Comparison of Diagnostic and Relapse Flow Cytometry Phenotypes in Childhood ALL: Implications for Residual Disease (MRD) Detection.

Flow cytometric analysis of MRD depends on detecting phenotypically abnormal populations. Phenotypic shifts are known to occur but there is little information to indicate how these shifts affect detection of MRD. We compared diagnostic and relapse bone marrow specimens in 36 children with precursor B-ALL studied with the two-tube panel CD19-APC/CD45-PerCP/CD10-PE/CD20-FITC and CD19/45/9/34 (Leukemia 13:558,1999). Small phenotypic shifts of either intensity or coefficient of variation of the distribution of at least one marker were seen in 34 cases, while 7 cases had shifts of 4 or more markers. Phenotypic shifts were complex and could not be explained simply by a change in maturation stage. CD19 was the most stable marker, changing in only 8 cases, CD45 changed in 10; CD10 in 13; CD9 in 14; CD20 in 16 and CD34 in 18. CD10 and CD20 generally lost intensity while CD45 mostly increased. CD34 changed from positive to negative (5 cases) or relapsed as the CD34-negative component when the diagnostic sample was partially positive (9 cases). In 26 cases there was a sufficient change in phenotype from diagnosis to relapse to determine if day 28 MRD more closely resembled the initial diagnostic or the subsequent relapse sample. In 13/19 MRD+ cases the day 28 phenotype resembled the diagnostic specimen and in 6 it resembled relapse. Surprisingly, time to relapse was similar in both groups. In 6 of the 7 MRD negative cases, retrospective review of the data did not identify an abnormal population that resembled either diagnosis or relapse. In the remaining case, in which both CD34 and CD10 were lost from diagnosis to relapse, it is possible that a MRD population resembling relapse might have been missed. We conclude that phenotypic shifts between diagnostic and relapse specimens in childhood ALL are relatively common, but that in general they do not interfere with the ability to recognize MRD. However, MRD analysis that adheres to rigid gating looking only for the precise abnormal phenotype seen at diagnosis might miss some positive cases.