Background: Conventional prognostic factors for relapse in patients with acute lymphoblastic leukemia (ALL) are the main basis of risk-stratified treatments.
Objectives: To analyze conventional risk factors for relapse and design a predictive model for relapse in our series, after 20 years of experience in treating ALL.
Patients and method: We performed a multivariate analysis of conventional prognostic factors in the treatment of ALL in our unit and compared them with the risk groups in the Berlin-Frankfurt-Münster (BFM-ALL) treatment protocols.
Results: Between 1984 and 2004, 232 children were diagnosed with ALL and treated according to the different versions of the BFM protocols (BFM83, BFM86, BFM90 and BFM95) at the Hospital Niño Jesús, Madrid, Spain. The event-free survival for all patients was 79.4 % (95 % CI: 72.7-85.4). Overall survival among patients who relapsed was 10.72 % (95 % CI: 6-27.3). The only significant prognostic factor for relapse identified by multivariate analysis was leukocyte [white blood cell (WBC)] count higher than 80,000/ml at diagnosis (hazard ratio [HR]: 4.63; 95 % CI: 1.61-13.3; p 5 0,004). The sensitivity and specificity of WBC in predicting relapses were 31.4 % and 87.5 %, respectively. The sensitivity and specificity of BFM risk group stratification in predicting relapses were 25 and 85.9 respectively.
Conclusions: A leukocyte count at diagnosis higher than 80,000/ml and BFM risk-stratified treatment have insufficient sensitivity and specificity to identify relapses.