#35 – Nir Barzilai, M.D.: How to tame aging

Barzilai's own account: he was prescribed metformin at pre-diabetes and started himself at 2,500 mg/day without titration; he experienced constant nausea for a month and lost significant weight. He now starts patients at 500 mg QHS and titrates slowly over weeks. The UK study used an average dose of just over 1,000 mg. The DPP (Diabetes Prevention Program) showed metformin prevents diabetes by 30%. UKPDS data show it prevents cardiovascular disease. Two Alzheimer's trials suggest improvement in name-recall domain at mild cognitive impairment stage.

Barzilai notes that the original caloric restriction studies in rats were actually closer to 23-hour daily fasting (food was given all at once and eaten immediately), which produced low IGF-1 — whereas the NIA-funded human CR study gave less food spread throughout the day and did not lower IGF-1. This distinction may explain why the animal CR results have not translated cleanly to human endpoints. Barzilai himself practices TRF: no eating from dinner until lunch the next day.

Barzilai's lab published a Nature paper (approximately one month before recording) showing IGF-1 receptor antibody administration to 22-month-old female mice (human equivalent ~70–75 years) produced 10% lifespan extension, cardiovascular protection, cognitive advantage, and functional improvement. The antibody was originally developed by Amgen against cancer targets. For males, the same treatment paradoxically increased inflammatory markers and showed no lifespan benefit. The sex difference is under active investigation — likely involves sex hormone interactions with the inflammatory pathway.

The Zucker diabetic rat experiment provided the mechanistic proof: as long as visceral fat was absent, the animals did not develop diabetes — even with the leptin receptor knockout that makes them perpetually hungry. Once visceral fat re-grew (it did so in 80% of animals), diabetes developed. This cleanly demonstrated that visceral fat presence, not caloric intake alone, is the gating variable for developing the metabolic syndrome.

The biopsy study was required because TAME's NIH reviewers challenged whether the animal biology would translate to humans. The human transcriptomic data directly addressed this concern by showing the same upstream regulatory pathway signatures (AMPK, mTOR, aging hallmarks) in human muscle and fat tissue. The finding that both tissues showed non-metabolic aging gene changes (BRCA1, myofibril genes) supports Barzilai's thesis that metformin works by fixing cellular aging directly, with metabolic improvements being downstream consequences.

Barzilai and colleagues — including Steve Austad as co-scientific director at the American Federation for Aging Research (AFAR) — spent years at NIH and FDA making the case that aging can be targeted. The FDA told them: design a study that shows you can delay a cluster of age-related diseases simultaneously, and we will listen. The composite endpoint was the solution: because once you survive one age-related disease, your risk of getting the others is about 10% each — disease-agnostic, it doesn't matter which comes first. TAME uses 1,500 mg/day of metformin (Merck GmbH providing matched placebo). The 70-million-dollar budget is split between NIA and private philanthropy through AFAR; NCI and NHLBI have indicated interest in co-funding. Enrollment target: early 2019.

The sulfonylurea arm of the same study showed higher mortality than non-diabetic controls — consistent with the hypothesis that it is metformin's anti-aging properties, not just glucose lowering, that drove the mortality benefit. The dose in the metformin group averaged over 1,000 mg. Barzilai notes the study was not stratified by dose, which he regards as a missed opportunity. Duration of metformin use could not be well characterized either, since only newly-diagnosed patients were enrolled.

The IGF receptor antibody (originally developed by Amgen for cancer) was tested in 22-month-old female mice (equivalent to ~70–75 years in humans). It extended lifespan by 10%, dramatically improved healthspan (cardiovascular, cognitive, functional markers), and reduced inflammatory markers. The antibody does not cross the blood-brain barrier, so peripheral IGF activity was reduced while central IGF-1 levels may have risen — consistent with the hypothesis that you want low peripheral IGF but preserved central IGF for neuroprotection. The same study did not show lifespan benefit in males, possibly because males had paradoxically increased inflammatory markers with the treatment.

Barzilai's team showed this through gene-to-gene interaction analysis using computational biology methods: centenarians with the 'bad' LPA genotype who survived to 100 were overwhelmingly also CTEP VV homozygous. CTEP VV phenotype: higher HDL, larger lipoprotein particle size, lower CTEP levels, cognitive protection. The CTEP VV genotype also increases after age 80 in a monotonic fashion — the biostatistical signature that indicates the genotype is conferring a survival advantage rather than being a marker of accelerated aging.

The offspring approach corrects the survivor bias problem: if you measure a biomarker in a 100-year-old, you don't know if you're measuring what helped them survive or what's contributing to their imminent death. The offspring at age 65 have the relevant phenotype without that ambiguity. Forty-four centenarians who underwent whole-genome sequencing were found to carry a total of more than 230 known pathogenic mutations from the ClinVar database — mutations that 'should' have caused diseases including Parkinson's, cancer, and APOE4-linked dementia. None had those diseases at 100 years old, suggesting the longevity gene network actively suppresses penetrance of disease-causing variants.

The resolution: insulin resistance is the muscle's way of pushing glucose elsewhere when glycogen is full — a physiological overflow valve. When this protective response becomes chronic because of constant visceral-fat-driven inflammation, it becomes pathological. The evolutionary conserved nature of this pathway explains why both CR (reducing visceral fat) and GH/IGF-1 pathway mutations (reducing IGF-1 signaling) extend lifespan: they both reduce signaling through this stress-response pathway.

Barzilai's practical position on NAD precursors: clinical studies are needed before confident recommendations. His concern about IV NAD is that it probably does not cross cell membranes efficiently either. His preferred future approach: sublingual or intravenous delivery of NR or NMN to bypass first-pass hepatic metabolism. The NAD-HDAC connection is relevant to metformin's mechanism: metformin inhibits complex I, which reduces NAD:NADH ratio, which in turn affects sirtuin activity — a potential overlap between the metformin and NAD precursor pathways.

Barzilai's 15-person crossover biopsy study (75-year-olds, 6 weeks metformin vs placebo, then crossed over) found changes in BRCA1, myofibril genes, and DNA-repair genes — not just metabolic genes — in both muscle and adipose tissue. When the transcriptomic data were analyzed via upstream regulator tools, all pathways converged on AMPK, mTOR, and aging-specific targets, confirming that the downstream tissue effects are secondary to fixing cellular aging, not the primary mechanism.