Summary by TwinMind · 2h 21m 19s

Summary by TwinMind · 2h 21m 19s

Journal Club with Dr. Peter Attia | Metformin for Longevity & The Power of Belief Effects

Journal Club with Dr. Peter Attia | Metformin for Longevity & The Power of Belief Effects

Andrew Huberman

Andrew Huberman

Introduction & Format

  • First collaborative journal club between Andrew Huberman and Dr. Peter Attia
  • Two papers discussed: metformin longevity study and nicotine belief effects
  • Format mirrors graduate/medical school journal clubs for paper critique

Metformin Study Analysis (Peter Attia)

Background Context

  • 2014 Banister study showed 15% reduction in all-cause mortality for diabetics on metformin vs non-diabetic controls
  • Study used "informative censoring" - excluded patients who progressed beyond metformin monotherapy
  • This created bias by only counting "successful" metformin patients

Current Study (Keys et al.)

  • Danish health registry: ~500,000 subjects analyzed
  • Two groups: matched singletons + discordant twins (genetic/environmental controls)
  • Tested with/without informative censoring

Key Findings

  • Crude mortality rates (per 1000 person-years):
    • Non-diabetic controls: 16.86
    • Diabetics on metformin: 24.93
  • Hazard ratios (unadjusted):
    • Singletons: 1.48 (48% higher death risk)
    • Twins: 2.15 (115% higher death risk)
  • With censoring: Only modest improvement (1.48 to 1.39)

Limitations & Confounders

  • Medication differences impossible to control (3x more lipid-lowering drugs, 2x more antihypertensives in diabetic group)
  • Epidemiological studies cannot eliminate all confounders
  • Missing control: diabetics not on metformin

Clinical Implications

  • No evidence metformin provides longevity benefit to overcome diabetes disadvantage
  • Animal studies (ITP) failed to show metformin lifespan extension
  • Personal experience: Attia stopped metformin due to elevated resting lactate (1.6 mmol vs normal <1.0)

Nicotine Belief Effects Study (Andrew Huberman)

Background on Belief vs Placebo Effects

  • Placebo effects: Binary on/off responses
  • Belief effects: Dose-dependent responses based on information/expectations
  • Examples: stress performance, milkshake satiety, hotel worker weight loss

Nicotine Neurobiology

  • Enhances signal-to-noise ratio in thalamus via nicotinic acetylcholine receptors
  • Key circuits: pedunculopontine nucleus → thalamus → ventromedial prefrontal cortex
  • Improves cognitive performance through enhanced attention/focus

Experimental Design

  • Experienced smokers given identical low-dose nicotine vape
  • Three groups told they received: low, medium, or high doses
  • fMRI during financial decision-making task
  • Measured: blood nicotine, subjective effects, brain activation

Results

  • Subjective effects: Scaled with told dose (not actual dose)
  • Thalamus activation: Weak correlation with belief (r=0.27)
  • Thalamus→prefrontal cortex connectivity: Strong dose-dependent activation matching beliefs
  • Reward pathways: No difference between belief groups
  • Performance: Scaled with believed dose

Significance

  • First demonstration of dose-dependent belief effects on brain physiology
  • Brain circuits respond to expectations, not just pharmacology
  • Implications for all drug treatments, including ADHD medications, antidepressants

Methodological Discussion

Paper Reading Strategy

  • Four key questions: What? How? Results? Conclusions justified?
  • Multiple readings required for complex papers
  • Focus on figures and statistical significance
  • Check supplemental materials

fMRI Limitations

  • Sub-centimeter resolution but not millimeter-level
  • Temporal resolution: hundreds of milliseconds
  • Movement artifacts (smokers coughing during scans)
  • Requires subtraction of resting state from activation

Statistical Considerations

  • Power calculations essential for interpretation
  • Large sample sizes in epidemiology compensate for inability to randomize
  • Confidence intervals more informative than point estimates
  • Multiple comparisons require correction