The role of inflammation in dementia using Mendelian Randomization: a systematic review

Abstract Background Whilst extensive evidence points towards an association between inflammation and dementia, the relationship between specific inflammatory markers and different subtypes of dementia remain poorly understood. Mendelian Randomization (MR) is a genetically informed causal inference method that is more robust against reverse causation than observational approaches and has been applied extensively in recent years. We conducted a systematic review of the existing MR-based evidence addressing the relationship between inflammation and dementia. Methods We searched 7 databases to identify studies that investigated the causal effect of inflammatory markers on any type of dementia. All results were systematically rated according to pre-determined statistical and methodological considerations. Results 37 studies were included, covering a total of 6795 causal effects, of which most assessed effects on Alzheimer’s disease, followed by vascular dementia and all-cause dementia. Nine distinct exposure – outcome pairs yielded robust or probable evidence for a causal effect, seven of which were assessed by more than one study. We found particularly strong and consistent evidence for a causal effect of the abundance of the transmembrane receptor CD33 (Siglec-3) and different myeloid-derived suppressor cells on Alzheimer’s disease risk. Conclusion Genetic evidence supported a robust potentially causal effect of CD33-related signalling and myeloid-derived suppressor cells on Alzheimer’s disease risk. Although numerous published MR studies report causal effects of inflammatory markers on dementia risk, many studies lack methodological rigour, standardized analysis methods and diversity of samples. Our review provides a systematic overview of current evidence and identifies key areas of improvement for future MR studies in this field.