Scientists identify a genetic “rescue effect” that lowers Alzheimer’s risk

Researchers from the Department of Brain Sciences at Imperial College London and the UK Dementia Research Institute revealed, in a new study, how TREM2 – a well-known Alzheimer’s risk gene – interacts with CD33, a gene that can reduce this disease risk.

They found that people carrying both the high-risk TREM2 variant and the protective CD33 variant had significantly lower levels of the protein that builds up in the brain in Alzheimer's disease (called beta-amyloid) than those with the risk variant alone. 

“It validates in humans a biological interaction previously only seen in animal models, it identifies CD33 as a potential therapeutic target, and it suggests that the genetic architecture of Alzheimer's disease risk is not simply additive. One protective gene can meaningfully counteract the effects of a high-risk gene, opening new avenues for gene-based or pharmacological strategies,” said Dr Johanna Jackson, Advanced Research Fellow in the Department of Brain Sciences at Imperial College London, Emerging Leader at the UK Dementia Research Institute, and the study’s corresponding author. 

A genetic counterbalance 

Alzheimer’s disease is the most common cause of dementia worldwide, and researchers have long sought to understand which genes increase the risk of developing it.  

One of the most significant factors in this quest is a gene called TREM2. It controls how well microglia (the brain’s immune cells) detect and clear away beta-amyloid, a protein that builds up in the brain and drives neuronal damage in Alzheimer’s disease. 

It validates in humans a biological interaction previously only seen in animal models, it identifies CD33 as a potential therapeutic target, and it suggests that the genetic architecture of Alzheimer's disease risk is not simply additive. Dr Johanna Jackson Advanced Research Fellow in the Department of Brain Sciences at Imperial College London | Emerging Leader at the UK Dementia Research Institute

When TREM2 doesn’t work properly, microglia struggle to clear beta-amyloid, and disease risk rises. Researchers are particularly interested in two rare TREM2 variants: R47H, which raises Alzheimer's risk almost 4-fold, and R62H, which raises it by around 40%. Whether they act the same way at the cellular level has remained unclear until now. 

The Imperial-led study, based on brain tissue from 58 donors, found that people carrying the high-risk R47H variant together with a protective variant of a second gene (CD33) had 2.5 times less beta-amyloid than those with the risk variant alone. In effect, a protective gene can partially compensate for the damage caused by a high-risk one.  

“It is the first demonstration in human brain tissue that the protective CD33 gene variant can partially rescue the increased Alzheimer’s disease pathology caused by TREM2 risk variants,” Dr Jackson noted, as this effect has previously been seen only in mice. 

This points to CD33 as a promising drug target for patients with high-risk TREM2 variants, and some CD33-targeting compounds are already in clinical testing for other conditions. 

Equally surprising is the finding that the two TREM2 variants (R47H and R62H), although both increase disease risk, damage the brain in entirely different ways. This is the first evidence that the two variants act through distinct biological mechanisms, meaning future treatments may need to be tailored to a patient’s specific genetic variant, an approach known as precision medicine. 

“What struck us most was not just the interaction between TREM2 and CD33, but the fact that the two TREM2 risk variants are driving disease through entirely distinct biological mechanisms,” said Dr Nurun Fancy, Edmond and Lily Safra Research Fellow at the Department of Brain Sciences at Imperial College London and the UK Dementia Research Institute and the study's first author. 

Illustration 1. The image shows more 'active' immune cells (marked with red arrowheads) and fewer 'resting' ones (marked with black arrows) in the brains of Alzheimer's patients, compared to healthy brains.

“When we looked at the single-nucleus transcriptomic data, it became clear that this is not a one-size-fits-all story. The cellular responses we observed across microglia, astrocytes, and neurons were variant-specific, which has implications for how we design future therapies. If we treat these variants as interchangeable, we risk missing the very patients who could benefit most from a targeted intervention," she added.  

Future impacts 

The study also revealed that the TREM2 gene affects the behaviour of other brain cells: astrocytes (cells that support neurons) and neurons themselves. This challenges the prevailing view that TREM2-related problems are confined to the brain's immune cells and shows that future therapies should take the whole network of brain cells into account, not just microglia. 

When we looked at the single-nucleus transcriptomic data, it became clear that this is not a one-size-fits-all story. Dr Nurun Fancy Edmond and Lily Safra Research Fellow at the Department of Brain Sciences at Imperial College London | UK Dementia Research Institute

“The finding that astrocytes and neurons, not just microglia, show distinct transcriptional responses depending on TREM2 genotype challenges the prevailing view that TREM2 pathology is purely a microglial story, and broadens the landscape of potential therapeutic targets,” Dr Jackson explained. 

"Our characterisation of non-cell autonomous effects of TREM2 variants provides a powerful demonstration of how cell networks contribute to functional expression of risk even for genes uniquely expressed in a single cell type,” noted Professor Paul Matthews, Edmond and Lily Safra Chair at the Department of Brain Sciences at Imperial College London, the study’s corresponding author. 

The findings support the potential of drugs targeting CD33 to compensate for TREM2-related dysfunction. This strategy is made more feasible by the fact that some CD33-targeting agents already have clinical experience in other conditions. 

The results are also directly relevant to ongoing trials of TREM2-activating antibodies, a proposed Alzheimer's treatment. The study suggests that both TREM2 and CD33 genotype should be factored into how such trials are designed, and that their effects on neurons, not just microglia, also need to be considered. 

Since R47H and R62H cause disease through different biological mechanisms, treatments effective for one variant may not work for the other, meaning future trials may need genotype-specific, precision-medicine approaches to accurately assess whether they work. 

“The study advances understanding of how glial-neuronal interactions contribute to neurodegeneration, with implications extending beyond AD to other diseases where microglial dysfunction plays a role,” Dr Jackson summarised. 

More 

• Fancy, N.N., Willumsen, N., Chau, V.M.N. i in., Mechanisms of increased Alzheimer's disease pathology with R47H and R62H TREM2 variants, Acta Neuropathologica (2026) 151:67, https://doi.org/10.1007/s00401-026-03036-z