A score at diagnosis for predicting length of remission in childhood acute lymphoblastic leukaemia.

Abstract

Thirty-two variables at diagnosis of acute lymphoblastic leukaemia (ALL) were studied in an unselected population-bases series of 209 children. Twelve variables had individually a statistically significant effect on the duration of first remission. A multivariate analysis using data on the 199 children who went into complete remission showed that all significant variation in remission times could be explained by only 3 variables acting simultaneously. These were the total white blood count (WBC) at diagnosis, the Franco-American-British (FAB) classification of blast morphology and the percentage of lymphoblasts with PAS+ coarse granules or blocks. A simple scoring system (for WBC add 1 if less than 20 X 10(9)/1, add 2 if 20 - 50 X 10(9)/1, add 3 if greater than or equal to 50 X 10(9)/1; for L2 or L3 leukaemia add 1; for PAS+ less than 5% add 1) separated patients into risk groups with widely different median lengths of first remission. Application of the risk score improves the prediction of the outcome of treatment, and the clinical trials, allows more accurate stratification, less extensive data collection and simpler analysis.