Defining new prognostic markers in childhood acute lymphoblastic leukemia

Examining MRD within the first few weeks will identify a particularly good risk group of patients who are likely to be cured with less aggressive therapy.

Issue: November 1, 2006

ByWilliam L. Carroll, MD

ByElizabeth A. Raetz, MD

Although most children with acute lymphoblastic leukemia are cured after receiving intensive therapy, the impressive outcomes are not without drawbacks, namely short- and long-term morbidities.

One of the greatest challenges in pediatric oncology is finding prognostic markers to identify children who will ultimately fail therapy. With better indicators, physicians can avoid over-treating these patients and instead start them on less aggressive protocols, which could have equally promising outcomes.

Refining risk

Pediatric patients with ALL are stratified into risk groups before they begin therapy. Most physicians use traditional, well-established clinical variables (age and white blood cell count) combined with blast cytogenetic features and measures of early treatment response, to tailor therapy. The Children’s Oncology Group, for example, classifies children into one of four risk groups: low, standard, high and very high. The therapy intensity is matched accordingly (see table).

Recent efforts have focused on refining risk group allocation through the identification of new prognostic markers. Examples of such initiatives include genomic studies that define gene expression profiles predictive of outcome and the identification of host gene polymorphisms in drug metabolizing enzymes and other modulators of treatment response. However, early treatment response has long been established as a major determinant of event-free survival. This variable is influenced by both the inherent chemosensitivity of the leukemia cell as well as host characteristics that may affect drug-target interaction (eg, drug absorption/metabolism).

Elizabeth A. Raetz, MD
Elizabeth A. Raetz

William L. Carroll, MD
William L. Carroll

Treatment response

Whether measured by peripheral blast clearance during the first week of chemotherapy, marrow blast regression by standard morphology during induction, or minimal residual disease (MRD) at various points during the initial phase of therapy, all studies have demonstrated the favorable prognostic importance of early disease regression. Improved techniques for measuring early treatment response have generated a slew of clinical questions:

What are the optimal methods and time point(s) to measure early treatment response?
Should disease status be serially monitored in all patients, or are there other markers at diagnosis that are more reliable and efficient predictors of treatment response?
Does early treatment response, which in theory should account for all aspects of the inherent biology of the blast and host, ultimately carry greater weight as a prognostic marker than other measures such as blast cytogenetics?
Does early disease regression have different meanings in the context of different genetically-defined subtypes of leukemia, or in genetically different hosts, or with different therapy?

Simple assays

Two recent papers by Elaine Coustan-Smith, MSc, and Dario Campana, MD, PhD, both from the St. Jude Children’s Research Hospital in Memphis, addressed some of these important questions. Coustan-Smith and colleagues described a simplified flow cytometry-based assay for assessing MRD at day 19 of therapy. The assay uses three antigens commonly expressed on B-precursor lymphoid cells (CD19, CD10 and CD34).

Non-leukemic B-precursor cells with this antigen profile were consistently undetectable after two weeks of induction therapy in patients with T-cell ALL, but the researchers showed that this antigen panel could reliably detect residual leukemic blasts in most children with B-precursor ALL. They also demonstrated a 99% concordance between results obtained with these simplified assays vs. more sophisticated flow cytometry and polymerase chain reaction-based MRD techniques.

They further demonstrated the prognostic importance of MRD at day 19 of induction therapy with this simplified assay. Among 84 uniformly treated patients, the 10-year incidence of relapse was 28.8% — 7.1% for the 59% of patients who had detectable MRD (>0.01% leukemic cells) at day 19 vs. 4.8% — 3.3% for the remaining patients without detectable MRD at day 19 (P=.003). The feasibility, efficiency and reproducibility of this assay were demonstrated in a separate analysis in 38 patients in Brazil, where there was no prior experience with these assays.

table

Early response

One of the advantages of early response measurements is that they account for unknown biological characteristics that could play a role in drug sensitivity. However, a mechanistic understanding of the in vivo drug response will undoubtedly lead to novel therapeutic efforts to modulate drug sensitivity in the future.

In a subsequent publication, the St. Jude researchers analyzed genes associated with MRD response. Gene expression profiles were obtained at diagnosis from 189 patients who were uniformly treated on hospital protocols, and who had MRD assessed on days 19 and 46 of induction using standard flow-cytometry. After adjusting for the expression of genes known to be associated with cytogenetic subgroups, researchers identified a group of 85 genes associated with a poor MRD response (>0.01% leukemic cells) at day 46. Seventeen of those genes were also associated with the persistence of MRD on day 19.

Low expression of CASP8AP2 at diagnosis was significantly associated with the persistence of MRD, and this gene was selected for further study because of its essential role in apoptosis. CASP8AP2 functions as a signaling complex that activates caspase 8, which is an initiator caspase in the cell death pathway. To validate these findings, the authors analyzed the prognostic importance of CASP8AP2 expression in a separate group of 99 patient samples. They showed that low CASP8AP2 expression predicted a lower event-free survival (P=.02) and a higher rate of relapse (P=.01). In a regression model that included all routine prognostic markers in childhood ALL, CASP8AP2 expression remained a significant predictor of outcome, superior to all markers except age (>10 years at diagnosis).

Residual disease

These studies raise important questions about how to best define new prognostic markers and incorporate them into existing risk algorithms. The feasibility and availability of a simplified flow-based MRD method has great appeal, as it is essential that prognostic factors be determined universally and evaluated within the context of different treatment regimens, to ultimately establish their value.

Although early treatment response has consistently been prognostic, determining the time point(s) for assessment is challenging. Examination of MRD within the first few weeks of therapy is likely to identify those that are MRD-negative. These patients are likely to be cured with less aggressive therapy and would not be candidates for further intensification. Those patients who were MRD negative at day 19 in the recent study, for example, had only a 5% likelihood of treatment failure.

Approximately 50% of patients were shown to have detectable MRD at day 19 using both the simplified as well as more sophisticated methods for flow cytometry-based assessment. These patients had approximately a 30% chance of relapse. Although some of these patients may benefit from therapy intensification, some MRD-positive patients are likely to be cured without exposure to the toxicity of augmented treatment. Assessment of MRD at a second time point might define a subset that would benefit from additional treatment intensification. Indeed those patients with residual leukemia at the end of induction or shortly thereafter have a significantly worse prognosis and therapy can be altered.

Gene expression signatures

The identification of gene expression signatures that predict slow response at diagnosis may improve patient outcomes. In their second publication, the St. Jude researchers identified a potential candidate gene whose expression predicted not only MRD response, but also event-free survival in an independent cohort of patients. Candidate makers such as these, if validated in future prospective studies, offer great promise for refining risk classification schemes. Moreover, earlier intensification up front, as opposed to later in therapy, theoretically could offer a greater advantage in preventing the emergence of drug-resistant clones.

Final questions raised by these recent publications include whether early treatment response can be used solely as a prognostic marker, or whether it will have different meaning within different cytogenetic subsets or with different therapies. Borowitz et al, for example, showed that despite a higher frequency of end-induction MRD in children with simultaneous trisomies of chromosomes 4 and 10, this subset of patients had some of the most favorable outcomes. Perhaps the kinetics of response differs among cytogenetic subsets, or exposure to certain agents at different time points in therapy ultimately translates into different responses in varying groups of patients. It is also important to consider differences in initial therapy when comparing the prognostic impact of MRD between studies, as well as the method of MRD determination. Although molecularly based MRD assays may offer greater sensitivity, they are more labor intensive and do not have short turn around times. This could be critical when assigning patients to risk groups early in therapy.

Editor’s note: The outstanding survival outcomes obtained in childhood ALL using modern therapies led some people to suggest that the pediatric oncology community should focus more on improving outcomes of patients with poor outcome cancers (eg metastatic sarcomas, stage IV neuroblastoma, pediatric brain tumors), rather than refining treatment for ‘curable’ cancers (eg ALL, Wilms’ tumor, Hodgkin’s lymphoma). However, the introduction of modern biologic (mostly molecular) markers to assess response has opened new research avenues for these so called ‘curable’ cancers. In Wilms’ tumor for example, LOH at 1p and 16q have been identified as poor prognostic indicators, and investigators are now studying whether more intensive treatment for patients with tumors that show LOH at these chromosomal sites will improve outcome. Similarly, the assessment of early response in ALL is currently used to determine the intensity of treatment. As Raetz and Carroll so aptly describe, the identification of particular ‘gene expression signatures’ may allow for further treatment stratification. While more effective treatment strategies for children predicted to do poorly are certainly needed, it is even more important to identify patients that can be cured with less intensive treatment. A recent study by Oeffinger et al., demonstrates that the term ‘cure’ for children with malignant diseases is a relative notion (NEJM. 2006;355:1572-1582). – Max J. Coppes, MD, PhD, MBA

For more information:

Elizabeth A. Raetz, MD, BS, is an Associate Professor of Pediatrics in the Departments of Pediatric Oncology and Hematology at New York University Cancer Institute. William L. Carroll, MD, is the Julie and Edward J. Minskoff Professor of Pediatrics, Head of the Division of Pediatric Hematology/Oncology and Director of the Stephen D. Hassenfeld Children’s Center for Cancer and Blood Diseases at NYU Cancer Institute and the NYU School of Medicine.

Coustan-Smith E, Ribeiro RC, Stow P, et al. A simplified flow cytometric assay identifies children with acute lymphoblastic leukemia who have a superior clinical outcome. Blood. 2006;108:97-102.

Flotho C, Coustan-Smith E, Pei D, et al. Genes contributing to minimal residual disease in childhood acute lymphoblastic leukemia: prognostic significance of CASP8AP2. Blood. 2006;108:1050-1057.

Borowitz MJ, Pullen DJ, Shuster JJ, et al. Minimal residual disease detection in childhood precursor–B-cell acute lymphoblastic leukemia: relation to other risk factors. A Children’s Oncology Group study. Leukemia. 2003;17:1566-1572.

Coustan-Smith E, Sancho J, Hancock ML, et al. Clinical importance of minimal residual disease in childhood acute lymphoblastic leukemia. Blood. 2000;96:2691-2696.

Schrappe M, Reiter A, Riehm H. Cytoreduction and prognosis in childhood acute lymphoblastic leukemia. J Clin Oncol. 1996;14:2403-2406.

Steinherz PG, Gaynon PS, Breneman JC, et al. Cytoreduction and prognosis in acute lymphoblastic leukemia–the importance of early marrow response: report from the Children’s Cancer Group. J Clin Oncol. 1996;14:389-398.

Gajjar A, Ribeiro R, Hancock ML, et al. Persistence of circulating blasts after 1 week of multiagent chemotherapy confers a poor prognosis in childhood acute lymphoblastic leukemia. Blood. 1995;86:1292-1295.

Reiter A, Schrappe M, Ludwig WD, et al. Chemotherapy in 998 unselected childhood acute lymphoblastic leukemia patients. Results and conclusions of the multicenter trial ALL-BFM 86. Blood. 1994;84:3122-3133.

Gaynon PS, Bleyer WA, Steinherz PG, et al. Day 7 marrow response and outcome for children with acute lymphoblastic leukemia and unfavorable presenting features. Med Pediatr Oncol. 1990;18:273-279.