It is unclear whether anticoagulation predisposes towards lobar location of hemorrhage or whether patients with AA-ICH are more likely to have underlying CAA.

We retrospectively assessed 844 consecutive patients with spontaneous ICH.  The Simplified Edinburgh criteria was used to assign high probability vs intermediate or low probability of CAA based on CT.  For patients who underwent brain magnetic resonance imaging (MRI) we used the Boston criteria version 2.0 to assign probable vs possible or no CAA.   

Of the 844 patients with spontaneous ICH, 33 were found to have a secondary lesion on MRI and were excluded.  Among the 811 patients with primary ICH [71±15 years, 356(44% female)], 177(22%) were associated with anticoagulation: 145(82%) with vitamin K antagonists (VKA), 27(15%) with direct oral anticoagulants (DOAC), and 5(3%) with heparin.   443(55%) patients underwent MRI of which 138(31%) had probable CAA by MRI and 60(14%) had high probability CAA by CT. Overall, 485(60%) of patients had strictly lobar ICH and 202(25%) had probable CAA by MRI or CT.   Compared to patients with non-AA-ICH, patients with AA-ICH were less likely to have lobar location of bleed (53% vs 63%, p=0.02), probable CAA by MRI or CT (28% vs 14%, p<0.0001) and probable CAA by MRI alone (34% vs 18%, p=.003).   Among patients with lobar ICH, those with AA-ICH were less likely to have probable CAA by MRI or CT (45% vs 27%, p<0.002) and probable CAA by MRI alone (43% vs 24%, p=.006) when compared to those with non-AA-ICH. 

In our study AA-ICH was inversely associated with lobar location and probable CAA.  These results suggest that anticoagulation may interact more strongly with hypertensive microangiopathy than amyloid angiopathy. These findings require replication in other cohorts.