Peter Attia MD204 - Centenarians, metformin, and longevity | Nir Barzilai, M.D.
CHAPTERS
- 0:02 – 6:38
Genes, Twins, and the Heritability of Lifespan
Peter Attia opens with centenarians and asks how much parents' genes determine offspring lifespan. Nir Barzilai explains why father-son comparisons and twin studies are problematic, citing small-for-gestational-age effects and the Barker hypothesis.
- •Attia notes genes seem to matter more for living to 90–100 than to 70–80
- •Barzilai argues that even a 20% genetic contribution is worth understanding in order to address the other 80%
- •Twins are often born small for gestational age, which predisposes them to early age-related disease
- •Catch-up growth can leave one twin obese, and obesity drives aging
- •Epigenetic imprinting from low birth weight is so far a description rather than a mechanism
- 6:38 – 12:02
Testing Three Hypotheses: Environment, Perfect Genome, or Longevity Genes
Barzilai lays out three hypotheses for why centenarians live so long and describes sequencing the first 44 centenarians against ClinVar variants. The centenarians carried disease variants, which pointed toward protective longevity genes.
- •Centenarians' behaviors were not exemplary: many smoked, were overweight and did not exercise
- •Each of 44 centenarians carried five to six ClinVar disease variants
- •Two centenarians were APOE e4 homozygous yet not demented at 100
- •The perfect-genome hypothesis was rejected, leaving protective longevity genes
- 12:02 – 20:00
CETP, HDL, and the Lipid Longevity Genotypes
Barzilai describes how very high HDL in centenarians' offspring led to CETP and APOC3 genotypes enriched in centenarians. Attia explains why CETP inhibitors failed and the two discuss HDL function, cognition, and Klotho.
- •CETP and APOC3 homozygosity rose from about 8–9% in controls to nearly 20% in centenarians
- •A rising genotype frequency with age signals a likely longevity genotype
- •Drug companies pursued CETP inhibitors, but Attia says HDL biology is poorly understood
- •Barzilai says the main HDL phenotype in his cohort was cognitive function rather than cardiovascular
- •Klotho showed a U-shaped pattern, disappearing by 85 and returning at 100
- 20:00 – 35:37
Growth Hormone, IGF-1, and Longevity Genetics
Barzilai explains that about 60% of centenarians carry genes impairing growth hormone/IGF signaling, including FOXO3A, and describes a growth hormone receptor deletion linked to taller stature and lower IGF-1. He also presents UK Biobank findings on IGF-1 and antagonistic pleiotropy.
- •He wrote a grant expecting the growth hormone–IGF hypothesis to be wrong and was wrong
- •A growth hormone receptor exon 3 deletion was 3% in controls and 12% in centenarians, with taller height and lower IGF-1
- •UK Biobank showed high IGF-1 protective in the young but harmful after 60, except for cancer
- •This illustrates antagonistic pleiotropy: what helps when young can hurt when old
- •Attia questions why the switch occurs so late, and Barzilai discusses longevity genes lost through reproduction
- 35:37 – 44:40
Growth Hormone as a Longevity Tool and Sex Differences
Attia describes his narrow use of growth hormone for injury recovery and asks about its use at longevity clinics. Barzilai says timing of any geroprotector is unknown, and shares data suggesting IGF-1's negative effects are mainly seen in females.
- •Attia prescribes growth hormone only for orthopedic healing and does not consider it a pro-longevity tool
- •Barzilai says he does not know when to start geroprotectors and suggests senolytics not before 70–80
- •Confounders prevent seeing any signal in people taking growth hormone alongside other interventions
- •Among female centenarians, lower-half IGF-1 predicted twice the survival, with better cognition and similar muscle function
- •Males showed only a nonsignificant trend, and Barzilai is unsure about sex differences in sensitivity
- 44:40 – 56:31
APOE e2, Lp(a), and High TSH in Centenarians
The conversation turns to other longevity genotypes. Barzilai calls APOE e2 the most validated one, discusses how centenarians with high Lp(a) are protected by CETP, and describes the correlation between high TSH and longevity.
- •APOE e2 is the most validated longevity genotype, though its mechanism is unclear
- •Centenarians with high Lp(a) were all CETP homozygous, suggesting a protective counterbalance
- •Attia suggests survivorship concentration and protective genes are not mutually exclusive
- •Centenarians and their children show TSH of about 5–8 with normal thyroid hormones
- •Barzilai says an elderly person with incidental TSH of 5–8 need not be treated straight away
- 56:31 – 1:09:26
Pathway Enrichment, Polygenic Approaches, and the 10,000-Centenarian Project
Barzilai explains the shift from single-genotype analysis to pathway enrichment of rare variants, which points to insulin, mTOR, and MAP kinase signaling. Attia compares this to polygenic risk for Alzheimer's, and Barzilai describes the plan to recruit 10,000 centenarians.
- •Rare variants in centenarians are grouped by pathway; the Nature paper found insulin/IGF, mTOR and MAP kinase signaling
- •Barzilai says animal models are good for aging because pathways are evolutionarily conserved
- •Attia describes APOE4 risk estimates falling over time and moving toward a polygenic approach
- •Ashkenazi Jews are used because their genetic homogeneity reduces the sample size needed
- •A $2.8 million donation will fund recruiting 10,000 centenarians, offspring and controls, with exome sequencing
- 1:09:26 – 1:14:56
Nature vs. Nurture Thought Experiment
Attia poses a thought experiment about 10,000 people with ideal genes randomized to different lifestyles. Barzilai declines to predict, discusses maximal human lifespan, and notes that environment and medical care change genetic risk.
- •Barzilai puts the maximal human lifespan at about 115 years
- •He says reaching 150 would require other mechanisms to break that limit
- •Barzilai says the narrowing APOE4 risk likely reflects changes in environment and treatment
- •Attia asks how much environment matters for those without centenarian genes
- •Barzilai frames his own question as whether exercising and changing the diet of centenarians would help or harm them
- 1:14:56 – 1:26:14
Compression of Morbidity and the Meaning of Healthspan
Barzilai explains that centenarians develop diseases 20–30 years later and are sick only briefly at the end of life. Attia challenges the interpretation and criticizes current healthspan definitions, and they discuss costs and economic value.
- •Centenarians delay disease by 20–30 years, with 30% disease-free after 100 versus 10% of 80-year-old controls
- •Attia's grim thought experiment shows compressed morbidity alone doesn't equal better healthspan
- •Attia says healthspan definitions miss differences such as VO2 max and strength
- •The last two years of life cost about a third as much in centenarians, per CDC data Barzilai cites
- •A paper by Andrew Scott describes the economic value of added healthspan
- 1:26:14 – 1:39:00
Centenarian Outlook, Loss, and Lessons for the Rest of Us
Barzilai tells a story of a 102-year-old who became positive only late in life and discusses how centenarians cope with losing spouses and friends. Both then draw takeaways: Barzilai says the lesson is longevity genes that could become drugs, while Attia emphasizes delaying chronic disease through ultra-primary prevention.
- •Older people recall positive images better, and centenarians' perspective may shift with physiology
- •Most centenarians have lost spouses and often children, yet focus on current activities
- •Barzilai says centenarians are not a lesson to skip exercise and nutrition
- •Attia sees their superpower as delaying disease 20–25 years
- •A German pathology study found centenarians die of the same causes, only much later
- 1:39:00 – 1:52:44
Metformin Background and the Evidence for Geroprotection
Attia introduces metformin, rapamycin, and NAD precursors as the topics he is asked about most. Barzilai covers metformin's history, safety, and diabetes-related evidence, and Attia points out confounding in the observational and DPP data.
- •Metformin derives from French lilac and was used for flu and malaria before diabetes; phenformin was too potent
- •Barzilai cites a strong safety record, with side effects mostly in the first week and persistent diarrhea in 3–5%
- •Observational studies show lower mortality in diabetics on metformin compared with other treatments
- •Attia notes selection and dropout confounders and says no clean human proof of geroprotection exists
- •Barzilai agrees studies are imperfect and says TAME targets aging agnostic of any single disease
- 1:52:44 – 1:59:30
Design of the TAME Trial
Barzilai and Attia discuss TAME's population, composite endpoint, and statistical design. Attia raises concerns about starting at 65 and about which outcomes are measured.
- •Participants are 65–79 with no type 2 diabetes and must show some sign of aging, such as slower walking speed
- •The primary outcome is a cluster of mortality, major cardiovascular events, cancer, and MCI or dementia
- •The design avoids disease-specific significance that could force the FDA to stop the study early
- •Attia worries a late start could yield a negative study; Barzilai says aging biology continues after a first disease
- •Healthspan outcomes like frailty are planned only in a separate program
- 1:59:30 – 2:07:27
Metformin in Animal Models and the ITP
Attia asks why metformin failed alone in the Interventions Testing Program. Barzilai points to dosing problems, says effects across studies were 7–10%, and argues that once TAME is underway further animal work is unnecessary.
- •Barzilai says some drugs weak in mice may be stronger in humans, and vice versa
- •Metformin is a weak mitochondrial inhibitor, and the 1% dose used in animals is toxic
- •Attia calls the 2013 de Cabo paper essentially a null result at the 0.1% dose
- •Barzilai says two ITP centers showed about 10% lifespan increases while another showed −2%
- •Barzilai says that with TAME there is no need to study metformin further in animals
- 2:07:27 – 2:26:29
TAME Biomarkers: Proteomics, Methylation, and Biological Clocks
The discussion turns to what TAME can reveal about biomarkers of aging. Barzilai describes proteomic findings from his cohort, while Attia criticizes biological clocks and urges collecting clock data in TAME.
- •TAME will bank plasma, DNA, and cells for omics, with NIA grants for the biomarker portion
- •Barzilai wants biomarkers that change with treatment within two to three months
- •Proteomics from 65 to 95 showed IGF-related proteins and breakdown of collagen and extracellular matrix, with women's proteome more stable
- •Attia calls biological clocks easily gamed and cites the astronaut twin methylation result
- •Barzilai favors proteomics over methylation, and TAME will collect genome-wide epigenetic data
- 2:26:29 – 2:39:36
Metformin, Exercise, and Defining the Right Indication
Attia explains why he stopped taking metformin after seeing elevated lactate and discusses data showing blunted cardiorespiratory fitness. Barzilai discusses muscle transcript data and personalization, and says TAME aims for an FDA indication for aging.
- •TAME will not include brain MRI, exercise testing, or biopsies, to keep elderly participants engaged
- •Attia now reserves metformin for people with an obvious indication such as insulin resistance
- •Barzilai says lactate and GDF15 rise with metformin and can serve as compliance markers
- •Barzilai says the MASTERS biopsies showed lower mTOR with metformin but transcripts shifting toward a younger pattern
- •Barzilai says a young or biologically young person who exercises heavily may not need metformin, and TAME aims for an FDA indication for aging
- 2:39:36 – 2:48:00
Are NAD Precursors Geroprotective?
Attia asks Barzilai about NAD precursors such as NR and NMN. Barzilai says he cannot trace their biology, notes weak animal evidence, and describes his own inconclusive experience with NMN.
- •Animals do better in some studies, but not in the ITP
- •Barzilai says the precursor cannot be measured or traced in the body
- •One hypothesis involves the microbiome, but he says he does not understand the biology
- •His REM sleep seemed to improve on NMN but did not return with a pricier preparation
- •He is not convinced and currently takes no NAD precursor
