I was born with about the worst cardiovascular genetics you can get without having a named disease.
One copy of APOE4. Homozygous risk at all three markers of 9p21, the strongest coronary artery
disease locus in the genome. Not one of the common protective variants at PCSK9, the LDL receptor
gene, SORT1, CETP or LPL.
Before I took anything, my LDL was extremely high and my HDL was extremely low. That is the whole
reason this protocol exists.
Today my LDL is 82.6, my inflammation is low, my EKG is clean, and my lipoprotein(a) is 3.43 on a
range of 2 to 53.
This is a record of what I take. It is not written as a recommendation and I give no doses. Talk to a physician who
knows your labs.
The tier that actually lowers LDL
Rosuvastatin
The strongest LDL evidence of anything I take. In JUPITER, 20 mg daily lowered LDL by 50% and cut
the primary cardiovascular composite by 44% in people without high cholesterol but with elevated
CRP (PMID 18997196). Across all statin trials, risk falls in proportion to the absolute LDL drop,
a rate ratio of 0.78 per 1.0 mmol/L (PMID 21067804).
What it does not do is touch Lp(a). Rosuvastatin 20 mg produced a median Lp(a) change of zero
(PMID 24243886). It may even nudge it up, a subject-level meta-analysis found an 11% relative
increase versus placebo (PMID 31111151), though a network meta-analysis disagrees (PMID 33166736).
Worth knowing rather than worrying about.
On the muscle question: across the randomised evidence, only about 1 in 15 muscle complaints in
statin-allocated patients was actually attributable to the drug (PMID 36049498).
Ezetimibe
Adds about 23.6% further LDL reduction on top of a statin (PMID 17659159), though only about 14%
if you are already on a high-intensity statin (PMID 33738013). IMPROVE-IT showed a real but small
event reduction over 7 years and no mortality benefit (PMID 26039521).
It does not lower Lp(a), ten placebo-controlled trials, no effect (PMID 30552391).
Very low side-effect burden, which is why I keep it.
Bempedoic acid
Honest answer: it did very little for me. My genetics do not respond well to it.
The trial data is real, 21.1 percentage points of LDL reduction and a 13% cut in major adverse
cardiovascular events in statin-intolerant patients (PMID 36876740). But no benefit on
cardiovascular death, all-cause death or stroke, and it does not lower Lp(a), nominally
raising it 2.4%, while also lowering HDL 8.3% and raising triglycerides 8% (PMID 36813656).
It is a prodrug activated by an enzyme absent from skeletal muscle, which is the mechanistic reason
it spares muscle (PMID 27892461).
The pharmaceutical tier I do not have, and want
PCSK9 inhibitors, evolocumab and alirocumab
The thing I actually need and have never used. Roughly $500 a month. I bought some over the counter
in another country and left them in a hotel refrigerator.
- LDL down 59% on top of a statin, and a 15% reduction in major cardiovascular events over a
median 2.2 years (FOURIER, PMID 28304224). - Alirocumab: 15% fewer recurrent events after acute coronary syndrome, with a nominal reduction
in all-cause death (ODYSSEY OUTCOMES, PMID 30403574). - And unlike everything above, they lower Lp(a). Evolocumab cut it a median 26.9%
(PMID 30586750). Pooled across all PCSK9-targeted agents, 25.8% (PMID 42025558). - Safety held over 8 years of open-label extension with no excess muscle events, new diabetes,
haemorrhagic stroke or neurocognitive events (PMID 36031810).
For a man whose problem is genetic LDL, this is the tier that matches the problem.
Inclisiran, the twice-yearly siRNA
Same target, different mechanism. Two injections a year after loading. LDL down about 50% in
both phase 3 trials (PMID 32187462), and about 22% lower Lp(a) (PMID 40618457).
The honest limit: it has no cardiovascular outcome data at all. The first outcomes trial,
ORION-4, does not report until 2027 (PMID 42567431). It lowers the number beautifully. Whether that
prevents events is unproven.
What lowers Lp(a) and what does not
Across 147 randomised trials, this is the cleanest summary available (PMID 40618457):
| Drug class | Effect on Lp(a) |
|---|---|
| Statins | No effect |
| Ezetimibe | No effect |
| Bempedoic acid | No effect |
| Omega-3 fatty acids | No effect |
| Fibrates | No effect |
| PCSK9 monoclonals | −29% |
| Inclisiran | −22% |
| CETP inhibitors | −46% |
| Niacin (extended release) | −37% |
Extended-release niacin lowers Lp(a) 22.9% in its own dedicated meta-analysis (PMID 27733255).
Note that this belongs to niacin, not to nicotinamide riboside or NMN, which are different
molecules and do not do this.
The caveat that applies to every row. No trial has ever tested Lp(a) lowering as its own
endpoint and reported results. The dedicated outcome trials are still running. Every one of these
numbers is a change in a marker. None of them is yet a proven change in outcome.
The research-compound tier
GW-501516 (cardarine), the biggest single change I made
This is the one I credit most, and the human lipid data behind it is better than most people
realise.
In the largest human trial, 268 subjects with low HDL over 12 weeks (PMID 22814748):
- HDL up 16.9%
- Triglycerides down 16.9%
- ApoB down 14.9%
- LDL down 7.3%
- Free fatty acids down 19.4%
A separate 2-week study in overweight men found triglycerides down 30%, apoB down 26%, LDL down
23%, with liver fat reduced, but HDL unchanged in that one (PMID 18024853). The first-in-man
trial did show an HDL rise (PMID 17110604). So the HDL effect is real but not uniform across trials,
and I would not state it flatly.
Now the part that has to travel with it.
GSK stopped developing this compound after rodent studies linked it to widespread tumour
development (PMID 24184294). The primary toxicology package is not in the indexed literature, so
the specific claim people repeat, cancers in multiple organs in rats, cannot be verified to the
standard of species, dose and duration. What is published:
- It accelerated intestinal adenoma growth in tumour-prone mice, a fivefold increase in polyps over
2 mm (PMID 14758356). - In mice engineered to overexpress PPAR-delta, it cut mammary tumour latency from 12 months to 5.
It did not produce tumours in normal mice on the same diet (PMID 23811944). - It enhanced colorectal tumorigenesis in APC-mutant mice (PMID 30679176).
- Counter-evidence exists: it inhibited growth of human breast cancer and melanoma cell lines in
vitro (PMID 18054822).
There is also a published human case of severe rhabdomyolysis and liver injury with cardarine taken
alongside ostarine, ALT 922, AST 2558, CPK 86,435 (PMID 34678947).
And nobody has ever measured it against Lp(a). Verified absence: zero records.
Banned by WADA at all times since 2009.
I take it with my eyes open. That is a different thing from telling you it is safe.
SR-9011
I take it. I will not pretend there is evidence.
There is no human data on SR-9011 of any kind, no trial, no safety dataset, no lipid numbers.
The only lipid observation in any species is a figure in a mouse paper with no numbers attached
(PMID 22460951). And the wider SR-90xx literature has a problem: SR9009’s effects persist in cells
lacking both REV-ERB receptors, so the mechanism is not what the label says (PMID 31127047).
Unmeasured means exactly that: nobody has looked. It leaves the question open in both directions.
Citrus bergamot
Helpful in my experience, less powerful than the cardarine. I did not find Lp(a) data on it.
What I would add from the catalogue, and what I would not
Some of these are things I take. Some are things the evidence supports for someone with a different
problem than mine. I have marked which is which.
L-carnitine, the most interesting entry, with a catch that applies to me specifically
This has the single best Lp(a) evidence of any supplement. Pooled across 7 randomised trials,
Lp(a) fell 8.82 mg/dL (95% CI −10.09 to −7.55), and 9.00 mg/dL for the oral form
(PMID 26754058). An independent meta-analysis by a different team reproduced it at −7.13 mg/dL
(PMID 30850271). In mixed hyperlipidaemia on top of simvastatin, 2 g/day cut Lp(a) 19.4% where
the statin alone did nothing (PMID 27914033).
There is even hard outcome data in a different setting: pooled across 13 trials in acute myocardial
infarction, L-carnitine was associated with 27% lower all-cause mortality and 65% fewer
ventricular arrhythmias (PMID 23597877).
Why it does not apply to me, and this matters. Every one of those trials enrolled people
selected for high Lp(a), entry criteria above 20, 30, or 40 mg/dL. The effect is reported as an
absolute milligram reduction. A 9 mg/dL drop is arithmetically impossible in a man whose Lp(a) is
3.43. And the largest trial of all, 229 statin-treated diabetics whose Lp(a) was not selected
for being high, median 9 to 11.5, was completely null (PMID 29696241).
So: if your Lp(a) is high, this is the most evidence-backed supplement on this page. If it is
already at the floor, there is nothing for it to do.
Three more honest limits. The effect does not scale with dose or duration (PMID 26754058). It
does not lower ApoB, the pooled estimate is a non-significant increase (PMID 31255604). And the
Lp(a) meta-analysis shows statistical evidence of publication bias, Egger p=0.04.
The TMAO argument, and the answer to it. Gut bacteria convert carnitine to TMAO, and TMAO
associates with cardiovascular events, 7.6% higher relative risk of death per 10 μmol/L
(PMID 29020409). Human supplementation at 1500 mg/day for 24 weeks raised TMAO roughly tenfold
(PMID 30485835). It is fully reversible four months after stopping (PMID 31200429).
But the causal step is not established. Mendelian randomisation does not support TMAO or
carnitine causing cardiovascular disease, the arrow appears to run backwards, with diabetes and
kidney disease raising TMAO (PMID 31167879). And eating fish raises TMAO 46 to 62 times more
than beef or eggs, without any of the associated risk (PMID 27377678). In dialysis patients, oral
carnitine raised TMAO and simultaneously lowered vascular injury markers (PMID 25636076).
I think the TMAO objection is weaker than it looks. I also think anyone repeating it should have to
explain the fish.
Telmisartan, if you also have blood pressure to manage
An ARB with a real metabolic side. It lowers blood pressure about 4.0/2.2 mmHg versus placebo
(PMID 18757085) and is equivalent to ramipril for major cardiovascular events (PMID 18378520).
It improves insulin sensitivity and glucose parameters in metabolic syndrome (PMID 23523138, PMID 19708000), and in type 2 diabetics over 12 months it lowered total cholesterol 9% and LDL
11.5% versus a calcium blocker (PMID 15476904).
The PPAR-gamma partial agonism everyone cites is in vitro and rat data only (PMID 15007034).
ARBs do not change Lp(a) (PMID 16532712).
Nebivolol, if you need a beta blocker and care about your panel
Beta blockers as a class raise triglycerides by about 31 mg/dL (PMID 7992988). Nebivolol appears to
escape that. Head to head against atenolol over 12 weeks, atenolol raised triglycerides 19% and
raised Lp(a) 30%, while nebivolol did neither (PMID 12808486).
It raises endothelial nitric oxide where atenolol does not (PMID 11479245), does not worsen insulin
sensitivity the way metoprolol does (PMID 22353614), and reduces death or cardiovascular
hospitalisation in elderly heart failure patients (PMID 15642700).
Be precise about the claim: it is lipid-neutral, not lipid-improving (PMID 9416986).
Thyroid hormone, powerful, and the one I would be most careful with
If you are genuinely hypothyroid, correcting it is one of the biggest lipid interventions there is.
In overt hypothyroidism, levothyroxine lowered LDL 41 mg/dL, ApoB 34 mg/dL and Lp(a) 5.6 mg/dL, though it also lowered HDL 4 mg/dL (PMID 32954428). In subclinical hypothyroidism the LDL benefit
is smaller, about 8.9 mg/dL (PMID 27699011).
The reason I flag it hardest. In euthyroid people, being in the highest quartile of free T4
within the normal range carried a 45% higher risk of atrial fibrillation (PMID 29061566). And
on long-term thyroxine, a suppressed TSH carried a doubled fracture risk and 37% higher
cardiovascular risk (PMID 19906785, PMID 26010634).
Fix a deficiency. Do not push a normal number.
Clomiphene and enclomiphene
I used clomiphene, alongside hCG and hMG, to have children while shut down. Worth knowing it also
touches lipids: clomiphene 25 mg/day roughly doubled endogenous testosterone and lowered total
cholesterol 11 mg/dL (PMID 22951175). Enclomiphene raises testosterone and preserves sperm counts
unlike topical testosterone (PMID 26496621), though it showed no significant lipid effect
(PMID 23875626).
Given the argument in my full panel article, that testosterone may lower Lp(a), raising your own production
rather than replacing it is a route worth asking your doctor about.
Oleoylethanolamide
Lowers triglycerides about 17.7 mg/dL pooled across 13 trials, plus waist circumference and fasting
glucose (PMID 40661161). It does not change total cholesterol, LDL or HDL (PMID 40469682). No
Lp(a) data exists.
A triglyceride and body-composition tool. Leave it out of the cholesterol column.
What I would not add to a cholesterol protocol
Being straight about the ones that do not earn a place here:
- NAD+ and nicotinamide riboside. It raises NAD+ dose-dependently and is safe (PMID 31278280),
but 2,000 mg/day for 12 weeks did nothing to insulin sensitivity, lipolysis or lipid
oxidation (PMID 29992272), and nothing to glucose tolerance (PMID 31390002). Take it for other
reasons. - Glutathione. Oral dosing does raise body stores (PMID 24791752), and one 16-person
sublingual trial lowered total cholesterol and LDL with no effect sizes published (PMID 28526381).
The LDL and Lp(a) reduction people cite is from mice (PMID 25463112). N-acetylcysteine
produced no change in human Lp(a) (PMID 1836221). - SR-9011, on the evidence. I take it anyway and I have told you exactly why that is a bet
rather than a conclusion.
The boring tier that I actually rely on
Fish oil, vitamin D, magnesium. And the thing I keep telling myself to do and do not do enough of:
cardio. It is genuinely excellent for heart health and it is a little ridiculous that I take
handfuls of supplements instead of doing it. I should be doing both.
Get these drawn
A full lipid panel with HDL. A CBC for hematocrit and hemoglobin. A liver panel, and take a
2-day break from training first, because exercise raises AST and ALT. LH and FSH, which almost
nobody orders. A metabolic panel with creatinine and eGFR. D-dimer if your hematocrit runs high.
And Lp(a), once in your life. Most doctors will not order it unless you ask for it by name.
Education and personal documentation. Not medical advice, not a protocol for you, and not a
recommendation to use any compound. No doses are given anywhere on this page deliberately. Several
compounds discussed are not approved for these uses by Western governments. Talk to a physician who
knows your labs.
The bloodwork and genetics this protocol answers to are in my full panel after 14 years on gear. For everything else on the hormone side, start at the testosterone hub.