HT-4253: The LRRK2 Drug That Copies APOE4 Carriers Who Never Get Alzheimer's

Carrying one copy of APOE4 lifts your lifetime Alzheimer’s risk from a baseline of roughly 10 to 15 percent to somewhere around 20 to 30 percent, a 2 to 4 fold jump over the common ε3 version. Two copies push it higher still. And yet a stubborn minority of ε4 carriers reach their 80s with sharp minds and clean scans. Researchers have spent nearly a decade asking what protects them, and one answer keeps surfacing: a gene called RAB10. Halia Therapeutics has now built a drug, HT-4253, meant to copy that natural protection in a once-daily pill, and at the 2026 Alzheimer’s Association International Conference it laid out the case in two posters. Clever idea. Also, so far, mostly an idea.

What Halia actually presented at AAIC 2026

The two posters are a preclinical package plus a trial blueprint, not a “the drug works” announcement. One gathers the lab evidence for HT-4253. The other maps a Phase 2a study designed to test whether the drug can delay or prevent Alzheimer’s in ε4 carriers. Having cleared a Phase 1 trial that raised no safety alarms, Halia built the next study around the United Arab Emirates’ national genetics program to find cognitively healthy ε4 carriers early, before symptoms show.

Read the fine print and the stage is clear. Phase 1 tells you a drug is tolerable and reaches the brain. It tells you nothing about whether it prevents dementia. What Halia brought to the conference is a safety readout, a mechanism, and a plan.

The mechanism: LRRK2, Rab10, and a lucky mutation

Picture your neurons running a busy recycling and waste-hauling operation. Small proteins called Rab GTPases act as dispatchers, routing cargo through the cell’s internal delivery and disposal system. One enzyme, LRRK2, tags a dispatcher named Rab10 with a phosphate group under stress, producing the switched form known as phospho-Rab10. When LRRK2 runs hot, phospho-Rab10 piles up, and the cell’s cleanup crews (the lysosomes, which act as internal recycling centers) and its immune sentries get overwhelmed. HT-4253 is a small molecule that crosses into the brain and blocks LRRK2, which drops phospho-Rab10 and is meant to calm both the overactive immune response and the backed-up disposal line.

Halia treats this hub as a control valve for inflammation and goes after it from two directions: LRRK2 from above and an enzyme called NEK7 from below. Both feed the NLRP3 inflammasome, a protein complex inside immune cells that switches on inflammatory signaling and has been tied to Alzheimer’s, cancer, and heart disease.

The RAB10 idea traces to a 2017 study by Perry Ridge and colleagues in Genome Medicine. They studied people who stayed cognitively sharp well past 75 despite the odds stacked against them, including age and the APOE ε4 allele. A rare variant in RAB10 conferred meaningful protection, with an odds ratio of 0.59 in the discovery group and 0.69 on replication in an independent group. In a dish, dialing RAB10 down lowered Aβ42 and the Aβ42/Aβ40 ratio (two amyloid readouts) in neuroblastoma cells, and RAB10 ran high in human Alzheimer’s brains.

One honest wrinkle: the original resilience data pointed at amyloid, less RAB10 meaning less Aβ42, while Halia’s current framing leans on neuroinflammation and the inflammasome. Both sit at the same Rab10 hub, so this reads less as a contradiction than as two windows on one pathway. But it does mean the exact “how it protects” story is still being written.

How strong is the “never get Alzheimer’s” claim?

Softer than the headline. An odds ratio of 0.59 is not immunity. It works out to roughly a 30 to 40 percent lower odds of Alzheimer’s, from a rare variant, in a modest sample, replicated once. That is a real and interesting signal. It is also a genetic difference present from birth, not proof that nudging the same pathway with a drug in midlife lands the same benefit.

Here is the risk landscape the drug is trying to bend, with the usual caveat that absolute lifetime figures are estimates that vary by age, sex, and ancestry:

Genotype	Relative AD risk vs ε3/ε3	Approx. lifetime risk (to ~85)
ε2/ε3	About half	~5 to 8%
ε3/ε3 (reference)	1x	~10 to 15%
ε3/ε4 (one copy)	~2 to 4x	~20 to 30%
ε4/ε4 (two copies)	~8 to 12x	up to ~50 to 60%

Carrying one ε2 roughly halves your risk, which is the mirror image of what HT-4253 hopes to manufacture: a smaller number, chemically. For more on how these numbers are built, see APOE4 and Alzheimer’s risk and, if you carry two copies, is Alzheimer’s inevitable for homozygotes.

What we still don’t know

Quite a lot, and it is the important part.

There is no human efficacy data. Phase 1 measured safety, tolerability, and drug levels, full stop. The Phase 2a study will read out blood biomarkers, not cognition or dementia diagnoses. Those who qualify take HT-4253 once daily for 48 weeks while researchers watch whether the biomarkers move in the right direction. That would be encouraging, but the graveyard of Alzheimer’s drugs is full of compounds that shifted a biomarker and did nothing for patients.

LRRK2 also carries baggage worth knowing. It is best known as the most common genetic cause of Parkinson’s disease, and LRRK2 inhibitors as a class have been pushed furthest there. That gives HT-4253 a head start on safety experience, but it also means the long-term effects of dialing this enzyme down for years are still being characterized. For the APOE4 and Parkinson’s overlap, see APOE4, Lewy body dementia, and Parkinson’s. Add the amyloid-versus-inflammation question above, and the tidy “mimics the resilient” narrative gets less tidy. This is a promising target with a plausible mechanism, early human safety, and zero proof of prevention. Hold it there.

The trial, and whether you can join

For now, almost certainly not, unless you live in the United Arab Emirates. The study, NCT07399171, is a multicenter, randomized, double-blind, placebo-controlled Phase 2a trial, listed as not yet recruiting with an estimated start in April 2026. The recruitment engine is unusual and, frankly, ambitious. Candidates come from the Emirati Genome Programme, the country’s population-wide genome sequencing effort, which lets researchers pull out symptom-free residents who carry ε4. Abu Dhabi’s Department of Health then handles outreach, texting the people who match and inviting them to screening. A blood test from C2N Diagnostics, PrecivityAD2, narrows the field further by flagging those whose results suggest amyloid is beginning to build. To understand what those tests actually measure, start with blood-based biomarkers for Alzheimer’s.

What to do now

You cannot buy HT-4253, and you should not wait on it. The good news is that the fire it targets, chronic neuroinflammation and sluggish amyloid clearance, responds to levers you already control. A reasonable carrier’s playbook:

• Know your exact genotype and your numbers. One ε4 versus two changes the stakes. Pair that with a lipid panel read the right way (how to read your lipid panel) and blood pressure, since vascular health feeds brain health.
• Train aerobically and lift. Exercise is the closest thing we have to a broad anti-inflammatory and metabolic drug for the brain. See exercise and APOE4 and Zone 2 and VO2max.
• Defend sleep. Deep sleep appears to be when the brain’s glymphatic system clears metabolic waste, including amyloid, though most of that evidence so far comes from rodents and the human data are thinner. Details in sleep and APOE4.
• Eat to lower the metabolic and inflammatory load. Mediterranean and MIND diets and watching saturated fat are the best-supported dietary moves for carriers.
• Ask a clinician about prevention trials generally, not just this one. Our guide is finding an Alzheimer’s prevention trial. To see where drugs like this sit in the wider effort, read the APOE-targeted therapies horizon.

Common questions

Is HT-4253 an anti-amyloid drug like lecanemab? No. It works upstream, quieting a trafficking-and-inflammation pathway rather than stripping plaque directly. Because it does not pull amyloid off blood vessels the way antibody drugs do, you would not expect it to cause the brain swelling and small bleeds (known as ARIA) those drugs can trigger. That is a reasonable inference from the mechanism, not something the trials have shown, so treat it as a hypothesis rather than a proven advantage. For that separate class, see anti-amyloid drugs and APOE4.

Do the “resilient” carriers really never get Alzheimer’s? That framing oversells it. The RAB10 variant tilts the odds, it does not zero them out. Ridge and colleagues themselves described RAB10 as a promising target for prevention, which is a hypothesis, not a finished result.

When will we know if it works? The Phase 2a study estimates primary completion around May 2027, and even then the endpoints are biomarkers, so a real answer on preventing dementia is years further out.