Subclassification of acute lymphoblastic leukaemia in children: analysis of the reproducibility of morphological criteria and prognostic implications.

Stained smears of aspirated bone marrow obtained at time of diagnosis from 223 children with acute leukaemia were reviewed independently by three observers in a double-blind fashion in order to assess the reproducibility and clinical significance of the French-American-British Cooperative Group Classification. In 170 cases of acute lymphoblastic leukaemia (ALL), triple agreement of 69.4% was reached in the subclassification into L1, L2, and L3 types. The closest degree of agreement between two observers was 86.8%. In children with blasts classified as L1, 57 OF 61 patients (93.4%) remained in haematological remission after 12 months, as compared with 14 of 20 (70%) in children whose blasts were typed as L2 morphology, a difference which was statistically significant (P less than 0.05). There was no difference when these groups were compared after 24 or 36 months. Children older than 7 years had an increased incidence of L2 type (P less than 0.05). We conclude that although there may be significant variation in individual interpretations of the criteria utilized for classification, blast morphology may nevertheless be a useful prognostic factor.