Comment on Breit et al, page 1151
In this issue, Breit and colleagues report that NOTCH1 mutations, present in approximately 50% of T-ALLs, are associated with an excellent early response to therapy and outcome, raising important questions about how clinical trials should be designed to evaluate agents that target Notch1.
In 1991, Ellisen et al1 identified TAN-1 (translocation-associated NOTCH), a gene fused to the T-cell receptor beta locus by a t(7;9)(q34;q34.3) in T-cell lymphoblastic leukemia (T-ALL). Notch was the gene product responsible for the notched wing phenotype in Drosophila and had been demonstrated to play a major role in cell fate determination of the fly nervous system.2 Subsequent functional studies showed that the rare T-ALLs with a t(7;9) expressed a truncated, activated form of TAN-1 (now termed Notch1) with leukemogenic properties.3 Notch1 was also demonstrated to be involved in regulation of thymocyte differentiation, and was essential for normal T-cell development.4 Biologic studies showed that NOTCH1 encodes a transmembrane receptor that undergoes a series of proteolytic cleavage steps, the last of which is catalyzed by γ-secretase to produce intracellular Notch1 (ICN) protein that is involved in transcriptional regulation.5
In 2004, Weng et al reported that more than 50% of human T-ALLs contained activating NOTCH1 mutations, suggesting that Notch might be a rational therapeutic target.6 Potential agents were already available because γ-secretase plays an important role in cleavage of amyloid precursor protein to amyloid beta-peptide, a major constituent of amyloid plaques in Alzheimer disease. γ-secretase inhibitors (GSIs) blocked transcriptional activation induced by mutated Notch1 polypeptides similar to those present in T-ALL.6
Taken together, these data created great interest in the use of GSIs in T-ALL, and phase 1 clinical trials were initiated. At the same time, other studies began to examine whether NOTCH1 mutations had prognostic significance. In the current issue of Blood, Breit and colleagues report analyses of 157 children with T-ALL enrolled in the ALL-BFM 2000 trial. Leukemia cells from 52% of T-ALL patients had NOTCH1 mutations, while no mutations were present in 50 B-precursor ALL controls. Most clinical characteristics were no different in those with/without NOTCH1 mutations, but there was a strong association between the presence of NOTCH1 mutations and an excellent early response to therapy, measured either by response to the 7-day prednisone prophase or levelsofminimalresidualdisease(MRD)presentat the end of induction or consolidation therapy. Furthermore, patients with NOTCH1 mutations had a significantly better event-free survival (EFS) than those without mutations (90% vs 71% at 4 years), due to a 4-fold higher risk of relapse among the NOTCH1 germ-line patients.
NOTCH1 mutation status may turn out to be a critical prognostic factor in T-ALL that can help differentiate between patients at high versus low risk of relapse. These results also raise the question of how molecularly targeted therapy can be integrated into therapeutic strategies for patients with an excellent expected treatment outcome. It is relatively straightforward to design a trial of targeted therapy in patients expected to have a poor outcome: one applies the experimental therapy and assesses response, either in terms of complete remission rate or perhaps by log-reduction in MRD. How can one evaluate targeted therapies in patients expected to have an excellent outcome? Should the goal be to improve response by adding a GSI to a multiagent chemotherapy backbone in NOTCH1-mutated T-ALL patients? To detect a change in EFS from 90% to 94%, one needs a clinical trial of approximately 1200 patients. It would take about 10 years for the Children's Oncology Group to accrue 1200 NOTCH1 mutant T-ALL patients. It seems more attractive to endeavor to replace components of therapy that have significant potential side effects with a novel agent such as a GSI and maintain equivalent outcome. How might a clinical trial be designed to answer this question? It has great relevance to other subtypes of childhood ALL.
Not all molecularly targeted agents will be tested in patients with bad outcomes. This is good, but challenges us to be creative as we evaluate such agents. Let's notch up our belts and get to work. ▪
