Bone Marrow Infiltration of Angioimmunoblastic T-Cell Lymphoma: Identification and Prognostic Impact of Histologic Patterns and Diagnostic Application of Ancillary Phenotypic and Molecular Analyses.

Abstract

Angioimmunoblastic T-cell lymphomas originate from T follicular helper cells and express respective markers (BCL6, CD10, CXCL13, ICOS, and PD-1). Although commonly present, bone marrow involvement by angioimmunoblastic T-cell lymphoma can be diagnostically challenging. Additionally, only little is known about the distribution of T follicular helper cells in healthy and reactively changed bone marrows or in samples affected by other lymphomas. To establish a diagnostic approach to reliably identify bone marrow infiltration of angioimmunoblastic T-cell lymphoma. We analyzed the morphologic infiltration pattern and the expression of T follicular helper-cell markers in 42 matched paired lymph node and bone marrow samples and applied comparative clonality testing. Furthermore, we studied the expression of BCL6 and PD-1 in a control cohort of healthy, reactively changed, and otherwise affected bone marrows. We identified 3 different bone marrow infiltration patterns correlating with overall survival (interstitial/micronodular infiltration with or without eosinophilia and diffuse infiltration with eosinophilia). The matched pairs showed a consistent (co)expression of PD-1 and BCL6 with a generally weaker expression in the bone marrow than in the lymph nodes. Comparative clonality testing was helpful in only a minority of cases. Infiltrates of the most important differential diagnoses contained either PD-1- or BCL6-positive tumor-infiltrating cells, but no coexpressing cells. Bone marrow infiltration by angioimmunoblastic T-cell lymphoma displays 3 different patterns that correlate with prognosis. BCL6 and PD-1 can be reliably used to identify lymphoma infiltrates and to help rule out several differential diagnoses. Comparative clonality testing rarely provides additional value and cannot replace morphologic and phenotypic analyses.