Reproducibility of histological diagnosis of lymphoma in bone marrow

Abstract

<h3>Background</h3> Many studies have assessed rates of histological bone marrow (BM) involvement in non-Hodgkin's lymphoma (NHL) but there are currently no data on the reproducibility of these results. This study was conducted to determine the rates of inter- and intra-observer variability in detection of BM involvement in NHL. <h3>Methods</h3> We randomly selected 40 cases each of diffuse large B-cell lymphoma (DLBCL) and follicular lymphoma (FL), the two most common types of NHL. The BM slides of these cases were independently examined by two haematologists using standardised reporting criteria on two occasions at least 2 weeks apart. Lin's concordance correlation coefficients and weighted kappa statistics were used for continuous and discrete measures, respectively. <h3>Results</h3> Weighted kappa results showed variable inter-observer agreement in DLBCL [trephine = 0.776; aspirate=0.654] and FL [trephine = 0.774; aspirate=0.479]. Similar results were noted for intra-observer variability in DLBCL [trephine = 0.616; aspirate=0.653] and FL [trephine = 0.95; aspirate=0.34]. <h3>Conclusion</h3> This study emphasises the role of trephine biopsy in histological staging of lymphoma. The variable levels of agreement in reporting BM involvement in NHL suggest that histological diagnosis is not adequate without the use of ancillary investigations such as flow cytometry, immunohis-tochemistry and molecular studies.