The French-American-British (FAB) classification of leukemia. The Pediatric Oncology Group experience with lymphocytic leukemia.

The Pediatric Oncology Group institutions initiated extensive subclassification of cases of acute lymphocytic leukemia (ALL) at diagnosis into laboratory-designated categories. Included was a French-American-British (FAB) classification of all new patients, which was reviewed by a central six-member committee. In addition, on the basis of immunologic criteria, patients were defined as having T-, B-, pre-B-, or "null" cell leukemia. Slides from 617 patients were reviewed. Five hundred forty-six (88.5%) were classified as L1, 51 (8.3%) were classified as L2, 9 (1.5%) were classified as L3, and the remainder could not be assigned. Concordance within the committee was good: in 71% of the cases the committee was unanimous, and in an additional 17% only one member disagreed. In only 11 cases (1.8%) was diagreement such that a majority classification could not be assigned. Institutions assigned L2 more frequently. There was a strong correlation with L3 for B-cell disease only. However, four patients had unequivocal B-cell disease and unmistakable L1 morphologic type, whereas one and had L3 morphologic features and had non-B-cell disease. There was no correlation between the other immunologic markers or periodic acid-Schiff stain and FAB classification, nor between L1 or L2 and risk factors. However, for the 248 null cell and pre-B-cell patients, L2 was more frequent among patients in the poor-risk group (P = 0.008). The time to first failure was significantly shorter for patients with L3 morphologic type. The induction failure rate of L2 patients was significantly greater than that of L1 patients (P = 0.016). With analysis of the duration of remission and adjustment for risk factors, the impact of L2 morphologic characteristics on outcome was not significant (P = 0.18) in null cell patients. Even unadjusted for risk factors, there was no impact of L2 morphologic type on outcome in the pre-B-cell phenotype. It can be concluded that other risk factors overshadow the impact of L1 and L2 morphologic features in predicting duration of remission.