Smoking interacts with APOE ε4 to influence dementia: evidence from UK biobank and validation by meta-analysis of longitudinal studies
Biological Psychiatry Global Open Science, 2026
Cong Y., Wang Y., Huang L., Yang Z., Tan L., Chu X., Xu W.
| Disease area | Application area | Sample type | Products |
|---|---|---|---|
Neurology Environmental Health & Toxicology | Pathophysiology | Plasma | Olink Explore 3072/384 |
Abstract
Background
Gene-by-environment interaction plays important roles in influencing dementia occurrence. Smoking is a well-established modifiable risk factor for all-cause dementia (ACD). However, its interaction effects by apolipoprotein E (APOE) ε4 as well as the underlying peripheral mechanisms are unclear.
Methods
Panoramic smoking burden (PSB) and smoking history score (SHS) were constructed for 173,366 non-demented participants from the UK Biobank (UKB). The associations of PSB and SHS with ACD risk (median follow-up = 13 years), as well as their interactions by APOE ε4 were explored in death-competing risk models. Next, the interaction relationships of smoking by APOE ε4 were validated by meta-analyzing epidemiological evidence searched up to May 2025. Finally, plasma proteomics, bioinformatics, and causal mediation analyses were conducted to unveil the potential mechanisms.
Results
Smoking indexes were associated with ACD risk, with significant multiplicative interactions of PSB (p < 0.001) and SHS (p = 0.031) by APOE ε4. The associations of smoking with ACD were more pronounced among ε4 non-carriers. Meta-regression of 207,887 participants validated that the relationship of smoking with ACD was modulated by APOE ε4 (p = 0.011). Current smoking was associated with ACD risk exclusively among APOE ε4 non-carriers (p < 0.001). Inflammation, angiogenesis modulation, and regulation of endopeptidase activity could be mediating mechanisms.ConclusionsThese findings highlighted the interplay of smoking by APOE ε4 in influencing dementia, which should be considered in the risk stratification and prediction models.