascular Mortality 10 Years after Supplementation with Selenium and Coenzyme Q10 for Four Years. PLOS ONE. 2015;10:e0141641. PMID 26624886; PMCID PMC4666408.
- Alehagen U, et al. Still reduced cardiovascular mortality 12 years after supplementation with selenium and coenzyme Q10 for four years. PLOS ONE. 2018;13:e0193120. PMID 29641571.
- Opstad TB, et al. Increased SIRT1 Concentration Following Four Years of Selenium and Q10 Intervention Associated with Reduced Cardiovascular Mortality at 10-Year Follow-Up. Antioxidants. 2023;12:759. PMID 36979007; PMCID PMC10045001.
- Ghosh N, et al. Urolithin A augments angiogenic pathways in skeletal muscle by bolstering NAD+ and SIRT1. Scientific Reports. 2020;10:20184. PMID 33214614.
- Harrison DE, et al. 17-α-estradiol late in life extends lifespan in aging UM-HET3 male mice; nicotinamide riboside and three other drugs do not affect lifespan in either sex. Aging Cell. 2021;20:e13328. PMID 33788371.
- Singh A, et al. Direct supplementation with urolithin A overcomes limitations of dietary exposure and gut-microbiome variability in healthy adults. European Journal of Clinical Nutrition. 2022;76:297–308. PMID 34117375; PMCID PMC8821002.










