Researchers led by Matthew J. Lennon at UNSW Sydney's Centre for Healthy Brain Ageing screened 12,224 potentially druggable genes against the genetics of vascular dementia and found four that passed their statistical threshold: TOMM40, APOE, ERAP1 and SAA1-4. The work, published in Alzheimer's & Dementia: Translational Research & Clinical Interventions on 22 April 2026, used two-sample Mendelian randomisation, a method that uses inherited genetic variants as stand-ins for lifelong differences in a protein or gene's activity. All four hits came from protein-level data; none of the gene-expression analyses produced a significant result.
The evidence was uneven across the four. TOMM40, measured in cerebrospinal fluid, was associated with vascular dementia (P = 3.67E-36) and with imaging markers of cerebral small vessel disease, including white matter hyperintensities (P = 0.0001) and fractional anisotropy (P = 0.0028). APOE, measured in brain tissue, was associated with vascular dementia (P = 1.90E-54) and with three small vessel disease markers. ERAP1 and SAA1-4 showed much weaker associations (P = 0.0163 for each), did not colocalise, meaning the authors could not show the same underlying genetic variant was driving both the protein change and the disease risk, and were not linked to any imaging marker. The authors describe ERAP1 and SAA1-4 as novel and involved in immune regulation and inflammation.
The finding that complicates the picture is where the strongest signals sit. TOMM40 and APOE are neighbours on chromosome 19 and are among the best established genetic risk factors for Alzheimer's disease, and the authors themselves frame the result as support for shared biology between the two conditions, particularly lipid metabolism, inflammation and mitochondrial function. Because vascular dementia cases in genetic datasets are diagnosed clinically rather than confirmed at autopsy, and because Alzheimer's changes and vessel disease very often coexist in the same brain, a signal at the APOE locus cannot be cleanly separated from Alzheimer's copathology by this design. The paper also notes that limited replication across tissues and smaller vascular dementia GWAS samples reduced its statistical power compared with earlier scans in Alzheimer's disease and small vessel disease. For scale, the largest published genome-wide meta-analysis of the condition, in 2024, drew on 800,597 individuals but only 8,702 vascular dementia cases.
There are no approved disease-modifying treatments for vascular dementia, which the researchers note accounts for a substantial share of dementia worldwide. Lennon said the study "provides an initial step by identifying biologically plausible targets", and the team stressed that turning genetic findings into safe, effective treatments is long and uncertain, with many candidate targets never producing a usable drug.