Association of the Presence and Pattern of MRI Markers of Cerebral Small Vessel Disease With Recurrent Intracerebral Hemorrhage

BACKGROUND Assessing the risk of recurrent intracerebral hemorrhage (ICH) is of high clinical importance. MRI-based cerebral small vessel disease (SVD) markers may help establish ICH etiological subtypes (including cryptogenic ICH) relevant for recurrence risk. METHODS We investigated the risk of recurrent ICH in a large cohort of consecutive ICH survivors with available MRI at baseline. Patients with macrovascular, structural or other identified secondary causes (other than SVD) were excluded. Based on MRI findings, ICH etiology was defined as probable cerebral amyloid angiopathy (CAA) according to the Boston 2.0 criteria, arteriolosclerosis (non-lobar ICH and additional markers of arteriolosclerosis, absent lobar hemorrhagic lesions), mixed SVD (mixed deep and lobar hemorrhagic changes) or cryptogenic (no MRI markers of SVD). Recurrent ICH was determined using electronic health records and confirmed by neuroimaging. Data from an independent multi-center cohort (CROMIS-2 ICH) was used to confirm core findings. RESULTS Of 443 patients with ICH (mean age 67±13 years, 41% female), ICH etiology was mixed SVD in 36.7%, arteriolosclerosis in 23.6%, CAA in 23.0%, and cryptogenic in 16.7%. During a median follow-up period of 5.7 years (IQR 2.9-10.0, 2682 patient-years), recurrent ICH were found in 59 individual patients (13.3%). The highest recurrence rate per 100 person-years was detected in patients with CAA (8.5, 95% CI 6.1-11.7), followed by mixed SVD (1.8, 95% CI 1.1-2.9) and arteriolosclerosis (0.6, 95% CI 0.3-1.5). No recurrent ICH occurred in patients with cryptogenic ICH during 510 person-years follow-up (97.5% CI, 0-0.7); this finding was confirmed in an independent cohort (CROMIS-2 ICH, n=216), in which there was also no recurrence in patients with cryptogenic ICH. In patients with CAA, cortical superficial siderosis was the imaging feature strongest related with ICH recurrence (hazard ratio 5.7, 95% CI 2.4-13.6). CONCLUSIONS MRI-based etiological subtypes are helpful in determining the recurrence risk of ICH; while the highest recurrence risk was found in CAA, recurrence risk was low for arteriolosclerosis, and negligible for cryptogenic ICH.