Supplements & Compounds: What the Primary Evidence Supports

This is an evidence hub for the compounds themselves. For each compound, the question is narrow: what does the primary literature support, what is merely claimed (analyst / expert opinion expert), and where does a conflict of interest sit? The wiki is unusually careful about COI, and so is this page.

The recurring lesson is "surrogate ≠ outcome." Several of these compounds reliably move a biomarker while failing to move a hard outcome, a pattern that repeats across the NAD⁺ precursors, niacin, and vitamin D.

NAD⁺ precursors, the biomarker moves, the outcome doesn't (for healthy people). NMN raises blood NAD⁺ dose-dependently [1]; a 300 mg RCT raised NAD⁺ with no change in walk distance, BP, insulin, or well-being [2]. But NMN does not raise muscle NAD⁺ [3], and VO₂max was unchanged, only submaximal aerobic thresholds moved [4]. NR tells the same story: 1 g/day raises blood NAD⁺ ~60% [5] yet delivers null function and null cognition in healthy adults and in MCI [6], and null systolic BP on top of exercise [7]. A landmark three-way RCT found the gut microbiome converts both NMN and NR to nicotinic acid before absorption, neither is itself the primary systemic precursor [8]. Where NR does earn : three genuine between-group anti-inflammatory RCTs, IL-6/IL-5/IL-2 in aged men [9], sputum IL-8 in COPD [10], and CRP/IL-17 in psoriasis [11], and arterial-stiffness improvement in Werner syndrome, a progeroid context that does not generalize [12]. COI flag: NR's leading advocate Charles Brenner is chief scientific advisor to ChromaDex/Niagen (SEVERE COI); his "8 anti-inflammatory RCTs" count audits down to 3 genuine qualifiers.

Resveratrol, animal-scoped, human-null, COI on both sides. It extends survival and activates SIRT1 in high-calorie-diet mice [13]. The widely cited "lean-mouse lifespan extension" is refuted by the paper Sinclair co-authored, standard-diet mice showed no significant extension [14], and the independent ITP found no extension in normal-chow mice, with resveratrol detected in plasma, defeating the bioavailability excuse [15]. COI: Sinclair (Sirtris/resveratrol history) and Brenner (NR, a commercial rival) hold opposite, financially loaded positions; the cleanest independent signal is the ITP null.

Niacin, strong evidence, mostly negative. Niacin lowers LDL and raises HDL [16, 17], but adds no cardiovascular benefit on a statin and causes net harm across two large RCTs [16, 17], and its terminal metabolite 4PY is mechanistically tied to vascular inflammation and tracks MACE [18]. The "niacin benefit" from older meta-analyses does not survive pooling: D'Andrea was null in the main analysis [19] and the Cochrane synthesis was null overall [20]. Verhoeven's analyst synthesis "it works vs non-statin comparators" reading is selective [21].

Cardiovascular botanicals, real lipid/BP effects, no proven plaque reversal. Berberine lowers LDL-C ~15-16% [22] and TMAO >30% [23]; its plaque-regression headline rests on a single open-label, non-randomized study and stays preliminary [24]. Nattokinase modestly lowers BP in hypertensives (DBP −2.3 mmHg, SBP −3.5 mmHg pooled) [25] with no lipid effect; Verhoeven's "4-5 mmHg" overstated the pooled figure [26]. Lycopene's one positive CIMT RCT combined it with lutein, so the effect cannot be pinned on lycopene [27], against a null short trial [28]. Potassium deficiency drives arterial calcification via smooth-muscle transdifferentiation in mice [29], preclinical only.

Omega-3, dose- and endpoint-dependent, with a real harm signal. All-cause mortality is unaffected [30], CV events drop modestly [31], the benefit tracks EPA specifically (REDUCE-IT and JELIS positive, EPA+DHA STRENGTH null) [32, 33, 34], and omega-3 raises atrial-fibrillation risk dose-dependently [35]; a UK Biobank cohort found ↑AFib but AF→MACE HR 0.92 in supplementers [36]. Omega-3 augments the muscle anabolic response in older adults (a surrogate, MPS rate) [37]. See Contested for the EPA-vs-EPA+DHA split.

Vitamin D, magnesium, sulforaphane, correct deficiency, don't expect magic. Vitamin D gave no cancer, CVD, or fracture reduction in replete adults [38, 39] but reduced autoimmune disease (HR 0.78) [40]. Magnesium modestly lowers BP (~2 mmHg), concentrated in the deficient [41]. Sulforaphane up-regulates NRF2 carcinogen-detox enzymes, a biomarker, not an outcome [42]; eat the cruciferous vegetables, don't treat the detox marker as cancer prevention.

GLP-1 agonists, outcome-proven, not a free lunch. Semaglutide drives ~15% weight loss [43] and cuts MACE ~20% in non-diabetic obese people with CVD, an outcome, not a surrogate [44]; ~40% of lost weight is lean mass [45], mitigable with resistance training and protein.

Geroprotector drugs, animal-strong, human-absent. Rapamycin is the most reproducible mouse-lifespan hit, extending lifespan even when started late [46], but has no human longevity data; PEARL established weekly-dose safety [47] while being dose-compromised. Acarbose extends lifespan in both sexes without weight loss, ruling out calorie restriction as the mechanism [48]. 17α-estradiol [49] and canagliflozin [50] are male-only ITP hits, preclinical only, no human longevity data. Metformin is explicitly null in the ITP, blunts exercise adaptation [51], and its human-longevity spark (Bannister) was refuted as a confound [52]; it remains excellent for diabetes [53].

Creatine, the best-evidenced supplement here. Creatine + resistance training adds ~1.4 kg real lean mass at the meta level [54], with genuine fibre hypertrophy on biopsy beyond water weight [55]; the acute DEXA signal is partly water [56]. Cognition is only under stress (sleep deprivation) [57, 58], not for general healthy memory [59]. Safety is solid: the creatinine/eGFR rise is an artifact, not kidney injury [60], and creatine does not cause hair loss [61]. Its lower cancer incidence is association-only [62].

Where the Evidence Is Contested

The question of whether NMN improves insulin sensitivity in humans remains open. One small randomized controlled trial by Yoshino reported a 25 percent improvement [3], but that result was subsequently challenged by a published critique questioning the randomization in Science. Neither side has been settled by a larger, cleaner trial, so the single positive finding sits under a methodological shadow.

The broader question of NMN's clinical efficacy is similarly unresolved. Independent randomized controlled trials have come back mostly null [63, 2], while Sinclair's industry-funded MIB-626 trial returned positive results [64], a finding that critics have met with a baseline-imbalance critique. Underneath that specific dispute lies a more fundamental mechanistic disagreement about how NMN even enters cells. Imai has argued for a dedicated transporter, Slc12a8 [65], while Brenner has argued for a phosphate-barrier model; the observation that gut conversion likely occurs [8] largely moots both positions in practice. Severe conflicts of interest exist on both sides of that mechanistic debate, which makes independent adjudication harder.

Resveratrol's status as a sirtuin activator is one of the more openly contested questions in the longevity space. Sinclair has argued, based on SIRT1 data in high-fat mice, that resveratrol meaningfully activates sirtuins. Brenner has called that claim "science fiction." The Interventions Testing Program returned null results, and commercial conflicts of interest are heavy on both sides, leaving the disagreement unresolved and the mechanism disputed.

Several other areas carry live uncertainty. On niacin and NAD precursor safety, evidence linking 4PY to vascular harm, alongside null results from multiple trials [18, 16, 17], has raised concern, but Verhoeven has argued those findings are being over-read. On omega-3 fatty acids and cardiovascular outcomes, EPA-only trials including REDUCE-IT and JELIS [34] reported benefit, while the EPA-plus-DHA STRENGTH trial was null and dose-dependent atrial fibrillation harm has been reported [35]. On NAD boosting and Alzheimer's disease, mouse models have shown NAD reversal in preclinical work, but human RCTs using NR and NMN have come back null [66, 6]. On creatine and memory, a meta-analysis by Prokopidis found a positive effect, while Eckert and Pascher raised a double-counting concern in a published letter. On metformin, the UKPDS established clear benefit in diabetes, but the Bannister observational data are considered confounded, the Keys analysis was null, the ITP was null in mice, and the TAME trial is still pending. On rapamycin, the ITP mouse data are considered solid, but human data from PEARL remain thin and dose-compromised. On GLP-1 drugs, SELECT demonstrated cardiovascular benefit alongside large weight loss, while concern persists about lean-mass loss during treatment and weight regain after stopping. On vitamin D, the VITAL trial found no benefit in replete populations, while the possibility of real benefit from correcting frank deficiency keeps the question from closing entirely.

References

1. Igarashi et al. 2022, npj Aging — NMN in older men
2. Huang — NMN 300 mg RCT (NAD↑, no function)
3. Yoshino et al. 2021, Science — NMN & muscle insulin sensitivity · PubMed 33888596
4. Liao et al. 2021, J Int Soc Sports Nutr — NMN & aerobic capacity
5. Martens et al. 2018, Nat Commun — NR 1g ↑NAD+ · PubMed 29599478
6. Martens CR et al. 2026 (Alz & Dementia) — NR in aMCI pilot (n=42, 12 wk, DB-RCT); NAD+ doubled (target engagement confirmed); cognitive, CBF, BP… · PubMed 42478598
7. Lin AL et al. 2025 (GeroScience) — NR 1000mg + exercise vs placebo + exercise in sedentary adults ≥55; systolic BP null; trend toward nighttime BP… · PubMed 40770531
8. Christen W et al. 2026 (Nature Metabolism) — Three-way NMN/NR/nicotinamide RCT (n=65, 14 days): gut microbiome converts both NMN and NR to NA… · PubMed 41540253
9. Elhassan 2019 (Cell Reports) — NR 1g/day in aged men; IL-6/IL-5/IL-2/TNF-α significantly reduced · PubMed 31412242
10. Norheim 2024 (Nature Aging) — NR in COPD; sputum IL-8 −52.6% (primary endpoint met) · PubMed 39548320
11. Han et al. 2026 (JCI Insight) — NR 1g × 4wk in psoriasis (n=27); CRP↓, SLIT2↑, IL-17↓ vs placebo; SLIT2/ROBO1 mechanism; NIH-funded; no Brenner COI · PubMed 42048163
12. Shoji M et al. 2025 (Aging Cell) — NR 1000mg/day in Werner syndrome; DB crossover RCT, 26wk/arm; CAVI arterial stiffness improved, skin ulcers… · PubMed 40459998
13. Baur et al. 2006, Nature — resveratrol in high-cal-diet mice · PubMed 17086191
14. Pearson 2008 (Cell Metab) — "without extending life span" in lean mice; standard-diet null; EOD+RSV = CR confound; Sinclair co-author · PubMed 18599363
15. ITP resveratrol (Miller 2011 + Strong 2013) — null at all doses/ages; plasma RES detected; Sinclair co-author on 2011 paper · PubMed 20974732
16. HPS2-THRIVE 2014, NEJM — ER niacin, no CV benefit + harm · PubMed 25014686
17. AIM-HIGH 2011, NEJM (Boden) — niacin futility · PubMed 22085343
18. Ferrell et al. 2024, Nature Medicine — niacin 4PY & CVD · PubMed 38374343
19. D'Andrea et al. 2019 (JAMA Network Open) — 17 RCTs; null vs statin comparator; supplemental non-statin analysis: ACS↓ (p=0.02), stroke↓, revasc↓↓… · PubMed 30977858
20. Schandelmaier et al. 2017 (Cochrane) — 23 RCTs; null overall; low-risk-bias subgroup: statins > niacin (p=0.05, HPS2-THRIVE dominant) · PubMed 28616955
21. Physionic — Niacin (B3) reduces heart disease (studies 198–201; 4 meta-analyses)
22. Berberine LDL-C meta-analysis (placebo-controlled) · PubMed 36941490
23. Berberine → TMAO reduction >30% (human + hamster) · PubMed 35794102
24. Berberine vs. statin plaque regression (open-label, 4-month) · PubMed 35794102
25. 39076715 (note: pooled DBP −2.3 mmHg, not 4–5 mmHg as analyst synthesis cited) · PubMed 39076715
26. Physionic — 4 factors to clear plaque (3+1 simultaneous)
27. Lycopene + lutein 1-year RCT — CIMT improvement (positive) · PubMed 24047757
28. Thies 2012 — lycopene/tomato diet RCT (12wk, n=225); NOT a CIMT study; all CV risk factor endpoints null; disease-free population "too healthy" · PubMed 22492370
29. Potassium deficiency + arterial calcification (VSMC, mouse) · PubMed 28978809
30. Cochrane 2020 (Abdelhamid) — omega-3 CVD, ACM null · PubMed 30521670
31. Hu 2019 (JAHA) — 13-RCT omega-3 CVD meta
32. REDUCE-IT (Bhatt 2019, NEJM) — EPA 4g, MACE ↓25% · PubMed 30415628
33. STRENGTH (Nicholls 2020, JAMA) — EPA+DHA 4g null · PubMed 33190147
34. JELIS (Yokoyama 2007), Lancet — EPA 1.8g, coronary events · PubMed 17398308
35. Gencer 2021 (Circulation) — omega-3 & AFib meta
36. Chen G et al. 2024 (BMJ Med 3:e000451) — UK Biobank; n=415,737; 11.9 yr; fish oil → ↑AFib (HR 1.13), ↑stroke (HR 1.05); BUT AF→MI HR 0.85, AF→MACE… · PubMed 38800667
37. Smith et al. 2011, AJCN — omega-3 & muscle protein synthesis · PubMed 21159787
38. Manson et al. 2019 (VITAL), NEJM — vitamin D & cancer/CVD · PubMed 30415629
39. LeBoff et al. 2022 (VITAL), NEJM — vitamin D & fractures · PubMed 35939577
40. Hahn et al. 2022 (VITAL), BMJ — vitamin D & autoimmune disease · PubMed 35082139
41. Zhang et al. 2016, Hypertension — magnesium & blood pressure · PubMed 27402922
42. Egner et al. 2014, Cancer Prev Res — broccoli sprout & detox · PubMed 24913818
43. Wilding et al. 2021 (STEP 1), NEJM — semaglutide weight loss · PubMed 33567185
44. Lincoff et al. 2023 (SELECT), NEJM — semaglutide & CV events · PubMed 37952131
45. GLP-1 body composition — ~40% lean-mass loss (STEP/SURMOUNT DXA) · PubMed 39996356
46. Harrison et al. 2009, Nature — rapamycin extends mouse lifespan (ITP) · PubMed 19587680
47. PEARL trial 2024, Aging — low-dose rapamycin healthspan RCT
48. Harrison et al. 2014 (Aging Cell) — ITP Acarbose; both-sexes lifespan extension (~22% male); NO weight loss at longevity dose → not CR mechanism;… · PubMed 24245565
49. Harrison et al. 2021 (Aging Cell) — ITP 17α-estradiol; male-specific lifespan extension (+11–19%); female blood conc. substantially higher (exact… · PubMed 33788371
50. Miller et al. 2020 (JCI Insight) — ITP Canagliflozin; male-specific (+14%); female blood conc. ~20-fold higher (not 3–10× as cited by Miller);… · PubMed 32990681
51. Konopka et al. 2019, Aging Cell — metformin blunts exercise · PubMed 30548390
52. Bannister et al. 2014, Diabetes Obes Metab — metformin survival (refuted) · PubMed 25041462
53. UKPDS 34, 1998, Lancet — metformin in overweight T2D · PubMed 9742977
54. Chilibeck et al. 2017 — creatine+RT lean mass meta (older) · PubMed 29138605
55. Volek et al. 1999, Med Sci Sports Exerc — creatine + RT, muscle-fibre hypertrophy (biopsy) · PubMed 10449017
56. Desai et al. 2025, Nutrients 17(6):1081 — creatine washout LBM RCT · PubMed 40292479
57. McMorris et al. 2006 — creatine, sleep deprivation & cognition · PubMed 17046034
58. Gordji-Nejad 2024, Sci Rep — creatine & sleep-dep cognition · PubMed 38418540
59. 2024 systematic review — creatine cognition (skeptical) · PubMed 38582412
60. Vilar Neto et al. 2020 (Toxicol Res) — creatine 3g/5g × 35d RCT; creatinine↑ + eGFR↓ confirmed as false positives; KIM-1/MCP-1/urea/protein all normal · PubMed 32670557
61. Lak et al. 2025 (JISSN) — creatine hair loss RCT (n=45, 12wk DB-RCT); null across all hair metrics; DHT/free-T ratio shift same in placebo · PubMed 40265319
62. Jiang et al. 2025 (Front Nutr) — NHANES dietary creatine + cancer (n=25,879, 11yr); inverse association at highest quartile; heavy confounding… · PubMed 39867555
63. Uthever NMN multicentre RCT (middle-aged/older)
64. MIB-626 (β-NMN, Metrobiotech) RCT — overweight/obese
65. Grozio/Imai 2019 (Nat Metab) — Slc12a8 NMN transporter in mouse small intestine; CONTESTED in same journal (Schmidt & Brenner 2019) · PubMed 31131364
66. Nicotinamide riboside in MCI — RCT (null cognition)