The outcome for childhood acute lymphoblastic leukemia (ALL) following marrow relapse remains bleak in spite of numerous approaches to further intensify therapy. Understanding the biological basis of relapse and chemoresistance, as well as identifying and validating potential new targets, are the goals of our study. Previously we determined global gene expression patterns of matched diagnosis and relapse leukemic blasts in 32 patients (64 samples) with childhood B-precursor ALL using Affymetrix U133A arrays (

We now have extended this analysis to 60 patients (120 samples). Thirty-six patients relapsed early (within 36 months of initial diagnosis), while 24 patients relapsed late. Within the TEL/AML1 subset (n=12 patients), time to relapse was inversely proportional to the correlation co-efficient of expression profiles of the diagnosis and relapse matched pair samples, suggesting that the later the relapse, the more distinct the relapse clone is from the diagnostic clone. A supervised pairwise analysis in all 60 patients identified 292 probesets that were differentially expressed between diagnosis and relapse (FDR < 10%). In a relative enrichment analysis, multiple genes mediating cell death were down-regulated at relapse (p=0.00003), suggesting that the leukemia cells had evolved mechanisms to enhance survival. These included p21, TNFPAI3, RIPK2, BCLAF1, STK17B. In concert, DNA replication genes were up-regulated at relapse (p=0.00002). Differences in pathways leading to early vs. late relapse were evident. Early relapse was characterized by an over-expression of cell cycle genes reflecting a proliferative state. At the time of relapse, a marked over-representation of genes involved in the progression through the M phase of the cell cycle was observed in early relapse compared to late relapse (p=1.3E-08). Late relapse was characterized by the over-expression of genes involved in nucleoside biosynthesis, particularly targets of antifolates (DHFR, MTHFD1, TYMS). A small number of gene expression patterns were common to both early and late relapse, including up-regulation at relapse of BIRC5 (survivin): an attractive target for therapeutic intervention. In conclusion, analysis of an expanded cohort of matched diagnosis/relapse pairs has validated and extended our previous findings that early relapse is associated with a proliferative gene expression signature. In addition we have now identified pathways operative in late relapse. Targeting these individual genes and pathways may lead to innovative strategies to treat or prevent relapsed ALL.

Disclosures: No relevant conflicts of interest to declare.

Author notes

Corresponding author