Abstract
BACKGROUND AND PURPOSE: Sarcoidosis is a granulomatous disease of unknown etiology affecting the central nervous system in up to 15% of the patients. Diagnosis of neurosarcoidosis is very challenging due to the heterogeneity of its clinical manifestation. This study intended to evaluate the distribution of cerebral lesion sites and the potential presence of specific lesion clusters in neurosarcoidosis patients using voxel-based lesion symptom mapping (VLSM). METHODS: Patients with neurosarcoidosis were retrospectively identified and included between 2011 and 2022. Cerebral lesion sites were correlated voxel-wise with presence and absence of neurosarcoidosis using non-parametric permutation test. Multiple sclerosis patients served as controls for the VLSM-analysis. RESULTS: Thirty-four patients (mean age 52 ± 15 years) of whom 13 were diagnosed with possible, 19 with probable and 2 with confirmed neurosarcoidosis were identified. Lesion overlap of neurosarcoidosis patients demonstrated a distribution of white matter lesions in all brain areas, with a periventricular predilection similar to multiple sclerosis. In contrast to multiple sclerosis controls, no propensity for lesions in proximity of the corpus callosum was observed. Neurosarcoidosis lesions appeared smaller and lesion volume was lower in the neurosarcoidosis cohort. The VLSM analysis showed minor associations between neurosarcoidosis and damaged voxels in the bilateral frontobasal cortex. CONCLUSIONS: The VLSM analysis yielded significant associations in the bilateral frontal cortex, suggesting that leptomeningeal inflammatory disease with following cortical involvement is a quite specific feature in neurosarcoidosis. Lesion load was lower in neurosarcoidosis than in multiple sclerosis. However, no specific pattern of subcortical white matter lesions in neurosarcoidosis was revealed.