Diagnostic concurrence in the subclassification of adult acute leukemia using French-American-British criteria

Publisher Website
Google Scholar
Abstract
Using FAB criteria, we retrospectively classified 195 adult acute leukemia cases seen over a six-year period. Acute lymphoblastic leukemia (ALL) was separated from acute nonlymphoblastic leukemia (ANLL) by negative peroxidase. The dysmyelopoietic syndrome was separated from ANLL when the percentage of immature leukemic cells was <30%. The 55 cases of ALL and 140 cases of ANLL thus defined were initially independently subclassified with Wright's stained smears by three observers. Cases were then jointly reviewed by using all available information, and a final consensus diagnosis was reached. For ALL, there was complete agreement (of initial independent observations) about subtype in 32 of 55 (58%); for only one case was there total disagreement. For ANLL there was complete agreement in 89 of 140 (64%) and total disagreement in 5 of 140 (4%). Frequent disagreement of independent observations with the consensus diagnosis were L1 interpreted as L2, M5 interpreted as M4, L3 interpreted as L2, L2 interpreted as L1, M1 interpreted as M2 and M4 interpreted as M5. Although there appears to be variation between observers, all but 6 of 195 cases had at least two observers in agreement. Disagreement appeared to be partly based on varying interpretations of subjective criteria and partly on the variability in determining percentages of cell types present. The use of nonspecific esterase stain or lysozyme appeared to improve diagnostic agreement over that observed using Wright's stained slides alone in approximately 5% of cases of ANLL.