AbstractBackgroundThere is large variation in estimates of the prevalence of cerebral amyloid angiopathy (CAA). CAA is associated with an increased risk of cognitive dysfunction and intracerebral hemorrhages, and it is tightly linked to immunotherapy‐related side‐effects in Alzheimer’s disease (AD). Thus, accurate estimates of the prevalence of CAA are important. CAA can be diagnosed neuropathologically (‘definite CAA’), or during life (‘probable CAA’) using the (modified) Boston criteria based on MRI markers.MethodWe performed a systematic literature search in PubMed and EMBASE for studies on CAA prevalence in cohorts of patients with AD, population‐based individuals, healthy elderly, patients with intracerebral hemorrhage (ICH), and patients with lobar ICH. Study inclusion criteria were: 1) a study population of at least 10 subjects, 2) a mean age of ≥ 55 years, 3) diagnostic criteria included use of neuropathology or MRI (T2* or SWI), 4) outcome measures included CAA prevalence according to neuropathological assessment or the (modified) Boston criteria, or the prevalence of the MRI markers used in the Boston criteria: strictly lobar microbleeds or cortical superficial siderosis.ResultWe included 170 studies including over 73,000 subjects. We found that in patients with AD, the estimated prevalence of CAA based on pathology (48%) is twice the prevalence based on the presence of strictly lobar microbleeds (22%). In the general population, this difference is threefold (23% vs 7%). However, both methods yield the same estimated prevalence of CAA in healthy elderly (5‐7%), in patients with ICH (19‐24%), and in patients with lobar ICH (50‐57%). In addition, we observed large heterogeneity in both neuropathology and MRI study protocols.ConclusionCAA is a very prevalent condition in AD and in the population. Since CAA is associated with the development of amyloid‐related imaging abnormalities in anti‐amyloid β immunotherapy, a faster decline of cognitive function in AD, and a growing spectrum of clinical symptoms, awareness of the high prevalence of CAA is important. In addition, we found that MRI markers seriously underestimate the prevalence of CAA in AD patients and population‐based individuals. Lastly, our results emphasize the need for standardized assessment and reporting of CAA in neuropathology and MRI studies.