Gene Chip Identifies Risk of Relapse in Children With ALL

May 1, 2002

ORLANDO-A gene-profiling chip might help identify children with acute lymphoblastic leukemia (ALL) who are at low risk of relapse and could be spared intensive therapy, or who are at high risk for treatment-induced acute myeloid leukemia (AML) and should not be treated with topoisomerase II inhibitors.

ORLANDO—A gene-profiling chip might help identify children with acute lymphoblastic leukemia (ALL) who are at low risk of relapse and could be spared intensive therapy, or who are at high risk for treatment-induced acute myeloid leukemia (AML) and should not be treated with topoisomerase II inhibitors.

In a plenary presentation at the 43rd Annual Meeting of the American Society of Hematology (abstract 1816), Allen Eng-Juh Yeoh, MD, said that a limited number of genes determine risk in these two areas, and that this might permit development of an inexpensive, simple test that could be used in developing countries as well as in the United States.

Carolyn A. Felix, MD, who introduced Dr. Yeoh’s paper, said that it "takes gene expression profiling to the next level by defining biologic variables that affect prognosis." Dr. Felix is associate professor of pediatrics, Children’s Hospital, Philadelphia.

Dr. Yeoh was presented with an ASH Merit Award for this study, which was done in the laboratory of John R. Downing, MD, at St. Jude Children’s Research Hospital, Memphis. Dr. Yeoh is a clinical fellow in the Departments of Hematology-Oncology and Pathology at St. Jude and assistant professor of pediatrics, National University of Singapore.

Dr. Yeoh said that the 80% cure rate for pediatric ALL achieved at St. Jude and other institutions is the result of stratifying patients by leukemia subtype and tailoring treatment accordingly, but, he added, this requires extensive laboratory studies.

The approach pioneered by his group opens the possibility that a single test might replace molecular screening, immunotyping, and related assays, and also enable the clinician to identify patients with a high risk of relapse and with a high vulnerability to treatment-induced secondary AML.

Study Methods

Dr. Yeoh and his colleagues used oligonucleotide microarrays (gene chips) to analyze the expression of 12,000 genes in diagnostic bone marrow blasts from 393 pediatric ALL patients. Samples had already been analyzed for immunopheno-type, karotype, DNA ploidy, and the presence of chimeric transcripts encoded by the t(12;21), t(1;19), t(9;22), and 11q23 gene rearrangements.

Dr. Yeoh next narrowed this to the 1,200 genes whose expression correlated most closely with the leukemia subtypes of interest. He applied learning algorithms, including support vector machines, to patterns in expression of these 1,200 genes, and was able to devise classifiers that would identify specific genetic and prognostic groups.

High vs Low Risk

When applied to bone marrow samples from 347 pediatric patients with ALL, the proposed classifiers identified distinct gene expression profiles for known ALL subtypes. These ALL subtypes include T-cell, E2A-PBX1, TEL-AML1, rearranged MLL, BCR-ABL, and high hyperdiploid cases.

"This single platform has the potential for replacing all of the exhaustive laboratory studies now required for risk stratification," Dr. Yeoh said.

In patients classified as high risk by the standard National Cancer Institute criteria, the classifiers are able to further divide the high-risk group into a large group of low-risk patients with a 96% cure rate and a very-high-risk group with only a 20% chance of cure (P < .0001), he commented.

Dr. Yeoh pointed out that no single gene expression profile predicted disease relapse across all ALL subtypes, but that an expression profile associated with a high risk of relapse could be identified for each specific subtype. "This ability to predict relapse might enable us to reduce the therapeutic intensity for children who are at low risk of relapse. This needs to be tested in a prospective study," Dr. Yeoh commented.

Predicting Therapy-Induced AML

Dr. Yeoh also identified gene expression profiles for each leukemia subgroup that predicted the development of therapy-induced AML with an accuracy of 99% (P = .031).

"By knowing this in advance, we could avoid the use of topoisomerase-II inhibitors in these patients and probably greatly reduce the incidence of treatment-induced AML," Dr. Yeoh said.

Interestingly, the genes involved in this marker include an ATP transferase, a suppressor of the Ras pathway, and a mismatch repair enzyme, he said.

This study used the Affymetrix HG-U95Av2 gene chip. Dr. Yeoh told ONI that prices for these gene chips have already dropped from about $2,000 each to about $400.

His goal is a mass-produced "McJude chip" built around the relatively small number of genes whose expression determines subtype and prognosis in pediatric ALL, and cheap enough for use in developing countries, which seldom have the resources for the elaborate assays now required.