John Kastelein on CETP inhibition, ApoE4, and a new angle on brain and heart risk

This is our independent, journal-club summary of Peter Attia’s conversation with John Kastelein (Episode 255). We distill what was said, separate solid evidence from opinion, and flag where it gets murky. You can check it against the episode yourself.

Why this one matters if you carry APOE4

Most APOE4 conversations stay parked in the brain. This one takes a longer route. It walks through lipid biology, lands on a drug class with a long history of failure, and then argues that this same class might do something useful for an APOE4 carrier’s brain and heart at the same time. Kastelein is a lipid researcher who has spent his career on familial hypercholesterolemia and lipid-lowering drug development. He is also a co-developer of the drug at the center of the back half of this episode, which matters when you weigh how upbeat he sounds.

First, a quick glossary, because this episode is dense with lipid jargon:

• CETP is a protein in the blood that moves cholesterol out of HDL particles and into LDL particles. Blocking it (a “CETP inhibitor”) keeps more cholesterol on HDL and lowers LDL.
• ApoB is the protein tag on the particles that drive plaque (LDL and its relatives). Lifetime ApoB exposure is the thing that builds atherosclerosis.
• ApoA1 is the main protein on HDL. It is small, and a lot of this episode hinges on the idea that it may slip into the brain.
• Lp(a) is a specially troublesome, largely genetic LDL-like particle that independently raises heart risk.

Familial hypercholesterolemia, briefly

Before CETP, the conversation grounds itself in familial hypercholesterolemia (FH), an inherited condition that drives lifelong sky-high LDL. The figures here are worth keeping. Kastelein estimates FH affects roughly 1 in 250 people, which in his framing would make it the most frequent autosomal dominant disorder in men. In children with high cholesterol, where other causes are rare, his group finds a causal mutation in about 95% of cases, versus only 50 to 60% in adults, where the clinical diagnosis is less reliable. He uses an LDL-C of 190 mg/dL as the cutoff, though his referred patients often arrived far higher, at 300+. Dutch guidelines treat children with definite FH from age 6, with a pediatric LDL target of about 130 mg/dL, a target Kastelein himself calls too conservative. In a case of severe FH, he notes a statin takes off roughly 45% of LDL. He also estimates that about 5% of FH patients, mostly active non-smoking women with high HDL, seem to escape disease entirely, though no one has found a clean biological reason. All of this is established within his large Dutch cohorts.

The CETP story, plainly

CETP shuttles cholesterol out of HDL and into LDL. In an energy-scarce world that was useful, because cholesterol is expensive to build (Kastelein puts the cost at 27 ATP per molecule) and the liver recycles it. Today, with sluggish LDL clearance, loading more cholesterol onto LDL works against us.

People born with naturally low CETP activity tend to live longer with less heart disease, less diabetes, and, per Kastelein, less Alzheimer’s, heart failure, and kidney disease. This is established as an association from human genetics and Mendelian randomization, a method that uses inherited gene variants as a natural experiment to test cause and effect. The size of the effect varies by endpoint, and the causal chain is not equally proven for all of them. Kastelein is candid that low CETP “is still by many people called a longevity gene,” and much of that framing is shared interpretation rather than a settled causal verdict.

The drug history is brutal, and it is the reason to keep your guard up. Pfizer’s torcetrapib raised HDL roughly 70% but, in Kastelein’s blunt telling, “killed more people than it saved” because of an off-target effect that drove up blood pressure (the ILLUMINATE trial, NEJM 2007, 15,067 patients, was stopped early for a 58% rise in all-cause mortality and a roughly 5 mmHg rise in systolic blood pressure). Roche’s dalcetrapib was a weak inhibitor that raised HDL but never moved the event curve, which he frames as the end of the HDL hypothesis (the dal-OUTCOMES trial, NEJM 2012, 15,871 post-acute-coronary-syndrome patients, found no change in cardiovascular events). Kastelein gave two slightly different figures for that HDL rise (30% and 35%); either way, his point is that lifting HDL by roughly a third produced not one fewer heart attack or stroke. Merck’s anacetrapib, in a 30,000-patient trial (the REVEAL trial, NEJM 2017, 30,449 patients on intensive statins), lowered LDL only about 17%. The baseline LDL was 60 mg/dL, so the on-treatment LDL landed around 11 mg/dL, and as Kastelein put it, that predicted a 9% reduction in major events, “and that’s exactly what they got” (major coronary events 10.8% vs 11.8%, p=0.004). He did not say whether that 9% was relative or absolute, so the real-world size of the benefit stays unclear; for the record, REVEAL’s 9% was the relative reduction, against an absolute gap of about one percentage point. The lesson Kastelein draws is firm: CETP inhibition only reduces events through LDL lowering, and it sits on the same line as statins, ezetimibe, and PCSK9 inhibitors. That part is well supported.

There is also a sepsis strand worth knowing. Two research groups, one in Vancouver and one in Leiden, report that people born with a CETP loss-of-function variant are better protected against dying from sepsis. The proposed mechanism is that low CETP keeps HDL high during infection, and HDL appears to act as a sponge for endotoxins, the bacterial toxins that drive septic shock. Kastelein notes the signal is real in the human genetics but that much of the supporting work is in cells and animals, so treat the mechanism as emerging.

The new drug, and the carrier-relevant claims

The current compound, obicetrapib, is described as far more potent than its failed predecessors. In phase 2, at 10 mg, the claims included LDL lowering of about 50% on top of high-intensity statins, an HDL rise of about 165%, and Lp(a) reduction of 56% at 10 mg (43% at 5 mg). Kastelein says no drug-related side effects have shown up across the phase 1 and phase 2 program, and the drug is cheap, projected at $36 per year to manufacture at scale.

Treat these numbers as emerging. They come from phase 2, and from the developer. Phase 3 trials (Broadway, Brooklyn, Prevail, plus an Alzheimer’s trial) are underway. Broadway is described as 2,400 secondary-prevention patients over one year, randomized two-to-one, and not powered for hard outcomes (about 120 events expected, with the team hoping only to see a trend). Brooklyn is a small heterozygous-FH study of 300 patients, which by its size is a biomarker-scale trial, not an outcomes trial. Prevail (about 9,000 secondary-prevention patients, target baseline LDL around 100, exclusion below 55) is the outcomes trial, and Kastelein’s own projection of better than a 20% reduction in major adverse cardiac events is an extrapolation off a regression line, not a result. He did not give that projected benefit as an absolute number, so read it as a directional estimate only.

On diabetes, a meta-analysis Kastelein cites found all four CETP inhibitors reduced new-onset type 2 diabetes by roughly 16 to 20% versus placebo. That is a relative reduction, and the absolute event rates in each arm were not given, so the real-world size stays unclear. The proposed mechanism: more ApoA1 and small HDL particles pulling cholesterol out of pancreatic beta cells, easing the fat overload that kills them. The meta-analysis signal is reasonably solid; the mechanism is emerging.

The APOE4 angle

Here is the part built for carriers. Kastelein’s framing: the ApoE4 protein is a poor cholesterol acceptor and a poor cholesterol deliverer inside the brain. It differs from ApoE3 by a single amino acid (an arginine swapped for a glutamine; the exact position was disputed in the conversation), and that one change reshapes the protein. Cholesterol gets stuck in neurons, oxidizes into oxysterols (oxidized cholesterol byproducts), and drives inflammation and cell death. His proposed workaround is ApoA1, which he says is small enough to slip through the blood-brain barrier where the larger ApoE cannot, and can take over those jobs. The drugs that raise ApoA1 the most are CETP inhibitors. A large Mendelian randomization study, he says, found that low CETP protects APOE4 carriers against Alzheimer’s.

The genetics he recites are familiar: in late-onset sporadic Alzheimer’s, about 65% of patients carry an E4 allele, and he cites roughly 16-fold higher risk for E4/E4 and about 4.5-fold for E3/E4 versus E3/E3. Read these as lifetime relative risks, not annual odds, and remember that penetrance is incomplete. Attia pushes back on the size, saying he reads it as closer to 10-fold and 2-fold, then waves off the gap as not worth arguing. Either way, both stress this is not deterministic: plenty of E4 carriers never develop Alzheimer’s, and a third of Alzheimer’s patients carry no E4. Both also note this is far less penetrant than familial hypercholesterolemia.

The brain trial is a six-month, single-center, proof-of-concept study measuring ApoA1, desmosterol, lathosterol, 24-hydroxycholesterol, and inflammation markers in cerebrospinal fluid. That is a biomarker study, not an outcomes study. The entire brain hypothesis is opinion and extrapolation at this stage, however biologically tidy it sounds.

Fair warning

• Conflict of interest, front and center. Kastelein co-develops this drug with Michael Davidson. The 50% LDL drop, 165% HDL rise, 56% Lp(a) reduction, and “zero side effects” all come from a developer describing his own phase 2 program. He says so plainly and even flags it himself, joking that he does not want to sound like a salesman. Hold the numbers loosely until phase 3 outcomes land.
• No outcomes yet. Every hard claim about preventing heart attacks, diabetes, or dementia rests on regression lines, genetics, and biomarkers, not on a completed event trial for this drug.
• The HDL trap. Notice the irony: this drug raises HDL enormously, yet Kastelein is emphatic that the HDL rise does nothing for events. Do not let a big HDL number stand in for benefit. The case rests on LDL, ApoB, and Lp(a).
• The Lp(a) caveat matters for carriers. A 56% Lp(a) reduction is striking, but it is not yet known whether that translates into fewer events. Nobody can explain the mechanism either, and the comparison numbers (other CETP inhibitors lowering Lp(a) by about 20%) come from the developer’s own work.

Bottom line for a carrier

There is no actionable drug here yet. What you can take away: CETP inhibition is finally being aimed at the right target (LDL and ApoB, not HDL), and there is a genetically plausible reason it might help APOE4 carriers specifically, by raising ApoA1 to do the brain cholesterol work that E4 does poorly. That is a hypothesis being tested, not a treatment. The established lesson is the durable one: lifetime ApoB exposure drives atherosclerotic risk, and for an E4 carrier that risk likely runs a bit higher than ApoB alone predicts, because of inflammation and lipoprotein handling. Watch the Prevail and Alzheimer’s readouts, and keep working the levers you already control.

This is general information, not medical advice, and our distillation rather than the guest’s exact words. Listen to the episode and talk to your own doctor.