Morphological subclassification of follicular lymphoma: variability of diagnoses among hematopathologists, a collaborative study between the Repository Center and Pathology Panel for Lymphoma Clinical Studies

A collaborative study between the Repository Center for Lymphoma Clinical Studies and the members of the lymphoma pathology subcommittee of the major cooperative oncology groups was undertaken in an effort to ascertain the reproducibility and the interobserver agreement for the cytologic diagnosis of follicular lymphomas. A group of 105 patients with follicular lymphomas were subclassified by seven hematopathologists according to two methods. In the first method, cases were subclassified according to the Rappaport, Lukes, and Collins, and Working Formulation systems. In these systems, follicular lymphomas are subclassified by estimation of the different cell populations without the actual counting of cells. With this method, great variability in diagnosis was noted. For example: (1) The consensus diagnosis was that of poorly differentiated lymphocytic lymphoma (PDL) in 39 cases, but among the individual pathologists the number of cases thus diagnosed ranged from 24 to 65; (2) In 40 cases, the consensus diagnosis was follicular lymphoma, mixed-cell type; however, all seven pathologists independently agreed on this subtype in only one case; (3) A major disagreement was noted in 39 cases (37%), in which both diagnostic extremes (small cleaved and large noncleaved) were expressed. In the second method, only precise counts of different cells were made, according to a modification of the method recommended by Berard. With this counting method, diagnoses were independently derived based on the counts provided by the seven pathologists for large cleaved, small noncleaved, and large noncleaved cells. The variability in the results was wide also with this second method. For example, the average number of large cells found by each pathologist was ascertained, and the ranges were determined. The average range was 28 cells, which was considered high. The same determinations were performed only for large noncleaved cells, and the range was found to be 15 cells, which was also considered high. When the diagnoses derived from counts of only large noncleaved cells were compared with the traditional, more subjective diagnoses, fairly close agreement was obtained. In summary, the great variability in diagnoses of follicular lymphomas among pathologists may be attributed to the difficulties inherent in accurate determination of cell size and of the precise percentages of different cells. Until solutions to these problems are developed, one can subclassify follicular lymphomas according to the Berard method or the estimation method.