chatGPT:
Speakers (4)
- Siim Land (Host) — “Youth Span Society” promo; moderates questions (your transcript spells him like “Seam” in places).
- Matt Kaeberlein — CEO of Optispan; aging biology background.
- Dr Brad Stanfield — GP in Auckland, New Zealand; guideline/primary-care lens.
- Dr Michael Lustgarten — “Aging or Die Trying”; heavy self-tracking/biomarker optimization lens.
Tidy transcript (cleaned, speaker-attributed)
0:00–0:41 Opening framing
Michael: The idea that “healthy lifestyle means nothing ever goes wrong” isn’t real. Lifestyle is top tier (leanness/low visceral fat, physical + cognitive function), but decline still comes; then you ask “what’s under the hood?” Over-targeting biomarkers can cause unintended harm.
Siim (Host): Welcome, excited to get everyone together. Today’s panel: bloodwork in health/longevity.
0:42–3:40 Introductions + frameworks
Siim: Introduce yourselves + your framework. Matt first.
Matt: Matt Kaeberlein, CEO Optispan. ~25 years in aging biology. Scientist lens, now translating into real-world health impact.
Brad: GP in Auckland, NZ. Bloodwork should be actionable—measure what you might change and that changes outcomes. Uses training + clinical guidelines (e.g., UpToDate, local pathways).
Michael: Former research scientist (~15 yrs). Biomarkers + “how to move the needle.” Approach: repeated testing over long time + tracking diet/supplements, looking for correlations; noisy but signal emerges with enough data.
3:41–6:58 Misconceptions: what bloodwork can/can’t tell you
Siim: Biggest misconceptions—what can bloodwork predict about future health/longevity?
Brad: Focus on markers where lowering them lowers hard outcomes (e.g., BP; ApoB). Beware overhyped areas like “biological age clocks” when people try to change the score without strong evidence.
Siim: But clocks are popular—worth discussing.
Brad: Great research tools; weak evidence for general-pop behavior changes aimed at “moving the clock.”
7:00–10:12 Michael’s “top tier first,” then “under the hood”
Siim → Michael: You track everything—what’s your take on bloodwork for future risk?
Michael: Bloodwork is a piece, not first: diet/exercise/sleep/air/water first. ROI of optimizing biomarkers beyond fundamentals is unknown. Start with basics (CBC + CMP ± hsCRP). Track trends, not just reference ranges. Exotic tests can be noise. Biological age scores: treat as QC; care about rate of change.
10:13–13:01 HbA1c nuance + “overtreatment” harm example
Siim: Extreme examples: diabetes (A1c ≥6.5) clearly bad; but small shifts (e.g., 5.0 vs 5.4) less clear.
Michael: Even “small” A1c differences might matter over decades; focus on flattening rate-of-change, not “artificially” lowering.
Brad: In type 2 diabetics on insulin/sulfonylureas, pushing A1c too low increased hypoglycemia harms (falls, etc.), so clinical targets moved higher; GLP-1s change this by reducing insulin need. Over-targeting can cause harm.
13:02–16:12 Systems view + “measure–intervene–measure again”
Matt: Context matters; bloodwork is one tool in “systems health.” Combine with cognition, strength/mobility, DEXA, etc. Measure what’s actionable, integrate data, then iterate: measure → intervene → measure again. Also: detect existing disease/pre-disease early.
16:13–18:15 Frequency + tracking confounders
Michael: Add: how often? Even 8 tests/yr may miss “true average.” Track lifestyle variables; otherwise interventions can cancel each other out and you conclude “nothing happened.”
Siim: Top 3 biomarkers—forced constraint.
18:16–20:16 Brad’s “top 3” (and his twist)
Brad: My top 3 aren’t blood: weight, blood pressure, how the person feels.
If blood-only: ApoB (or lipid panel proxy), creatinine (cost), full blood count. (He’d like A1c too—hard to pick only 3.)
Matt → Brad: Creatinine vs cystatin C for eGFR?
Brad: Cystatin C is better, but more costly.
20:17–27:18 Matt’s list + debate: population hormone screening
Matt: For metabolic function: A1c is great, but without insulin you miss information—so include insulin. Also thinks hormones are under-considered (women definitely; also middle-aged men) for quality of life.
Brad (pushback): Testosterone testing should be for clinical suspicion (symptoms/signs). Population-wide screening lacks utility; guidelines reflect this.
Matt: Total testosterone alone isn’t that useful; and guidelines are population-level and can lag; individualized assessment matters.
Brad: Online guideline systems update fast (example: GLP-1s). Trials in hypogonadal men were underwhelming; not enough justification for generalized testing / TRT in normal range.
27:19–30:20 Michael reframes “andropause” + his top 3 CVD markers
Siim: Testosterone has more nuance than ApoB/insulin.
Michael: “Andropause” isn’t just testosterone; broad androgen pathway declines strongly with age in metabolomics. DHEA-S is on his radar.
Michael (top 3): Prevent what kills you: CVD. He’d do ApoB + Lipoprotein(a) + VLDL-related risk (notes MR data suggesting VLDL and Lp(a) can be more atherogenic per particle than LDL). If Lp(a) is very low, de-prioritize. Claims Lp(a) isn’t fully fixed—he says he has many tests showing it moved.
30:21–32:19 Biomarkers as “impetus” for imaging
Matt: Lipid biomarkers help both (1) intervene to reduce risk and (2) motivate follow-on diagnostics (plaque imaging).
Siim: “Misleading” biomarker?
Matt: Creatinine can look bad if you take creatine. Worse: epigenetic “biological age” tests that claim to measure age directly.
Brad: Also thinks “biological age” is overused and not decision-grade.
32:20–37:25 Biological age tests: what’s valid vs hype
Michael: Pushback: PhenoAge uses real clinical markers; track rate-of-change. Avoid “chasing” a single score.
Matt: Agrees PhenoAge / similar composite clinical panels are actionable because you can see which markers drive the score. Main complaint: none measure “biological age” directly; they estimate mortality risk correlated with age.
37:26–38:25 Host promo interlude
Siim: Promotes Youth Span Society community/courses.
38:26–44:10 Lipids deep-dive: ApoB, LDL, discordance
Siim: Lipids—start with ApoB: why better than LDL?
Brad: ApoB ≈ total atherogenic particle count. In NZ, ApoB often not funded; LDL/non-HDL used as cheaper proxies. Discordance happens, especially in metabolic syndrome/high A1c, but “relatively rare.”
Michael: More in the weeds: ApoB can look “good” but risk differs if much is driven by Lp(a) etc.
Matt: No biomarker is perfect; discordance fuels LDL denialism—still probabilistic.
44:11–52:24 HDL: “good cholesterol” and the J-curve + reverse causation
Siim: HDL—thoughts?
Brad: Raising HDL pharmacologically failed in trials; HDL alone isn’t very useful; mainly for calculations.
Michael: HDL tends to decline with age; there’s a J-shaped association (too low or very high associates with risk). He’d avoid age-related drift.
Brad: Even if HDL drops, doesn’t mean trying to raise HDL extends life; unclear ROI.
Michael: Looks at within-person drift; if HDL falls with diet/weight stable, something changed physiologically; try endogenous fixes, not pharma.
Michael: Reverse causation example: low LDL in elderly can look “bad” if models don’t control for illness (cancer/cachexia/liver disease) that lowers LDL.
52:25–58:20 How often to test + “precision health” vs practicality
Brad: Frequency depends. For himself (34, on lipid meds): every 6–12 months. Similar for healthy patients.
Matt: Population-level: annual makes sense, not optimal; personally likes more frequent (e.g., quarterly) as age/risk rises.
Michael: If you want the truth fastest, measure a lot; for suspected issues (kidney), he’d even do daily urine markers for 30–60 days. He frames it as building a personal “precision diet/exercise” recipe with lots of data.
58:21–1:07:34 Glucose markers: A1c, insulin, OGTT/CGM
Siim: Blood sugar markers—why insulin and A1c matter?
Matt: Metabolic dysfunction drives many chronic diseases; catch early; A1c/insulin are cheap and actionable.
Michael: Insulin follows an inverse-U pattern with age; low insulin can mislead; he values insulin less and likes 2-hour glucose patterns.
Siim: CGM spikes—how important?
Michael: Spikes aren’t inherently evil; problem is slow return to baseline. CGMs didn’t stick for him; he finger-pricks.
Matt: CGMs can be excellent education/behavior change tools; but DTC companies sometimes scared people (“every spike is bad”).
Brad: For diabetics/pre-diabetics, great. For non-diabetics, he’s seen harm: people avoid bananas/apples/oats; more anxiety; evidence of benefit is weak.
1:07:35–1:12:12 Food avoidance debate + keto “context” trap
Matt: Is it harmful if people swap spike foods for broccoli? Depends. In the US many are pre-diabetic.
Brad: Restrictive diets often fail; avoiding fruit/oats can backfire into worse habits.
Michael: Always evaluate the net effect across markers (kidney/liver/inflammation), not one number.
Michael: Keto crowd avoids spikes but can drive ApoB up; then some deny ApoB matters. Context again: optimize glucose and ApoB and other risks.
1:12:13–1:21:52 Liver markers + “don’t measure everything if it makes you anxious”
Michael: Liver: CMP markers (ALT/AST-ish, albumin, bilirubin, ALP) but he likes GGT (more linear with aging/mortality associations).
Matt: Optimization vs disease detection are different. If liver markers off despite healthy lifestyle—what do you do?
Michael: Uses diet tracking + correlations; starts with literature (e.g., milk thistle), then iterates. Notes false positives and correlation pitfalls but follows top correlations iteratively.
Matt: Know yourself—if measurement causes anxiety, back off; perfection not required.
Michael: Agrees; he’s “obsessed” but not anxious; sees it like elite fixation (Jordan/LeBron analogy).
1:21:53–1:26:34 Cancer early detection (Galleri), cost, closing
Matt: Thoughts on multi-cancer early detection tests like Galleri; mentions NHS trial didn’t hit primary endpoint.
Brad/Michael: Haven’t studied it deeply; cancer/neurodegeneration biomarkers are “future,” expensive; Michael watches indirect patterns (IL-6, TNF-α, hsCRP; albumin/lymphocytes/platelets patterns).
Matt: Cost: you don’t need $1,000 panels; there’s an 80/20 (or 90/10) core set a few times per year.
Michael: Even CBC + CMP can be ~$35 (US), plus BP cuff—don’t go crazy.
All: Where to find them; goodbye.
Summary (what they actually agree on)
- Lifestyle first, then use biomarkers to detect drift, risk, or disease.
- Actionability matters: measure things you can change in ways that improve outcomes (BP, ApoB, glycemia).
- Context > single numbers: trends, comorbidities, medications, diet pattern, and tradeoffs across organs matter.
- Beware Goodharting: chasing a score (especially “biological age”) can mislead if you don’t understand what drives it.
- Measurement cadence should match purpose: population pragmatism vs individual optimization.
Critique (what was strong, what was shaky, what was missing)
What was strong
- Clear “measure → intervene → measure again” loop (Matt), and the emphasis that monitoring is not the same as health.
- Clinical realism (Brad): overtreatment harms (hypoglycemia), adherence issues, and the difference between research tools and decision-grade tools.
- Systems tradeoff framing (Michael): optimizing one marker (e.g., TG/HDL on keto) can worsen others (ApoB, kidney markers).
Where the reasoning gets shaky
-
N=1 correlation optimism (Michael):
Repeated self-testing helps, but “diet variable → biomarker change” is still confounded by sleep, stress, infection, training load, seasonality, lab variance, regression to the mean, and multi-comparison effects. You can feel like you found signal when you found noise—especially with many biomarkers and many inputs. -
Lp(a) “you can move it” (Michael):
Lp(a) is largely genetically determined; some movement happens with acute inflammation, assay variability, or certain therapies, but lifestyle-driven large durable shifts are not the norm. His claim may be true for him (or reflect measurement/physiology noise), but it’s risky to generalize as “false that you can’t move it.” -
CGM in non-diabetics:
Matt is right it can be educational; Brad is right it can drive unnecessary restriction/anxiety. The missing piece is: what behavioral program and what decision rules make CGM net-beneficial rather than orthorexic fuel?
What was missing (would have improved the panel)
- Pre-analytic standardization (they only brushed it): fasting duration, time-of-day, exercise in prior 24–48h, alcohol, infection, hydration, menstrual cycle phase, supplements (biotin!), and medication timing can move results a lot.
- Biological vs analytical variation: which markers need repeat confirmation (TSH, ferritin, hsCRP) and which are stable enough (Lp(a), HbA1c).
- A concrete “core panel” list (80/20) with frequency guidance by age/risk tier.
- Risk framing beyond CVD: cancer screening strategy, smoking history, family history, and imaging (CAC/CTA) got mentioned but not structured.
If you want, I can also produce:
- a one-page “core bloodwork panel” distilled from their stated priorities (with “why,” “how often,” and “what to do if abnormal”), and
- a separate mini-critique just on the three contentious zones: hormone screening, CGMs for non-diabetics, and biological age tests.