#19 – Dave Feldman: stress testing the lipid energy model

KEYPOINTS

• 01

  Peter opens already mostly unconvinced by Dave's lipid-energy model.

  He estimates he entered ~20% open to being persuaded and left unconvinced. His null hypothesis remains the lipid/apoB causal framework, even while granting Dave's n-of-1 experiments are interesting and technically earnest. He repeatedly separates respect for Dave's curiosity from assent to Dave's clinical conclusions.

• 02

  Dave's personal 'catharsis' was a late-2015 lipid spike on low-carb.

  About seven and a half months into the pattern, total cholesterol hit roughly 329 mg/dL with LDL-C around 200–250. That shock launched the software engineer into systematic self-experimentation and the 'cholesterol code' public project. Friends on similar diets had milder moves; Dave's magnitude made denial impossible and launched the public 'drop roll call' community.

• 03

  Lean mass hyper-responders are the phenotype under debate.

  Lean people on ketogenic/low-carb diets who show very high LDL-C/LDL-P with high HDL and low triglycerides. Dave hypothesizes energy trafficking via lipoproteins explains the rise; Peter doubts the triglyceride-export math can account for the extra cholesterol mass and particle counts. Peter also argues that even garden-variety type 2 diabetes does not generate the LDL burden people casually attribute to 'energy delivery,' weakening the model's scaling claim.

• 04

  Inversion experiments: huge dietary fat can drop LDL short-term.

  Dave reports multi-day high-fat runs where LDL-C fell ~73% and LDL-P dropped on the order of ~1,000+ nmol/L in three days. Peter finds the phenomenology striking but insufficient to overturn chronic risk models built on apoB exposure over years. If fat influx acutely lowers measured LDL, Dave reads that as particles being used; Peter reads it as an acute redistribution that still fails the decades-long apoB risk literature.

• 05

  Mass balance is Peter's first objection.

  Where does the extra cholesterol in all those LDL particles come from if not synthesis or delayed clearance? Without a closed account of cholesterol flux, the energy model feels incomplete to him—even if Dave later refined answers after the taping. Until synthesis, absorption, and clearance are quantified together, a storytelling model remains a hypothesis.

• 06

  Genetic FH lessons cut against 'exclude my phenotype' special pleading.

  Peter notes ~2,000 vaguely identified genetic routes to familial hypercholesterolemia phenotypes. He is unwilling to throw out that causal body of evidence for a low-carb subgroup that has not yet shown clean outcome data. Excluding those mutations from the debate because they are 'genetic' while treating diet-induced highs as a different species of LDL is, to Peter, special pleading.

• 07

  PCSK9 and receptor biology still dominate his clinical priors.

  Even on max-dose atorvastatin (80 mg), some patients need more particle-lowering. PCSK9-related FH subsets (~3–5% in his framing) and receptor pathway genetics keep him anchored to apoB reduction as the lever with outcome trials. Fourier and ODYSSEY-type PCSK9 outcome data sit in the background of that prior even when cost-effectiveness is a separate argument.

• 08

  Dave wants atherosclerosis imaging and more stratified cohorts.

  He argues trouble should show up as rising CAC or carotid intima-media thickness before panic, and that studies stratifying high HDL/low TG alongside high vs low LDL-C are the fair test. Peter agrees better data would help while refusing to treat absence of powered studies as reassurance. Dave grants imaging lag and noise; Peter answers that waiting for plaque to prove a hypothesis is a harsh experiment to run on yourself.

• 09

  Composition experiments (SFA cuts, mono swaps) are practical middle ground.

  In one early case with Tom Dayspring, a keto loyalist with extreme LDL-P was urged to cut saturated fat toward ~25 g/day and lean on mono sources (olive oil, macadamia)—hard to do without creeping carbs. Lipid-P moved; ideology was the obstacle. Olive-oil-and-macadamia thought experiments show how hard pure mono eating is at weight-stable calories without sneaking carbohydrate back in.

• 10

  Remnants, oxidized LDL, and kinetics remain measurement gaps.

  Both admit clinical labs poorly resolve true remnant particles without apoC-III-aware methods, and oxidized LDL assays are imperfect. Those gaps cut both ways: they limit Dave's reassurance and Peter's mechanistic certainty in any single blood draw. Better assays would help both men; they would not, by themselves, rewrite Mendelian and trial evidence favoring lower apoB.