TREM2 is a receptor on microglia — the brain's resident immune cells — and rare changes in the gene that encodes it raise Alzheimer's risk substantially. The best-studied variant, p.R47H, roughly triples to quadruples a carrier's risk, placing one copy on about the same footing as one copy of APOE ε4, the common risk gene most readers have heard of. That effect is large but it reaches very few people: R47H is carried by roughly one in 400 people of European ancestry and is close to absent elsewhere. For families, the practical meaning is less about prediction than about biology — TREM2 variation tells researchers that the brain's immune response to amyloid plaques is part of the disease, not a bystander.
Medical information disclaimer: This article is for general educational purposes only and does not provide medical advice, diagnosis, or treatment. Always consult a physician or other qualified health professional about symptoms, medications, tests, or treatment decisions.
Table of Contents
- How much risk does the R47H variant actually carry?
- Two copies are not simply double the risk
- What the variant does to microglia
- Who the finding applies to
- Losing TREM2 entirely causes a different disease
- Why "boost TREM2" has not become a treatment
- Frequently Asked Questions
How much risk does the R47H variant actually carry?
The signal showed up in case-control sequencing. As summarized on Alzforum's TREM2 R47H mutation page, the variant appeared in 22 of 1,091 Alzheimer's cases against 5 of 1,105 controls — an odds ratio of 4.5 — with a replication series at 4.59 and an Icelandic population study at 2.92. Pooling the data narrowed the estimate.
A meta-analysis of coding variants in TREM2 covering 32,598 subjects across four studies put R47H's odds ratio at 4.11 (95% confidence interval 2.99–5.68). Odds ratios are relative, not absolute. A fourfold increase on a lifetime risk in the low teens still leaves most carriers without a diagnosis, and most people who develop Alzheimer's carry no TREM2 variant at all.
Two copies are not simply double the risk
People who inherit R47H from both parents appear to face a far steeper risk than a doubling. A 2023 research letter in the New England Journal of Medicine modeled the odds ratio for homozygous carriers at 97.1 under a full model, against 5.6 under the simple multiplicative assumption.
Read that number with its interval: 23.5 to 401.1. It comes from a model fitted to very few homozygotes, because two copies of a variant this rare is an extremely uncommon genotype. The direction of the finding — a non-multiplicative jump — is more solid than the point estimate.
What the variant does to microglia
R47H is a partial loss of function. In reporter assays described by Wang and colleagues in *Cell*, the variant sharply reduced TREM2 activation by anionic lipids — phosphatidic acid, phosphatidylserine, phosphatidylinositol and sulfatides — without changing how much receptor reached the cell surface.
The receptor is still there; it senses damaged membranes less well. That matters because TREM2 signaling is what keeps microglia massed around amyloid plaques. In R47H models, fewer microglia cluster at plaques, activation is weaker, amyloid-β clearance is poorer and plaque burden grows — a line of work reviewed in Frontiers in Immunology.
Who the finding applies to
Ancestry decides whether R47H is relevant at all: So a negative R47H result from a consumer genetic test is close to uninformative for someone of non-European ancestry, and only mildly reassuring for anyone else, since the baseline carrier rate is already under 1%.
- Allele frequency is about 0.23% in control groups and 0.25% in gnomAD, the large public variant database.
- The Alzheimer's association replicates in European-ancestry cohorts but not in Asian or African-American samples, where the variant is nearly absent.
- In Han Chinese populations a different TREM2 variant, H157Y, carries the risk instead, with an odds ratio of 11.01 (95% CI 1.38–88.05) in a small cohort — note how wide that interval runs.
Losing TREM2 entirely causes a different disease
It is tempting to assume that no TREM2 means severe Alzheimer's. It does not.
Homozygous null variants — p.Q33X, whose transcript is destroyed by nonsense-mediated decay, and c.549del — cause Nasu-Hakola disease (PLOSL2), a recessive leukodystrophy described in *Acta Neuropathologica*. It combines early-onset dementia with polycystic bone lesions and fractures. The distinction is clinically useful: bone involvement alongside young-onset cognitive decline points somewhere entirely different from late-onset Alzheimer's, and the inheritance pattern is recessive rather than risk-modifying.
Why "boost TREM2" has not become a treatment
The loss-of-function story suggests an obvious fix, and it has been tried. Alector announced in November 2024 that AL002, a TREM2 agonist antibody, missed its primary endpoint — change in CDR-Sum of Boxes — in the 381-person Phase 2 INVOKE-2 trial in early Alzheimer's, despite confirmed target engagement. ARIA, brain swelling and microbleeds visible on MRI, occurred mainly in treated participants, and the extension study was stopped. Carrier biology is also less straightforward than the label implies.
A review of soluble TREM2 in cerebrospinal fluid notes that R47H carriers show *elevated* CSF soluble TREM2, possibly a compensatory inflammatory response. A carrier's biomarker panel cannot be read as "less TREM2 activity," which is one reason a simple agonist may not be enough. For a family weighing testing, that is the honest state of play: a confirmed variant changes no treatment decision today, and the usual modifiable factors — blood pressure, hearing correction, sleep, exercise, vascular health — remain where the leverage is. Anyone considering a TREM2 result should take it to a genetic counselor rather than a direct-to-consumer report, because the risk figures above are population odds ratios and say little about one person's timeline.
Frequently Asked Questions
Is TREM2 R47H as strong a risk factor as APOE ε4?
One copy is roughly comparable. The pooled odds ratio of 4.11 puts a single R47H allele about on par with one APOE ε4 allele, though R47H is hundreds of times rarer.
Does carrying R47H mean I will develop Alzheimer's?
No. It raises relative risk several-fold, and most carriers never receive a diagnosis.
Should I ask for TREM2 testing?
There is no treatment that depends on the result, and no approved therapy targets TREM2 — AL002 failed its Phase 2 endpoint in 2024. Discuss any testing with a genetic counselor first.





