MRI-Based Prediction of Macrovascular Causes of Intracerebral Hemorrhage: The MACRO Score.

Background and objectives: Although most spontaneous intracerebral hemorrhages (ICHs) are due to cerebral small vessel diseases (SVDs), between 1 in 7 and 1 in 10 are due to a macrovascular cause. Rapid diagnosis has important therapeutic and prognostic implications but sometimes requires digital subtraction angiography (DSA), an invasive procedure which cannot be performed in all patients. MRI provides optimal sensitivity for markers of SVD but was not included in previous risk stratification scores. We aimed to create and validate a risk stratification score for macrovascular causes of ICH including MRI findings to guide diagnosis and further investigations.

Methods: We pooled data from 2 large observational study cohorts (London/United Kingdom and Graz/Austria) of consecutive patients with ICH who had brain MRI and at least 1 angiographic modality within 90 days of symptom onset. The primary outcome was a macrovascular cause of ICH (arteriovenous malformation/dural arteriovenous fistula, aneurysm, cavernoma, or cerebral venous thrombosis), with the diagnosis based on neurovascular multidisciplinary meetings. Using lasso logistic regression, we built the MRI Assessment of the Causes of intRacerebral haemOrrhage (MACRO) score to assess the probability of a macrovascular cause. We performed internal validation using bootstrapping and external validation in an independent cohort (Bern/Switzerland).

Results: We included 1,043 patients with ICH (mean age 66 years, 42% female), 78 of whom had a macrovascular cause (7.5%). The final score includes age (0-39, 40-69, or ≥70), location of ICH (lobar, deep, or infratentorial), and SVD markers on MRI (≥1 microbleed, ≥1 lacune, presence of cortical superficial siderosis, or white matter hyperintensities using the Fazekas scale). The MACRO score showed an optimism-adjusted c-statistic of 0.90 (95% CI 0.88-0.93), superior to existing CT-based scores (p < 0.001). In external validation, the c-statistic was 0.87 (95% CI 0.80-0.94). MACRO scores ≥6 (59.5% of patients) indicated a very low risk of a macrovascular cause (0.2%), while scores ≤2 (9% of patients) indicated a high risk (48.9%).

Discussion: The MRI-based MACRO score shows excellent performance in predicting the likelihood of macrovascular causes of spontaneous intracerebral hemorrhage, making it useful in guiding further investigations. Important limitations include the observational study design and the performance of DSA in a minority of patients.