In trial ALL-BFM 2000, high-risk (HR) acute lymphoblastic leukemia (ALL) is defined by inadequate initial response to induction treatment [poor prednisone response on treatment day eight (PPR), non remission on treatment day 33, and/or a high load of minimal residual disease (MRD, ≥10E-3) after 12 weeks of treatment (TP2) and/or by positive cytogenetics for a t(4;11) or t(9;22)]. Recently, we presented data on the prospective evaluation of MRD at additional time-points under HR-treatment and showed that patients with a persistently high MRD load after the application of three intensive HR blocks were neither curable by chemotherapy alone nor by the addition of stem cell transplantation (very highly resistant leukemia, VHRL,

Schrauder et al.,
Blood
,
2007
;
110
:
585
). Thus, VHRL patients are a target group for the implementation of experimental treatment approaches. Unfortunately, despite MRD analysis being an excellent tool for the identification of this group of patients, the late time point of identification does not allow an early experimental intervention. In addition, MRD measurements do not yield any information on potentially targetable molecular mechanisms of resistance. Therefore, we developed a strategy for

  1. early identification of VHRL patients and

  2. their subsequent rational treatment based on the exploration of treatment targets.

In practice, we apply different genome-wide screening approaches for a comprehensive molecular characterization. Here, we present data on gene expression profiling of HR patients performed in parallel to the ongoing extensive MRD analysis. All HR-patients with sufficient spare initial diagnostic specimens and at least 80% of blasts in bone marrow or peripheral blood were included. In addition, non-HR patients were profiled for the purpose of comparison. Microarrays containing more than 39,000 distinct cDNA clones (SFGF, Stanford, CA) were used. Data were analyzed using the entire set of genes and separately focusing on genes involved in apoptosis. Several publicly available analysis tools were applied. At the time of analysis, gene expression data of 106 patients with B-precursor immunophenotype and negative for TEL-AML1, MLL-AF4 and BCR-ABL rearrangements were analyzed: 43 Non-HR patients; 28 HR patients with a PPR, but low or absent MRD at TP2 (isolated PPR); 17 patients with a high MRD load at TP2 but good MRD reduction under HR blocks (MRD-HR); and 18 patients with persistently high MRD levels after three HR blocks (VHRL). By unsupervised clustering two cases with an unfavorable E2A-HLF rearrangement were identified. Applying Significance Analysis of Microarrays (

SAM,
PNAS
2001
,
98
,
5116
–5121
), only 15 genes (all low expressed in PPR) were identified to be differentially expressed comparing isolated PPR and Non-HR patients (1000 permutations, FDR < 5%, fold change >1.5). In contrast, comparing VHRL patients (excluding the E2A-HLF-positive cases) to the entire group of Non-HR, isolated PPR and MRD-HR patients, 135 genes (57 down-, 78 upregulated in VHRL) were identified. Of interest, in addition to their E2A-HLF-specific pattern, the two E2A-HLF-positive patients did show the same pattern of resistance-associated genes as did VHRL patients without known molecular background. Focusing on genes involved in apoptosis in each single VHRL patient, at least one potentially targetable anti-apoptotic gene with an expression of at least 2-fold the median was identified. Connectivity Map searches identified several substances with highly significant connections to VHRL and Non-VHRL signatures, respectively. In summary, we have seen clear differences in gene expression patterns dependent on the degree of resistance. VHRL patients seem to be characterized by a specific GEP independent of their heterogeneous molecular background. The present data are currently evaluated in a prospective study and will be complemented by those obtained through additional other genome-wide screening approaches (e.g., CGH analysis and epigenetic profiling). The resulting information will then be utilized to build up an integrated VHRL classifier and to discover new treatment options with the ultimate goal of individualized experimental treatment of VHRL patients based on their specific molecular characteristics.

Disclosures: No relevant conflicts of interest to declare.

Author notes

Corresponding author

Sign in via your Institution