Obeticholic Acid for Heart Disease and Alzheimer's

A drug that failed in 2006 for raising blood pressure and killing patients is, two decades later, the reason a small biotech company thinks it can treat both heart disease and Alzheimer's with the same pill. That is the strange arc at the center of this conversation between Peter Attia and cardiologist and lipidologist Michael Davidson, founding CEO of New Amsterdam Pharma. Davidson has spent forty years chasing cholesterol, starting with a personal wound: his father died of a heart attack at forty-seven, when Davidson was sixteen (04:01). His brother, diagnosed with the same abnormal lipids at fourteen, delayed treatment and needed bypass surgery at forty-four. Davidson started statins early and never had an event. He calls this his best natural experiment, and it shapes everything else he says.

The organizing idea is what Davidson calls the eight-gram rule: a rough calculation suggesting that a lifetime cumulative cholesterol exposure of about eight grams is the threshold past which heart disease becomes likely (06:54). Run the math at 200 mg/dL for forty years, or 100 mg/dL for eighty years, or 80 mg/dL for a hundred years, and you land in the same place. The practical translation: keep LDL cholesterol under 80 mg/dL for life, and atherosclerosis mostly does not happen (07:15). Davidson frames this as primordial prevention, stopping plaque before it forms, rather than the more familiar primary and secondary prevention, which intervene after disease is already present.

"It's a lot easier to stop it from forming than it is to reversing it when it already exists." — Michael Davidson [06:05]

Twenty years of a good idea failing

The middle of the conversation is a forensic history of CETP inhibitors, a drug class built around blocking cholesteryl ester transfer protein, an enzyme that shuttles cholesterol out of HDL (the particle often called "good cholesterol") and into LDL. Block it, and HDL rises while LDL falls (16:57). Animals that naturally lack CETP, like dogs and rats, have low LDL, high HDL, and resist atherosclerosis; give a mouse the CETP gene and it develops the human-like lipid profile and starts forming plaque (17:53). That logic drove Pfizer's torcetrapib into massive trials in the mid-2000s. It raised HDL by 75% (21:08), which should have been protective. Instead the trial was stopped for excess mortality. The drug had an off-target effect: it stimulated the adrenal glands to overproduce aldosterone, spiking blood pressure independent of CETP biology at all (22:25). Three more CETP inhibitors followed over the next decade, each proving progressively less dangerous and less useful: Roche's dalcetrapib was safe but had no effect on LDL and no cardiac benefit; Lilly's evacetrapib lowered LDL by 15 to 20% and raised HDL 75%, but the trial was stopped early for futility, a decision Davidson thinks was premature given that similar drugs like ezetimibe took more than two years for benefit curves to separate (28:49). Merck's anacetrapib finally proved the concept: in a 30,000-patient, four-year outcome trial (32:17), a 17% LDL reduction produced a 9% relative risk reduction in cardiac events, in line with what LDL-lowering trials predict regardless of mechanism. The drug itself was shelved anyway, because it accumulated indefinitely in fat tissue (33:09).

This is the graveyard obeticholic acid, New Amsterdam Pharma's drug, is walking out of. It is a far more potent CETP inhibitor, dosed at 10 mg rather than the 100-plus mg range of its predecessors, and in early trials it lowered LDL by roughly 45% while raising HDL by 150% (40:18, 40:41). In patients with familial hypercholesterolemia already on maximal statin therapy, it added a 40% further LDL reduction (46:41). Crucially, large genetic studies (Mendelian randomization) had already shown that raising HDL through most other mechanisms does nothing for cardiovascular risk. The benefit obeticholic acid may offer has to come from its LDL-lowering and metabolic effects, not from HDL itself (41:05). Davidson's own math: for every 38.8 mg/dL drop in LDL, expect about a 22% relative risk reduction in cardiac events, a relationship that holds whether cholesterol is lowered by genetics, statins, or four different drug mechanisms (35:46). New Amsterdam's own outcome trial, called PREVAIL, is underway now and expected to read out within the next year or two (48:47).

The brain has its own cholesterol economy

The second half of the conversation moves from arteries to neurons, and this is where the episode earns its stranger claims. The brain is 2% of body weight but holds roughly 50% of the body's cholesterol (69:13). Critically, none of that cholesterol arrives via LDL: the blood-brain barrier walls off brain lipid metabolism entirely, so the brain relies on its own local system built around ApoE and HDL-like particles (70:26). ApoE comes in three common variants, E2, E3, and E4. Carrying two copies of ApoE4 raises Alzheimer's risk roughly tenfold; one copy raises it about threefold (71:41). Davidson's explanation: ApoE4 impairs the ability of brain support cells (astrocytes) to package and clear cholesterol from neurons, and that stalled clearance eventually triggers the toxic cascade of amyloid plaques and tau tangles that define Alzheimer's pathology (72:05).

This is where CETP loss of function becomes interesting. A study of elderly Ashkenazi Jews (the Bronx Aging Study) found that people who lived into their nineties and beyond were disproportionately likely to carry a CETP loss-of-function mutation, and later work showed that this mutation appeared to erase the added Alzheimer's risk normally conferred by ApoE4 (64:44). Davidson describes mouse experiments in which giving animals amyloid precursor protein produces dementia, adding the CETP gene makes it worse, and then blocking CETP with a drug improves it (65:56). The proposed mechanism is that raising HDL helps clear toxic cholesterol byproducts and deliver antioxidants across the blood-brain barrier, partially compensating for what ApoE4 fails to do (74:33).

New Amsterdam ran a small pilot study of thirteen patients with mild cognitive impairment (MCI) and ApoE4, and measured cerebrospinal fluid markers before and after obeticholic acid. Toxic oxysterols dropped, antioxidant markers rose, and cognition looked stable in a study too small to draw firm conclusions from (77:51). A larger, pre-specified biomarker analysis within the BROADWAY cardiovascular trial then looked at p-tau 217, a blood biomarker that closely tracks amyloid buildup seen on PET scans and predicts progression from normal cognition to MCI (83:17). Across the whole trial population, p-tau 217 fell modestly on the drug; in patients who were ApoE4 homozygotes, the effect exceeded 20% versus placebo, alongside improvement in several other biomarkers (87:37). Davidson is careful to flag the limits here: p-tau 217 is not yet an accepted regulatory endpoint, and biomarker movement is not the same as proven clinical benefit. He raises his own counterargument, that HDL research once looked exactly this promising and turned out not to matter for cardiovascular outcomes (91:26). The company's next study will enroll patients earlier than MCI, on the logic borrowed directly from cardiology: waiting for functional impairment is like waiting for heart failure before treating cholesterol, too late for the intervention to help (92:31).

The conversation closes on two smaller but concrete threads. One concerns omega-3 fats: DHA, the fatty acid that dominates brain tissue, crosses the blood-brain barrier via a specific transporter (MFSD2A) that strongly prefers a particular chemical form of DHA called Lyso-PC, meaning ordinary fish-oil supplementation may not raise brain DHA the way it raises blood levels (99:06, 99:34). The other concerns drug development itself: an average drug now costs three to four billion dollars to reach approval (109:26), and Davidson thinks artificial intelligence's most useful application is not diagnosis but making trials radically cheaper and faster, potentially by isolating patients who actually took their medication as directed rather than diluting results with everyone who did not.