Cortical Superficial Siderosis in Memory Clinic Patients: Further Evidence for Underlying Cerebral Amyloid Angiopathy.

BACKGROUND Cerebral amyloid angiopathy (CAA) is associated with many cases of spontaneous symptomatic lobar intracerebral haemorrhage in older individuals and is emerging as an important contributor to cognitive impairment. Cortical superficial siderosis (cSS) is an increasingly recognized haemorrhagic neuroimaging manifestation of CAA. We sought to investigate its prevalence and its association with underlying CAA among memory clinic patients. METHODS We included consecutive eligible patients who presented to the out-patient memory clinic at the Massachusetts General Hospital from 2007 to 2010 and had appropriate MRI, including blood-sensitive sequences. We analyzed the prevalence and topography of cSS according to demographic, clinical, APOE and MRI data. RESULTS Our cohort consisted of 339 memory clinic patients: Alzheimer's disease (n = 86); mild cognitive impairment (n = 162); vascular dementia/mixed dementia (n = 18); other dementia/undetermined (n = 42); and subjective cognitive complains (n = 31). cSS was detected in 10 patients (3%; 95% CI 1.4-5.4): in 7 cases cSS was focal and in 3 cases, it was disseminated. In multivariable logistic regression analysis, the presence of cSS was associated with lobar microbleeds (OR 1.08; 95% CI 1.03-1.13; p = 0.001, per each additional microbleed) and severe white matter hyperintensities (Fazekas score 5-6, OR 4.43; 95% CI 1.21-26.28; p = 0.028) after adjusting for age. These associations were not influenced by the clinical diagnosis. In patients with APOE data, the APOE ε4/ε4 genotype was overrepresented among subjects with vs. without cSS. In the subgroup of patients with probable CAA (n = 68; 9 with cSS) based on the presence of strictly lobar microbleeds, cSS was also associated with a higher prevalence of severe white matter hyperintensities (66.7 vs. 10.2%; p = 0.001), high centrum semiovale perivascular spaces burden (88.9 vs. 52.4%; p = 0.041) and higher counts of lobar microbleeds (median 13; IQR 10-36 vs. median 1; IQR 1-2; p < 0.00001), compared to patients without cSS. CONCLUSIONS Our data provide further evidence supporting the hypothesis that cSS is a manifestation of advanced CAA in memory clinic populations. Future longitudinal studies should explore any direct effect of cSS on cognition or haemorrhage risk and disease progression.