Richard Isaacson and Lauren Miller Rogen on personalizing Alzheimer's prevention for an APOE4 carrier

This is our independent, journal-club style summary of Peter Attia’s conversation with Lauren Miller Rogen and Richard Isaacson (Episode 138). We distill what was said, separate solid evidence from opinion, and flag where it gets murky. You can and should check it against the episode yourself.

Rogen watched Alzheimer’s take her grandfather, grandmother, mother, and uncle. Isaacson runs the Alzheimer’s Prevention Clinic at Weill Cornell. The episode is two things at once: a hard family story, and a working tour of how one clinic builds a personalized prevention plan for someone who carries APOE4.

Why this matters for carriers

Lauren is an APOE3/4 carrier. Her mother developed symptoms around age 52 and attended Lauren’s wedding at 60 in advanced disease. Her uncle, five years older, did not show symptoms until his late 60s or early 70s. Both turned out to be APOE4/4. That contrast is the spine of the episode: the same high-risk genotype, roughly a 15-year gap in onset, and a long conversation about why.

Isaacson’s framing is the useful part for carriers. He stresses, repeatedly, that APOE4 is a risk factor, not a verdict. The genes that actually guarantee Alzheimer’s, presenilin 1, presenilin 2, and amyloid precursor protein, are a different and far rarer category. He says he has seen those in roughly two hands’ worth of patients across 1,500 to 2,000 seen over more than 15 years. Attia adds that these deterministic genes account for far less than 1 percent of all cases. APOE4 is the common one, and in Isaacson’s telling it is also the one he feels he can understand and push back against.

One reframe is worth sitting with. Isaacson argues Alzheimer’s is not one disease but an umbrella, the way “cancer” is. Symptoms track wherever the pathology lands, whether memory, executive function, or visual processing, and different people travel different roads: a lipid road, an inflammation road, a possible infectious road, a head-trauma road. For a carrier, the takeaway is that “APOE4” describes a road, not a fixed destination.

The one hard number to carry with you

Most of this episode trades in relative figures. The speakers give exactly one usable absolute risk, and carriers should anchor on it. Isaacson describes three stages of the disease: preclinical (pathology present, no symptoms), mild cognitive impairment due to Alzheimer’s (noticeable glitches, more than a standard deviation and a half below norm, but the person can still care for themselves), and dementia. From the MCI stage, he puts conversion to dementia at roughly 12 to 16 percent per year depending on the study. That is an absolute annual risk, and it is the concrete number to hold against everything else. It says progression is a yearly probability, not a switch, which is exactly why a multi-year head start on prevention has room to matter.

He also cites a striking figure: 46 million Americans with Alzheimer’s pathology but no symptoms. He flagged it himself as a study estimate he would hand to Attia to “tear apart,” so treat it as a claimed number, not a hard count. And in the closing exchange he updates a slogan. Where he once said one in three cases may be preventable, he now says four in ten cases “may be preventable if that person does everything right.” Notice the conditional. He attaches no absolute baseline risk to either fraction, so these are population framings, not a personal probability.

The carrier-relevant specifics

Women and APOE4. Isaacson says roughly two out of three Alzheimer’s brains are women’s, and that longer lifespan does not fully account for it. His leading hypothesis is the perimenopause transition, a period of bioenergetic shifts in the brain he calls a window of opportunity to intervene. He is careful to label this his interpretation of emerging evidence, and to say perimenopause does not mean a woman will get Alzheimer’s. In Lauren’s mother’s case he points to surgical menopause around age 46 or 47, ovaries removed, abrupt estrogen withdrawal, stacked on top of APOE4/4. Read the hormone-protection idea as a hypothesis he believes strongly, not as established.

Waist, not weight. Isaacson presents, as known research, a 39 percent increased risk of dementia in women with enlarged waist circumference over a certain degree, and frames visceral fat, not weight and not cellulite, as the thing to watch. As the belly grows, he says, the memory center shrinks. He asserts this as established data. One gap: he gives the 39 percent as a relative increase with no absolute baseline, so on its own it is hard to translate into what it means for any one woman. He also suspects muscle mass may matter more in men and visceral fat more in women, but flags that as an open question his own imaging study aims to answer.

Exercise. Presented as the single highest-priority lever, and one he believes is preferentially effective in APOE4 carriers. He separates general activity from structured exercise and singles out high-intensity interval training as possibly the only thing that moves certain cognitive measures in carriers. The general benefit of fitness is well supported. The “preferentially effective in E4” claim is emerging, not nailed down.

Omega-3s. Here the numbers are concrete. He argues carriers need higher doses and that eating fish alone may not be enough, citing work he attributes to Hussein Yassin (a spinal-fluid study) pointing to a need to reach at least 2 grams of DHA alone to get enough into the central nervous system. Lauren takes roughly 2,000 mg of DHA plus EPA, titrated to a red blood cell omega-3 index he wants at least 12 to 14, higher probably better (Lauren’s runs 14-plus). This is one of the better-supported carrier-specific moves discussed.

Vitamin D. Lauren started around 17 to 19, which he calls insufficient. He targets closer to 40 to 50, and possibly 50 to 70 in APOE4/4 carriers, citing a European Journal of Nutrition paper, and notes the practical detail of taking it with a fat-containing meal. The optimal-range numbers are his clinical judgment on top of limited evidence.

Homocysteine. Treated as genuinely modifiable: high homocysteine is associated with brain shrinkage and worse memory, and B-complex (B12, B6, folate) can slow atrophy. He is specific that this works only in people who actually have elevated homocysteine and sufficient omega-3s. That conditional is the point.

The genetics beyond APOE4

Isaacson resists naming a single culprit and leans on “polygenic risk.” Two ideas are worth taking away. First, not all E4s are equal. He points to TOMM40, the gene sitting next to APOE, as something that may make one person’s E4 behave worse than another’s, and he explicitly calls this his belief rather than a certainty. Second, in Lauren he found a TNF-alpha-related variant that, combined with E4, may raise risk further, which is part of why her plan leans on curcumin (for its possible anti-TNF effect) and heavy exercise. Attia draws the parallel to Lp(a) in heart disease, where the same elevated marker tears through one family and barely touches another. The honest summary: this is early, semi-quantified detective work that Isaacson himself calls imprecise.

Lauren’s trajectory

Isaacson reports that Lauren’s memory function started at about 0.87 standard deviations below the mean and is now roughly a standard deviation above it. Separately, he says she scores above the 90th percentile across a range of tests and that her processing speed has improved markedly. These are different measures, described in a somewhat tangled exchange, not one clean before-and-after on a single test, so the fair reading is broad improvement across several domains rather than a single tidy delta. He argues this cannot be fully explained by practice effects, partly because, as Lauren notes, APOE4 carriers are more resistant to practice effects. He is candid about the limits: cognitive testing is imprecise, a bad day (she describes one genuinely terrible visit) drags scores down, and these are not done daily. The blood-based and calculated-risk improvements are the firmer evidence. The cognitive gains are encouraging but, by his own admission, harder to separate from confounders.

Fair warning

This episode is also, in part, a tour of a supplement stack. Lauren is on roughly 21 interventions. Several carry specific, brand-leaning recommendations: a “nanoparticle” curcumin, reserved for higher-risk patients, justified by a single small UCLA study (Gary Small) showing less amyloid accumulation, which Isaacson is careful to call small and whose exact effect size he does not know; cocoa flavanols; magnesium L-threonate, which he himself ranks mid-list on evidence; and high-dose omega-3s. To his credit, he repeatedly states he has nothing to disclose from supplement companies, and he applies a do-no-harm filter: since some of these patients may never get Alzheimer’s, nothing risky should be added. Still, 21 interventions is a lot, and most of their individual effects are unproven in prevention.

Other places where enthusiasm runs ahead of the evidence: the perimenopause-as-prevention-window thesis is a hypothesis; the TOMM40 and TNF-alpha refinements are his clinical interpretation; the “vitamin D to 50-70 in E4/4” targets are judgment calls. Hearing loss is flagged by the 2020 Lancet Commission as a major modifiable risk factor, but Isaacson openly admits he does not know the mechanism, and a noisy correlation with age is hard to rule out. On THC and CBD he is genuinely undecided, calling it a “precision medicine answer” and noting that for every study showing benefit you can find one showing harm. On head trauma in carriers he describes his own view as still shifting. These are honest hedges. Read them as hedges.

One structural caveat: this is a clinic’s own protocol, presented by the clinician who designed it and the patient who is its showcase. That is not a knock. It is a conflict of incentives to keep in mind when a single patient’s improvement is offered as proof of concept.

Bottom line for a carrier

Hold the one hard number first. From mild cognitive impairment, the speakers put conversion to dementia at roughly 12 to 16 percent per year, which means progression is a yearly probability you may be able to push against, not a fixed sentence. The durable, better-supported messages follow from that: APOE4 raises risk but does not seal fate; women carriers appear to face higher risk, with the menopause transition a plausible (not proven) lever; visceral fat, fitness, glycemic control, homocysteine, and omega-3 status are modifiable; and carriers likely need higher omega-3 doses (aim for an RBC index of at least 12-14, which may take roughly 2 g of DHA) and probably higher vitamin D than the lab “normal.” The rest, the 21-item stack, the gene-by-gene tuning, the cognitive-score rebound, is promising and personalized but built on thin or single-study evidence. Bring the omega-3, vitamin D, homocysteine, exercise, and body-composition pieces to your own clinician with real numbers. Treat the boutique supplements as experiments, not conclusions.

This is general information, not medical advice, and our distillation rather than the guests’ exact words. Listen to the episode yourself and talk to your doctor before changing anything.