Reproducibility of the morphological diagnostic criteria for acute myeloid leukemia: the GIMEMA group experience.

Diagnostic reproducibility of the FAB morphological subtypes of acute leukemia is a basic step in the assessment of the clinical outcome in multicenter trials. Unusual cytologic variables and slightly different interpretations of the FAB morphological criteria have been the most significant factors affecting the overall diagnostic concordance rate among the various centers. An evaluation of the diagnostic concurrence between 35 institutions of the Italian Cooperative Study Group GIMEMA and two reviewers of the ad hoc morphological committee has been performed on 377 patients entering the AML 8A and AML 8B GIMEMA protocols. Overall concordance rate was 62.6%. The most significant differences were observed for M2 vs M4, M4 vs M5, M1 vs M2, and M2 vs M5 subtypes. In order to minimize the impact of some diagnostic deviations on the mean cytologic concordance rate, a distinction between "major" and "minor" discrepancies in the diagnostic procedures has been proposed. When the results of the single institutions were corrected by considering the "major" discrepancies only, a mean diagnostic agreement of 78.1% was reached.