Abstract
OBJECTIVE: The association between cerebral microbleeds (CMBs), an indicator of microvascular damage on MRI, and incident dementia (ID) remains inconclusive. We aim to investigate the interplay of CMB presence, number, and location with ID in a large community-based elderly cohort. METHODS: The study included 1532 dementia-free participants (aged ≥ 65 years) derived from baseline examinations (2012-2013) of the Atherosclerosis Risk in Communities Neurocognitive Study (ARIC-NCS). The ID cases were diagnosed at visits 6 (2016-2017), 7 (2018-2019), and 8 (2019-2020). Cox proportional-hazards models were deployed to assess the association of CMB presence, number, and location with ID. RESULTS: Over a nine-year follow-up (median = 8.1 years), 390 participants developed ID. Microbleed (MB) presence was related to an increased ID risk (HR = 1.24, 95% CI = 1.02-1.51). A stronger association was observed in those with mixed (lobar+subcortical) MBs (HR = 1.60, 95% CI = 1.10-2.30). Compared to participants without MBs, those with ≥ 2 CMBs in any MBs, any lobar MBs, and any subcortical MBs were associated with 57%, 174%, and 74% higher ID risks. Furthermore, those with a pattern of only lobar MBs or superficial siderosis had the highest ID risk (HR = 1.80, 95% CI = 1.26-2.65). INTERPRETATION: In this large community-based elderly cohort, we identified that the presence of MBs or a high MB count (i.e., ≥ 2), with some specificity for location, was independently related to an increased ID risk over a nine-year follow-up.