Chapter 26 — Nutrition and Chronic Disease: Which Numbers Respond to Food, and by How Much

The Hook: Walt's panel, with a column added

Walt Prosser is 68. Type 2 diabetes for eleven years, on metformin. Chapter 11's fibre ramp, Chapter 13's B12, Chapter 16's supplement audit — his A1c went 7.4% to 6.9% and stopped.

Here is his most recent panel. ⚠️ And here is the column nobody puts on a results letter.

⚠️ Walt had been told to "watch his diet" for eleven years, and nobody had ever told him which of his numbers that would actually move.

⚠️ That's the chapter. Not "is diet important" — it obviously is — but for which outcome, by how much, and compared to what else is available.

Two failure modes this chapter exists to prevent:

1. ⚠️ Under-claiming. Type 2 diabetes remission is real, reproducible and dose-dependent, and a great many people with the condition have never been told it is possible.

2. ⚠️ Over-claiming, which does more damage. Diet does not treat cancer. It does not fix Lp(a). And "diet-controlled" is not a synonym for "treatable without medication."

🏃 Fast Track: §26.1 (the responsiveness table), §26.3 (⚠️ remission), §26.7 (blood pressure), §26.10 (⚠️ cancer). Thirty-five minutes.

🔬 Deep Dive: §26.6 (ApoB), §26.8 (sodium, finally), §26.11 (⚠️ kidney disease — the second inversion).

26.1 The responsiveness hierarchy

⚠️ Ordered by how much food actually moves them.

💡 ⚠️ Notice the shape: the things that respond fastest are the metabolic ones — liver fat, triglycerides, glucose, blood pressure.

These are the numbers of a person's current metabolic state. ⚠️ Structural and genetic variables — Lp(a), an LDL-receptor mutation, an autoimmune process — do not respond, because food is not what set them.

A great deal of unnecessary guilt is generated by people trying to move tier 4 numbers with tier 1 interventions.

26.1b One process, five diseases

⚠️ Chapter 25 §25.10b told bone as one story across sixty years. This is the equivalent for metabolism, and it explains why the same intervention keeps appearing.

Look at what §26.1's tier 1 contains: liver fat, triglycerides, glycaemia, blood pressure, weight.

⚠️ Those are not five independent problems. They are five readouts of one process.

When energy intake persistently exceeds what subcutaneous fat tissue can comfortably store, fat accumulates where it shouldn't — liver, pancreas, muscle, viscera. ⚠️ That ectopic and visceral fat is metabolically active, drives insulin resistance, and produces every number in tier 1.

💡 ⚠️ This is why one intervention — losing ectopic fat — improves five apparently separate conditions at once, and why the dose-response curves in §26.3 and §26.9 have the same shape.

It also explains something that puzzles patients: why weight loss "fixed" a liver problem, a glucose problem and a blood pressure problem simultaneously. ⚠️ It didn't fix three things. It reduced one exposure that was producing three readouts.

⚠️ Two important qualifications, because this framing can be over-extended:

1. ⚠️ Not everyone with a given condition has this process. There are lean people with type 2 diabetes, non-metabolic causes of hypertension, and cancers with no metabolic contribution at all. The framing is a common pathway, not the only one.

2. ⚠️ And where you store fat is substantially genetic (Chapter 24 §24.1b). Two people with the same BMI can have very different amounts of liver and visceral fat — which is why BMI is a poor individual predictor and why some people develop metabolic disease at weights that look unremarkable.

⚠️ Which is also why waist measurement adds information that BMI doesn't, and why it takes ten seconds and almost nobody does it.

26.2 Type 2 diabetes: the most food-responsive chronic disease there is

What the disease actually is: ⚠️ insulin resistance plus progressive beta-cell dysfunction, and the dominant modifiable driver is excess fat in the wrong places — particularly liver and pancreas (Chapter 18 §18.9).

⚠️ That last point is the mechanism that makes §26.3 work, and it's the twin-cycle model: excess liver fat drives hepatic insulin resistance and raises hepatic glucose output; fat accumulating in the pancreas impairs beta-cell function. Both are reversible.

✅ Prevention is established. The Diabetes Prevention Program achieved ~7% weight loss and reduced progression to type 2 diabetes by about 58%, outperforming metformin (Chapter 24 §24.8).

26.3 ⚠️ Remission, and the number that should be better known

🔬 Claim → Evidence → Verdict

The claim: "Type 2 diabetes can go into remission through weight loss."

What the evidence shows: ⚠️ The DiRECT trial (Lean et al., published in The Lancet, 2018 and 2019) is the flagship. Primary care patients with type 2 diabetes of up to six years' duration were randomized to a structured programme: a total diet replacement — a formula low-energy diet — for several months, followed by structured food reintroduction and long-term maintenance support, with diabetes medications withdrawn at the start.

⚠️ At 12 months, remission was achieved by roughly 46% of the intervention group; at 24 months, roughly 36%.

And the finding that matters most:

⚠️ REMISSION WAS STRONGLY DOSE-DEPENDENT ON WEIGHT LOST. In the intervention group at 12 months, remission was achieved by around 86% of those who lost 15 kg or more, roughly 57% of those losing 10–15 kg, around 34% for 5–10 kg, and only about 7% for those losing under 5 kg.

⚠️ That gradient is the single most useful set of numbers in this chapter. It converts "try to lose some weight" into a quantified prediction.

📉 Evidence quality: Randomized, in primary care, with a hard biochemical endpoint and two-year follow-up.

Verdict: ✅ Well supported.

⚠️ Four qualifications that matter clinically:

1. ⚠️ Duration of diabetes matters. Remission is substantially more likely earlier in the disease course, when beta-cell function is more recoverable. DiRECT recruited people diagnosed within six years.

2. Remission is not cure. ⚠️ Weight regain generally means relapse, and the definition — commonly HbA1c below the diagnostic threshold for at least three months without glucose-lowering medication — requires ongoing monitoring.

3. ⚠️ This is a supervised intervention, not a self-directed one. Medications are withdrawn as part of the protocol. Someone on insulin or a sulfonylurea who starts a very-low-energy diet without adjusting medication is at real risk of hypoglycaemia (Chapter 21 §21.11).

4. And the route matters less than the loss. ⚠️ Bariatric surgery produces higher remission rates still (Chapter 24 §24.12), and GLP-1-based approaches are producing losses in the range DiRECT associated with high remission rates. The agent is the weight loss and the metabolic change that accompanies it, not the method.

⚠️ The clinical implication I'd most want carried out of this chapter: a substantial proportion of people with type 2 diabetes have never been told remission is possible, or what magnitude of weight loss it would take. Both are knowable and specific.

26.4 What to eat with type 2 diabetes

⚠️ The answer is less exciting than the argument about it, and Chapter 10 predicted it.

Multiple dietary patterns improve glycaemia: lower-carbohydrate, Mediterranean, low-fat, vegetarian and vegan patterns have all shown benefit in trials. ⚠️ Major diabetes organizations no longer specify a single eating pattern, and instead recommend individualizing.

What consistently helps:

⚠️ And the safety point that is not optional:

⚠️ Reducing carbohydrate while taking insulin or a sulfonylurea without adjusting the dose causes hypoglycaemia. This is a prescriber conversation before, not after.

Chapter 21 flagged this as the one place in the book where "try it and see" is genuinely dangerous. It applies here with more force, because low-carbohydrate diets are widely and reasonably recommended for glycaemic control.

26.5 Cardiovascular disease: what LDL-C actually is

🔬 Claim → Evidence → Verdict

The claim: "LDL cholesterol causally contributes to atherosclerotic cardiovascular disease."

What the evidence shows: ⚠️ This is one of the most thoroughly established causal chains in all of medicine, and it does not rest on observational epidemiology.

It rests on convergence (Chapter 2 §2.7) across:

• ⚠️ Mendelian randomization — people with genetic variants causing lifelong lower LDL-C have lower cardiovascular risk, across many independent genes. The same instrument that made Chapter 12's alcohol case
• Randomized trials of multiple drug classes — statins, ezetimibe, PCSK9 inhibitors — all reducing events roughly in proportion to the LDL-C reduction achieved, regardless of mechanism
• Familial hypercholesterolaemia, where very high lifelong LDL-C produces very early disease
• Animal and mechanistic work on the artery wall (Chapter 9)

⚠️ Different genes, different drugs, different mechanisms, same relationship between LDL-C lowering and event reduction. That is what a causal finding looks like.

Verdict: ✅ Well supported. ⚠️ And note the contrast with almost everything in Part IV: this is what the evidence looks like when a question actually has been settled.

⚠️ Now the honest part about diet's magnitude:

⚠️ Both halves of this need saying.

Diet moves LDL-C meaningfully — the portfolio evidence is real and 20–30% is not trivial.

⚠️ And for someone at high cardiovascular risk, declining a statin in favour of diet is choosing the smaller intervention. They are not alternatives; they are additive, and the drug is the larger one.

Dietary patterns and hard outcomes

⚠️ Everything above is about a number. The question people actually care about is events, and the trial evidence there is thinner and worth stating precisely.

The strongest is PREDIMED — a Spanish randomized trial of a Mediterranean dietary pattern supplemented with either extra-virgin olive oil or mixed nuts, against a control advised to reduce dietary fat, in people at high cardiovascular risk. ⚠️ It reported reduced major cardiovascular events in the Mediterranean groups.

⚠️ And it needs its caveat stated, because this book's whole method requires it: PREDIMED was retracted and republished in 2018 after randomization irregularities were identified at some sites. The reanalysis, using more conservative methods, reported broadly similar conclusions — but the episode is a legitimate reason to hold the result somewhat less firmly than its citation frequency suggests.

⚠️ How I'd weight it: 🟢 the Mediterranean pattern probably reduces cardiovascular events in high-risk people. It is the best trial evidence available for a dietary pattern and a hard outcome, and it is not as strong as the LDL-C causal chain in §26.5.

⚠️ Which is the honest shape of the field: we know far more about how diet moves biomarkers than about how diet moves events, because the trials required for the second are enormous, long and largely unfundable (Chapter 2 §2.6).

And two other pattern-level findings worth knowing: the DASH pattern's blood-pressure effect (§26.7) is a controlled feeding result rather than an events trial, and ⚠️ §26.7's salt-substitute trial is the rare dietary intervention with randomized EVENT evidence — which is exactly why it gets a section rather than a sentence.

26.6 ApoB, and why HDL isn't a target

Two refinements that change how you read a lipid panel.

ApoB

⚠️ Every atherogenic lipoprotein particle carries exactly one apolipoprotein B molecule. So ApoB counts particles, while LDL-C measures the cholesterol carried inside them.

Usually these agree. ⚠️ When they disagree — discordance — ApoB is the better predictor of risk.

And discordance is common in exactly the people this chapter is about: ⚠️ insulin resistance, raised triglycerides, and metabolic syndrome produce many small, cholesterol-depleted LDL particles. LDL-C looks acceptable; particle number is high; risk tracks the particle number.

🔬 Verdict: 🟢 Probably true that ApoB is a better risk marker than LDL-C where they diverge, and several guidelines now recommend measuring it. ⚠️ If you have raised triglycerides or type 2 diabetes and a "normal" LDL-C, ask about ApoB. Walt's situation exactly.

HDL

⚠️ Higher HDL-C is associated with lower risk. Raising it pharmacologically has repeatedly failed to reduce events — CETP inhibitor trials being the clearest example.

⚠️ HDL-C is a marker of metabolic state, not a target to be moved.

Chapter 2's reverse-causation problem, in one number: low HDL travels with insulin resistance, raised triglycerides and adiposity — and it's those that carry the risk.

Which means "raise your HDL" is a 🟡 instruction, and "improve the metabolic state that produced a low HDL" is the useful version.

⚠️ And Lp(a), which belongs here: an inherited, largely genetically determined lipoprotein that is an independent cardiovascular risk factor and ❌ does not meaningfully respond to diet. ⚠️ It should be measured at least once in a lifetime — because a raised Lp(a) changes how aggressively everything else should be managed, and because people spend years trying to diet away a number that food cannot move.

26.7 Blood pressure

One of the most food-responsive numbers there is, and the evidence is unusually good.

🔬 Claim → Evidence → Verdict

The claim: "Dietary pattern substantially lowers blood pressure."

What the evidence shows: ✅ the DASH trials. A dietary pattern rich in fruit, vegetables, low-fat dairy, wholegrains, nuts and legumes, and lower in saturated fat, red meat and sweets, lowered blood pressure substantially in a controlled feeding trial — with larger effects in people with hypertension. ⚠️ And DASH-Sodium showed the effects of the pattern and of sodium reduction were additive.

Verdict: ✅ Well supported.

What else moves blood pressure, roughly in order:

⚠️ Salt substitutes — the most important trial most people haven't heard of

🔬 Claim → Evidence → Verdict

The claim: "Replacing regular salt with a potassium-enriched salt substitute reduces cardiovascular events."

What the evidence shows: ⚠️ A large randomized trial in rural China (the SSaSS trial, reported in the New England Journal of Medicine, 2021) replaced household salt with a potassium-enriched substitute in whole villages, in people at elevated stroke risk. It reported reductions in stroke, major cardiovascular events and death.

⚠️ This is outcome evidence — events, not blood pressure — from a randomized trial of a dietary substitution costing almost nothing.

Verdict: ✅ Well supported in the population studied. ⚠️ The caveat is essential: potassium-based substitutes are NOT safe in advanced chronic kidney disease or in people on potassium-retaining medications, because of hyperkalaemia risk (§26.11). Check before recommending.

26.8 Sodium, finally

⚠️ Chapter 14 deliberately left this unresolved. Here is where I land, and why.

What is well established: ✅ Reducing sodium intake lowers blood pressure, dose-dependently, with larger effects in people who are older, hypertensive, or salt-sensitive. This comes from randomized trials and is not seriously disputed.

What is contested: ⚠️ whether reducing sodium to very low levels improves outcomes in everyone.

The contrarian evidence comes largely from observational cohorts (PURE most prominently) reporting a J-shaped relationship — higher mortality at both high and low estimated intakes.

⚠️ Why I don't weight that heavily:

1. ⚠️ The intake estimates in these cohorts rely largely on spot urine samples with prediction equations, which are a poor proxy for true intake and produce systematic error.

2. Reverse causation. ⚠️ People who are ill eat less of everything, including sodium.

3. ⚠️ And the randomized outcome evidence points the other way — §26.7's salt substitute trial being the strongest example.

🔬 Where I land: ✅ reducing sodium lowers blood pressure and this is not in doubt. 🟢 population-level sodium reduction reduces cardiovascular events. ⚠️ 🟡 whether individuals with normal blood pressure benefit from pushing to very low intakes.

⚠️ And the practical version, which sidesteps most of the argument: most dietary sodium comes from processed and restaurant food, not the salt cellar (Chapter 22). Reducing ultra-processed food reduces sodium as a side effect, and a potassium-enriched substitute at home does the rest.

26.9 Fatty liver disease

⚠️ Now most commonly termed MASLD — metabolic dysfunction-associated steatotic liver disease — following a nomenclature change in 2023, replacing NAFLD. You will meet both terms.

It is common, frequently undiagnosed, and ⚠️ among the most diet-responsive conditions in this chapter.

⚠️ Dose-dependent, like §26.3, and the same shape.

What else matters: ⚠️ sugar-sweetened beverages (Chapter 18 §18.6's hepatic material lands here) · alcohol, which compounds it · physical activity, which improves liver fat independent of weight loss · and the Mediterranean pattern, which has the best trial support of the named patterns.

🟢 And an affirmative worth having: coffee. ⚠️ Coffee consumption is consistently associated with lower rates of liver disease progression and fibrosis — across multiple cohorts, with a plausible mechanism, and it is one of the few genuinely enjoyable recommendations in this chapter.

⚠️ Walt's ALT of 61 is the number nobody had explained to him, and it is tier 1 on §26.1's table.

26.10 ⚠️ Cancer: where nutrition claims do the most damage

Handled carefully, because this is the section where getting it wrong costs the most.

✅ What is established

⚠️ The World Cancer Research Fund / AICR recommendations are the reference document and they are free.

⚠️ And the magnitude honesty: dietary factors matter for cancer risk, and they matter considerably less than not smoking. Individual-level effect sizes are modest, and the biggest dietary lever is alcohol.

❌ What does the damage

Chapter 18 §18.10 promised this section and here it is.

⚠️ The harm is specific, and it is not hypothetical.

1. Delayed or abandoned effective treatment. ⚠️ The most serious, and it happens.

2. ⚠️ Malnutrition during treatment. Restrictive "anti-cancer" diets adopted during chemotherapy or radiotherapy cause weight loss and reduced protein intake at exactly the point where nutritional status affects treatment tolerance, complication rates and outcomes.

3. And the moral injury of implied blame — the suggestion that a person's diet caused their cancer, or that better discipline would cure it.

⚠️ Nutrition DURING cancer treatment inverts

Chapter 25 §25.12's inversion, arriving again — and this catches people out badly.

⚠️ Cancer cachexia — the loss of muscle and weight driven by the disease process — is not reversed by eating more alone, and it requires specialist input. But under-nutrition on top of it is avoidable and common.

⚠️ If you take one thing from §26.10: the person most likely to be harmed by nutrition advice in this book's subject area is someone undergoing cancer treatment who has been given a restrictive diet by someone who meant well.

26.11 ⚠️ Kidney disease: the second inversion

Chapter 25 inverted for age. ⚠️ Chronic kidney disease inverts for a different reason, and several of this book's recommendations reverse.

⚠️ Three things worth knowing even if you don't have kidney disease:

1. ⚠️ Protein restriction in CKD has more modest evidence than its prominence suggests. The question of how much it slows progression has been studied — the MDRD study being the best known — and the effects are smaller and more debated than the confidence of the advice implies. ⚠️ And restriction carries a real risk of protein-energy wasting, particularly in older patients (Chapter 25 §25.13's territory, in direct conflict).

2. ⚠️ On dialysis, the recommendation REVERSES AGAIN — protein requirements go UP, commonly to around 1.0–1.2 g/kg or more, because dialysis removes amino acids.

3. ⚠️ Phosphate additives are far more bioavailable than naturally occurring food phosphate. A person told to "reduce phosphate" who cuts dairy and legumes while continuing to eat processed food containing phosphate additives has made the problem worse and their diet poorer. ⚠️ Read the ingredients list for phosphate additives — this is a practical, high-yield instruction that is rarely given.

⚠️ CKD nutrition is individualized, changes by stage, and is a renal dietitian's job. This section exists so that you recognize when general advice stops applying — which is the single most useful thing a general text can do here.

⚠️ The pattern to notice across §26.10 and §26.11

Both sections are about populations where general dietary advice stops applying — and they share a failure mode worth naming, because it will recur in Chapters 28 and 29.

⚠️ A person with a serious diagnosis is unusually motivated, unusually receptive, and unusually likely to be handed a restrictive diet by someone who means well.

And in both cases the restriction is the harm:

⚠️ The general rule, and it is the most transferable thing in these two sections:

In a person with a serious illness, the question is not "what should they cut out?" ⚠️ It is "are they eating enough, and of what?" — and the restriction, if any, should be specific, targeted, and issued by someone who knows the disease.

Chapter 25 §25.12's inversion, Chapter 20 §20.13's stopping-rule problem, and Chapter 34's territory, all meeting in the same clinical room.

26.12 Gout, and the honest magnitude

Purines raise urate; urate crystallizes; the joint hurts. ⚠️ And the dietary story is more modest than its cultural prominence.

What raises urate: ⚠️ alcohol — particularly beer · fructose, including in sugar-sweetened beverages (Chapter 18) · organ meats and some seafood · ⚠️ and adiposity, which is the biggest dietary-adjacent factor.

What lowers it: weight loss · reducing alcohol and SSBs · 🟡 dairy, coffee and vitamin C show modest associations · 🟡 cherries, where the evidence is thin and popular.

⚠️ The honest magnitude: dietary modification typically produces small reductions in urate compared with urate-lowering therapy such as allopurinol.

Which means the common experience — "I've cut out everything and I still get attacks" — is not a failure of adherence. It is the expected result of using a small lever on a large problem.

⚠️ Diet is worth doing and is not a substitute for treatment where treatment is indicated.

26.13 The word "inflammation"

⚠️ A short section, because the word does an enormous amount of unearned work.

What's real: chronic low-grade inflammation is associated with cardiometabolic disease, and CRP is a measurable marker.

What actually lowers CRP: ⚠️ weight loss · physical activity · smoking cessation · treating infection or inflammatory disease · improving sleep.

⚠️ What "anti-inflammatory diet" describes: a high-fibre, high-plant, oily-fish-containing, lower-ultra-processed pattern — ⚠️ which is Chapter 10 §10.5's pattern with a different label. The label adds a mechanism it hasn't demonstrated.

⚠️ And Chapter 19 §19.5 is the cautionary case: the omega-6 chain predicted that changing dietary fat composition would change inflammatory markers, and direct testing did not confirm it.

⚠️ Verdict: 🟡 — the dietary pattern is good, the label is doing work it hasn't earned, and CRP responds mostly to weight, activity and smoking rather than to food composition.

26.13b How fast do the numbers move?

⚠️ Almost nobody is told this, and it is the single most common reason people abandon a change that was working.

💡 ⚠️ Two practical consequences.

1. ⚠️ Do not recheck HbA1c at four weeks. It cannot have moved, the result will disappoint, and people quit on the basis of a measurement that was always going to look like that. This happens constantly.

2. ⚠️ And notice which numbers move fastest — triglycerides, glucose, blood pressure, liver fat. Those are §26.1b's tier 1, and they are fast because they reflect a current state rather than an accumulated structure.

Which gives an unusually useful rule: ⚠️ if you want early evidence that something is working, measure a fast number. If you want to know whether it mattered, measure a slow one.

26.14 Who this chapter is about

⚠️ How firmly I hold these

26.15 What to actually do

1. ⚠️ Get your numbers, and get the responsiveness column (§26.1). Most people have never been told which of their results food can move.

2. ⚠️ If you have type 2 diabetes: ask about remission, and ask what weight loss it would take. §26.3.

3. ⚠️ If you're on insulin or a sulfonylurea: talk to your prescriber before changing carbohydrate.

4. ⚠️ Measure Lp(a) once, and stop trying to diet away a number food cannot move.

5. If your triglycerides are up and your LDL-C looks fine, ask for ApoB.

6. ⚠️ For blood pressure: the DASH pattern, potassium, less alcohol, weight — and consider a potassium-enriched salt substitute (⚠️ not if you have significant kidney disease or take potassium-retaining drugs).

7. ⚠️ For fatty liver: this is the most reversible condition in the chapter. 7–10% weight loss, no sugar-sweetened beverages, less alcohol, more activity — and coffee is fine.

8. ⚠️ Take a statin if you're advised to. Diet is additive, not alternative, and the drug is the larger intervention.

9. ⚠️ If you or someone you love is having cancer treatment: eat. Do not restrict. Get a dietitian if there's weight loss.

10. And if you have kidney disease, get individualized advice — ⚠️ and read the ingredients list for phosphate additives.

🧾 What it costs

⚠️ Chapter 24 broke the twelve-chapter streak. This chapter partly restores it, and the exceptions are instructive.

⚠️ Two observations.

The single best cost-per-outcome item in this chapter is a potassium salt substitute at about twenty dollars a year, with randomized event data behind it — and almost nobody has heard of it.

⚠️ And generic statins and metformin are among the cheapest interventions on this page while being among the most effective. A person spending $600 a year on supplements to avoid a $50 medication has made an expensive decision, and Chapter 16 §16.9's questions were designed to prevent exactly it.

What we don't know

⚠️ Long-term remission durability beyond a few years. Whether very low sodium targets benefit normotensive people (§26.8). How much protein restriction actually slows CKD progression, against its wasting risk (§26.11). ⚠️ And how much of the residual diet–cancer association survives full control for adiposity, alcohol and smoking — which is a question nobody has answered cleanly.

Spaced Review

1. (Chapter 2) Why is LDL-C's causal status held so much more confidently than most nutrition claims?

⚠️ Convergence across methods with non-overlapping weaknesses (Chapter 2 §2.7). Mendelian randomization across many independent genes · randomized trials of multiple drug classes reducing events roughly in proportion to LDL-C lowering regardless of mechanism · familial hypercholesterolaemia · mechanistic artery-wall work. ⚠️ Different genes, different drugs, same relationship. That's what a settled question looks like.

2. (Chapter 24) DiRECT found remission strongly dose-dependent on weight lost. ⚠️ Why does that matter more than the headline remission rate?

Because it converts a vague instruction into a quantified prediction. ⚠️ "Try to lose weight" becomes "at 15 kg or more, around 86% achieved remission; under 5 kg, about 7%." That is actionable, it sets expectations honestly, and it tells someone whether the effort is likely to reach the outcome they want.

3. (Chapter 25) Name two places in this chapter where the general advice inverts, and why.

⚠️ Cancer treatment (§26.10) — goals become maintaining weight and lean mass; restriction is harmful when nutritional status affects treatment tolerance. And ⚠️ chronic kidney disease (§26.11) — protein often restricted, potassium may need limiting, and salt substitutes are contraindicated. ⚠️ Both are Chapter 25 §25.12's structure: a population where general recommendations stop applying.

Project Checkpoint: Your Risk Audit

Component twenty-six. Get your actual numbers, and sort them.

Step 1 — Collect what you have. ⚠️ Most people can request these; many already have them.

Step 2 — Add the column nobody gives you. ⚠️ For each, write the tier from §26.1.

Step 3 — Identify your tier 1 numbers. ⚠️ These are where food does the most, fastest.

My most responsive abnormal number is: __ The intervention with the largest effect on it is: __

Step 4 — ⚠️ Identify what food will NOT fix.

Tier 4 items on my panel: __

⚠️ Write them down and stop spending effort there. If Lp(a) is raised, the response is to manage everything else more aggressively — not to eat differently at it.

Step 5 — The medication question.

⚠️ Am I on insulin or a sulfonylurea? _ (If yes: prescriber before any carbohydrate change.) Have I been advised a statin and declined it in favour of diet? _ (⚠️ §26.5. They're additive.)

Step 6 — And if you have type 2 diabetes, one question to take to your next appointment:

⚠️ "Am I a candidate for remission, and how much weight loss would it take in my case?"

⚠️ Most people with type 2 diabetes have never asked it, because most have never been told it was a question.

Next checkpoint (Chapter 27): your fibre and fermented food audit — before anyone sells you a probiotic.

Chapter Summary

⚠️ The organizing question: not "does diet matter" — it does — but FOR WHICH OUTCOME, BY HOW MUCH, AND COMPARED TO WHAT ELSE.

§26.1's four tiers: ⚠️ highly responsive (liver fat, triglycerides, HbA1c, blood pressure, weight) · moderately responsive (LDL-C, ApoB, urate, CRP — usually 5–20%) · indirectly responsive (eGFR, bone, cancer risk) · ⚠️ NOT meaningfully responsive (Lp(a), FH's underlying defect, type 1 diabetes, coeliac genetics, age, family history).

⚠️ And the shape of the field, stated once: we know far more about how diet moves BIOMARKERS than about how diet moves EVENTS, because the trials required for the second are enormous, long and largely unfundable (Chapter 2 §2.6). §26.7's salt-substitute trial is the rare exception, which is why it gets a section rather than a sentence.

⚠️ §26.10's central warning: the person most likely to be harmed by nutrition advice in this book's subject area is someone undergoing cancer treatment who has been given a restrictive diet by someone who meant well. During treatment the goals INVERT: maintain weight and lean mass, adequate protein, eat what you can tolerate.

⚠️ §26.11's second inversion: in CKD, protein is often restricted, potassium may need limiting, and the salt substitute in §26.7 is contraindicated. And phosphate ADDITIVES are far more bioavailable than food phosphate — read the ingredients list.

⚠️ §26.5's honest magnitude: diet moves LDL-C 5–15%, and a full portfolio approach 20–30%. Statins do 30–50%+. They are additive, not alternative, and declining the drug in favour of diet is choosing the smaller intervention.

⚠️ §26.1b's unifying frame: liver fat, triglycerides, glycaemia, blood pressure and weight are not five problems — they are five readouts of one process. Ectopic and visceral fat accumulation drives fatty liver, type 2 diabetes, atherogenic dyslipidaemia, hypertension and several cancers, ⚠️ which is why one intervention improves five conditions at once and why §26.3 and §26.9 have the same dose-response shape. (⚠️ Not everyone with a given condition has this process, and where you store fat is substantially genetic — which is why waist measurement adds what BMI misses, takes ten seconds, and almost nobody does it.)

⚠️ §26.13b, and it prevents more abandoned attempts than anything else here: triglycerides move in 1–2 weeks, blood pressure in 2–4, liver fat in weeks, LDL-C in 4–6 weeks — and HbA1c takes 8–12 weeks minimum because it reflects three months of glycaemia. ⚠️ Do not recheck HbA1c at four weeks. It cannot have moved, and people quit on a measurement that was always going to look like that.

The one thing to remember: ⚠️ Walt had been told to "watch his diet" for eleven years, and nobody had ever told him which of his numbers that would actually move.

What's Next

Chapter 27 enters the area with the largest gap in this book between scientific excitement and clinical usefulness.

What the microbiome actually is and what sequencing can and can't tell you. Short-chain fatty acids and why fibre keeps reappearing. ⚠️ Probiotics — where the evidence is genuinely good, which is a much shorter list than the shelf suggests. Prebiotics, fermented foods, and the trial that gave fermented food a better result than fibre. Faecal microbiota transplant, which works spectacularly for one condition and unimpressively for the others. Microbiome testing kits, and what your $200 actually buys.

⚠️ And IBS and the low-FODMAP diet — which is the most clinically useful thing in the chapter and the part most often done badly.