Diet: what actually lowers ApoB

Diet is a real lever on cholesterol, but the effect sizes are honest — moderate, not miraculous — and the strongest moves aren't the ones the internet argues about. Here's what actually lowers ApoB, and roughly by how much.

The proven levers

What moves LDL/ApoB, and by how much

Swap saturated fat for unsaturated: replacing fatty/processed meat, butter, and coconut oil with olive oil, nuts, seeds, and fatty fish lowers LDL/ApoB (a ~10–15% swing is achievable). Viscous fibre: oat β-glucan ≥3 g, psyllium 7–10 g, legumes → ~5–10% LDL. Plant sterols ~2 g → ~7–12%. Cut refined carbs and sugar to lower triglycerides and remnant cholesterol and shift LDL toward the larger, less-atherogenic pattern. These stack: the combined 'Portfolio Diet' reaches ~20–30% LDL reduction. Evidence ●●●●○

Patterns that work

The pattern beats any single food

Mediterranean eating has the best outcome evidence — PREDIMED found ~30% fewer cardiovascular events with extra-virgin olive oil or nuts. DASH and, broadly, whole-food, high-fibre, plant-forward diets do the same job. Fatty fish helps triglycerides. No single 'superfood' matters as much as the overall pattern you eat most days.

The honest controversies

Nuance without dogma

Dietary cholesterol (eggs) matters far less than saturated fat for most people. Low-carb/keto can send LDL and ApoB sharply up in some people ('lean-mass hyper-responders') — so if you eat that way, don't assume it's heart-safe: measure your ApoB. Saturated fat's role is real but more modest than 1980s dogma, and the key is the replacement: swapping it for unsaturated fat helps, swapping it for refined carbs doesn't. And processed meat carries a consistent risk signal. The theme throughout: verify with a blood test rather than trusting the theory.

The highest-yield dietary move

For most people: swap saturated fat for unsaturated, add viscous fibre (oats, psyllium, legumes), cut refined sugar — then re-measure ApoB in ~8–12 weeks. Let your own number, not the diet-tribe argument, tell you whether it worked.