Working Paper · Nutritional Biochemistry
What metabolic health actually is, what breaks it, and the order in which to fix it — the hub paper the mechanism deep-dives link back to.
The practical end of this whole series, in plain language. Nothing here needs a lab, a prescription, or much money. Each item links to the paper that makes the full case and carries the same evidence grade used everywhere else on this site, so you can see which ones are firm and which are reasoned.
#start-here on the end.
The short list above is the stable, shareable part of this page. Everything below it is a working structural draft, published as-is rather than held back. Sections marked TODO are open decisions, not oversights — several are flagged explicitly in Limitations and Open Structural Questions below. It does not yet carry a numbered citation list; claims that need one are noted inline.
This paper is the hub. It states what metabolic health is, what breaks it, and the order in which to fix it. Where a mechanism deserves its own treatment, this paper states the conclusion and links out:
→ ahr.html
planned
Evidence grades follow the same six-label set used in the linoleic acid paper: Strong / Moderate / Animal–in-vitro / Inferred / Contested / Speculative. Grades attach to specific claims, not sections. Caveats live in Limitations at the end, not scattered through the argument.
Body weight is a lagging, low-resolution readout. The thing underneath it is the capacity to move fuel where it belongs, when it belongs there. Four properties define that capacity.
How much insulin it takes to clear a given glucose load. The target isn't low insulin or low glucose in isolation — it's a small insulin release producing fast clearance. Those two moving together is the signature.
The ability to switch cleanly between fat and glucose oxidation as availability changes. Rigidity in either direction is dysfunction. A person locked in fat oxidation who can't handle carbohydrate is not metabolically healthy — they've adapted around a broken system rather than fixing it.
Enough mitochondrial density, and enough NADPH-regenerating capacity downstream, to handle substrate flux without leaking oxidants. Capacity is the variable that determines whether a given fuel load is handled or becomes a stressor.
Every one of the above varies across the day. Insulin sensitivity and β-cell responsiveness are highest in the morning and decline into the evening — an identical meal produces a larger glucose excursion at dinner than at breakfast. A protocol that ignores timing is leaving a real lever unused. Moderate
The single largest compositional change to the food supply in the last century. High membrane and adipose LA content creates a substrate pool primed for peroxidation, and pushes flux down the desaturase chain toward inflammatory metabolite output. Both routes run together; only the weighting shifts by tissue.
The dose is the argument. LA is essential, but the requirement is small — roughly 1–2% of calories — and its benefits saturate there. Seed oils pushed typical intake to ~8–10%. And most of that doesn’t come from the bottle at home: it comes from restaurant, fried and packaged food, where soybean oil is the default. Frying is a separate, additive exposure — oil held hot for days delivers the oxidation products pre-formed, rather than waiting for them to form in the body. Strong for the intake figures and for heated oil accumulating aldehydes.
Full argument: linoleic-acid.html. This paper states the conclusion and moves on — don't re-litigate here.
Environmental and dietary ligands hold the aryl hydrocarbon receptor in a chronically activated state, dysregulating adipogenesis, hepatic lipid handling, and inflammatory tone. Several of the same food-supply changes that raised LA intake also raised AhR ligand exposure.
Full argument: ahr.html.
Not carbohydrate as a class — carbohydrate without a break, and without the mitochondrial capacity to oxidize it. The distinction matters and Part III is built on it.
The relevant failure is that a large glucose load in someone with reduced insulin sensitivity is disposed of poorly: less enters muscle for oxidation, more remains in circulation or partitions toward de novo lipogenesis and storage. The same load in a sensitive system is largely oxidized, and — per the PDHC–NNT literature — the fraction that is oxidized appears to drive proton-motive-force-consuming redox cycling that dissipates energy as heat rather than storing it.
Animal–in vitrofor the NNT circuit. Moderatefor the disposal difference itself.
Worth stating plainly because it cuts against the low-carb default: human overfeeding studies do not show carbohydrate converting efficiently to body fat. Whole-body de novo lipogenesis stays low even under substantial carbohydrate overfeeding, including in overweight subjects, because glycogen storage absorbs the excess first and the remainder is preferentially oxidized. Carbohydrate is metabolically self-limiting in a way fat is not.
Moderate— overfeeding studies are short and don't settle habitual free-living intake.
The implication for the protocol: carbohydrate is not the enemy. Carbohydrate in a system that can't dispose of it is the enemy, and that's a fixable condition.
Light at night, eating late, irregular sleep. Degrades the same insulin sensitivity rhythm described in Part I, and the effect is independent of diet quality. This is where an otherwise clean diet still fails.
Never being cold. Never being hot. A permanently narrow thermal band removes a stimulus the system evolved to expect, with measurable costs to brown adipose tissue and glucose disposal.
Full argument: cold-exposure.html.
Insufficient UVB (vitamin D synthesis), insufficient full-spectrum morning light (circadian anchoring), insufficient infrared. Also a thermal-comfort-shaped problem: modern indoor life removes an expected input.
Full argument: sunlight.html.
Skeletal muscle is the largest glucose sink in the body. Resistance training increases both the size of that sink and its insulin-independent glucose uptake. Under-muscled is under-buffered.
The modern diet runs short on two inputs to vascular function. Potassium intake sits well below adequate for most people. Nitrate-rich vegetables, with arugula and beets at the top, have mostly dropped out of the diet, and nitrate is the substrate for the oral-bacteria route to nitric oxide that bypasses the eNOS pathway metabolic disease damages.
Salt is the one mineral where the usual advice points the wrong way for the people this protocol is written for. Processed and restaurant food is where most sodium comes from. Someone who follows Step 1 and cooks everything at home has removed that source and can easily end up under-salted. Mortality studies show a J-shaped curve for sodium, not a line where lower is always better. The trial results that drive low-salt guidance are blood-pressure reductions of a few mmHg, which is a surrogate endpoint. They aren’t outcomes.
Strongfor nitrate → NO. Moderatefor potassium adequacy. Contestedfor the sodium curve — the low end of the J is argued over, largely on reverse causation.
Related: sodium.html, potassium.html, beetjuice.html. statins.html VII.4–5 takes a more mainstream line on sodium.
This paper doesn’t list LDL under “what breaks it”, and the omission is deliberate. The statins paper concedes that apoB-containing particles cause atherosclerosis — the genetic evidence is too strong to argue with — and then argues that for most people outside familial hypercholesterolaemia and established heart disease, serum lipids are substantially a readout of metabolic terrain rather than an independent lever. Restoring insulin sensitivity moves triglycerides, HDL, particle size and number, blood pressure and visceral fat together. The same LDL-C number means something different in an insulin-resistant person than in a lean athlete — and in the first, LDL-C probably understates the particle count, which is why apoB is the better test.
Exercise is the one place this has direct anatomical support: in the CERT trial, supervised high-intensity intervals regressed coronary plaque on intravascular ultrasound with no lipid target at all — layered on top of standard care, not replacing it. Strong for that result; Speculative that dietary terrain levers alone beat the drug for most people. statins.html Part VII.
The sequencing is the point. Most protocols fail because they apply a step-2 intervention to a step-1 problem.
Goal: be able to handle a large carbohydrate load without a large insulin excursion. Everything in this step either lowers the load or raises the capacity. Nothing in this step is about weight.
| Intervention | Mechanism | Grade |
|---|---|---|
| Remove seed oils / lower dietary LA toward ~1–2% of calories | Reduces peroxidation substrate and desaturase flux. In practice: cook at home, cut fried, restaurant and packaged food | Moderate |
| Oily fish; MUFA and saturated fat in place of PUFA | Omega-3 competes with LA for D6D and membrane slots; oleic acid displaces PUFA from membranes via ACSL3 | Moderate |
| Resistance training, 2–3×/week | Expands muscle glucose sink; glycogen depletion creates draw | Strong |
| Zone 2 / aerobic base | Mitochondrial density; intramyocellular lipid clearance | Strong |
| High-intensity intervals (4×4 min, 85–95% peak HR, 2×/week) | Regressed coronary plaque on IVUS vs. guideline care (CERT) — on top of standard therapy, in post-PCI patients | Strong |
| Adequate sleep | Insulin sensitivity, appetite regulation, GH/autophagy window | Strong |
| Cold exposure | BAT recruitment, insulin-independent glucose uptake, lowered fasting glucose | Moderate |
| Morning light / evening dark | Anchors the insulin sensitivity rhythm | Moderate |
| Post-meal walking (10–15 min) | Blunts postprandial excursion directly | Strong |
| Meal sequencing (protein/fiber/veg before starch) | Slows gastric emptying and absorption | Moderate |
| Dietary nitrate (arugula, beets); no antiseptic mouthwash | Oral bacteria reduce nitrate → nitrite → NO, bypassing damaged eNOS; vasodilation and blood flow | Strong |
| Adequate potassium; salt to taste when cooking at home | Potassium is broadly under-consumed; cutting processed and restaurant food removes most dietary sodium | Moderate |
| Whole fruit as staple; orange juice as a supplement | Fibre and matrix intact in whole fruit; juice supplies vitamin C, folate and potassium | Moderate |
| Reduce alcohol | Consumes NAD⁺ via ADH/ALDH; acutely worsens insulin sensitivity | Moderate |
| Front-load carbohydrate earlier in the day | Aligns intake with the sensitivity peak | Moderate |
Step 1 needs a measurable endpoint, not a duration. Candidates: fasting insulin, HOMA-IR, a post-load glucose curve, or continuous glucose monitoring showing a tight excursion and fast return. Pick the marker and set the number.
Two distinct protocols exist in the literature and they serve different purposes. Sustained moderate cold acclimation over ~10 days produced the large insulin-sensitivity improvement in type 2 diabetics — that's the corrective shape, and it belongs here in Step 1. Brief repeated exposure (a few minutes, 2–3× weekly, ending on cold) is the maintenance shape and belongs after. Don't confuse the two.
The maintenance literature explicitly finds that longer single sessions stop adding benefit.
Once disposal capacity is restored, carbohydrate becomes the preferred fuel rather than the liability it was in Step 1:
Timing rule: weight carbohydrate toward the earlier part of the day, lighter at dinner. This is the same chronobiology from Part I, applied.
Source rule: whole-food carbohydrate, low in LA by construction. The failure mode of “carb-forward” is that it becomes a license for industrial food. It isn't.
TODO — define what “carb-forward” means numerically: a ratio, a floor, or a per-meal target?
Seasonal and circadian variation rather than a fixed prescription. Cold in the cold months, light in the light months, carbohydrate weighted toward periods of higher activity and longer days.
TODO — decide how much seasonality is defensible vs. inherited framing. This is the section most likely to overreach.
Supplements fill identified gaps. They do not substitute for Part III. Ordered by confidence, not popularity.
The first question isn't which supplement to take. It's whether there's an actual gap to fill. The target for vitamin and mineral status is adequate — clearing RDA/AI thresholds with a reasonable margin — not deficient, and not maximized either. Pushing past adequacy has its own cost profile (see the zinc/copper interaction under Tier 1) and isn't supported by the same evidence base as correcting a real shortfall. Guessing which nutrients are short is unnecessary when it can be measured directly.
Cronometer (cronometer.com) logs everything eaten against a full micronutrient panel — not just calories and macros — and shows exactly which vitamins and minerals are running below RDA/AI across a tracked window. It turns Part IV from guesswork into a gap list: log a representative week or two of normal eating, then supplement what the data actually shows is low. Re-check periodically rather than assuming the gap stays fixed as intake changes.
Speculativefor the specific logging-window length — reasonable, but not something with trial support behind it.
Magnesium glycinate. Repletion of a mineral that is genuinely under-consumed on modern diets and involved in hundreds of ATP-dependent reactions. Glycinate form for absorption and GI tolerance; the glycine component is itself useful for sleep and glutathione synthesis. Strong for repletion; Moderate for the downstream metabolic claims.
Vitamin D3, October–March. Above roughly 37°N, cutaneous synthesis effectively stops in winter. This is a seasonal supplement, not a year-round one — the summer replacement is sunlight. Strong
Vitamin K2 (MK-7). Directs calcium to bone and away from vascular tissue via MGP and osteocalcin carboxylation. Requires daily dosing for consistent activation — MK-7's long half-life is often overstated as license for intermittent dosing. Moderate
Freeze-dried oyster. Excellent whole-food zinc and copper in roughly the ratio the body wants — the main advantage over isolated zinc supplementation, which can drive copper deficiency at chronic high doses. Moderate
L-5-MTHF. Bypasses the MTHFR conversion step. Justified where there's a documented folate gap or known variant; not a default. Moderate
Oysters are not a strong iodine source. They're outstanding for zinc and copper and reasonable for selenium and B12, but iodine content is modest and variable. If iodine is a documented gap, the honest source is seaweed (kelp, kombu, nori) or a measured potassium iodide dose — not oysters. Worth fixing rather than repeating.
Iodine — separate from the oyster entry. TODO decide source and dose. Kelp is potent and variable; dosing by “a sheet of nori” isn't a dose. Consider testing before supplementing.
Whole-food vitamin C (Pure Radiance / VitaCherry). Real advantage: comes with the flavonoid and cofactor matrix rather than isolated ascorbate. Real limitation to state plainly: these products deliver tens of milligrams, not grams. If the goal is food-equivalent vitamin C with intact cofactors, they work. If the goal is gram-level ascorbate for a specific pharmacologic effect, they don't — pretending otherwise is the main way this category is oversold. Moderate for the food-matrix rationale.
NAC, episodically before alcohol. Glutathione precursor ahead of a known oxidative and NAD⁺-consuming load. Episodic use avoids the chronic-antioxidant question. Inferred
Milk thistle (standardized silymarin). Standalone over blended “liver support” formulas, where dosing is opaque. Moderate
Glycine. Underrated. Collagen-derived amino acid missing from muscle-meat-dominant diets, rate-limiting for glutathione, and useful for sleep onset via core temperature drop. Bone broth or supplemental. Moderate
Black seed (whole, ground fresh), dandelion root, taurine, creatine.
TODO — creatine probably deserves promotion to Tier 1. The muscle and cognitive evidence is stronger than everything else in Tier 3. Decide.
Optional additions are evaluated one at a time on the Supplements index, and most sit at the weak end of the grading scale. The ones that bear directly on this paper:
Someone still in Step 1 has limited disposal capacity, so large daily volumes of juice work against the goal. A glass is a supplement, not a staple. For beet and pomegranate, juice is the form the trials used, not necessarily the form that makes sense: whole beets, arugula and pomegranate arils carry the same actives with the fibre intact.
Anything whose claimed benefit is already delivered by Part III. Metabolic health is not a supplement problem, and a long stack is usually a sign that a foundation is being worked around rather than fixed.
Without markers, this is belief. Minimum useful panel:
TODO — add target ranges. Decide whether to state “optimal” ranges vs. lab reference ranges — the gap between them is a point worth making.
This paper does not yet carry a numbered citation list — per the reader's guide above, structure came first, prose second, citations last. Claims above that most need one: the cold-acclimation insulin-sensitivity trial (III.1), the overfeeding/de novo lipogenesis studies (II.3), and the chronobiology postprandial-glucose figures (Limitations, above).