Peter Attia MD281 ‒ Longevity drugs, aging biomarkers, and updated findings from the Interventions Testing Program
CHAPTERS
- 0:00 – 8:51
How the Interventions Testing Program works and what it costs
Richard Miller explains that the ITP began about 20 years ago under the National Institute on Aging and runs at three sites. It screens nominated drugs for lifespan effects. He and Peter Attia also cover the number of nominations, the budget, and how philanthropy could help.
- •Three labs (Michigan, Texas, Jackson Labs) test five to seven drugs a year
- •The program has four published significant hits plus smaller and pending ones
- •28 nominations this year; slots are also used for retesting doses and later-start dosing
- •About $1 million direct costs per site, roughly $4.5 million per year in total
- •Philanthropists could set up independent arrangements with the universities
- 8:51 – 17:28
Why genetically heterogeneous UM-HET3 mice beat inbred Black 6 mice
Miller explains why the ITP uses genetically heterogeneous UM-HET3 mice instead of the inbred Black 6 strain used in most research. Attia reinforces the point with a human analogy about inbreeding. They discuss why the field still defaults to Black 6.
- •97% of aged-mouse requests are for the single-genotype Black 6 strain
- •Inbred mice are homozygous and often have disease, deafness or blindness
- •UM-HET3 mice share half their genes with any other mouse and give reproducible heterogeneity
- •Gene mapping with these mice has already identified lifespan-linked genes
- •Researchers default to Black 6 because of mentors and limited funds, like lemmings
- 17:28 – 28:59
Lifespan metrics, statistical design, and site differences
Miller describes the primary outcomes: proportional hazard and median lifespan, plus a pre-specified test at the 90th percentile for maximum lifespan. He explains the power analysis and mouse numbers. He also describes persistent site differences in control males and how they are handled statistically.
- •Median lifespan is the age at which half the mice have died
- •Maximum lifespan is tested at the 90th percentile of the pooled group, fixed in advance to avoid false positives
- •Design gives 80% power to detect about a 12% effect even if one site fails
- •Controls are doubled and reused across drug comparisons
- •Michigan males live 5–10% longer and Michigan mice are about 10% lighter; analysis is site-stratified
- 28:59 – 36:57
Formulating drugs in food and verifying dosing
Miller explains how drugs are put in food, measured, and pilot-tested before a lifespan study. He covers rapamycin encapsulation and the sex differences in blood levels of rapamycin and canagliflozin. He also explains why dosing is limited to food and cannot be tracked per mouse.
- •Food batches are assayed for drug content and pilot-fed for eight weeks with tissue levels checked
- •Liver mRNA is checked to confirm the drug did something
- •Rapamycin had to be encapsulated because about 90% was degraded in the stomach
- •Blood rapamycin and canagliflozin levels are about threefold higher in females
- •Individual food intake can't be reliably measured, and injected drugs are impractical
- 36:57 – 39:55
Rapamycin's success and late-life starts
Miller recounts how rapamycin became the first ITP hit in 2009, and the size of its lifespan effect. He explains why it was nominated and the decision to give it to 20-month-old mice. He also notes that several other drugs work when started in middle age.
- •Top drugs give 15–20% median lifespan increases, more than abolishing cancer or heart attacks would
- •Rapamycin was nominated because TOR inhibition lengthened life in invertebrates
- •Starting at 20 months worked as well as starting young in both sexes
- •17α-estradiol, acarbose and canagliflozin also work when started in middle age
- •Rapamycin can boost some immune responses such as vaccine response
- 39:55 – 48:50
Sex differences in drug metabolism and dosing in people
Attia asks why rapamycin and canagliflozin are more concentrated in female mice. Miller says xenobiotic metabolizing enzymes differ by sex, though acarbose is not absorbed. Attia notes that human drug dosing rarely accounts for sex differences.
- •The reasons for higher female levels have not been studied drug by drug
- •Xenobiotic metabolizing enzymes differ between males and females in mice and people
- •Acarbose stays in the gut, so excretion is not the explanation for its sex difference
- •Attia says human dosing is based on weight or nothing at all, not sex
- •Miller says a pharmacologist could answer this
- 48:50 – 59:18
What late-start rapamycin tells us about aging, and healthspan measures
Miller and Attia discuss what late-life effectiveness implies about aging biology. Miller describes the stage 1 and stage 2 study structure and the secondary healthspan tests. He also covers tissue sharing with outside labs and plans for cognition testing.
- •Late-start benefit implies some drug-sensitive aging processes continue even at 20 months
- •Grip strength and rotarod improve in mice started on 17α-estradiol or acarbose in middle age
- •Stage 2 studies add tests like hearing, vision, glucose control and pathology
- •Frozen 22-month tissues are shared with outside labs, though NIA is changing the program
- •Cognition is the next frontier, with a new neurobiologist at Michigan
- 59:18 – 1:04:40
Aging rate indicators versus biomarkers of aging
Miller distinguishes biomarkers of aging, which are like an odometer, from aging rate indicators, which are like a speedometer. He describes how they were found by looking for changes shared across slow-aging mice. He gives examples including UCP1, fat macrophages, BDNF and doublecortin.
- •Both Miller and Attia are skeptical that good assay-based biomarkers of aging exist
- •Aging rate indicators change in the same direction in every slow-aging mouse, even when young
- •Slow-aging models include four mutants, calorie restriction and four drugs
- •UCP1 rises in three fat depots, and only in males for the two male-specific drugs
- •Inflammatory fat macrophages fall while BDNF and doublecortin rise in the brain
- 1:04:40 – 1:12:33
Measurement noise and humane euthanasia
Attia asks how noisy UCP1 measurements might be, for example from exercise or time of day. Miller says noise is built into the observed differences and explains the standardized sampling. He explains why the lab uses rapid carbon dioxide euthanasia.
- •UCP1 is known to change with chronic exercise; whether cramming exercise would fool the test is unknown to Miller
- •Mice are euthanized between 9 and 10 AM, though ad lib feeding adds some variation
- •Rapid CO2 exposure causes unconsciousness within five seconds
- •Gradual CO2 over about seven minutes would raise adrenaline and glucose and lower blood pH
- •The animal care committee agreed to keep the older method
- 1:12:33 – 1:25:58
GPLD1, translation control, and why proteomics matters
Miller describes GPLD1 as another indicator, linked to exercise and cognition, and ties it to cap-independent translation. He then explains that RNA and protein changes with age correlate only about 30%. This means transcript-only data can miss most of what controls protein levels.
- •GPLD1 is elevated in all slow-aging mice and is controlled by cap-independent translation
- •A Jackson Labs study found only 30% concordance between age-related RNA and protein changes
- •Miller says transcript-only data is incomplete and mostly wrong
- •Protein levels are shaped by translation, RNA stability and location, and protein degradation
- •Attia sums up the possibilities as at least five non-exclusive routes between transcription and protein
- 1:25:58 – 1:37:24
Using aging rate indicators: speed of switching, persistence, and bridging to humans
Miller outlines open questions about aging rate indicators, including whether new drugs shift them and how quickly. He describes the Ames dwarf growth hormone experiment showing lasting reversion. He also covers plans to find plasma markers and test human samples.
- •Rapid switching could allow screening of many drugs before lifespan studies
- •Brief growth hormone in infant Ames dwarf mice permanently reverted the indicators, but starting at four weeks did not work
- •Human muscle and fat biopsies from the SOMA study will be tested
- •Irisin and GPLD1 rise in plasma of slow-aging mice
- •Planned mouse work will look for plasma metabolites that predict tissue-level indicators
- 1:37:24 – 1:44:32
Irisin, GPLD1 as drugs, and what counts as an anti-aging drug
Attia raises the irisin hype and whether such molecules could be pills. Miller explains the irisin assay problems and that irisin and its precursor FNDC5 move together. He says lifespan extension is necessary but not sufficient evidence of slowed aging, which needs broad functional benefits.
- •Early irisin papers overestimated levels by about 100-fold due to a poor assay
- •Both irisin and GPLD1 are peptides, so would likely need injection unless a small molecule can induce them
- •Miller says anti-aging drugs like rapamycin and acarbose already raise FNDC5
- •A drug extending lifespan is a critical step but must also preserve many youthful functions
- •The case is strong for several mutants and diets and building for rapamycin and others
- 1:44:32 – 1:49:55
What mice die of and healthspan without lifespan
Attia asks about causes of death in the mice. Miller says about 80% die of neoplasia, though small necropsy numbers limit conclusions. They also discuss whether drugs improve healthspan without lifespan, which the ITP has not looked for.
- •Jackson Lab strains were historically selected for cancer
- •About 80% of UM-HET3 mice die of cancer, with different types by sex
- •Necropsy groups of about 60 are too small to detect shifts in cause of death
- •Only one statistically significant change was seen, and it may be a fluke
- •Healthspan-without-lifespan drugs haven't been screened for; system-specific benefits are easy but not the goal
- 1:49:55 – 1:58:34
17α-estradiol and the surprising sex-specific results
Miller explains how 17α-estradiol differs from 17-beta-estradiol and why it was nominated. It lengthens male lifespan beyond that of females with no benefit in females. He also reports that rapamycin plus acarbose gave a 29% male increase and that 16-hydroxyestriol helps males but shortens female life.
- •17α-estradiol is a stereoisomer that binds classical estrogen receptors poorly
- •It pushes male lifespan beyond that of females and does not affect female longevity
- •Rapamycin plus acarbose gave a 29% male increase, the largest so far
- •Males given 17α-estradiol made estriol compounds, and hydroxyestriol worked in males
- •Hydroxyestriol is the first drug found to reduce lifespan in females
- 1:58:34 – 2:09:13
Failures: metformin, resveratrol and nicotinamide riboside
Attia asks about drugs widely hyped that failed in the ITP. Miller discusses metformin, resveratrol and nicotinamide riboside, giving a different degree of remaining optimism for each. He also says aging rate indicators might raise the hit rate by pre-screening.
- •Metformin did not extend mouse lifespan in the ITP, though Miller thinks a human case is plausible
- •Miller considers resveratrol largely hype, citing the red wine dose and the million-fold dose disagreement among experts
- •Nicotinamide riboside failed alone; NMN or combinations may still be tried
- •Failure in mice does not mean failure in people
- •Current hit rate is about 10% big effects and 15% significant, which pre-screening might raise
- 2:09:13 – 2:16:01
Over-the-counter winners: meclizine and astaxanthin
Attia opens with his own meclizine seasickness story, and Miller reveals meclizine and astaxanthin are in the latest ITP paper. Both extended male lifespan by about 10% and are available without a prescription. Miller discusses possible TOR involvement and the regulatory implications.
- •Meclizine was suggested by Gino Cortopassi after a screen for TOR inhibitors
- •Meclizine and astaxanthin each raised male lifespan about 10% and did not affect females
- •Neither changed the maximum lifespan test, and higher doses will be tried
- •Astaxanthin is what turns farmed salmon pink
- •These are the first over-the-counter winners, which raises questions about FDA positions on anti-aging claims
- 2:16:01 – 2:26:35
Fisetin fails, and Miller's view of senescent cells
Miller reports that fisetin did not extend lifespan or reduce senescence markers in any tissue tested. Attia presses him on his skepticism of senescent cells. Miller argues the term is vague and tells the story of how rare beta-gal positive cells were in the classic skin study.
- •Fisetin was tested at two dose regimes suggested by Jim Kirkland, with no effect on lifespan
- •Blinded tissue tests showed no change in p16, p21 or beta-galactosidase in brain, liver, muscle or kidney
- •Miller says 'senescent cell' is defined differently across labs
- •The skin study's highest score corresponded to about one positive cell in 10,000
- •Quercetin plus dasatinib was proposed but not accepted, and Miller can't discuss it
- 2:26:35 – 2:31:20
Looking ahead: cognition, collaboration and funding
Attia and Miller list what remains to be followed up, including repeat studies, stage 2 results for new drugs, aging rate indicator validation and cognition testing. Miller says the ITP aims to stimulate work in other labs. Attia closes by suggesting philanthropists could help fund the program.
- •Follow-ups include 17α-estradiol plus rapamycin, meclizine and astaxanthin stage 2 studies
- •Miller wants to know whether the drugs slow cognitive decline
- •The ITP hopes each result triggers work in many labs using shared tissues
- •Miller says the last podcast led to useful collaboration requests
- •Attia argues philanthropy could double throughput of a roughly $4.5 million per year program
