For a decade, nicotinamide riboside and nicotinamide mononucleotide have been sold as the first real anti-aging molecules. Rodent data looked spectacular. Human trials, repeatedly, did not. The usual explanation is that something is wrong with the trials. A harder look suggests the bigger problem is that the field built its case on a premise — that NAD+ decline drives aging — which is shakier than the marketing implied, and then measured the wrong endpoint in the wrong people for the wrong duration. This piece tries to untangle what we actually know from what we merely hope.

This article expands on a longer technical brief: The NAD+ Paradox — Deep Research Brief (PDF). The brief goes deeper on pharmacokinetics and trial design; this article focuses on the harder question of whether the underlying hypothesis is even right.

1. The paradox, stated honestly

Here is the puzzle, in one sentence: oral NR and NMN reliably double whole-blood NAD+ in human volunteers, yet almost no human trial has shown a meaningful functional benefit in healthy adults. The 2026 head-to-head trial by Christen and colleagues in Nature Metabolism is the cleanest demonstration of the first half — 14 days of NR (1 g/day) or NMN (1 g/day) approximately doubled baseline whole-blood NAD+ in 65 healthy adults, while nicotinamide alone did not. The second half — the null functional outcomes — is documented across more than two dozen randomized trials, including the systematic reviews and meta-analyses published in 2024–2025.

The dominant narrative, including in our own earlier brief, is that the trials are flawed and the molecule is fine. That is partly true. But it lets a bigger question off the hook: was the preclinical evidence ever as strong as the field claimed?

2. The "miracle in mice" premise is weaker than advertised

The translational gap assumes there was something to translate. Look closely at the two foundational mouse papers and the assumption frays.

Mills 2016: NMN improved healthspan, not lifespan

The Mills et al. 2016 paper in Cell Metabolism — the most-cited NMN mouse study — administered NMN to mice for 12 months and reported that it "effectively mitigates age-associated physiological decline." What it did not report is a lifespan extension. The improvements were in markers: energy metabolism, physical activity, insulin sensitivity, eye function, bone density, immune function. These are real, but they are healthspan measures. A 2024 bioRxiv preprint from a separate group reported a lifespan extension with long-term NMN, but a preprint is not a confirmed finding, and the failure of the original Mills study to show lifespan is the fact that matters.

Mitchell 2018: NAM improved healthspan, not lifespan

Mitchell et al. 2018, also in Cell Metabolism, was even more direct: "chronic NAM supplementation improves healthspan measures in mice without extending lifespan." The NIA issued a press release confirming this. The mice showed better glucose metabolism and less hepatosteatosis on a high-fat diet, but they did not live longer. This is the Sinclair group's own data, and it is the field's most honest result: the precursor helps some age-related measures in a metabolically stressed mouse, and does not extend life.

Why the mouse-to-human leap is harder than it looks

Three structural problems undermine the "miracle in mice" framing:

1. Dose scaling. Mouse NMN doses of 300 mg/kg/day scale to roughly 2–3 grams per day for a human, before accounting for the faster mouse metabolism. Many human trials used 250–1,000 mg/day. If the mouse effect is dose-dependent and the human dose is sub-therapeutic, the null is unsurprising — but it also means the "translational gap" is partly a dose gap, not a biology gap.
2. Chow and baseline niacin status. Mouse chow has a defined niacin content; human diets vary enormously. A mouse on standard chow is a controlled-baseline subject; a human trial participant walks in with unknown niacin status and unknown NAD+ reserve.
3. Model-specific effects. The most dramatic NAD+ precursor results come from extreme models: high-fat-diet obesity, progeria (BubR1, XPA, CSB), chemically induced DNA damage, or genetic defects (BESTO diabetic mice, ataxia-telangiectasia models). In normal aged mice, the effects are smaller and confined to healthspan markers. The human trial population — usually healthy or mildly metabolically abnormal adults — resembles the normal-aged-mouse condition, not the extreme models. So the most-cited mouse data may not even apply to the population being tested.

This reframes the paradox. If the preclinical signal is healthspan-only, dose-dependent, and model-specific, then human trials showing modest or null effects in healthy adults are consistent with the preclinical data, not contradictory to it. The "gap" may be smaller than the field assumed.

3. The dual role of NAD+ — and why it matters for what we measure

The field has never agreed on which job of NAD+ is the aging-relevant one. This ambiguity is not a footnote; it determines what endpoint a trial should measure, and almost no trial has measured the right one.

Job 1: Redox cofactor (the Krebs cycle role)

NAD+ and NADH shuttle electrons through glycolysis, the TCA cycle, fatty-acid β-oxidation, and Complex I of the electron transport chain. This is the textbook role, and it is the role most people picture when they hear "NAD+ is for energy." But biochemically, this job is recycled, not consumed. The NAD+ is reduced to NADH and then regenerated when NADH delivers its electrons to the electron transport chain. What matters for this job is the NAD+/NADH ratio, not the total pool size.

More importantly, the dehydrogenases that use NAD+ as a cofactor — isocitrate dehydrogenase, malate dehydrogenase, α-ketoglutarate dehydrogenase — generally have K_m values well below physiological NAD+ concentrations. They run saturated. A 20–30% drop in total NAD+ does not obviously throttle the Krebs cycle, because the enzymes were already operating at near-maximal substrate occupancy. This is the weakest link in the "NAD+ decline causes energy failure" story, and it is why simply flooding cells with more NAD+ should not be expected to speed up respiration in healthy tissue. The cofactor was not the limiting factor.

Job 2: Consumed substrate (the signaling role)

This is where the aging hypothesis actually lives. Three enzyme families burn NAD+ as a substrate — they cleave the molecule and release nicotinamide as a byproduct — and their activity is sensitive to the size of the available pool:

• Sirtuins. SIRT1 (nuclear, regulates PGC-1α and mitochondrial biogenesis), SIRT3 (mitochondrial matrix, deacetylates respiratory-chain and antioxidant enzymes), SIRT6 (chromatin, DNA repair). Sirtuins have K_m values for NAD+ in the range where physiological fluctuations matter, which is why they are described as "NAD+ sensors."
• PARPs. PARP1 is activated by DNA single-strand breaks and consumes large amounts of NAD+ to synthesize poly(ADP-ribose) chains that recruit repair proteins. Under heavy DNA damage, PARP1 can deplete cellular NAD+ to the point of starving the sirtuins — this is the logic behind work in Cockayne syndrome, xeroderma pigmentosum, and ataxia-telangiectasia.
• CD38 and SARM1. CD38 is an ecto-NADase whose expression rises with inflammation and age; SARM1 is a destructase that triggers axon self-destruction when NAD+ falls below a threshold. These are the inflammation and neurodegeneration arms.

The standard hypothesis chain

The most-cited version of the hypothesis runs like this: accumulating DNA damage and chronic inflammation raise NAD+ consumption by PARP1 and CD38 → less NAD+ is available for SIRT1 and SIRT3 → SIRT1 loss contributes to a "pseudohypoxic" state (HIF-1α stabilization) that disrupts nuclear-to-mitochondrial signaling → mitochondrial function declines. In this chain, mitochondrial dysfunction is downstream of DNA damage and inflammation, not upstream. The Krebs cycle is mostly a bystander. DNA repair is arguably the trigger.

The Camacho-Pereira et al. 2016 Cell Metabolism paper (1,217 citations) is the foundational evidence for the CD38 arm: CD38 expression and activity increase with age in mice, CD38 knockout mice preserve NAD+ and mitochondrial function into old age, and the effect is mediated at least partly through SIRT3. But this is mouse data. The human evidence that CD38 drives age-related NAD+ decline is thin by comparison, and the claim that CD38 makes precursor therapy "futile" in older adults is a hypothesis, not a measured fact.

4. The four barriers — honestly assessed

Our earlier brief identified four barriers between the swallowed pill and the intracellular NAD+ pool. Three of them hold up well; one is overconfident.

Barrier 1: Presystemic clearance and microbial reprogramming (solid)

The Christen 2026 Nature Metabolism trial is the empirical foundation. The key finding: NR and NMN each doubled whole-blood NAD+ over 14 days, but intact NR and NMN were barely detectable in circulation. Instead, the Preiss–Handler pathway marker NAAD rose, and urinary excretion of nicotinic-acid riboside (NAR) and nicotinuric acid climbed. Ex vivo fermentation of NR and NMN with human fecal microbiota from multiple donors showed rapid conversion: NR → NAM → NA within 8–24 hours, and NMN → NAMN → NA through the same deamidation route.

In plain terms: the NAD+ boost from oral NR and NMN is largely a nicotinic-acid boost mediated by gut bacteria. Nicotinamide itself, which cannot be deamidated by the microbiome to NA, produced only a transient 1-hour NAD+ spike with no chronic elevation. This is the opposite of what the field assumed for a decade.

For the fraction of NR that escapes microbial deamidation, purine nucleoside phosphorylase (PNP) hydrolyzes it to free nicotinamide and ribose-1-phosphate. This is fast — intact NR has a whole-blood half-life of approximately three minutes (Trammell 2016). The liver then methylates the surviving nicotinamide via NNMT to MeNAM, 2PY, and 4PY for renal excretion. Peripheral tissues are exposed to systemic NAM and methylated catabolites, not intact parent nucleoside.

Barrier 2: Transport and compartmentalization (solid, with a recent breakthrough)

NR crosses the plasma membrane via ENT1/ENT2 and is phosphorylated by NRK1 (ubiquitous) or NRK2 (muscle, heart, adipose). NMN, a charged nucleotide, cannot cross by passive diffusion; it is either dephosphorylated to NR by cell-surface CD73 (functionally equivalent to dosing NR) or transported intact by Slc12a8, which is intestine-dominant. This means different tissues see different metabolite mixes, and blood NAD+ is a poor proxy for tissue NAD+.

The compartmentalization problem goes deeper than the brief stated. Cells maintain separate nuclear, cytosolic, and mitochondrial NAD+ pools, and the mitochondrial pool was long thought to be insulated. That changed in 2020, when two groups (Girardi et al. and Lu et al.) independently identified SLC25A51 as the long-sought mammalian mitochondrial NAD+ transporter. SLC25A51 selectively imports oxidized NAD+ into the mitochondrial matrix and is required for sustained cell respiration. The 2024 Lu paper in Cell Metabolism showed that SLC25A51 "decouples the mitochondrial NAD+/NADH ratio" from the cytosolic pool under certain conditions. This means raising cytosolic NAD+ does not guarantee the mitochondrial pool moves in lockstep — and the mitochondrial pool is the one SIRT3 cares about. Any trial that measures only whole-blood or PBMC NAD+ is blind to the compartment that matters most for the sirtuin hypothesis.

Barrier 3: The CD38 sink (real in mice, unproven in humans)

CD38 is a type-II transmembrane ectoenzyme on immune cells and endothelium that converts NAD+ to cyclic ADP-ribose, releasing nicotinamide. Its K_m for NAD+ is lower than most other NAD+-consuming enzymes, so it is constitutively active. The Camacho-Pereira 2016 paper showed CD38 protein abundance rises 2- to 5-fold with age in mouse liver, spleen, and muscle, and that this increase accounts for the majority of age-related NAD+ decline in those tissues. CD38-knockout mice preserve youthful NAD+ into advanced age.

This is compelling mouse biology. The human evidence is thinner. There is some flow-cytometry data showing CD38 is elevated on PBMCs from older adults, and the Tarragó 2018 Cell Metabolism paper showed the synthetic CD38 inhibitor 78c restores aged-tissue NAD+ and SIRT3 activity in mice — but 78c has no human pharmacokinetic data. The dietary flavonoids apigenin and quercetin inhibit CD38 in cell studies at low-micromolar concentrations, but their oral bioavailability is poor (single-digit percent for quercetin) and variable, and there is no human trial showing they meaningfully lower tissue CD38 activity at achievable doses. The brief's suggestion of "apigenin 50 mg or quercetin 500 mg" as a CD38-inhibitor protocol was speculation, not an evidence-based recommendation.

Barrier 4: Trial design (solid, and arguably the biggest problem)

This barrier is the one the critique rightly said is "retrofittable" — you can always explain a null result by claiming the dose was wrong, the duration too short, or the cohort too healthy. That criticism is fair. But the structural problems are real regardless of whether they are applied post hoc:

• Surrogate marker misalignment. The dominant endpoint is steady-state NAD+ in whole blood or PBMCs. This measures compliance, not efficacy. Blood NAD+ is mostly erythrocytes, which have their own kinetics and do not reflect what is happening in muscle, heart, or brain mitochondria. Worse, blood NAD+ itself does not reliably decline with age — so a rise after supplementation tells you even less about whether you corrected a deficit.
• Cohort ceiling effects. Most trials enrolled healthy or mildly metabolically abnormal adults. In bioenergetically intact people, homeostatic feedback maintains tight control over intracellular NAD+. The system actively resists surplus — Dollerup's high-dose trial may have shown exactly this, though the specific NAMPT-feedback claim requires primary-source verification we have not been able to confirm.
• Duration too short for the endpoint being measured. Mitochondrial biogenesis — new mitochondria, expanded networks, upregulated OXPHOS complexes — operates on a timescale of 8–16 weeks in human skeletal muscle. A 6-week trial cannot detect a structural mitochondrial endpoint. A 12-week trial is borderline. Most trials were 4–12 weeks.
• Monotherapy design. The salvage pathway is upregulated by exercise (NAMPT induction in skeletal muscle). The consumption side is gated by CD38 (suppressible by flavonoids, in principle). The signaling side requires functional sirtuins (regulated by circadian rhythm and substrate availability). Testing NR or NMN as a monotherapy in sedentary, inflammaging older adults is testing the molecule under conditions designed to make it fail.

5. The "wrong arm" problem — the deeper issue the field has not faced

Here is the problem the brief underweighted. If NAD+ does two jobs — redox cofactor and consumed substrate — and the aging hypothesis lives in the second job, then the question is which arm of the second job matters most in humans. The field has never agreed, and the trials overwhelmingly measured only one arm.

Almost every published human trial measured the mitochondrial/metabolic arm: skeletal-muscle mitochondrial respiration ex vivo, insulin sensitivity by clamp or HOMA-IR, body composition, aerobic capacity, lipid profile. These are the endpoints Remie, Dollerup, Yoshino, and most others chose. They are easy to measure because skeletal muscle is easy to biopsy.

Almost no trial measured the DNA damage arm: γH2AX foci, 8-oxo-dG, comet assay, PARylation levels, or SIRT6 target acetylation. Almost no trial measured the inflammation arm: cytokine panels, SASP markers, CD38 activity on PBMCs by flow cytometry. Almost no trial measured the neurodegeneration arm: axonal markers, SARM1 activity, neurofilament light chain.

If the real mechanism in humans runs mainly through DNA repair or inflammation, the trials measured the wrong arm and the nulls are unsurprising. If it runs through mitochondria, the trials measured the right arm and the nulls are real. The data cannot yet distinguish the two readings.

The Fang 2016 evidence for the DNA damage arm

The strongest case that the DNA damage arm matters comes from Fang et al. 2016 in Cell Metabolism, which showed that NAD+ replenishment improves lifespan and healthspan in ataxia-telangiectasia (A-T) model mice. A-T is caused by ATM mutation, a DNA double-strand-break sensor; the disease produces progressive cerebellar ataxia and neurodegeneration. The mechanism: defective DNA repair → chronic PARP1 hyperactivation → NAD+ depletion → sirtuin suppression → mitophagy failure → neuronal death. NAD+ replenishment reduced neuropathology, normalized neuromuscular function, delayed memory loss, and extended lifespan. This is the cleanest demonstration that in a DNA-repair-deficient context, NAD+ is genuinely limiting and supplementation helps.

The implication for trials: if your participants do not have heavy DNA damage burden, you may be testing a mechanism that does not apply to them. A-T is an extreme case, but the principle generalizes — the DNA damage arm should produce benefit only in people with elevated DNA damage (chronic inflammation, radiation exposure, certain chemotherapies, mitochondrial DNA disorders, advanced age with significant senescent-cell burden). Most trial cohorts did not enrich for these conditions.

6. Is NAD+ even limiting in healthy human aging?

This is the question the field has preferred not to ask directly, because the answer undermines the entire translational case. Two recent findings make it unavoidable.

Human tissue NAD+ decline: real but small, and possibly not functional

The Janssens 2022 review (PMC9417223) found that skeletal-muscle biopsies from older adults show lower NAD+ than biopsies from younger adults, and that physically trained older adults preserve more. This is the evidence the field uses to claim "NAD+ declines with age in humans." It is real, but the magnitude is modest — typically 10–30% — and the correlation with function is not strong.

The Chubanava et al. 2025 paper in Cell Metabolism directly challenged the functional interpretation. Using an inducible genetic system, they showed that NAD+ depletion in mouse skeletal muscle, even exceeding the magnitude seen in normal aging (10–30%), does not impair muscle function or accelerate its deterioration with aging. Their conclusion: "NAD depletion does not contribute to age-related decline in skeletal muscle function." If a 30% NAD+ drop does not measurably harm muscle, then correcting a 20% drop with a precursor should not be expected to help muscle — which is exactly what the trials showed.

Blood NAD+ does not decline with age at all

Multiple cohort analyses (summarized in 2026 reviews) found that whole-blood NAD+ does not significantly decline with age in healthy adults. This is doubly damaging to the standard trial design: not only is blood NAD+ a poor proxy for tissue NAD+, but the blood pool itself is not the thing that declines. Measuring a rise in blood NAD+ after supplementation is measuring a deviation from a non-declining baseline — it tells you nothing about whether you corrected a deficit.

What this means

If NAD+ is not substantially limiting in most healthy older adults — if the age-related decline is small and possibly non-functional, and the body's homeostatic system resists surplus — then "supplement doesn't help healthy people" is the correct result, not a design flaw. The hypothesis predicts benefit only where NAD+ is genuinely limiting: heavy DNA damage, high CD38/inflammation, primary mitochondrial disease, severe metabolic dysfunction. That is a much smaller market than "everyone over 40," and it is exactly the population the trials mostly did not test.

7. Four landmark trials, re-examined honestly

Remie 2020 — NR in overweight/obese adults

Six-week double-blind RCT of NR 1,000 mg/day versus placebo in overweight and obese adults, with paired skeletal-muscle biopsies. Blood NAD+ rose; muscle NAD+ metabolites (NAAD, MeNAM) rose; sleeping metabolic rate ticked up modestly; fat-free mass increased slightly. Ex vivo mitochondrial respiration, intramyocellular lipid, and insulin sensitivity were unchanged. Honest read: the dose may have been too low to overcome presystemic clearance and CD38 consumption (Barriers 1 and 3), six weeks was too short for mitochondrial biogenesis (Barrier 4), and the cohort was bioenergetically intact enough to hit a ceiling (Barrier 4). The trial does not refute NR; it refutes low-dose, short-duration NR monotherapy in metabolically borderline adults.

Dollerup 2018 and 2020 — high-dose NR in obese, insulin-resistant men

The 2018 American Journal of Clinical Nutrition paper: 12 weeks of NR 2,000 mg/day in obese, insulin-resistant men. Safe, but no improvement in insulin sensitivity or whole-body glucose metabolism. The 2020 Journal of Physiology paper: NR did not alter mitochondrial respiration, content, or morphology in skeletal muscle from the same population. Despite doubling the Remie dose and doubling the duration, the muscle mitochondrial endpoint stayed flat. (Note: an earlier version of our brief attributed a NAMPT-decrease finding to Dollerup. We have not been able to verify this against the primary papers and are retracting it. The verified findings are the null insulin sensitivity and null mitochondrial respiration.)

Yoshino 2021 — NMN in prediabetic postmenopausal women

This is the trial most often cited as evidence that NAD+ supplementation "works," and it is the one that most clearly demonstrates what right looks like. Ten-week RCT of NMN 250 mg/day in postmenopausal women with prediabetes, using the hyperinsulinemic-euglycemic clamp (gold standard for insulin sensitivity). The result: a roughly 25% improvement in skeletal-muscle insulin sensitivity, with upregulation of muscle-remodeling pathways mediated by PDGF signaling. Crucially, NMN did not directly elevate skeletal-muscle NAD+ — it elevated NMN metabolites in muscle and NAD+ in PBMCs. The benefit was strictly localized: hepatic and adipose insulin sensitivity, blood pressure, and plasma lipids were unchanged.

The trial succeeded because it stratified into a tissue-specific deficit (prediabetic muscle), chose a precursor whose transport profile favors muscle, used a low dose matched to uptake capacity rather than a high dose that might trigger feedback suppression, and ran long enough for muscle signaling to shift. It is a proof-of-concept that barrier-aware design produces barrier-crossing results — but it is also a single trial in a narrow population, and "NMN helps prediabetic postmenopausal women's muscle insulin sensitivity" is a much narrower claim than "NAD+ supplementation fights aging."

Christen 2026 — the mechanism revealer

Fourteen-day open-label RCT in 65 healthy adults comparing NR 1 g/day, NMN 1 g/day, NAM 0.5 g/day, and placebo. NR and NMN each approximately doubled whole-blood NAD+; NAM did not. Ex vivo fermentation with human fecal microbiota confirmed that the NAD+ rise was mediated largely by microbially produced NA via the Preiss–Handler pathway, not by direct tissue uptake of intact NR or NMN. This trial retroactively explains why every prior NR/NMN blood-NAD+-positive trial produced tissue-function-negative results: the wrong metabolite was being measured in the wrong compartment. It is the empirical foundation of Barrier 1.

8. What would actually settle this

The field has run enough monotherapy trials in healthy cohorts. Four things would move the question:

1. Isotope-tracer flux studies. The ongoing NCT06882096 trial uses 13C- or 2H-labeled NMN and NAM to map flux from oral precursors into blood, muscle, urine, and stool in young versus older adults. This is the trial design that can actually answer "did the precursor reach the tissue and contribute to the intracellular NAD+ pool?" Once isotope-tracer flux becomes standard, the field can distinguish pharmacokinetic failures (precursor never reached tissue) from pharmacodynamic failures (precursor reached tissue but did not translate to functional benefit).
2. Stratified enrichment cohorts. Enroll only participants with documented baseline deficits: elevated CD38 on PBMCs (measurable by flow cytometry), elevated DNA damage markers (γH2AX, 8-oxo-dG), confirmed mitochondrial DNA disease, or severe metabolic dysfunction. Stop testing NR/NMN in healthy 30-year-olds.
3. Multi-arm endpoints. Measure mitochondrial function (respirometry, SIRT3 target acetylation), DNA damage (γH2AX, PARylation), and inflammation (cytokine panel, SASP) simultaneously. If the benefit lives in the DNA repair or inflammation arm, single-arm mitochondrial trials will keep coming back null.
4. Reduced precursors (NRH/NMNH) with safety data. NRH is a neutral, PNP-resistant molecule that crosses membranes by passive diffusion and uses the ADK pathway. Preclinical head-to-head studies show 2- to 5-fold higher intracellular NAD+ than NR or NMN. But a 2026 chronic-supplementation study (Rumpler et al., Nature Communications) showed toxicity signals at high dose. Human PK and safety trials are in progress; no published efficacy data yet. NRH is the most plausible pharmacokinetic fix, but it is not a drop-in replacement for NR.

9. Bottom line: if you wanted to act on this today, what can we do

For a healthy adult under ~50 with normal glucose tolerance, normal lipids, and regular exercise habits: the expected benefit from NR or NMN supplementation is small, possibly zero. The bioenergetic ceiling is real, the homeostatic system resists surplus, the CD38 sink is not yet overactive in your tissue, and your mitochondria are not yet dysfunctional. The most reliable NAMPT inducer known — endurance exercise — is free and has a vastly larger evidence base. Spend your money there first.

For an adult with documented baseline deficits — prediabetes, dyslipidemia, primary mitochondrial DNA disease, significant inflammaging, or elevated DNA damage burden — the evidence is more encouraging but still narrow. NMN at 250 mg/day for 10 weeks improved muscle insulin sensitivity in prediabetic postmenopausal women (Yoshino 2021). The benefit was tissue-specific and population-specific. It is not a general anti-aging effect. If you fit a deficit population, the most defensible protocol given current evidence is:

• Precursor: NMN 250–500 mg/day or NR 300–600 mg/day. Higher doses do not produce proportionally larger tissue effects and may trigger feedback suppression.
• Timing: Morning, aligned with the circadian NAMPT peak. Avoid bedtime dosing, which may disrupt the salvage rhythm.
• Co-therapy (speculative but biologically rational): A CD38 inhibitor — apigenin (50–100 mg) or quercetin (500–1,000 mg), taken separately from the precursor to avoid absorption competition. Be aware that flavonoid bioavailability is poor and variable, and the human CD38-inhibition evidence is weak. This is the part of the protocol most likely to be revised as data comes in.
• Exercise: The single most reliable way to upregulate NAMPT in skeletal muscle. A precursor plus exercise is biologically more rational than a precursor alone, and the exercise effect size dwarfs the precursor effect size.
• What to track: Whole-blood NAD+ confirms you absorbed the pill. It does not confirm efficacy. The only currently accessible proxies for tissue-level effect are functional — continuous glucose monitoring for insulin sensitivity trends, aerobic capacity (VO2 max), body composition. If you want to know whether the protocol is working at the tissue level, push for isotope-tracer endpoints in any clinical context. Demand better biomarkers, not just more pills.

For anyone deciding whether to fund, prescribe, or take NAD+ precursors: the honest summary is that NR and NMN are probably not harmful, reliably raise blood NAD+, and probably do little in healthy adults. Modest benefits in specific deficient populations remain plausible but unproven at scale. The biology says the hypothesis should work where NAD+ is genuinely limiting. The execution — trial design, biomarker choice, cohort selection — is what has been missing. The next decade will be defined by whether the field adopts the fixes (isotope tracers, stratified cohorts, multi-arm endpoints, reduced precursors) or continues to run low-dose, short-duration, monotherapy trials in healthy cohorts and report null results as if they were informative.

The mystery is not solved. But the question is now clearer: not "does NAD+ supplementation work?" but "where, when, for whom, and how would we know?" The answer to that question is the only one worth acting on.

Sources

1. Imai S, Guarente L. NAD+ and sirtuins in aging and disease. Trends Cell Biol. 2014;24(8):464–471. https://pmc.ncbi.nlm.nih.gov/articles/PMC4112140/
2. Yusri K, Jose S, Vermeulen KS, et al. The role of NAD+ metabolism and its modulation of mitochondria in aging and disease. npj Metab Health Dis. 2025;3:26. https://www.nature.com/articles/s44324-025-00067-0
3. Christen S, Redeuil K, Goulet L, et al. The differential impact of three different NAD+ boosters on circulatory NAD and microbial metabolism in humans. Nat Metab. 2026;8:62–73. https://www.nature.com/articles/s42255-025-01421-8
4. Mills KF, Yoshida S, Stein LR, et al. Long-Term Administration of Nicotinamide Mononucleotide Mitigates Age-Related Physiological Decline in Mice. Cell Metab. 2016;24(6):795–806. https://pubmed.ncbi.nlm.nih.gov/28068222/
5. Mitchell SJ, Bernier M, Aon MA, et al. Nicotinamide Improves Aspects of Healthspan, but Not Lifespan, in Mice. Cell Metab. 2018;27(3):667–676. https://www.cell.com/cell-metabolism/fulltext/S1550-4131(18)30112-8
6. NIA News. Nicotinamide supplement in aging mice shows some health improvements, not increased longevity. https://www.nia.nih.gov/news/nicotinamide-supplement-aging-mice-shows-some-health-improvements-not-increased-longevity
7. Camacho-Pereira J, Tarragó MG, Chini CC, et al. CD38 Dictates Age-Related NAD Decline and Mitochondrial Dysfunction through an SIRT3-Dependent Mechanism. Cell Metab. 2016;23(6):1127–1139. https://www.cell.com/cell-metabolism/fulltext/S1550-4131(16)30224-8
8. Schultz MB, Sinclair DA. Why NAD+ Declines during Aging: It's Destroyed. Cell Metab. 2016;23(6):965–966. https://www.cell.com/cell-metabolism/fulltext/S1550-4131(16)30244-3
9. Tarragó MG, Chini CC, Kanamori KS, et al. A Potent and Specific CD38 Inhibitor Ameliorates Age-Related NAD+ Decline and Metabolic Dysfunction in Aged Mice. Cell Metab. 2018;27(5):1081. https://www.sciencedirect.com/science/article/pii/S1550413118301943
10. Girardi E, et al. Epistasis-driven identification of SLC25A51 as a regulator of mitochondrial NAD+. 2020. https://pmc.ncbi.nlm.nih.gov/articles/PMC7708531/
11. Lu MJ, et al. SLC25A51 decouples the mitochondrial NAD+/NADH ratio. Cell Metab. 2024. https://www.cell.com/cell-metabolism/fulltext/S1550-4131(24)00013-5
12. Yoshino M, Yoshino J, Kayser BD, et al. Nicotinamide mononucleotide increases muscle insulin sensitivity in prediabetic women. Science. 2021;372(6547):1224–1229. https://www.science.org/doi/10.1126/science.abe9985
13. Remie HME, et al. Nicotinamide riboside supplementation alters body composition and skeletal muscle acetylcarnitine concentrations in healthy obese people. Am J Clin Nutr. 2020;111(6):1229–1241.
14. Dollerup OL, et al. A randomized placebo-controlled clinical trial of nicotinamide riboside in obese men: safety, insulin-sensitivity, and lipid-mobilizing effects. Am J Clin Nutr. 2018;108(2):343–353. https://pubmed.ncbi.nlm.nih.gov/29992272/
15. Dollerup OL, et al. Nicotinamide riboside does not alter mitochondrial respiration, content or morphology in skeletal muscle from obese and insulin-resistant men. J Physiol. 2020;598(3):581–594. https://pubmed.ncbi.nlm.nih.gov/31710095/
16. Trammell SAJ, et al. Nicotinamide riboside is uniquely and orally bioavailable in mice and humans. Nat Commun. 2016;7:12948. https://www.nature.com/articles/ncomms12948
17. Fang EF, et al. NAD+ replenishment improves lifespan and healthspan in ataxia telangiectasia models via mitophagy and DNA repair. Cell Metab. 2016;24(4):566–581. https://pmc.ncbi.nlm.nih.gov/articles/PMC5777858/
18. Janssens GE, et al. NAD+ to assess health in aging humans. 2022. https://pmc.ncbi.nlm.nih.gov/articles/PMC9417223/
19. Chubanava S, et al. NAD depletion in skeletal muscle does not compromise muscle function and does not exacerbate age-related muscle decline. Cell Metab. 2025. https://www.cell.com/cell-metabolism/fulltext/S1550-4131(25)00212-8
20. ClinicalTrials.gov NCT06882096: Tracing the metabolic flux of orally administered NAD+ precursors. https://clinicaltrials.gov/study/NCT06882096
21. Zhang J, et al. Efficacy of oral nicotinamide mononucleotide supplementation: systematic review and meta-analysis. Crit Rev Food Sci Nutr. 2025;65(11):2387–2402. https://www.tandfonline.com/doi/full/10.1080/10408398.2024.2387324
22. Prokopidis K, et al. The effect of NMN and NR supplementation on sarcopenia-related outcomes: systematic review and meta-analysis. J Cachexia Sarcopenia Muscle. 2025;16(3):e13799. https://onlinelibrary.wiley.com/doi/full/10.1002/jcsm.13799
23. Rumpler M, et al. Therapeutic potential of dihydronicotinamide riboside. Nat Commun. 2026. https://www.nature.com/articles/s41467-026-70965-4

This article is a companion to the longer technical brief The NAD+ Paradox — Deep Research Brief (PDF), which covers pharmacokinetics, transport biology, and trial-design mechanics in greater depth. Where this article and the brief differ on confidence levels, this article reflects the more conservative reading. Specific claims about CD38 in humans, NAMPT feedback suppression, and flavonoid CD38-inhibition dosing are hypotheses requiring further human data, not established facts.