Taking childhood leukemia personally

The concept of biologic unity and the feasibility of “one-fits-all” treatment strategies has become outdated for most malignancies and certainly for acute lymphoblastic leukemia (ALL) of childhood. Conventionally defined malignant diseases represent a collection of molecularly distinct entities with characteristic features that define their response to treatment as well as their prognosis. Clinically significant heterogeneity of childhood ALL became apparent when disparity in treatment responses and differences in the cytogenetic makeup of leukemia cells were identified. Even today, with molecular subtyping of leukemias, these seemingly archaic features remain important components of the individual risk stratification of modern treatment protocols that allow us to successfully individualize treatment according to risk.¹ 

The concept of risk-adapted therapy has been one of the cornerstones in achieving the current 80+% cure rates of childhood ALL in developed countries. This is all very well, unless you belong to the 20% of patients who eventually relapse or show treatment resistance. Furthermore, identifying those children who would be cured with a less intensive regimen with less acute and long-term side effects is difficult yet equally important.

The advent of mRNA expression and genomic high-resolution DNA analyses has revolutionized the technical potential to define the functional genetic composition of malignant cells. Various studies of expression arrays have defined previously undetectable subgroups of childhood ALL patients with different risk levels.^2,3  However, mRNA expression analyses are prone to preanalytical changes,⁴  which is also reflected by different expression analyses identifying different components of risk signatures. The more robust genomic DNA analyses by array–comparative genomic hybridization, single nucleotide polymorphism arrays, or deep sequencing have therefore proved to be most useful to complement expression studies in finding correlations between genetic lesions and prognostic biomarkers and, more importantly, in finding functionally critical drivers of leukemogenesis.^5-7 

In this issue of Blood, Harvey and coworkers now confirm that children with B-precursor ALL, who have been classified as high risk on the basis of the simple clinical National Cancer Institute criteria age and high white blood cells at the time of diagnosis, can be subdivided into subgroups by a combination of mRNA expression analyses and assessment of genomic DNA copy number aberrations.⁸  Two of these subgroups coincided with known cytogenetically defined risk groups. Two other subgroups were remarkable in that they identified patients with an excellent prognosis of 4-year event-free survival (EFS) of > 90% or a particularly bad prognosis of 4-year EFS of ∼ 30%. The favorable subgroup was defined by the high expression of 5 outlier genes and ERG deletions. The unfavorable group was characterized by (1) high expression of 4 genes and (2) DNA deletions, rearrangements, and mutations of another 6 gene loci including CRLF, RAG, JAK2, and IKZF1 (thus confirming the prognostic significance of these genes and a BCR-ABL–like expression signature in pediatric ALL^5,6,9-12  and also confirming the power of combined mRNA expression and DNA analyses in determining individual risk profiles).

Although the definition of the subgroups represents a remarkable success, critical questions remain. The first relates to the influence of therapy on the risk profile. This is well illustrated by the case of pediatric T-ALL, where the favorable impact of activating NOTCH1 mutations depends on the subtleties of the different treatment protocols.^13-15  While the role of CRLF has been documented in both Childhood Oncology Group^8,10  and Berlin-Frankfurt-Münster protocols,¹¹  the general role of the other changes in different treatment protocols still requires testing. A second issue relates to the small size of the subgroups of a disease that is already rare. Tailoring and testing treatment concepts in the context of such a personalized medicinal approach will require studies that are designed on a wide international scale: this will challenge the different cultures that underlie medicine in general, and the organization of clinical studies including different regulatory agencies in particular. Finally, while it is helpful to define very-high-risk groups in pediatric ALL, identifying drugable driver pathways of high-risk ALL and developing logical new treatments for this most difficult group of patients will be essential.

In this issue, Harvey and colleagues state that they will focus on identifying hitherto unidentified altered kinases that may contribute to the BCR-ABL–like expression profile of high-risk childhood ALL that may also represent new targets for therapeutic intervention. Alternatively, one may want to look at the potential of epigenetic strategies to reverse the block in B-cell differentiation in this group of patients. In any case, finding new options for the unfortunate 20% of children with ALL is what future patient generations will expect from us.