Cerebral microbleeds predict infectious intracranial aneurysm in infective endocarditis

Background and purpose Magnetic resonance imaging (MRI) features such as cerebral microbleeds and sulcal susceptibility‐weighted imaging (SWI) or gradient‐echo T2* lesions in infective endocarditis (IE) have been associated with the presence of infectious intracranial aneurysm (IIA). Our aim was to validate these MRI predictors for IIA in order to better assist in assessing the appropriate indications for digital subtraction angiography (DSA). Methods The derivation cohort comprised IE patients with neurological evaluation, MRI and DSA at a single tertiary referral center from January 2015 to July 2016. Validation was performed in a cohort of IE patients who underwent MRI and DSA at the same center from 2010 to 2014. Results Of 62 patients in the derivation cohort, 10 (16%) had IIAs. Of 129 in the validation cohort, 19 (15%) IIAs were identified. The MRI predictors for IIA consist of (i) contrast enhancement with microbleeds, (ii) cerebral microbleeds >5 mm or sulcal SWI lesions and (iii) any MRI hemorrhages. The sensitivity for the presence of IIA in each group of the derivation cohort was 90%, 80% and 100%, respectively. The sensitivity in the validation cohort was 47%, 68% and 94% respectively. The specificity in the derivation cohort was 87%, 85% and 18%. In the validation cohort, the specificity was similar at 87%, 75% and 27%. Conclusions The absence of MRI hemorrhages may not necessitate the need for DSA.