Longevity & the Biology of Aging: What the Primary Evidence Supports

Aging has a shared vocabulary, not a proven master switch. The dominant framework describes aging as a set of "hallmarks", processes that appear with age, accelerate aging when aggravated, and slow it when ameliorated, expanded from nine to twelve categories across two landmark reviews [1, 2]. The framework is a taxonomy, not a causal proof: hitting a hallmark in a model organism does not establish that an intervention extends healthy human life.

Cellular senescence is one hallmark with genuine causal teeth, in mice. Senescent cells are stably arrested cells best identified by multiple markers in combination (p16/p21, SA-β-gal, lamin B1 loss) [3], and they secrete a pro-inflammatory cocktail, the SASP [4], which links senescence to sterile inflammation through the cGAS, STING, NF-κB axis [5]. Clearing p16⁺ cells delays age-related pathology in progeroid mice [6] and, critically, extends median lifespan ~24% in naturally-aged mice while improving function [7]. Pharmacologic senolytics reproduce this: dasatinib+quercetin extends remaining lifespan and improves physical function in old mice [8], and the FOXO4-DRI peptide does so by a distinct mechanism [9]. Human senolytic outcome data do not yet exist, see Blockers.

Inflammaging is a measurable, prognostic feature of human aging. Improved IL-6 independently predicts all-cause mortality in community-dwelling older adults [10], and a rise in inflammatory markers over time predicts mortality better than any single snapshot [11]. TNF-α signaling predicts mortality too, though honestly it is sex- and frailty-context-specific: significant in 80-year-old men but not women [12], dominant at extreme age [13], captured reliably by its soluble receptor [14], and null in the frailest [15]. Exercise is a solid well-supported lever on this inflammation, aerobic training lowers IL-6 and CRP in older adults versus controls [16].

The strongest modifiable longevity signal in humans is fitness, not any pill. Cardiorespiratory fitness tracks all-cause mortality more powerfully than any drug or supplement in the evidence base: in 750,302 veterans the least-fit had ~4× the mortality of the most-fit, consistently across age, race, and sex [17]. The dose-response is large at the inactive-to-active transition, meeting the ~150 min/week guideline cuts mortality ~31%, with no upper harm threshold observed [18]. Prolonged sitting raises mortality chiefly in the inactive; high activity volumes largely erase the penalty [19]. Grip strength, a functional-reserve marker, predicts death better than systolic blood pressure [20]. Caloric restriction has human RCT support, for risk factors, not lifespan: two years of CR in healthy non-obese adults improved cardiometabolic risk and slowed an epigenetic-aging clock, with no serious harms [21].

Longevity genetics converges on two robustly replicated loci. Alongside APOE, common FOXO3 variants (notably rs2802292) replicate for human longevity across populations, Japanese-American, German, Italian, Danish, and Han Chinese cohorts [22, 23, 24, 25, 26], reaching genome-wide significance with a pooled OR ~1.17 per allele [27]; the longevity allele is functional, creating an HSF1 binding motif that boosts the stress response [28]. The same insulin/IGF-1 axis shows a growth-versus-longevity trade-off: severe lifelong IGF-1 deficiency nearly abolishes cancer and diabetes without extending overall lifespan [29]; a GH-receptor exon-3 deletion associates with ~10 extra years, male-specific [30]; partial IGF-1-receptor resistance is enriched in exceptional longevity, concentrated in women [31]; and in the oldest-old, lower IGF-1 predicts longer survival [32].

Epigenetic reprogramming reverses aging markers, in mice, locally, with caveats. Cyclic, short-term expression of Yamanaka factors ameliorates aging hallmarks and extends lifespan in progeroid mice while improving regeneration in wild-type mice [33], and OSK reprogramming restores vision and resets the methylation clock in mouse retina [34]. This is a world away from systemic, safe, durable human age-reversal, which is not demonstrated. The CR-versus-lifespan question is itself contested at the primate level (see Contested), and the drug side of longevity, rapamycin, metformin, NAD boosters, stays preliminary or animal-only (see Contested and Blockers).

Where the Evidence Is Contested

Does caloric restriction extend primate lifespan? This remains one of the most debated questions in aging research, because two studies using the same species reached opposite conclusions. The Wisconsin cohort, led by Colman and colleagues, found a 67% reduction in age-related mortality among restricted animals [35]. The NIA study by Mattison and colleagues, conducted under comparably rigorous conditions, found no survival benefit at all [36]. The leading explanation for this divergence points to differences in control-diet quality. That explanation has not fully satisfied either camp, and the disagreement remains live.

Rapamycin presents a different kind of impasse. Its ability to extend mouse lifespan through the Interventions Testing Program is among the most reproducible findings in the field, and there is also evidence of immune rejuvenation in humans [37, 38]. The PEARL trial moved rapamycin into a human randomized controlled trial and found the drug safe. But dosing limitations compromised the trial, and no human lifespan data exist [39]. Kaeberlein, who has championed rapamycin research, acknowledges that the gap between mouse results and actionable human evidence has not been closed.

Metformin's reputation as an anti-aging intervention has eroded considerably, even as its value as a diabetes drug remains secure. Its efficacy in type 2 diabetes is well established through the UKPDS [40]. The observational case for extending healthy lifespan, however, has been challenged on confounding grounds by Keys and colleagues, who identified the Bannister confound as a major problem with earlier analyses [41]. The Interventions Testing Program found no lifespan effect in mice, and there is evidence that metformin blunts the benefits of exercise [42]. The TAME trial was designed to test the drug prospectively in healthy older adults. It has not yet reported results, leaving the central question unanswered.

The controversy over epigenetic clocks cuts closer to the foundations of aging science itself. These clocks are widely regarded as the best available biological age biomarkers, and they respond meaningfully to interventions: clock age moves with caloric restriction [21] and with cellular reprogramming [34]. Their inventors have conceded, however, that causation has not been established. Studies with replication samples of 8 to 9 years or fewer generate noise that undermines confidence in the measurements [43]. Kaeberlein has gone further, describing some clock-based longevity claims as "snake oil," a characterization that reflects genuine scientific frustration rather than mere rhetoric.

Does time-restricted eating work through a specific mechanism, or by reducing total calorie intake? This question has not been settled either. Proponents point to autophagy induction and circadian alignment as plausible pathways distinct from calorie restriction alone. But isocaloric randomized controlled trials have consistently failed to find an independent benefit from the eating window itself. Alternate-day fasting performed worse than continuous caloric restriction [44]. Adding a restricted window produced no additional benefit over twelve months [45]. A 16:8 schedule showed no edge over unrestricted eating when calories were matched [46]. The pattern across these trials suggests that whatever benefit time-restricted eating produces tracks the calorie deficit, not the timing.

References

1. López-Otín C et al. 2013 (Cell) — "The Hallmarks of Aging"; 9 hallmarks taxonomy; ~14,000 citations; the foundational framework · PubMed 23746838
2. López-Otín C et al. 2023 (Cell) — "Hallmarks of aging: An expanding universe"; adds disabled macroautophagy, inflammaging, dysbiosis; 9→12 hallmarks · PubMed 36599349
3. Gorgoulis 2019, Cell — cellular senescence consensus definition (review) · PubMed 31675495
4. Coppé 2008, PLoS Biol — SASP characterization · PubMed 19053174
5. Dou Z, Ghosh K, Ivanova IG et al. 2017 (Nature) — Cytoplasmic chromatin triggers inflammation in senescence and cancer; IMR90 fibroblasts + primary… · PubMed 28976970
6. Baker DJ, van Deursen JM et al. 2011 (Nature) — INK-ATTAC clearance of p16+ cells in BubR1 progeroid mice; delayed cataracts/sarcopenia/fat… · PubMed 22048312
7. Baker DJ, van Deursen JM et al. 2016 (Nature) — p16-3MR in naturally-aging mice; selective p16+ elimination → ~24% median lifespan extension from… · PubMed 26840489
8. Xu M, Kirkland JL et al. 2018 (Nature Medicine) — D+Q in naturally-aged mice (24 months); +36% median remaining lifespan; improved… · PubMed 29988130
9. Baar 2017, Cell — FOXO4-DRI targeted senolytic · PubMed 28340339
10. Harris TB, Ferrucci L, Tracy RP et al. 1999 (Am J Med) — Iowa 65+ Rural Health Study; n=1,293 healthy non-disabled older adults; mean 4.6y… · PubMed 10335721
11. Alley DE, Crimmins EM, Bandeen-Roche K, Guralnik JM, Ferrucci L et al. 2007 (JAGS) — InCHIANTI; n=736, two-wave; 79 deaths/2,079 person-years;… · PubMed 17727645
12. Bruunsgaard H et al. 2003 (Clin Exp Immunol) — Glostrup 1914 Birth Cohort; n=333, age 80, 6y follow-up; TNF-α HR 1.15/unit in men (p=0.03), NOT… · PubMed 12653832
13. Bruunsgaard H et al. 2003 (Am J Med) — Danish Centenarians; n=126, ~age 100, 5y follow-up; TNF-α HR 1.34/SD (95% CI 1.12–1.60, p=0.001) in both… · PubMed 12967692
14. Carlsson AC et al. 2014 (Atherosclerosis) — PIVUS (n=1,005, age 70) + ULSAM (n=775, age 77); sTNFR1 predicts ACM: PIVUS HR 1.37/SD (1.17–1.60),… · PubMed 25255422
15. Volpato S et al. 2011 (JAGS) — ilSIRENTE; n=362 frail ≥80y, 4y, 150 deaths; TNF-α HR 1.26 (0.74–2.15) — NOT significant; IL-6 HR 2.18, CRP HR 2.58… · PubMed 21883115
16. Kohut ML et al. 2006 (Brain Behav Immun) — 10-month parallel-group RCT; older adults ≥64y; aerobic exercise vs flexibility/strength control;… · PubMed 16504463
17. Kokkinos P, Faselis C, Samuel IBH et al. 2022 (JACC) — n=750,302 US veterans; 5 CRF groups; HR for ACM: least-fit vs most-fit = 4.09 (95% CI… · PubMed 35926933
18. Garcia L, Pearce M, Abbas A et al. 2023 (BJSM) — Dose-response meta-analysis; 196 articles; >30 million person-years; 150 min/wk moderate PA… · PubMed 36854652
19. Ekelund U, Steene-Johannessen J, Brown WJ et al. 2016 (Lancet) — Individual-level meta-analysis; n=1,005,791; sitting ≥8hr/day + low PA → 59%… · PubMed 27475271
20. Leong et al. 2015, Lancet (PURE) — grip strength & mortality · PubMed 25982160
21. Kraus et al. 2019, Lancet D&E (CALERIE) — 2yr caloric restriction · PubMed 31303390
22. Willcox BJ et al. 2008 (PNAS) — FOXO3A rs2802292 G allele; OR=2.75 for longevity in Japanese-American men; IIS pathway; replicated across multiple… · PubMed 18765803
23. Flachsbart F, Nebel A et al. 2009 (PNAS) — FOXO3A rs2802292 G allele; German centenarians/nonagenarians (n=1,762); OR ~1.26; first European… · PubMed 19196970
24. Anselmi CV, Puca AA et al. 2009 (Rejuvenation Res) — FOXO3A Italian replication; Southern Italian centenarians; rs2802288 OR=1.51 (1.15–1.98),… · PubMed 19415983
25. Soerensen M et al. 2010 (Aging Cell) — Danish 1905 birth cohort; n=1,089 oldest-old (143 centenarians) + 736 middle-aged controls; 15 FOXO3A… · PubMed 20849522
26. Li Y, Wang WJ, Cao H et al. 2009 (Human Mol Genet) — Han Chinese centenarians; discovery n=761 centenarians+1,056 controls; independent geographic… · PubMed 19793722
27. Broer 2015, J Gerontol A — CHARGE GWAS; only APOE+FOXO3 replicate · PubMed 25199915
28. Grossi 2018, NAR — rs2802292→HSF1 functional FOXO3 mechanism · PubMed 29733381
29. Guevara-Aguirre J, Longo VD et al. 2011 (Sci Transl Med) — 99 Laron syndrome subjects (Ecuador) vs 1,266 relatives; 17-year follow-up; 0 cancer… · PubMed 21325617
30. Ben-Avraham D, Barzilai N et al. 2017 (Science Advances) — GHRd3 longevity; 4 cohorts n=841; d3/d3 homozygosity increases linearly with age… · PubMed 28630896
31. Suh 2008, PNAS — IGF1R mutations in Ashkenazi centenarians · PubMed 18316725
32. Milman 2014, Aging Cell — low IGF-1 predicts oldest-old survival · PubMed 24618355
33. Ocampo 2016, Cell — in vivo cyclic partial reprogramming ameliorates aging hallmarks + extends lifespan (progeroid mice); WT regeneration · PubMed 27984723
34. Lu et al. 2020, Nature — OSK reprogramming restores vision (mouse) · PubMed 33268865
35. Colman RJ, Weindruch R et al. 2009 (Science) — NHP caloric restriction; 76 rhesus monkeys; 20% CR × 20 years; 67% reduction in age-related disease… · PubMed 19590001
36. Mattison JA, de Cabo R et al. 2012 (Nature) — NIA NHP CR; 121 rhesus monkeys; CR did NOT improve survival; age-related health markers improved;… · PubMed 22932268
37. Harrison et al. 2009, Nature — rapamycin extends mouse lifespan (ITP) · PubMed 19587680
38. Mannick et al. 2014, Sci Transl Med — mTOR inhibition & immune function · PubMed 25540326
39. PEARL trial 2024, Aging — low-dose rapamycin healthspan RCT
40. UKPDS 34, 1998, Lancet — metformin in overweight T2D · PubMed 9742977
41. Keys et al. 2022 (Int J Epidemiol) — Danish registry; ~500K subjects; metformin diabetics vs healthy non-diabetics; IRR ~1.52–1.90 higher mortality… · PubMed 36287641
42. Konopka et al. 2019, Aging Cell — metformin blunts exercise · PubMed 30548390
43. Higgins-Chen et al. 2022, Nat Aging — clock reliability (PC-clocks) · PubMed 36277076
44. Templeman et al. 2021, Sci Transl Med — isocaloric ADF vs CR · PubMed 34135111
45. Liu et al. 2022, NEJM — TRE + CR vs CR alone · PubMed 35443107
46. Lowe et al. 2020, JAMA Intern Med (TREAT) — 16:8 TRE · PubMed 32986097