Best Diet for High Cholesterol? Why the Answer Isn't a Food List — It's a Test

In the same week, two patients handed me lab reports I'd have been pleased to see on anyone.

Irakli and Khatia are composites built from a pattern seen often enough in clinic — two people whose diets contradict each other and whose numbers both look excellent — not any single identifiable patient.

Two eating patterns that agree on almost nothing. Two sets of biomarkers a preventive cardiologist would sign off on without hesitation. Neither diet is "correct" in any universal sense — and that isn't a diplomatic dodge. Their biomarkers are the most useful answer available, because a diet is a hypothesis and a lab panel is how you test it. That measured response is not a photograph of either patient's arteries, and it doesn't prove either diet is right for anyone else — but it's the best evidence either of them has to act on.

What the Two Diets Actually Have in Common

Line the two patterns up and four things are identical: neither eats much ultra-processed food — the category large prospective cohorts have repeatedly linked to higher cardiovascular event rates in a dose-dependent way, largely independent of macronutrient split.¹ Neither eats added sugar in any quantity. Both eat a large volume of vegetables. And both cook — dinners that share almost no ingredient, assembled by a person standing at a stove.

A fifth similarity matters more than any of them and isn't about food at all: both have been doing this for years, without willpower. Most dietary trials fight adherence — participants drift toward each other's assigned diets as months pass. So the real contrast here is narrower than it looks: these two patterns differ enormously in macronutrients and hardly at all in food quality. That's the less-discussed axis, and it's arguably the more important one.

Comparing Three Evidence-Based Patterns, Fairly

Three eating patterns dominate the evidence a cardiologist reasons from. Each has a strongest honest case and a real limitation — and the winner is determined per reader, in a lab, not per article.

Here's what the comparison actually supports, and it's a defensible reading of the evidence, not a shrug: when these patterns are compared head-to-head under controlled conditions, the differences between patterns are consistently smaller than the differences between individuals within any one pattern. The DIETFITS trial randomized 609 adults to a healthy low-fat or healthy low-carbohydrate diet — both eliminating added sugar and refined flour — and found no meaningful difference in average weight change between groups. Neither a pre-specified genetic score nor baseline insulin secretion predicted who did better on which diet.¹⁰ What both arms showed was a very wide spread of individual results around the group average — some of that spread is measurement noise, and telling the two apart in a given person is exactly what a retest is for.

The Saturated Fat Question, Answered Honestly

Not contested: replacing saturated fat with polyunsaturated fat lowers LDL-C and ApoB, on average, under controlled feeding conditions — one of the most reproduced findings in nutritional biochemistry.¹¹

Also not contested: what you replace it with determines what happens next. Pooled cohort data consistently finds that swapping saturated fat for polyunsaturated fat associates with lower coronary risk, while swapping it for refined carbohydrate shows no benefit at all.¹² This distinction explains much of the apparent contradiction in the public debate.

Genuinely contested: whether saturated fat intake, on its own, predicts cardiovascular events in free-living populations. Several large cohort meta-analyses have found no significant association.¹³ Meanwhile, a Cochrane review of randomized trials that actually reduced saturated fat intake found a 17% reduction in combined cardiovascular events.¹⁴ Both bodies of evidence are real — they measure different things (association vs. assignment) and neither is fraudulent.

Three Things Nobody Is Seriously Arguing About

Fiber, specifically viscous soluble fiber

Oats, barley, legumes, psyllium, apples, citrus. Soluble fiber forms a gel that binds bile acids and interrupts their reabsorption, forcing the liver to pull cholesterol out of circulation to make more — a mild version of the mechanism behind ezetimibe. Total fiber intake shows a dose-response relationship with cardiovascular outcomes across large prospective datasets, with no threshold at typical intakes — meaning more has kept helping across the observed range, and most people are nowhere near the top of it.²³

The omega-3 index

A blood test, not a food — and one of the more useful nutritional biomarkers available. It reports EPA and DHA as a percentage of the fatty acids in your red blood cell membranes, integrating over months rather than days, the way HbA1c does for glucose.²⁶ Two people eating identical amounts of fish can land at very different index values, because absorption, metabolism, and genetics all intervene between plate and membrane.²⁷ Large randomized trials of omega-3 supplementation have been largely null for major events in people without established cardiovascular disease.²⁹ One trial of a prescription-strength EPA preparation (REDUCE-IT) did show a substantial reduction in a high-risk, high-triglyceride population; a differently formulated trial (STRENGTH) did not — the discrepancy remains unresolved.³⁰ The index is worth knowing; a supplement decision belongs to your physician, not a store shelf.

Protein, scaled to you

The familiar 0.8 g/kg/day figure is an adequacy floor designed to prevent deficiency — not an optimum for someone protecting muscle mass with age.³¹ Expert groups working on aging (PROT-AGE) recommend roughly 1.0–1.2 g/kg/day for healthy older adults, and at least 1.2 g/kg/day for those training regularly — a real portion of protein at each of three meals, not an exotic intake.³² In people with normal kidney function, higher intakes within this range haven't been shown to affect kidney function trajectory.³³ Kidney function is the real gate: anyone with reduced eGFR should have their protein target set by their physician, not a general range.³⁴

Test, Don't Guess: A Framework You Can Run Yourself

This is the deliverable — a loop, not a plan, designed to be run more than once.

1. Establish a baseline. The three columns this framework runs on, all reachable from an ordinary blood draw: the lipid column (ApoB or LDL-C + triglycerides, plus a once-in-a-lifetime Lp(a)), the metabolic column (fasting glucose, HbA1c, fasting insulin or the TyG Index), and the inflammatory column as context (hs-CRP). Add a waist measurement and, if available, an omega-3 index. See our complete guide to what to ask for if you're starting from zero.
2. Choose a pattern you can imagine still eating in a year. Not the optimal one — the livable one. If your metabolic column is the concern, carbohydrate-restricted patterns have the most direct mechanism; if the lipid column is, plant-forward and Mediterranean patterns have the more reliable effect. Adherence beats optimization past about eight weeks.
3. Run it for twelve weeks, changing one major variable — not four at once.
4. Retest under matched conditions — same lab, same fasting state, not during active weight loss or acute illness, both of which can distort the comparison independent of diet.
5. Read all three columns, not just the one you were hoping about.

• All three improved: you've found a pattern producing a favorable measured response in your biology — not proof it minimizes lifetime risk, but the best evidence available to act on. Keep it, retest annually.
• Metabolic improved, lipid worsened: the most common conflict. Not a reason to panic or to ignore it — a reason for a specific conversation with your physician about your overall risk picture, family history, and whether the ApoB rise is modest or dramatic.
• Lipid improved, metabolic unmoved: common when shifting plant-forward while keeping refined carbohydrate high. Usually fixable inside the same pattern rather than by abandoning it.
• Nothing moved: information, not failure — it redirects effort toward the levers that do move your numbers.

Two boundaries worth holding onto. Lp(a) will not move on any dietary arm — a flat Lp(a) after a diet change is Lp(a) behaving exactly as expected, not a failed diet. And if your ApoB needs to fall by 40–50% to reach target, diet alone was never going to close that gap — this framework finds out how much of the gap is yours to close, and closes that part.

Bringing This to Your Physician

A defined plan gets a different answer than "should I try keto?" Three questions worth bringing to your next visit:

• "Can we add an ApoB to my next draw, so I have something precise to compare against after I change how I eat?"
• "I'm making one defined change to how I eat for twelve weeks. Which markers should we recheck, and when should I book the draw?"
• "What's my eGFR, and does it put any limit on how much protein I should be eating?"

One reader in this framework — call her Rusudan, 49 — had unremarkable lipids, an ApoB that was fine rather than excellent, and a slightly elevated inflammatory marker from a prior visit. She chose a Mediterranean pattern, changed exactly two things (olive oil as her cooking fat; lentils and oats where bread and potato used to be), and kept everything else the same. Twelve weeks later: ApoB down about a tenth, triglycerides down, hs-CRP essentially unchanged. Unspectacular, and useful — she now has two numbers, twelve weeks apart, produced by two changes she chose. That's what this framework delivers: not the right diet, the ability to find out.

Frequently Asked Questions

Sources

Related reading