Amyloid in the walls of brain blood vessels, known as cerebral amyloid angiopathy or CAA, was present in roughly four of every five brains examined in a post mortem study of 1,938 people who had been followed in longitudinal studies of ageing, according to a paper by Rupal I Mehta, Tianhao Wang, Sue E Leurgans, David A Bennett and Julie A Schneider in the August 2026 issue of Brain. Participants had a mean age at death of 89.8 years (standard deviation 6.6). On examination, 415 (21.4 per cent) had no CAA, 795 (41.0 per cent) had mild CAA and 728 (37.6 per cent) had moderate to severe CAA. This is autopsy data analysed with regression models, not a trial, and it describes associations rather than causes.
The odds of moderate to severe CAA were higher in people who were older (odds ratio 1.34 per 10 years, 95 per cent confidence interval 1.22 to 1.63), in APOE e4 carriers (odds ratio 3.62, 2.90 to 4.52) and in those with Alzheimer's disease pathology, defined as intermediate to high Alzheimer's disease neuropathologic change (odds ratio 4.14, 3.28 to 5.23). The two conditions did not track together perfectly: 117 of 1,216 participants with Alzheimer's pathology (9.62 per cent) had no CAA at all, and nearly one in five of those with moderate to severe CAA (18.59 per cent) had none to low Alzheimer's disease neuropathologic change. Moderate to severe CAA was associated with neuritic plaques (odds ratio 1.27, 1.09 to 1.48) and neurofibrillary tangles (odds ratio 1.52, 1.32 to 1.76), with a statistical interaction between tangles and neuritic plaques in the whole group (P = 0.047).
The genotype-specific results are the part the field is discussing. Among participants without Alzheimer's disease, the odds of severe CAA were about 28 times higher in APOE e2 carriers when neuritic plaque and tangle loads were higher, an interaction that also reached significance in that subgroup (P = 0.039). The same pattern was not found in e2 carriers who did have Alzheimer's disease, nor in e4 carriers with or without it. Related dementia pathologies as a group were not related to CAA once Alzheimer's disease and APOE e4 were accounted for. On the question that partly motivated the work, whether vascular amyloid co-segregates with diffuse plaques as recent laboratory experiments have suggested, the authors modelled diffuse plaque counts alongside neuritic plaques and tangles, but the published abstract reports its associations and its conclusion in terms of neuritic plaques and tangles only.
In a scientific commentary published in the same issue on 10 July 2026, Gargi Banerjee of University College London and Susanne J van Veluw of the University of Edinburgh write that the absence of a consistent CAA to plaque and tangle relationship across genotypes suggests vascular amyloid is not merely a marker of advanced Alzheimer's disease, and may in some circumstances progress along its own path. They also place the finding in the context of anti-amyloid antibody treatment, where APOE e4 carriers and people with haemorrhagic imaging markers of CAA are known to be at higher risk of amyloid-related imaging abnormalities. The counterpoint is that these were brains of people who died at a very advanced average age, in cohorts of research volunteers, and CAA severity was graded at a single point: the end of life.