OBJECTIVEMonitoring of minimal residual disease (MRD) is proven to be increasingly valuable for predicting relapse and outcome of childhood acute lymphoblastic leukemia (ALL) and is used to identify patients' risk groups in several current clinical trials. However, the limitation is that most studies focused on the cut-off value at 10(-4) and the time point after induction. The aim of this study was to investigate the predictive values of different MRD levels detected at different chemotherapy phases in childhood ALL.

METHODSOne hundred and two patients were enrolled in this study from January 2002 to December 2004 in our hospital. All the patients were treated with modified National Protocol of Childhood ALL in China 1997. MRD levels were detected on the 15th day, 29th day, at 3 months, 6 months and 12 months after initial chemotherapy. All samples were stained with a panel of four colour combinations of fluorochrome conjugated monoclonal antibodies according to the leukemia-associated immunophenotype (LAIP) defined at diagnosis and analyzed by multi-parametric flow cytometry. CD45CD19CD34CD10, CD45CD19CD34CD20 and CD45CD19CD10CD20 were the most common combinations in B lineage ALL, while CD45CD2CD3CD7 and CD45CD2CD3CD34 were the most frequently used immunophenotypes for T lineage ALL. The median follow-up time was 63.3 months ranged from 40.6 to 87.5 months.

RESULTSOf the 102 patients, 64 were male and 38 were female, with a median age of 5.7 (0.2 - 14.8) years. Eighty-eight cases were diagnosed as B lineage ALL and the remaining 14 were T-ALL. The 5-year overall survival (OS) rate and event free survival (EFS) rate for this cohort were (86.9 +/- 3.4)% and (79.9 +/- 4.0)%, respectively. Twelve patients underwent relapse. Among the 102 patients, 14.3% had negative MRD (MRD < 10(-4)) on day 15, 43.9% on day 29, 39.1%, 39.7% and 45.6% had negative MRD at the third, sixth and twelfth month after chemotherapy. Patients who could achieve negative MRD within one year had superior outcome to the others [5-year EFS rates: (92.5 +/- 3.2)% vs. (58.3 +/- 8.6)%, P < 0.001]. The EFS for patients based on MRD levels measured at different stages of therapy were compared by Kaplan-Meier analyses. MRD was predictive of outcome at all 5 time points at a range of thresholds. The optimum threshold, selected for each time point on the basis of log rank analysis, progressively dropped from 10(-2) of day 15 [5-year EFS rates (79.8 +/- 10.3)% vs. (28.6 +/- 17.1)%, P < 0.001], to 10(-3) of day 29 [5-year EFS rates (88.3 +/- 4.9)% vs. (51.3 +/- 14.4)%, P < 0.003], to 10(-4) at 3 [5-year EFS rates (92.4 +/- 5.1)% vs. (65.5 +/- 7.5)%, P < 0.015], 6 [5-year EFS rates (96.3 +/- 3.6)% vs. (65.4 +/- 7.5)%, P < 0.003] and 12 [5-year EFS rates (100.0 +/- 0.0)% vs. (67.7 +/- 8.4)%, P < 0.002] months. And the hazard ratios for relapse and death at higher MRD level groups were 5.91 (95%CI: 1.9 - 18.9), 5.02 (95%CI: 1.5 - 16.5), 5.21 (95%CI: 1.2 - 22.9) and 11.10 (95%CI: 1.5 - 84.5) on day 15, day 29, at month 3 and month 6, respectively. And MRD >or= 10(-2) on day 15 was proven to be an independent predictor by multivariate Cox proportional-hazards regression model.

CONCLUSIONDynamic MRD detection by multi-parametric flow cytometry is highly predictive of outcome for childhood ALL, and the cut-off values at different time points were different.