NAD+ precursors in food vs supplements: equivalence (or not) and correlation with public-class food-log data

Scientific delta of the report

The English twin freezes, in one factual chain, the same post-adversarial reranking and design cut as the Spanish report. This revision corrects the recorded bilingual QA failure: the background heading now contains mechanism, and confidence-interval wording is locked to the Spanish token inventory. L9-6’s title does not describe two doses of one NAD exposure. Mixed food and a pharmacological bolus split across five enzymatic doors (NAMPT, NAPRT-NA, NRK, QPRT, trigonelline-NAPRT) plus an NNMT methyl sink. At ~21 mg/d preformed niacin in U.S. adult women, NAMPT already captures nanomolar NAM and NNMT is first-order: Fashe 2024 gives NAM Cmax 170 µM after 1 g (0.40× hepatic Km 0.43 mM); the WWEIA pulse predicts ~3.6 µM systemic / ~6 µM portal, two orders below that Km. Therefore the 21 mg food–bolus 2×2 does not discriminate a methyl map by Tmax. The smallest discriminator is Q0-SPILL (n=12; 21 vs fractionated-21 vs 95 vs 250 mg): titrate R_spill against Cmax at matched NAM AUC, split H10R-diet from H10R-pharm, and collapse D0-FOOD-EQ to two periods if Tmax dies at food dose. NAM+NA-matched NAD+ new (H1R) remains coprimary H0. Cup trigonelline is absorbed in women (Lang 2010) and does not license a fifth NAD coin: COMP-H7 (98.6% of drinkers with reconstructed P5_hat moles ≥ that day’s niacin) is HPLC arithmetic; Cmax 6.55 µM sits ~48× below the HSMM EC50 of 315 µM. The food-log is a split instrument (OPEN + DR1TNIAC–coffee blindness), never Lua user data. G8/healthspan remains H0.

1. Executive scientific abstract

Food and supplement “NAD+ precursors” are not interchangeable by label milligram or by a diary score. USDA/AOAC tables report preformed niacin or niacin equivalents that hydrolyze NR/NMN to NA and collapse Trp×1/60. Food supplies NAM (meat), NA (cereal, often bound), protein Trp, NAD/NR/NMN at µg–mg/100 g, and trigonelline from coffee/fenugreek. A clinical NR/NMN/NAM bolus operates two to three orders above those dietary riboside moles [1–5].

In U.S. women ≥20 years, WWEIA 2015–2016 gives 21.2 mg/d preformed niacin (DR1TNIAC), not NE; the female EAR is 11 mg NE. The sufficiency slack is larger than the initial map attributed [6,7]. A public-class food-log does not recover that exposure: OPEN, 223 women aged 40–69, underreports energy 16–20% (24HR) and 34–38% (FFQ); raw 24HR–TEE r = 0.24 [8,9]. The trigonelline–B3 recall null is male (MEMOSA/Bushehr) and does not anchor female H3 [10].

Public computation in the same cycle closes female table-to-coffee blindness: n=2602, weighted mean 21.33 mg; Spearman coffee-g vs DR1TNIAC ρ=0.034 (all) and 0.042 after subtracting FNDDS niacin of coffee codes. With frozen HPLC factors, 53.7% of those women and 98.6% of drinkers have reconstructed trigonelline moles ≥ that day’s niacin moles [11–13]. That is not NAPRT occupancy. Lang 2010 (350 mL, six women): trigonelline Cmax 6.55 µM, parent urinary recovery 46.2±7.4%/8 h, plasma 2Py/4Py unchanged [14]. The HSMM EC50 is 315 µM (110 µM +FK866); a portal bound of ~3–29 µM stays below [10,15]. H9 (P5-as-NAD at cup dose) is weakened. Roast nicotinate (4–6% of initial trigonelline → NA) is the only coffee NAD door still live at table dose [16].

H4 is no longer a survey ceiling. Its molecular identity: NAMPT (Km NAM 5 nM +ATP; 855 nM −ATP) is already occupied at basal NAM 0.16 µM; extra NAM is methylated by unsaturated NNMT (Km 0.43 mM human liver) to MNAM + SAH [17–19]. Fukuwatari 2008 (n=6 university women): urinary recovery 37.9±4.8% → 71.9±10.1% on stepping from 27.6 to 95.0 mg NE/d; intercept ~18 mg NE, sitting on the WWEIA day [20]. NR-SAFE (3 g NR, n=10 mixed Parkinson) raises Hcy +1.66 µM without moving whole-blood SAM/SAH [21]. A gram bolus moves 2Py to 41 µM; a cup does not [14,17]. That distinguishes species; it does not prove spillover at 21 mg.

No human crossover of weighed diet vs bolus at matched absorbed NAM+NA with NAD+ new exists (EM-C6) [22]. H1R remains H0. Blood is not tissue (L9-3). Function without muscle NAD (Yoshino) and genetic ANM ≠ net longevity (Ruth) close G6–G8 [23,24]. Lua food-logs are forbidden: the instrument is qualified with public 24HR/tables.

Executive conclusion: maturity H1 for NAMPT/NNMT/NAPRT enzymology and female cup PK; H2 for the DR1TNIAC⊥coffee proxy and reconstructed P5 moles; H0 for NAD+ new equivalence, Tmax spillover at 21 mg, P5-as-NAD at cup dose, proximal function and healthspan. The smallest discriminator is Q0-SPILL before spending isotope on D0. A clean Tmax kill at food dose, or a clean instrument-H3 win, is scientific success. Nothing is prescribed.

2. Scientific question and relevance

Question

What mechanisms connect NAD+ precursors in food vs supplements — equivalence (or not) — and correlation with a public-class food-log to women’s health and longevity, and which new hypotheses can be falsified?

The question decomposes into gates that must not be fused:

1. Identity. Which species does food supply versus the bolus, in moles, not NE?
2. Coin. Do absorbed NAM+NA equalize late NAD+ new (H1R), or does residue remain from Trp, intact NR/NMN, trigonelline, bound niacin or Tmax?
3. Ceiling. Does diet operate in sufficiency (P1 occupied, NNMT spillover) and the bolus in a pharmacological range?
4. Instrument. Does a 24HR/FFQ reconstruct NAM+NA, coffee-P5, both, or neither?
5. Function and distal. Does a flux change mediate an independent endpoint and, only then, an organ event with competing risks?

“NAD foods” is not an exposure. “Correlation with Lua food-log” is not an estimand: current user data are forbidden. Later menopause is not healthspan.

3. Scope, population and life stage

Biological population: adult women in natural reproductive aging, STRAW+10, not age alone. Habitual-diet reference: late reproductive (−3b/−3a). Causal D0 window: one +1c stratum (50–60 years, 3–6 years post-FMP), L9-3’s anchor, so cycle is not mixed with Tmax. Q0-SPILL uses non-pregnant 18–45 (Fashe anchor) and is not called STRAW. Q0-INSTRUMENT: two blocks of 12 (transition vs +1c).

Inferential unit = woman. Diary days, meals, cups and aliquots are subsamples.

Out of scope: individual prescription and dosing; Lua food-logs, PHI, PII or genomes; NMN vs NR ranking (L9-3); CD38/flavonoids (L9-4); granulosa NAD/ovulation (L9-5); oocyte SIRT6 (L9-2); coformulations; p>0.05 as equivalence; male evidence transported without proof; fusion of fertility, CVD, brain, bone and mortality.

LATAM/nixtamal diets are a transportability priority (Carter n=3 men does not authorize magnitude; Goldsmith pellagra n=7 women is deficiency, not 2026 sufficiency).

4. Background knowledge and mechanism map

4.1 What the object is not

Niacin equivalents, DR1TNIAC, total blood NAD(H), NAAD, plasma NAM, urinary trigonelline, an “NAD foods” score and an app diary are not compartmental NAD+ nor function. Intact precursor <LLOQ does not prove failed delivery (L9-3-K3). Function without muscle NAD, or NAD without function, does not identify equivalence (L9-3-K6).

4.2 Original chain to destroy

“NAD-rich” food ≈ or < capsule → woman’s NAD+ → better diary / menopause / longevity.

That chain is inference today. L9-3 already rejected that the precursor’s name identifies what circulates. L9-6 extends the rejection to food and to the instrument.

4.3 Revised model: five doors and one sink

MIXED FOOD (moles, matrix, low Tmax)     NAMED BOLUS (100–1000 mg, high Tmax)
        │                                            │
        ▼                                            ▼
 Q0-CHEMISTRY / 24HR (AOAC collapses species)
        │                                            │
 LIGHT + SURFACE
  NAD/NADP → NAM; NMN → NR; NR → NAM
  bound niacin: alkali yes, gastric acid no
  trigonelline (N-methyl-NA) ──NAPRT→ ?NAD
        │                                            │
 FIVE DOORS
  P1 NAMPT   NAM → NMN → NAD+     (salvage; nM Km)
  P2 NAPRT   NA  → NAMN → NAD+    (Preiss–Handler)
  P3 NRK     NR  → NMN            (diet µg–mg; bolus 100–1000 mg)
  P4 QPRT    Trp → Kyn → QA       (60:1 is not a life-stage coefficient)
  P5 NAPRT   trigonelline → NAD+  (path at 300 mg kg−1; cup dose H0)
        │                                            │
 CEILING / SPILLOVER
  extra NAM ──NNMT + SAM──► MNAM → urinary 2Py/4Py
  NNMT Km 0.43 mM: first-order at 21 mg; toward Km ≥250–1000 mg
        │
 PBMC NAD+ new  ≠  muscle  ≠  ovary  ≠  healthspan
 Food-log: can rank coffee-P5 and fail NAM+NA at once

NA is not NAM in absorption: SLC5A8 transports nicotinate, not nicotinamide; GPR109A mediates NA flushing, not NAM/NR/NMN/trigonelline flushing [25,10]. Matching NAD+ new can leave receptor and methyl map discordant.

5. Evidence method

Scoping, map, primary verification (DOI/PMID/PMC), synthesis, compute decision (decision: use only NHANES+HPLC), public execution, hypothesis generation, adversarial review and experimental design were integrated. Classes: H human, A animal, V in vitro/ex vivo, C computational/tables, I inferred. Verification verdicts: V+, V±, R, Q.

NMN vs NR, CD38 and granulosa UDP-sugars are not reopened. Lua data are not used. RNA ≠ flux. p>0.05 ≠ equivalence. Unit = woman. Three PMIDs from the initial map were incorrect and are corrected here (Carter 6215855; Fukuwatari 2008 18635909; Fukuwatari 2004 15895512).

6. Evidence map

7. Contradictory evidence and null findings

7.1 H8 v1 was a conjunction

The package required NAD+ new in equivalence (not measured), higher trigonelline parent on coffee (definitional), higher 2Py on the bolus, and cup isotope <MDC, all at once. Adversarial dissolved it: H10R + H8b + H1R. Package confidence 0.18.

7.2 The cup plasma 2Py null does not discriminate

Female 2Py baseline ~1.13 µM; Lang LLOQ 20 nM; n=6; 8 h; NAPRT does not drain to 2Py. A tens-of-nM flux is lost in that background. Compatible with H8 and with H9. Spillover is titrated in 24–48 h urine of MNAM+2Py+4Py distinct from trigonelline parent and from N-methyl-2-pyridone-5-carboxylic acid [14,17,31].

7.3 COMP-H7 does not misspecify H1 by itself

One LOW cup (240 g × 0.14 mg/g = 0.245 mmol) already exceeds 0.172 mmol from 21.2 mg. The 98.6% of drinkers is prevalence × factor. If H9 loses, matching NAM+NA is again well specified; coffee is a compliance and P2 factorial, not a second coin.

7.4 48× Cmax/EC50 does not kill female hepatic NAPRT

6.55 µM is cup plasma; 315 µM is a male myotube bath. HepG2 does not convert trigonelline (low NAPRT); trigonelline is stable 72 h in serum [10]. The portal bound does not raise, it dilutes. H8b survives as order-of-magnitude underexposure, not as an enzymatic kill.

7.5 The 8-h urinary gap is not NAD+

Gebhardt: ≤64.3% parent/8 h; ~10% pyridone-carboxylic; ~30% late urine is plausible (t½ ~5.5 h predicts ~0.36 residual at 8 h). Human conversion to NA is “only assumed” [31]. H9 converts a non-closing mass balance into Preiss–Handler flux.

7.6 Fukuwatari is not a NAMPT ceiling nor food-versus-bolus

It adds encapsulated NAM to a Japanese diet; NE includes Trp×1/60. Recovery 38→72% is metabolite spillover, not NAD+ new, not Tmax vs AUC. The ~18 mg NE intercept lands on WWEIA; a Δ U_me at 21 mg can sit <MDC and weaken population H4 without proving equivalence.

7.7 Blood / tissue / function collage

NR/NMN/NAM raise blood NAD without intact precursor (L9-3-K3). Yoshino improved a postmenopausal clamp without raising muscle NAD [23]. That inherits rival F: the bolus can move non-NAD function; food can move function via caffeine/CGA/protein.

7.8 Mills vs Alegre

Broccoli NMN differs ~10×. Presence ≠ molar equivalence. Q0-CHEMISTRY does not average.

8. Multiscale mechanism synthesis

8.1 Molecular

Five substrates, five enzymes, one table number. H1R, if true, closes late P1 and the P2 fraction that goes through NA. It leaves P4, P5, bound niacin and Tmax outside.

NAMPT operates in the nanomolar. At basal NAM 0.16 µM it already exceeds both published Km values. The 21 mg dietary pulse is not read as “filling salvage”. NNMT, by contrast, has Km 0.43 mM: at predicted Cmax ~3.6 µM it is first-order (U_me ∝ AUC, not ∝ Tmax). Toward 250–1000 mg the portal bound (68–273 µM) enters the Km range. That is why H10R-diet and H10R-pharm do not collapse (ED-C1, ED-C2, ED-C4).

Trigonelline arrives already N-methylated. It does not spend SAM to exist. Bolus NAM does spend SAM to leave. The opposite methyl sign (MS-C2) survives as biochemistry; it does not survive as a fifth door at cup dose.

Trp→NAD is not 60:1 across reproductive life. Factor 67±~25% in 10 young women; ×2.3 in pregnancy; HT moves KA in +1c without NAD [28–30]. P4 is stage-modified, not diary-modified.

8.2 Cellular and first pass

Oral NR/NMN largely degrade to NAM; microbiota can deamidate (pncA); liver rebuilds via Preiss–Handler (L9-3-K4). In-vivo fraction in women: open. Food-versus-bolus asymmetry: food mixes species and fibre; the bolus is a pulse. G4-pncA is not first; it opens only if dietary NR/NMN are material (not unresolved broccoli µg) or if H9 lives.

Enterocyte: SLC5A8/GPR109A for NA; NAM by diffusion/ENTs. A cereal-NA menu is not a NAM bolus even if pyridine matches.

Adipocyte: WAT NNMT competes with NAMPT for NAM (Kraus 2014 animal; Kannt 2015 mixed). A sink candidate in transition; not measured against bolus vs food, not STRAW.

8.3 Compartment

PBMC is an accessible pharmacology matrix, not muscle, ovary, liver or brain. PBMC–tissue discordance closes biopsy (L9-3-Q7). Mural-high NAMPT (L9-5) does not titrate intake. Ovary closed as primary (SCOPE-C2).

8.4 Systemic and life course

No organ event is mediated by food-, bolus- or P5-derived NAD+. Ruth: later ANM improves bone/T2D and increases hormone-dependent cancers, with nulls for lifespan/CVD/Alzheimer [24]. ELITE moves kynurenine, not B3. Membrez grip–trigonelline is male. Coffee will associate via caffeine/CGA/confounding, not NAPRT, until a cup tracer wins. G8 closed (AR-C6, HG-H12).

8.5 Laboratory lineage

9. Computational layer

Decision: use of public NHANES/WWEIA 2015–2016 microdata + frozen HPLC factors before seeing results. GTEx, ovarian GEO and BioNeMo omitted: RNA and embeddings do not observe NAMPT occupancy, NNMT flux or food-versus-bolus NAD+ new.

Execution (2026-09-21): women ≥20, reliable Day-1, WTDRD1>0, n=2602. Weighted mean DR1TNIAC 21.33 mg (design CI 20.55–22.10) vs Table 37 21.2 mg. Spearman coffee-g vs niacin: ρ=0.034 all; 0.110 drinkers (n=1370); 0.042 after subtracting FNDDS niacin. Leave-one-code-out of the three most frequent codes remains |ρ|<0.20. FNDDS brewed 92101000 = 0.191 mg niacin/100 g (~2% in moles vs 21.2 mg).

MID factors: drip 0.25 mg/g; espresso 61.7 mg/25 g. P(P5_hat ≥ P1): 53.7% (50.2–57.1%) population; 98.6% (97.8–99.4%) drinkers. LOW already gives 50.0% / 91.9%.

What changes: female H3 of table P5-blindness is no longer OPEN (COMP-H6, H2 of proxy). D0 treats coffee as a factorial stratum. What does not change: not NAPRT flux, not serum, not equivalence, not Lua. Adversarial degraded COMP-H7 to an arithmetic threshold (AR-C2): it does not misspecify H1R unless H9 lives.

10. Primary hypothesis

L9-6-ED-H10R-diet v1 — NAMPT overflow at 21 mg, not NNMT saturation

Falsifiable statement: in women, at matched plasma NAM AUC0–24, a 21 mg NAM bolus produces a higher 24 h (urinary MNAM+2Py+4Py)/AUC_NAM ratio than the same amount split into ≥4 isocaloric doses (or a 3 h infusion); coffee does not move that ratio nor PBMC NAD+ new.

Parents: AR-H10R (split); MS-H5; SCOPE-H4; Aksoy 1994; Fashe 2024; Burgos 2008; Fukuwatari 2008; ED-C1–C4.

Proposed mechanism: at WWEIA dose, NAMPT is at relative Vmax versus extra NAM and NNMT is first-order. A Tmax contrast that survived would not be occupancy of the 0.43 mM Km (unreachable: 21 mg predicts ~3.6 µM systemic). It would be first-pass or a transient salvage overflow. If bolus/fractionated R_spill falls in 0.80–1.25, H10R dies as non-equivalence at population intake, not as gram pharmacology.

Predictions:

1. R_spill(21 bolus)/R_spill(fractionated 21) ∈ 0.80–1.25 with valid analytical Q0 (kill).
2. U_me/dose recovery rises from 21 to 95 mg (replicates Fukuwatari; spillover by dose, not form).
3. Coffee OFF in Q0-SPILL: P5/P2/caffeine are not mixed with Tmax.
4. If Δ U_me at 21 mg vs basal <MDC, population H4 weakens; the result is undecidable for Tmax, not “equivalence”.

Evidence for: nM NAMPT Km vs basal 0.16 µM [17,18]; NNMT Km 0.43 mM vs predicted µM Cmax [19]; female spillover feeding [20].

Evidence against: linear 1 g→21 mg scaling is a ceiling, not a measurement (Fashe cites non-linearity >1 g); first-pass portal unmeasured; n=6 Fashe/Fukuwatari is not +1c.

Kill: IC90% of the ratio inside 0.80–1.25 at 21 mg, valid Q0.

Status/maturity: proposed_primary_revised_split. H1 enzymology / H0 factorial at 21 mg. Confidence 0.28 (down from 0.44 for AR-H10R: the AR mechanism was NNMT Km, unreachable here).

Nested — L9-6-ED-H10R-pharm v1: the same contrast leaves the margin at 250–1000 mg, where portal (68–273 µM) approaches 0.43 mM. If it lives, the finding is methyl cost of a clinical bolus, not “food does not equal supplement” at 21 mg. Confidence 0.40.

11. Competing hypothesis

L9-6-ED-H1R v1 — NAD+ new at matched absorbed NAM+NA

Falsifiable statement: with Tmax and coffee fixed (the level that survives Q0-SPILL/D0a), PBMC NAD+ new AUC from a weighed diet vs a NAM bolus matching absorbed NAM+NA falls in 0.80–1.25 and MDC95<20%.

Parents: SCOPE-H1; AR-H1R; EM-C6; L9-3-K4.

Proposed mechanism: if P5 is not a door at cup dose and Tmax does not move R_spill at 21 mg, the late NAM+NA coin predicts synthesis. An out-of-margin residue, with A_NA inside, favours matrix, roast P2, Trp or pncA, not trigonelline-as-NAD.

Predictions:

1. Geometric food/bolus N_new ratio with IC90% in 0.80–1.25 if H1R is true.
2. If A_NA leaves the margin, the period is discarded: matching failure, not biology.
3. An N_new residue that vanishes on covariable Trp favours kynurenine.
4. p>0.05 is not called equivalence.

Evidence for: late NAM/NA convergence in mouse (Yaku; L9-3); NA —not NR/NMN— raises NAD in human blood ex vivo (Christen); population P1 sufficiency.

Evidence against: the contrast has not been run (EM-C6); bound niacin; espresso P2 4.20–8.80 mg [13]; stage P4; asymmetric luminal hydrolysis.

Kill: IC95% of the ratio outside 0.80–1.25 in ≥2 periods, valid species and isotope Q0.

Status/maturity: reinstated_coprimary. H0. Confidence 0.22.

Coffee-NAD competitor, not of the core — L9-6-ED-H2R v1: any extra NAD+ new on the coffee arm is abolished by matching infusion LC–MS nicotinate without matching trigonelline. FNDDS ~0.46 mg/240 g; Caprioli 4.20–8.80 mg/espresso. If the extra exceeds that P2, H9 or an unlisted species survive. Confidence 0.30.

P5 remnants — L9-6-ED-H8b / H9: cup isotope ≤MDC (H8b, confidence 0.36) vs >MDC and a material fraction (H9, 0.16). Gate: D-ISO does not open until extra coffee NAD after P2 match.

12. Translational hypothesis

L9-6-ED-HT2R v1 — the public instrument is qualified by split species

Falsifiable statement: in women, a 24HR reconstructed with public tables will (i) fail CCC≥0.70 against weighed food for NAM+NA; (ii) CCC(trigonelline_hat, parent urine) is labelled coffee compliance, not NAD qualification; (iii) neither DR1TNIAC nor trigonelline_hat predicts NAD+ new or urinary 2Py better than energy+protein. A single NE fails (i) and (iii).

Parents: SCOPE-H3; HG-HT2; AR-HT2R; OPEN; COMP-H6/H7; Lang 2011.

Proposed mechanism: three non-redundant failures. AOAC collapses species. OPEN kills energy/protein. The diary can “work” for a discrete item (coffee; Lang 2011, trigonelline/NMP urine 48–72 h) and fail the B3 coin. HT2-ii, if it wins, does not rescue H3 for NAD.

Predictions:

1. CCC(DR1TNIAC, weighed NAM+NA) IC95% below 0.70 (OPEN r=0.24 for energy sizes rejection of 0.70 if the true value ≤0.40).
2. trigonelline_hat ranks parent urine among drinkers with MID factors; fails if 0.25 mg/g is applied to a milk latte (compute rival B).
3. No score predicts N_new or 2Py better than energy+protein.
4. Forbidden: Lua food-logs. The test uses public class. HUMAN_QA_REQUIRED. Not called a biomarker.

Kill: an NE or DR1TNIAC predicts NAD+ new out of sample better than trigonelline_hat and better than energy+protein, with CCC≥0.70 vs weighed.

Status/maturity: parked_translational_test. H2 of table P5-blindness; H1 OPEN of E/P; H0 of trigonelline_hat CCC. Confidence 0.50 / 0.22 / 0.40.

Life course — L9-6-HG-H12 v1: no perturbation that matches PBMC NAD+ or MNAM, nor that moves serum trigonelline, will identify event-free years once global diet, energy, smoking and E2/P4 trajectory are matched. G8 closed. Confidence 0.11 for mediation; 0.72 for not inferring longevity from this project.

13. Falsifiable predictions and kill criteria

Quantitative rule: operational equivalence 0.80–1.25 and MDC95<20% (inherited from L9-3). If MDC95≥20%, the result is inconclusive, not equivalence and not a ceiling.

14. Discriminating experiment

Sequence (conditional; isotope is not dosed if Q0 leaves the contrast <MDC)

Q0-CHEMISTRY  →  Q0-SPILL n=12 (21 vs 21-frac. vs 95 vs 250 mg; coffee OFF)
                    ├ H10R-diet dies     → D0b two periods (H1R)
                    ├ H10R-diet lives    → D0a 2×2 form × coffee, then D0b
                    └ only ≥250 mg       → H10R-pharm; D0b at 21 mg without titrating foods
Q0-INSTRUMENT HT2R n=24  parallel; not a gate of causal D0; HUMAN_QA_REQUIRED
D-ISO / D-HEP / G4-pncA  conditional
G5–G6 / G8               inherited and closed

Q0-CHEMISTRY — hard stop

LC–MS of NAM, NA, NR, NMN, NAD, Trp, trigonelline, NMP, roast nicotinate, MNAM, 2Py, 4Py and N-methyl-2-pyridone-5-carboxylic acid in weighed food, bolus and infusion. Baseline resolution trigonelline (m/z 138) vs MNAM (m/z 137). Recovery 85–115%; CV ≤10–15%; MDC95 of U_me and PBMC NAD+ <20%. Niacin equivalents are not the assigned. Failure → STOP.

Q0-SPILL — the true smallest discriminator

Four-period Williams, n=12 completers (recruit 14). Women 18–45, coffee/cocoa/fenugreek OFF 48 h. Identical low-B3 background diet. Primary: R_spill(21 bolus)/R_spill(21 fractionated) ratio. Urine 0–8, 8–24, 24–48 h. No single plasma 2Py. No isotopic PBMC. If blinded SD after 8 completers requires n>20 for TOST, H10R-diet = undecidable and proceed to D0b for H1R only.

Q0-INSTRUMENT

n=24, two STRAW blocks. Three weighed-food days; AMPM 24HR before revealing weights; split urine. One day with 350 mL coffee; one day without. Forbidden: Lua.

D0a-SHAPE (only if H10R-diet lives)

Within-woman 2×2, n=16, +1c stratum, no NAD tracer. Coprimaries: U_me, C_NAM/R_spill, P5_parent, P2_cup. Coffee is not matched to NAM+NA.

D0b-FOOD-EQ

Two periods: weighed food vs NAM matching A_NA. Coffee and Tmax fixed. n=24 completers, blinded SD re-estimation at 12, cap 36. Coprimaries: PBMC N_new AUC and A_NA. NAM microtracer conditioned on ethics. PBMC is not muscle, ovary or healthspan. NR/NMN arm = later add-on (does not reopen L9-3 as primary).

Conditionals

D-ISO n=8: [13C]trigonelline at measured cup moles (typical MID 0.438 mmol; not 300 mg kg−1). D-HEP: female primary hepatocyte 3/30/110/315 µM ± FK866; localization, not STRAW magnitude. G4-pncA only if Q0 finds material dietary NR/NMN or H9 lives.

15. Biomarkers and stratification

There is no validated biomarker. Experimental candidate readouts:

NAAD is a secondary Preiss–Handler localization readout. Whole-blood Hcy/SAM/SAH are recorded at ≥250 mg; they are not called L6-2 vascular injury (NR-SAFE did not move whole-blood SAM/SAH). Erythrocyte NAD (Jacob/Fu, men in depletion) is a Q0 deficiency candidate, not NAD+ new.

Strata: STRAW on the instrument (two blocks) and one +1c stratum on D0; habitual coffee as characterization (the factorial imposes period coffee); BMI/insulin and eGFR as covariates; nixtamal/ethnicity recorded, not sized; Lua never.

16. Individual variability

Magnitude and sign can vary by:

• basal preformed-niacin intake and supplement use (WWEIA: 25% of women report supplemental niacin; 60+ years: 39%);
• coffee habit, roast, machine and milk (rival B: latte grams ≠ infusion);
• STRAW stage, HT and Trp→kynurenine deviation (ELITE);
• WAT NNMT and insulin (Kannt mixed);
• eGFR (renal M2PY; Fashe; Lenglet: 2Py rises with age, no reported sex difference);
• cereal vs nixtamal vs fortification matrix (Carter ♂; Goldsmith deficiency ♀);
• polymorphisms not titrated here (NNMT, NAPRT, NAMPT, SLC5A8);
• ancestry and LATAM transportability unknown at 2026 magnitude.

Heterogeneity will not rescue a null primary post hoc. Modifiers are prespecified after Q0.

17. Pharma relevance and maturity

H5 is not marked. HUMAN_QA_REQUIRED for any partnering language.

The potential asset is not an oral NAD booster, an “NAD food” or cup-dose trigonelline (space occupied by Membrez/Nestlé at 300 mg kg−1; COI declared). NAM is GRAS to ~3 g/d. NR/NMN already have clinical programs. No L9-6 arm generates composition exclusivity.

Go to internal evaluation (P0, post-D0b memo): U_me/R_spill qualifies as methyl PD in +1c; N_new at 21 mg is sensitive; tissue gaps listed.

No-go: claim “food–supplement equivalence”, “diary NAD”, “longevity”, ovary, or a cup-trigonelline IND. A null plasma 2Py after coffee is not promoted. COMP-H7 is not promoted to enzymatic occupancy.

Current maturity: H1 enzymology and cup PK; H2 of dietary proxy and reconstructed moles; H0 causal for equivalence, Tmax at 21 mg, P5-as-NAD at cup dose, function and healthspan. No partnering readiness. H5 is not marked.

18. Limitations

1. Linear 1 g → 21 mg Cmax scaling overestimates if NAM PK is non-linear >1 g; the 3.6 µM bound is a ceiling. Q0-SPILL replaces it.
2. Portal bound 1 L min−1 × 15–150 min is inference (AR-C3), not a measured concentration.
3. NAMPT Km in vitro; Ki NAD+ 0.14 µM collides with cellular mM NAD (compartment). “NAMPT occupied at basal” is H1 enzymology, not in-vivo STRAW occupancy.
4. Lang n=6, Cmax = reported maxima, not STRAW, one cup; Fashe 18–45 not +1c; D0 +1c does not inherit their Cmax.
5. Fukuwatari mixes Japanese diet + capsule; NE includes Trp×1/60; r=0.957 does not qualify USDA.
6. COMP-H7 uses HPLC factors, not LC–MS of each NHANES infusion; mixed drinks can overestimate grams.
7. Membrez: male 300 mg kg−1, male HSMM, MEMOSA/Bushehr men, Nestlé COI. The path is not the dose.
8. Carter n=3 men; Goldsmith n=7 women in pellagra; modern nixtamal outside.
9. Alegre vs Mills ~10× broccoli NMN remains; it is not averaged.
10. NR-SAFE is 3 g, mixed Parkinson, not dietary range.
11. OPEN n=223 is not replicated; Q0-INSTRUMENT n=24 is sized to reject CCC 0.70, not for epidemiology.
12. Equivalence 0.80–1.25 is metrological, not clinical or tissue.
13. ERα→NAMPT (Park/Hu 2026) is mainly male; not imported as G7.
14. No food-versus-bolus crossover with NAD+ new exists at cutoff.
15. It is not claimed that coffee, a dietary pattern or a supplement prolongs reproductive life or longevity. Lua data are not used. Nothing is prescribed.

19. Conclusions

L9-6 converts a commercial narrative — foods have NAD, the supplement is the same exposure in more milligrams, the diary proves it, and that touches women’s longevity — into a falsifiable sequence that can terminate early with an informative no-go.

What is usefully demonstrated is narrow. Tables collapse species. U.S. adult women are already in preformed-niacin sufficiency. The 24HR does not recover energy/protein and is blind to coffee as a trigonelline proxy. Cup trigonelline is absorbed in women, excreted largely intact, and does not resemble a NAM bolus, but that is not yet NAD+ new non-equivalence. Human NNMT is not saturated at 21 mg. NAMPT does capture nanomolar NAM. The causal food-versus-bolus contrast with NAD+ new still does not exist.

What is not demonstrated is the subtopic title. There is no measured equivalence or non-equivalence at matched NAM+NA. There is no Tmax spillover at food dose. There is no P5-as-NAD at cup moles. There is no food-log that measures precursor exposure. There is no mediation to an organ event.

The highest-value hypothesis is no longer “match foods to NMN”. It is more original and narrower: at the WWEIA dose, NNMT is first-order and the food–bolus 2×2 does not discriminate a methyl map; R_spill must be titrated against Cmax at matched AUC before spending isotope; if Tmax dies at 21 mg, remaining non-equivalence — if any — is of synthesis (H1R) or of instrument (HT2R), not of trigonelline nor of NNMT Km; and none of that authorizes healthspan. A clean refutation of the title is scientific success.

20. References

1. Mills KF, et al. Long-term administration of nicotinamide mononucleotide mitigates age-associated physiological decline in mice. Cell Metab. 2016;24:795–806. DOI: 10.1016/j.cmet.2016.09.013. PMID 28068222. Table 1: NMN mg/100 g.
2. Alegre GFS, Pastore GM. Occurrence of nicotinamide riboside and nicotinamide mononucleotide in edible plants. CyTA J Food. 2025. DOI: 10.1080/19476337.2025.2458753. Broccoli 13.059 mg NMN/100 g FW.
3. Çatak J. Determination of nicotinic acid and nicotinamide forms in some foods by HPLC. J Anim Sci Technol. 2019;61:138–146. DOI: 10.5187/jast.2019.61.3.138. PMID 31333870.
4. Ummarino S, et al. Simultaneous quantitation of nicotinamide riboside, nicotinamide mononucleotide and nicotinamide adenine dinucleotide in milk. Food Chem. 2017;221:161–168. DOI: 10.1016/j.foodchem.2016.10.032. PMID 27979136.
5. Alegre GFS, Pastore GM. NAD+ precursors nicotinamide riboside and nicotinamide mononucleotide. Curr Nutr Rep. 2023. DOI: 10.1007/s13668-023-00475-y. PMID 37273100. AOAC 944.13 hydrolyzes NR/NMN to NA.
6. USDA ARS. What We Eat in America, NHANES 2015–2016, Table 37. Women ≥20: food 21.2 mg niacin. PDF.
7. IOM. Dietary Reference Intakes for Thiamin, Riboflavin, Niacin, Vitamin B6, Folate, Vitamin B12, Pantothenic Acid, Biotin, and Choline. 1998. Surveys = preformed niacin, without Trp. Female EAR 11 mg NE.
8. Subar AF, et al. Using intake biomarkers to evaluate the extent of dietary misreporting in a large sample of adults: the OPEN study. Am J Epidemiol. 2003;158:1–13. DOI: 10.1093/aje/kwg092. PMID 12835280. 223 women 40–69.
9. Kipnis V, et al. Structure of dietary measurement error: results of the OPEN biomarker study. Am J Epidemiol. 2003;158:14–21. DOI: 10.1093/aje/kwg091. PMID 12835287.
10. Membrez M, et al. Trigonelline is an NAD+ precursor that improves muscle function during ageing and is reduced in human sarcopenia. Nat Metab. 2024;6:433–447. DOI: 10.1038/s42255-024-00997-x. PMID 38504132. PMC 10963276. HSMM EC50 315 µM (110 µM +FK866); 300 mg kg−1 male; MEMOSA/Bushehr men.
11. This project, stages/compute.md. NHANES 2015–2016 women ≥20, n=2602; coffee–niacin ρ 0.034/0.042; P(P5≥P1) MID 53.7%/98.6% drinkers.
12. Park J, Kim J, Kim J. Trigonelline contents in coffee. Foods. 2020;9:1746. DOI: 10.3390/foods9121746.
13. Caprioli G, et al. Quantification of caffeine, trigonelline and nicotinic acid in espresso coffee. Int J Food Sci Nutr. 2014;65:465–469. DOI: 10.3109/09637486.2013.873890. PMID 24467514.
14. Lang R, Wahl A, Skurk T, et al. Development of a hydrophilic liquid interaction chromatography–HPLC–tandem mass spectrometry based stable isotope dilution analysis and pharmacokinetic studies on bioactive pyridines in human plasma and urine after coffee consumption. Anal Chem. 2010;82:1486–1497. DOI: 10.1021/ac902616k. PMID 20073472. 350 mL; 6♀/7♂; Cmax ♀ 6547 nM; parent 46.2±7.4%/8 h.
15. Lang R, Wahl A, Stark T, Hofmann T. Urinary N-methylpyridinium and trigonelline as candidate dietary biomarkers of coffee consumption. Mol Nutr Food Res. 2011. DOI: 10.1002/mnfr.201000656. 5♀+4♂; 48–72 h.
16. Lang R, Yagar EF, Eggers R, Hofmann T. Quantitative investigation of roast-associated nicotinic acid formation. J Agric Food Chem. 2008. DOI: 10.1021/jf802838s. PMID 19007232. 4–6% of initial trigonelline → NA.
17. Fashe MM, Le TV, Gower MN, et al. Impact of pregnancy on the pharmacokinetics of nicotinamide. Clin Pharmacol Ther. 2024;115:556–564. DOI: 10.1002/cpt.3146. PMID 38093631. PMC 11250906. n=6 non-pregnant: NAM Cmax 170 (141–184) µM; 2Py Cmax 41.0 µM; predose NAM 0.16 µM.
18. Burgos ES, Schramm VL. Weak coupling of ATP hydrolysis to the chemical step of NAMPT. Biochemistry. 2008;47:11086–11096. DOI: 10.1021/bi801198m. Km NAM 855 nM (−ATP) → 5 nM (+ATP).
19. Aksoy S, Szumlanski CL, Weinshilboum RM. Human liver nicotinamide N-methyltransferase. J Biol Chem. 1994;269:14835–14840. PMID 8182091. Km NAM 0.43 mM.
20. Fukuwatari T, Shibata K. Urinary water-soluble vitamins and their metabolite contents as nutritional markers. J Nutr Sci Vitaminol. 2008;54:223–229. DOI: 10.3177/jnsv.54.223. PMID 18635909. n=6; 27.6→95.0 mg NE/d; recovery 37.9±4.8% → 71.9±10.1%.
21. Berven H, et al. NR-SAFE: nicotinamide riboside in Parkinson’s disease. Nat Commun. 2023;14:7793. DOI: 10.1038/s41467-023-43514-6. PMID 38016950. Hcy +1.66 µM at 3 g.
22. This project, stages/evidence-verification.md (L9-6-EM-C6): Christen NCT05517122 compares boluses, not weighed food; no NAM+NA-matched crossover with NAD+ new at cutoff 2026-09-20.
23. Yoshino M, et al. Nicotinamide mononucleotide increases muscle insulin sensitivity in prediabetic women. Science. 2021. DOI: 10.1126/science.abe9985. PMID 33888596. Function without muscle NAD. Inherited L9-3.
24. Ruth KS, et al. Genetic insights into biological mechanisms governing human ovarian ageing. Nature. 2021;596:393–397. DOI: 10.1038/s41586-021-03779-7. PMID 34349265. ANM ≠ net longevity.
25. Gopal E, et al. Sodium-coupled and electrogenic transport of B-complex vitamins by SLC5A8. Biochem J. 2005;388:589–596; Gopal E, et al. 2007. PMID 17245649. NAM is not an SMCT1 substrate.
26. Carter EGA, Carpenter KJ. The bioavailability for humans of bound niacin from wheat bran. Am J Clin Nutr. 1982;36:855–861. DOI: 10.1093/ajcn/36.5.855. PMID 6215855. n=3 men; 24/62/89%.
27. Goldsmith GA, Sarett HP, Register UD, Gibbens J. Studies of niacin requirement in man. I. Experimental pellagra in subjects on corn diets. J Clin Invest. 1952;31:533–542. DOI: 10.1172/JCI102638. PMID 14938464. n=7 women 25–54.
28. Horwitt MK, Harvey CC, Rothwell WS, Cutler JL, Haffron D. Tryptophan-niacin relationships in man. J Nutr. 1956. 60:1 convention, men.
29. Fukuwatari T, et al. Conversion ratio of tryptophan to niacin in Japanese young women. J Nutr Sci Vitaminol. 2004;50:385–391. PMID 15895512. n=10; factor 67.
30. Balasubramanian R, et al. Hormone therapy and kynurenine metabolites in early postmenopausal women (ELITE). Climacteric. 2025. PMID 40512395. n=80; not oral B3.
31. Gebhardt L, et al. Trigonelline: pharmacokinetics and metabolism. Nutrients. 2023. PMC 10146819. t½ ~5.5 h; ≤64.3% parent/8 h; conversion to NA “only assumed”.
32. Lenglet A, et al. N-methyl-2-pyridone-5-carboxamide (2PY) in CKD. Toxins. 2016;8:339. PMC 5127135. Healthy plasma 2Py ~0.39–4 µM.
33. Trammell SAJ, et al. Nicotinamide riboside is uniquely and orally bioavailable in mice and humans. Nat Commun. 2016;7:12948. PMID 27721479. NAD metabolome in PBMC; n=12 (6♀); no administered NAD+ new tracer.
34. This project: stages/scoping.md, evidence-map.md, evidence-verification.md, mechanism.md, compute-decision.md, compute.md, hypotheses.md, adversarial-review.md, experiment-design.md.

21. Claims and hypotheses of this report

Claims added: L9-6-EN-C1, L9-6-EN-C2, L9-6-EN-C3.

• L9-6-EN-C1: the food-versus-supplement title is frozen as potential non-equivalence of a methyl map by dose/form, not of cup trigonelline-NAD: Q0-SPILL must titrate R_spill against Cmax at matched AUC before the 2×2 at 21 mg, because Fashe 170 µM / 1 g is 0.40× NNMT Km and 21 mg predicts ~3.6 µM.
• L9-6-EN-C2: COMP-H7 (98.6% of drinkers) and Lang’s null plasma 2Py remain rejected as a fifth-coin licence or as spillover; NAM+NA-matched NAD+ new H1R is reinstated coprimary H0; H9 is weakened (0.16).
• L9-6-EN-C3: Pharma and healthspan do not inherit a dietary NAD asset; they inherit exposure metrology (U_me, split 24HR) and closed G8 (Ruth/Yoshino). The food-log is public-class; Lua data forbidden.

Hypotheses touched: L9-6-ED-H10R-diet v1, L9-6-ED-H10R-pharm v1, L9-6-ED-H1R v1, L9-6-ED-H2R v1, L9-6-ED-H8b v1, L9-6-ED-H9 v1, L9-6-ED-HT2R v1, L9-6-AR-H10R v1, L9-6-AR-H1R v1, L9-6-HG-H8 v1, L9-6-HG-H9 v1, L9-6-HG-H12 v1, L9-6-MS-H5 v1, L9-6-COMP-H6 v1, L9-6-COMP-H7 v1, L9-6-SCOPE-H1 v1, L9-6-SCOPE-H3 v1, L9-6-SCOPE-H4 v1.

Nothing is written into memoria agentes/Lua Labs/reportes/. Laboratory memory is not consolidated. Nothing is prescribed.