Unicentric Castleman disease: multidetector computed tomography classification with surgical and pathologic correlation

Background This study aimed to correlate multidetector computed tomography (MDCT) classification of unicentric Castleman disease with the surgical and pathologic features. Methods The imaging manifestations of 63 cases of unicentric Castleman disease confirmed by pathology were retrospectively analyzed. Every patient underwent an MDCT examination. Classification based on imaging manifestations, surgical, histopathological, and imaging features were simultaneously reviewed and analyzed by two radiologists, with any disagreements resolved by consensus. Results Sixty-three patients with unicentric Castleman disease were divided into I-IV types by imaging manifestations: type I, single mass with smooth margin (n=5); type II, single mass with irregular or lobulated margin (n=33); type III, single invasive mass with blurred margin (n=20); and type IV, multiple fused masses (n=5). Thirty-eight cases of type I and type II were diagnosed as hyaline-vascular type by pathology after complete surgical resection; 20 cases were type III, in which eight cases were partially resected, 17 cases were pathologically diagnosed as hyaline-vascular type, and the remaining three cases were a mixed type. In five cases of type IV that could not be completely resected, four cases were hyaline-vascular type, and one case was plasma cell type. Conclusions MDCT is an excellent tool for the detection and diagnosis of unicentric Castleman disease. Classification by MDCT features is a reliable method for evaluating tumor infiltration and growth mode, which helps surgical optimization.