Abstract
PURPOSE: The prevalence of susceptibility effects (SE) on T2*-weighted imaging (WI) and susceptibility-WI (SWI) in primary large B-cell lymphoma (IP-LBCL) of the central nervous system (CNS) and diffuse LBCL (DLBCL) with secondary CNS lymphoma (SCNSL) remains debated. This study aimed to clarify SE prevalence and their associations with primary versus secondary manifestations, immune status, corticosteroid treatment, and structural MRI features. METHODS: This retrospective, single-center study included histologically confirmed DLBCL cases (WHO ICD-10 C83.3) with intracerebral involvement (March 2011-November 2023). Subjects without cranial MRI or diagnostic susceptibility-based sequences were excluded. T2*WI and SWI were independently reviewed by two neuroradiologists for presence or absence of SE. Identified SE were classified into five types: punctate, linear, confluent, conglomerate, and/or ring-like. Lesion focality and morphology were assessed on T2WI and contrast-enhanced T1WI. Clinical data included extracranial lymphoma history, immune status, and corticosteroid initiation. RESULTS: Among 128 cases (median age: 70 years [58-75]; 65 men), 119 (93%) had IP-LBCL and 9 (7%) had SCNSL. 110 (85.9%) subjects were immunocompetent. T2*WI was available in 90 (70.3%) datasets and SWI in 38 (29.7%). SE detection was higher on SWI (71.1%) than T2*WI (47.8%; P = 0.03). No association was found between SE and lymphoma type (P = 1.00). In IP-LBCL, immunosuppression was significantly associated with SE presence (P = 0.001; OR = 16.1, 95% CI: 2.89-304.79), while age, gender, and corticosteroid use (18.5%) were not. SE showed no significant associations with structural imaging features, including necrosis. CONCLUSION: SE are common in both IP-LBCL and SCNSL, particularly on SWI, and present with variable patterns unrelated to structural MRI features. In IP-LBCL, immunosuppression, but not pre-existing corticosteroid treatment, is significantly associated with the presence of SE.