Here is a contradiction worth sitting with.
One camp says the case is closed: elevated apoB-containing lipoproteins are "no longer a hypothesis, but an established fact" as a cause of atherosclerosis. The other camp has just run the most sophisticated imaging study its side has ever produced — and found that in their specific population, apoB did not predict how plaque grew over a year.
The trick is understanding why the second one does not rescue the gamble people want to make with it.
Start with the body count, because lipid mythology has one.
A 70-year-old reader of Peter Attia's *Outlive* wrote in with a line that should stop you: "you saved my life." Her lipid panel looked clean except for one thing — HDL cholesterol over 100 mg/dL for five years.
She had been taught, like nearly everyone, that this was the "good" number, the free pass. Per E5, she reported no chest pain, no shortness of breath, no high blood pressure.
She pushed for a CT angiogram anyway. She suspected the cardiologist agreed to it "just to shut me up," since she had none of the classic symptoms.
The scan showed 50% stenosis of the left anterior descending coronary artery. Both she and her doctor were surprised.
That is what lipid folk-wisdom costs. "Good cholesterol" was a 1970s shorthand for a real population association: low HDL-C tracks with higher cardiovascular risk. But the HDL-C on your panel says nothing about how well those particles function, and — critically — every attempt to lower risk by raising it has failed.
Niacin, fibrates, and multiple CETP inhibitors including torcetrapib all raised HDL-C substantially in large outcome trials without delivering the expected drop in cardiovascular events. High HDL-C mostly reflects a favorable metabolic context; it does not manufacture protection. The reader reportedly nearly learned that the hard way.
Now hold that story next to the newest, most refined version of the same wishful thinking.
The settled science first, so the stakes are clear.
The claim that apoB-containing lipoproteins cause atherosclerosis is not one line of evidence. It is a triangulation, and each leg would persuade on its own.
Borén and Williams put it flatly in 2016: "Today, it is no longer a hypothesis, but an established fact, that increased plasma concentrations of cholesterol-rich apolipoprotein-B (apoB)-containing lipoproteins are causatively linked to ASCVD." That is not advocacy language. That is a field describing a solved problem.
The mechanism is boringly simple: apoB is the single protein that wraps each atherogenic particle — LDL, VLDL, IDL, chylomicron remnants. Roughly 90 to 95 percent of the circulating apoB pool rides on LDL particles. Those particles get retained in the arterial wall, and atherosclerosis follows. That relationship holds regardless of how or why the particles got elevated.
Then come the human experiments. Ezetimibe and anti-PCSK9 antibodies — both non-statins — lower events to the degree you'd predict from the LDL lowering alone. If cholesterol lowering only worked through statins, that convergence would be impossible.
But the cleanest natural experiment belongs to Helen Hobbs. She went looking for people born with broken PCSK9, the protein that destroys LDL receptors, and found them.
Carriers of less severe (loss-of-function) mutations had just a 28% reduction in LDL-C — and an 88% reduced risk of heart attack, cardiovascular death, or revascularization.
Read that ratio again. A modest lifelong reduction in apoB burden produced a near-elimination of coronary events. That is the dose-duration effect of exposure integrated over decades, and no randomized drug trial can replicate it because no trial runs for sixty years. It is as close to a genetic proof as cardiovascular biology offers.
A modest lifelong reduction in apoB burden produced a near-elimination of coronary events — as close to a genetic proof as cardiovascular biology offers.
So when someone argues that high LDL is safe for a given person, they are not arguing against a fragile hypothesis. They are arguing against Mendelian randomization, against two independent drug classes, and against people whose arteries have been protected since birth.
Which brings us to the pushback — and here I want to be fair, because the LMHR camp just did something their critics should respect.
The lean mass hyper-responder phenotype is real as a description. It emerged from a 2021 survey: lean people who adopt a carbohydrate-restricted diet and develop LDL-C ≥200 mg/dL alongside HDL-C ≥80 and triglycerides ≤70. The rise is large — reportedly averaging 143 mg/dL per the Mindrum commentary — and it clusters in the metabolically healthy, the inverse of typical dyslipidemia.
Critics pushed back hard and correctly on the original paper. TG:HDL ratio is not an established causal factor the way apoB is; BMI measures obesity risk, not leanness; and the framing risked leaving readers with an impression of safety the data could not support.
The obvious rebuttal to all of it was: stop surveying, go image the arteries. So they did.
The Keto CTA study followed 100 LMHR and near-LMHR participants for one year with coronary CT angiography, using prespecified plaque quantification and blinded AI confirmation. Fifteen percent of participants showed plaque regression despite sustaining elevated LDL-C (mean 242 mg/dL) and ApoB (mean 180 mg/dL).
The headline finding deserves an honest hearing: neither apoB nor cumulative LDL-C exposure predicted plaque progression in this group. Baseline plaque did. Fifteen percent of participants actually showed regression despite those sky-high numbers.
That is real signal. It is the strongest thing this camp has produced, and pretending otherwise is intellectually dishonest. In this cohort, over this window, the particle count did not track the plaque change.
Now the epistemics — because this is exactly where the gamble gets sold.
First, this is a surrogate endpoint. Plaque progression on CT over one year is not a heart attack. Hobbs's 88% is measured in events and deaths, integrated over lifetimes. You cannot overturn hard-outcome, decades-long data with a one-year imaging change in 100 people.
Second, look at what actually predicted progression: baseline plaque. That is not a rebuttal of apoB causality — it is a confound that swallows the study. These participants were relatively young and metabolically healthy; 57% had zero coronary calcium at baseline. A one-year window in low-baseline-plaque people simply may not be long enough for particle burden to declare itself, which is precisely why the genetic data — sixty years of exposure — matters so much.
Third, "apoB didn't predict progression within this narrow, high-apoB, low-plaque cohort" is a very different claim from "high apoB is safe." Restricted variance in a selected subgroup routinely flattens associations that are overwhelming across the full population. That is a statistical artifact, not a biological exemption.
So the honest verdict is neither mockery nor vindication. It is this: a contested subgroup, with genuinely interesting one-year imaging data, and no outcome data at all.
The authors themselves note it is methodologically inappropriate to assume general-population risk models transfer cleanly to this phenotype. Fine — but that blade cuts both ways. It is equally inappropriate to assume exemption from a causal biology proven across every other population ever studied.
The whole argument reduces to an asymmetry of the wager.
If the consensus is right and you treat your apoB as safe, you accumulate plaque you could have prevented — and, per the reader with the 50% LAD lesion, you may not feel a thing until you do. If the LMHR optimists are right and you lower apoB anyway, you lose essentially nothing; peripheral lowering doesn't starve tissues, and per E4, the brain runs its own separate cholesterol economy through apoE, so it isn't deprived either.
One side of that bet has an 88% risk reduction and a century of retention biology behind it. The other has a one-year plaque study with heavy baseline confounding and zero events.
That is not a close call. Take the phenotype seriously as science. Do not bet your arteries on it.