How I cut my cholesterol and ApoB nearly 40% (no statin)

Hi Phoenix friend,

My ApoB started at 115 mg/dL. It's around 70 now, a drop of nearly 40 percent. No statin, ever.
For an APOE4/4 carrier, ApoB may be the single most important number to move against Alzheimer's, and there's a cheap, generic, non-statin way to move it that not enough folks are aware of…

Here's what we will cover today

• First, why cholesterol is a different problem if you carry APOE4 (it changes what your numbers mean).

• Then the three levers in order, with the evidence behind each one and roughly what each is worth.

• Finally, my actual panels, the honest caveats (including where I'm still short), and the APOE4 target ranges you can take to your own doctor.

If you'd rather watch than read, the full walkthrough is on YouTube:

Otherwise, let's get into it.

Why cholesterol is a different problem if you carry APOE4

Here's the part most people miss. APOE4 doesn't just raise your numbers. It changes what they mean.

ApoE is your body's main lipid taxi. ApoE3 and ApoE2 prefer small HDL particles; ApoE4 prefers large, triglyceride-rich VLDL, a different trafficking pattern entirely (Husain et al. 2021). The result: APOE4 carriers tend to run higher cholesterol, and the largest Alzheimer's cohort in the world found total cholesterol significantly higher in carriers (Dunk & Driscoll 2022).

The finding that should change how you think about it: high total cholesterol is a stronger Alzheimer's risk factor in APOE4 carriers than in non-carriers. The authors called low total cholesterol "a critical aspect of preventative care for AD, particularly for APOE ε4 allele carriers."

It cuts both ways. In 2025, Nature Medicine tracked 5,705 people and isolated APOE4/4 homozygotes as their own subtype. In that group, Mediterranean diet adherence modulated dementia-related metabolites more effectively than in the general population (Liu et al. 2025).

Translation: if you're APOE4/4, your food has more leverage, not less. So does every other lever.

My baseline, December 2024

I ran a full panel at a lab in Petaling Jaya. The reference ranges below are the Phoenix Bloodwork Module's APOE4-specific optimal columns, the same numbers our members see when they upload a lab PDF. They're tighter than what your lab flags, because APOE4/4 isn't most people.

Hand this panel to most doctors and you'd hear "borderline, see you in a year."
For a APOE4/4 carrier, borderline is the dangerous word.
Two copies of APOE4 carry roughly 15 times higher Alzheimer's risk (and as an Asian it can go up to 33x), and high cholesterol amplifies that risk more in carriers than in non-carriers.
My ApoB at 96 meant real atherogenic particle burden circulating, possibly feeding amyloid pathology decades before any symptom.

Add the HbA1c at 5.9, officially prediabetic. Carriers are prone to insulin resistance, and insulin resistance amplifies AD risk. Two problems converging. So I built a plan with three levers.

Lever 1: Diet

Mediterranean-leaning. Not the branded program, the actual eating pattern the research tracks. Four moves:

• Cut saturated fat hard. Butter, cheese, fatty red meat down to occasional. Carrier cholesterol responds more to dietary fat than non-carrier cholesterol does (Dunk & Driscoll 2022), so this lever pulls harder for us.

• Upgrade the fats. Extra virgin olive oil as the default. Avocado, nuts, seeds.

• Soluble fiber in every meal. Oats, lentils, beans, barley, apples with the skin on.

• Drop liquid calories and refined carbs. This one probably moved my HbA1c more than anything else I did.

Honest caveat: there's no randomized trial of Mediterranean diet in APOE4/4 homozygotes specifically. We have the 2025 Nature Medicine cohort plus consistent observational evidence. Not RCT-grade proof. But the alternative, eating a standard Western diet with two copies of APOE4, is an experiment I won't run on myself.

Lever 2: Psyllium husk

This is the lever most people overlook. And most people buy the wrong kind.

Psyllium is a viscous soluble fiber. Mixed with water it forms a gel, and in your small intestine that gel traps bile acids. Your body makes bile acids from cholesterol and normally recycles about 95 percent of them. Psyllium breaks the recycling and escorts them out. Your liver has to make more, which means pulling cholesterol out of your blood to do it.

The detail the supplement aisle won't tell you: this only works with viscous fibers. McRorie & McKeown (2017) are blunt that gel-forming fibers (psyllium, beta-glucan, raw guar) lower cholesterol, while non-viscous ones (inulin, wheat dextrin) "do not provide these viscosity-dependent health benefits." If your fiber supplement is inulin, you're not getting this.

How much it moves the needle: a 28-RCT meta-analysis (Jovanovski et al. 2018) found psyllium at around 10 g/day dropped LDL about 13 mg/dL and ApoB about 5 mg/dL, both highly significant. An umbrella review of 108 systematic reviews puts it on the same evidence tier as plant sterols (Schoeneck & Iggman 2021).

The bonus for me: at 5.9 HbA1c, psyllium before meals was also a glucose play. In diabetics it cut HbA1c by nearly a full point, scaling to how high you start (Gibb et al. 2015). Two levers in one scoop.

My protocol: about 10 g/day (two scoops), split before meals, always with a full glass of water, and any medication separated by two hours or more.

Lever 3: Ezetimibe 10 mg (the standout)

I'll say it plainly. Ezetimibe has been the single best lever I've pulled. Stacked with diet and psyllium, it's the biggest reason my ApoB fell from 115 into the 70s in this window. If I had to drop two of my three levers tomorrow, this is the one I'd keep.

It's a once-daily drug, FDA-approved since 2002, now generic and cheap. And it is not a statin. Statins work in the liver, blocking the enzyme that makes cholesterol. Ezetimibe works in the gut, blocking a protein called NPC1L1 that absorbs cholesterol into your blood (Garcia-Calvo et al. 2005). Different mechanism, different side-effect profile. (Worth knowing if you've heard the statin muscle stories: muscle weakness is a statin class effect, not an ezetimibe one.)

Does it work? On its own, ezetimibe drops LDL about 18 percent (Pandor et al. 2009). In the one monotherapy outcomes trial, EWTOPIA 75, it cut cardiovascular events 34 percent in patients aged 75+ (Ouchi et al. 2019). And a 2007 study confirmed it lowers lipids just as well in APOE4 carriers as in everyone else (Mark et al. 2007).

The brain signal, with the caution it deserves. In 2024, a database analysis linked ezetimibe use to a sevenfold lower dementia risk (Ganne et al. 2024). On its own I'd file that under "interesting, unproven." But it didn't stay on its own. In 2025, Alzheimer's & Dementia ran a Mendelian randomization across 1,091,775 people. Those whose genes mimic lifelong ezetimibe (lower NPC1L1 activity) had an all-cause dementia odds ratio of 0.18 per 1 mmol/L lower non-HDL cholesterol (Nordestgaard et al. 2025).

Stay honest about what that is. It's target validation, not proof the pill cuts your dementia risk, and the effect was strongest for vascular dementia, weaker for Alzheimer's specifically. But it lights up the same target as the Ganne signal, from a completely different direction.

And lowering cholesterol in your blood doesn't starve your brain. Your brain makes its own cholesterol behind the blood-brain barrier. As Mahley (2016) put it, "there is essentially no cholesterol that enters the brain from the peripheral circulation." That's why I'm comfortable on ezetimibe as a 4/4. It works on my blood, not my brain's supply.

The 2026 update, fast. Three things shifted this year. The 2026 ACC/AHA Dyslipidemia Guideline (the first US update since 2018) moved ezetimibe to a first-line non-statin add-on and endorsed ApoB to guide therapy. The Ez-PAVE trial (NEJM, 2026) randomized 3,048 patients to LDL under 55 versus under 70; the aggressive arm cut events 33 percent, with ezetimibe the cheap route there. And the SWEDEHEART registry tied skipping ezetimibe to a 1.83 times higher cardiovascular death rate. The last two are secondary-prevention populations, not me, but the direction holds: earlier, lower, ezetimibe-inclusive.

What changed for me wasn't the drug. It was the confidence: I started in late 2024 on far thinner evidence than the same choice has behind it today.

The 8-month snapshot (December 2024 to August 2025)

This is a snapshot of one window, not where I am today. Three panels: December 2024, May 2025, August 2025. Redacted for privacy, otherwise untouched.

ApoB is the number I watch most closely. It counts every atherogenic particle, not just LDL (Sniderman et al. 2019), and for APOE4 carriers it may be the most accurate risk marker we have.

How honest am I being about "optimal"? Against the Phoenix APOE4 ranges, I'm fully inside optimal on only three markers: HDL, triglycerides, and Lp(a). On ApoB, LDL, total cholesterol, and HbA1c I'm close but not in range. Every one would earn a "looks great" from a standard panel, which is exactly why we use tighter ranges.
In this window the stack took my ApoB from 115 to 74.7, with under 60 the target.

Honest caveats

This is one APOE4/4 protocol. It worked for me. It is not a prescription for you. The best way to know if it works for YOU specifically, is to run your own experiment with it.

Don’t know where to start? The Phoenix App was exactly built to help you run your experiments!

Now here are the caveats:

• Three levers at once means I can't cleanly split the credit. Ezetimibe alone usually drops LDL about 18 percent, psyllium adds 5 to 10, diet is variable. My 33 percent is consistent with the stack.

• I'm not on a statin, and that's not a knock on statins. In APOE4 carriers, statin use cut Alzheimer's risk 40 percent; in non-carriers it didn't (Rajan et al. 2024). If your doctor recommends one and you're APOE4, that's evidence-based.

• My ApoB still isn't where I want it. As a 4/4 I aim under 60, and I'm not there yet (today it hovers around 70).

• One number I'm watching: my ALT came back at 64, mildly elevated. Ezetimibe occasionally nudges liver enzymes. Not concerning yet, but I'm monitoring it. If yours rises, tell your doctor.

The APOE4 lipid blueprint

If you carry APOE4, one copy or two, here's what I'd hand you.

Test this panel: full lipids, plus ApoB, plus Lp(a) once (it's genetic, it won't change), plus HbA1c, fasting insulin, hs-CRP, and homocysteine. If your doctor won't order ApoB, push. The 2026 guideline now formally endorses it.

Know your targets. These are the Phoenix APOE4 optimal ranges, tighter than generic lab cutoffs on purpose:

Those guidelines tier risk by heart disease and diabetes, not genotype. I'm a primary-prevention 4/4 aiming low on genotype reasoning. Your cardiologist may set different targets. Bring them your actual numbers.

Start with levers you can hold for years. Diet is lever zero. Psyllium is cheap and well-backed. Ezetimibe is a conversation with your physician. And don't fear statins out of proportion to the data; in APOE4 carriers the fear is louder than the evidence.

Where I am now

That snapshot ended in August 2025. Since then I've gone more aggressive on the same levers (still no statin), and my ApoB is now around 70. The goal hasn't moved: under 60.

I'm sharing these numbers openly. Not to prescribe for anyone. To show you can move the dial without aggressive pharmacology, and that the evidence is stronger than the online noise suggests.

You are not your genotype. You are what you do with it.

If you're just starting, two places to begin: the bloodwork blueprint at apoe4.co/bloodtest walks through which markers to test and how to track them, and the rest of us are inside The Phoenix Community, all carriers, comparing notes and real data.

Talk to your doctor before changing any medication, supplement, or protocol. And tell them you carry APOE4.

Kevin

Citations (full list)

All 30 peer-reviewed citations referenced across this protocol, with PMIDs/DOIs:

1. Jovanovski et al. 2018 · Psyllium + LDL/ApoB meta-analysis (AJCN): 30239559

2. Gibb et al. 2015 · Psyllium + HbA1c (AJCN): 26561625

3. Zhu et al. 2024 · Plantago updated meta: 38688104

4. Ghavami et al. 2023 · Soluble fiber dose-response: 36796439

5. McRorie & McKeown 2017 · Viscous fiber mechanism: 27863994

6. Schoeneck & Iggman 2021 · GRADE umbrella review: 33762150

7. Brum et al. 2018 · Psyllium + statin stack: 30078477

8. Juhász et al. 2023 · Fiber network meta T2DM: 36811560

9. Liu et al. 2025 · Med diet × APOE4/4 (Nature Medicine): 40855194

10. Mark et al. 2007 · Ezetimibe × APOE: 17559752

11. Ganne et al. 2024 · Ezetimibe + ADRD: 39263528

12. Pandor et al. 2009 · Ezetimibe meta: 19141093

13. Ouchi et al. 2019 · EWTOPIA 75: 31434507

14. Garcia-Calvo et al. 2005 · NPC1L1: 15928087

15. Gorbunova & Seluanov 2024 · Ezetimibe/AD commentary: 40124644

16. Rossi et al. 2025 · LLT + cognition: 41400788

17. Rajan et al. 2024 · Statins × APOE4 AD: 38447103

18. Westphal Filho et al. 2025 · Statin + dementia: 39822593

19. Banach et al. 2025 · Statin+ezet mortality: 40126455

20. Cannon et al. 2015 · IMPROVE-IT: 26039521

21. Awad et al. 2018 · Ezetimibe + Lp(a): 29396832

22. Husain et al. 2021 · APOE + AD: 33679311

23. Mahley 2016 · Brain cholesterol independence: 27174096

24. Dunk & Driscoll 2022 · TC + APOE4 AD: 34958023

25. Sniderman et al. 2019 · ApoB superiority: 31642874

26. Lee et al. 2021 · Ezetimibe + high-intensity statin: 33738013

27. 2026 ACC/AHA/Multisociety Dyslipidemia Guideline · Guideline on the Management of Dyslipidemia. Circulation, March 2026. DOI: 10.1161/CIR.0000000000001423

28. Lee Y-J et al. 2026 (presented by Kim B-K) · Ez-PAVE: Randomized Comparison of LDL-Cholesterol Targeting <70 vs <55 mg/dL in ASCVD. NEJM, March 2026 (ACC.26 Late-Breaker), trial NCT04626973. ClinicalTrials.gov | tctmd.com

29. Leosdottir M et al. 2025 · Early Ezetimibe Initiation After Myocardial Infarction in the SWEDEHEART Registry. JACC, April 2025. PMID 40240093 | JACC

30. Nordestgaard LT et al. 2025 · Cholesterol-lowering drug targets reduce risk of dementia: Mendelian randomization and meta-analyses of 1 million individuals. Alzheimer's & Dementia, Oct 2025. PMID 41059729 | DOI: 10.1002/alz.70638