🧬 Aging Wiki

nad-blood-biomarker

Properties1
aliaseswhole-blood NAD+, blood NAD, circulating NAD+, NAD biomarker, plasma NAD

10 min read

Whole-Blood NAD+ as an Aging Biomarker (refuted, 2026)

Whole-blood nicotinamide adenine dinucleotide (NAD+) concentration was proposed across multiple 2016–2025 studies as a candidate biomarker of human biological aging — the rationale being that NAD+ declines with age in rodent tissues, and that blood is the most clinically accessible matrix in which to measure it. As of 2026 this hypothesis is directly refuted at the whole-blood level by the most rigorous human assessment to date: Trętowicz et al. 2026, Nature Metabolism (n=303 across 7 independent cohorts) finds whole-blood NAD+ is stable with age (all six independent age comparisons null: P=0.24–0.62, R²=0.012–0.051) and stable across lifestyle interventions (exercise, protein supplementation) in older adults. The same UHPLC-HRMS assay readily detects pharmacological NAD+ elevation from NR supplementation, ruling out a sensitivity floor.

This page documents both the original biomarker hypothesis and why it failed, since the negative result is itself informative and the prior-literature misclassification is structurally instructive about pre-analytical artifacts in metabolomic aging biomarkers.

Bottom line for clinical use: A blood NAD+ measurement in 2026 carries essentially no information about an individual’s biological age, and a single blood NAD+ value should not be used to advise NAD+ precursor supplementation or to track aging-intervention response in healthy adults.

Why blood NAD+ was proposed as an aging biomarker

Five lines of evidence motivated the original hypothesis:

  1. Strong rodent data. NAD+ declines substantially in mouse liver, white adipose tissue, and skeletal muscle with age (Yoshino 2011, Mills 2016, Camacho-Pereira 2016; reviewed in nad-precursors). The magnitude varies by tissue but reaches 50–90% in some compartments.
  2. Mechanistic plausibility. Several aging-associated processes converge on NAD+ depletion: CD38 upregulation (NAD+-degrading NADase, rises with inflammaging), PARP1 hyperactivation (substrate-consuming DNA repair under accumulating genotoxic stress), and NAMPT decline (rate-limiting salvage enzyme expression decreases with age).
  3. Limited human tissue data was confirmatory. Massudi 2012 (n=49) reported age-associated NAD+ decline in human skin biopsies (r=−0.71, p<0.001 in males); Janssens 2022 (n=88) reported a positive association between healthy aging and muscle NAD+ abundance.
  4. Several blood-NAD+/age studies reported positive correlations. Chaleckis 2016, Yang 2022, Wang 2023, Breton 2020 each reported an age-related decline in some matrix of blood NAD+.
  5. Clinical accessibility. Compared to muscle biopsies or skin punches, blood draws are routine. A blood-based aging-NAD+ biomarker would have been ideal for trial PD monitoring and population screening.

The class-level NAD+-precursor supplementation field (nad-precursors) developed largely on premise #1 (rodent decline) plus premise #5 (clinical accessibility), with the implicit assumption that blood NAD+ would track tissue NAD+ as the biomarker.

Why the hypothesis failed: Trętowicz 2026

Trętowicz et al. 2026 is the definitive negative result. Three features of the design make it more rigorous than the prior positive literature:

1. Seven independent cohorts (n=303 total)

Each cohort is independently powered for an age comparison; the meta-evidence is consistent across diverse populations (frail, athletic, post-cardiac-event, long-lived-family) and geographies (Netherlands, Spain, Finland):

CohortnPopulationAge comparisonResult
Aging cohort4020 young <30y + 20 older >60y, NLOlder vs YoungerP = 0.24 (NS)
CardioHT26Mixed-age cardiology referrals, NLAge vs NAD+ regressionR² = 0.012
ELITE47Young controls + older controls + athletes, NLOlder vs AthleteP = 0.50 (NS)
LLS (Leiden Longevity Study)70Older adults 63–87y, NLAge within older rangeR² = 0.051
TEAMS65Older adults >65y, exercise±protein RCT, NLPre vs post interventionP = 0.62 (NS)
MEJNES201931Frail older adults >65y, supplement±exercise RCT, ESPre vs post interventionP = 0.62 (NS)
Twin-pair NR cohort24Healthy 33–41y, NCT03951285, FINR supp vs baseline (positive control)Significant rise (per abstract)

The age-comparison cohorts span the most-tested human populations for blood-NAD+/age claims; all are null.

2. UHPLC-HRMS assay rigorously controlled for pre-analytical artifacts

The authors trace much of the conflicting prior blood-NAD+/aging literature to pre-analytical handling effects. From the paper’s own source data:

  • Both freezer temperatures degrade NAD+, not just −20°C (Extended Data Fig. 1b): fresh ~40 nmol/mL → −80°C ~31 nmol/mL (P=0.0015) → −20°C ~24 nmol/mL (P=0.023 vs fresh; −80°C-vs-−20°C P=0.38). Storage at −20°C for 4–5 days destroys ~52% of whole-blood NAD+; −80°C destroys ~22%. No standard freezer protocol preserves NAD+ fidelity — and most prior studies used standard −80°C archive storage.
  • Repeated freeze-thaw cycles further degrade NAD+: loss is highly variable across donors (Fig. 1f shows two representative donors with cumulative losses of −4% vs −32% after 3 cycles; Extended Data Fig. 1c includes 4 additional donors with losses up to ~69%).
  • Methanol preservation before thawing recovers NAD+ to within ~1 SD (3.5 nmol/mL) of fresh values across n=15 paired technical replicates — but most prior studies did not use this protocol.
  • Plasma NAD+ is 50–100× lower than whole-blood NAD+ and falls at or below the LOQ (0.69 nmol/mL) for most samples (n=10 donors: 0–0.58 nmol/mL). NAD+ is overwhelmingly intracellular. Any plasma-NAD+ aging claim is methodologically suspect.

Assay specifications (Trętowicz 2026 Methods § Validation of quantitative NAD+ measurement): UHPLC-HRMS (Waters Acquity + Bruker Impact II Q-TOF) with ¹³C₅-NAD+ internal standard spiked into 10 µL whole blood. LOD 0.21 nmol/mL · LOQ 0.69 nmol/mL · intra-assay CV 12.5% · inter-assay CV 16.1% · linearity R²=0.9996 across 10–100 µL input · spike-recovery 108–112% across 500/1000/2000 pmol additions.

Implication: prior reports of blood-NAD+ decline with age may primarily reflect age-dependent pre-analytical handling differences in sample collection cohorts (e.g., older participants more likely to be hospitalized → samples handled differently → apparent NAD+ deficit), rather than true biological age effects.

3. Positive control rules out sensitivity floor

The twin-pair NR cohort (n=24 monozygotic BMI-discordant twins, ages 33–41 y, NCT03951285, 5-month NR escalated from 250 mg/day → 1 g/day, Finland) was specifically included as a positive-control assay-sensitivity test. The same UHPLC-HRMS assay detected a ~2-fold NR-induced NAD+ elevation: median 32 → 62 nmol/mL pre→post, P = 1.59 × 10⁻⁹ (linear regression adjusted for sex; Extended Data Fig. 3). The null age-NAD+ effect (P-values 0.24–0.62) is therefore not an artifact of insufficient assay sensitivity — the contrast in P-values spans ~10⁹, and the assay can clearly distinguish pharmacological from putative age-related effects when the latter exist.

Power analysis: how large must a study be to detect a “true” blood NAD+ age effect?

From the paper’s own power analysis (Extended Data Fig. 4, source data MOESM6):

  • To detect Δ = 1 nmol/mL (~5% of typical baseline) with 80% power: N ≈ 786 per group once analytical variability is included.
  • To detect Δ = 2 nmol/mL (~10% of baseline): N ≈ 197 per group.
  • To detect Δ = 5 nmol/mL (~25% of baseline): N ≈ 14–31 per group depending on noise assumptions.

Most prior individual-cohort blood-NAD+/aging studies (n = 20–100 per group) had power to detect only very large (>3 nmol/mL) effects. The current paper’s 7-cohort design (303 total) is the largest pooled assessment to date and is well-powered to detect biologically meaningful age effects if they exist.

What the negative result does NOT rule out

The Trętowicz 2026 result is specific to whole-blood NAD+ as an aggregate-pool aging biomarker in adults aged ~19–87. It does not falsify the broader NAD+/aging story:

  • Tissue NAD+ decline remains supported by Janssens 2022 (human muscle, n=88) and Massudi 2012 (human skin, n=49). These tissue findings are not directly contradicted; only the blood-tissue extrapolation is.
  • NAD+ flux / turnover (vs static pool size) is not measured. If NAD+ consumption rises with age but salvage compensates, pool size could be stable while functional NAD+-dependent signaling (sirtuin activity, PARylation, CD38 activity) deteriorates. No human assay currently captures NAD+ flux at the whole-blood level. 1
  • Sub-population NAD+ deficits may still exist. The current cohorts include frail community-dwelling older adults (MEJNES2019 baseline) but not, e.g., subjects with confirmed CHIP, severe sarcopenia, COPD, PAD, or progeroid syndromes (Werner). Disease-specific NAD+ deficits remain plausible — and consistent with the positive disease-specific RCT signals on NR (Heggelund 2024 COPD, McDermott 2024 PAD, Takeda 2025 Werner).
  • NR-induced blood NAD+ elevation is real and reproducible — confirmed by the Trętowicz twin-pair positive-control arm. Blood NAD+ remains a valid pharmacodynamic biomarker for NR engagement (i.e., “did the participant take the supplement?”), even though it is not a valid aging biomarker.
  • Pediatric / inter-generational NAD+ dynamics were not assessed. Cohorts spanned 19–87y; childhood and adolescent dynamics are open.

Sources of confusion in the prior literature

Several recurring sources of confusion in pre-2026 blood-NAD+/aging studies:

  1. PBMC NAD+ ≠ whole-blood NAD+ ≠ plasma NAD+. Different matrices, different distributions, different age relationships. Many citations conflate these. Trętowicz 2026 measures whole blood; the Yoshino 2021 NMN trial measured PBMC; Massudi 2012 measured skin biopsy. Direct comparisons across these are not warranted.
  2. Frozen samples lose NAD+ rapidly without methanol preservation. Any study using archive freezer → bench → analysis workflow without methanol-pre-freeze loses ~30–80% of NAD+. Older cohorts may have been over-represented in archives with longer freezer time → spurious “age-related decline.”
  3. NADH and other pyridine nucleotides interconvert during sample handling. NAD+/NADH ratio measurements are particularly susceptible to handling-induced shifts.
  4. NADP+ shows a genuinely different age pattern from NAD+ (Trętowicz 2026 Extended Data Fig. 5: NADP+ Older-vs-Younger P=0.031 in the aging cohort, R²=0.237 in CardioHT — see study page). Studies measuring “total NAD(P) pool” may have detected the NADP+ signal and attributed it to NAD+.
  5. Mendelian randomization not yet applied to circulating NAD+. Whether genetically-determined NAD+ levels predict aging outcomes via MR-instrumentable variants is untested. The negative Trętowicz result limits the prior probability of a meaningful MR signal at the blood level.

Implications for the wiki and for aging-intervention practice

For the nad-precursors intervention class: the “restore-an-age-related-deficit” rationale is undermined at the blood level. The class can still proceed on tissue-level (muscle/skin) or pathological-state (COPD, PAD, Werner, MCI) rationales — and the positive disease-specific RCT signals from 2024–2026 (Heggelund, McDermott, Takeda) provide that grounding. But the simplest version of the supplementation story (“blood NAD+ declines with age → supplement to restore it”) is no longer defensible.

For clinical practice: do not order whole-blood NAD+ as an aging biomarker. There is no validated reference range linking blood NAD+ to biological age, no actionable threshold, and no evidence that lifestyle interventions move it.

For trial design: blood NAD+ remains valid as a pharmacodynamic biomarker (confirming NR/NMN ingestion) but is invalid as a primary biological-age endpoint. Trials previously designed around blood NAD+ as a surrogate for “biological aging” need redesign with tissue-level or composite-clinical biomarkers (dunedinpace-2022, grimage-2019, phenoage-2018).

Comparison with other “refuted” or “weak” aging biomarkers

The biomarker-refutation pattern is not unique to blood NAD+:

  • Serum klotho: Initially proposed as an aging biomarker following the klotho-knockout mouse phenotype, but human serum klotho shows weak/inconsistent age correlations and limited intervention responsiveness.
  • Plasma 25-OH vitamin D: Strong age-related decline exists but is largely attributable to sunlight exposure and lifestyle confounders rather than aging biology per se; supplementation does not extend lifespan in MR or RCT.
  • Salivary cortisol diurnal slope: Promising in cross-sectional studies, weak in longitudinal designs.

The lesson from Trętowicz 2026 generalizes: aging biomarkers fail validation more often when pre-analytical handling is not rigorously controlled and when the underlying biological assumption (tissue-of-decline → blood-of-decline) is unexamined.

Footnotes

Footnotes

  1. no-mechanism — Whole-blood NAD+ flux/turnover is not currently a routine analyte. Isotope-tracer studies (e.g., D2-nicotinamide infusion) can measure NAD+ flux in animal models and have been piloted in humans for pharmacokinetic purposes, but a clinical-grade NAD+ flux biomarker for aging is not established. If functional NAD+-dependent signaling deteriorates with age while pool size is stable, flux is the missing measurement. ↩

Graph View

Backlinks