NEW YORK, March 29 - Microarrays can be used to predict patient responses to treatment for pediatric acute lymphoblastic leukemia, a group of researchers at Memphis' St. Jude Children's Research Hospital have found.
In fact, microarrays could be much easier and potentially more accurate than conventional methods even though the cost could be roughly equivalent, the researchers write in a recent paper. Additionally, similar tests with new and denser gene chips will probably uncover additional genes related to ALL subtype and prognosis.
But best of all, the researchers report, identifying genes associated with ALL relapse can point the way to new therapies for varieties of the disease that don't respond to traditional treatment.
Since pediatric ALL is genetically heterogenous, effective treatment depends on identifying which subtype of the disease a patient has developed. Using pathology and cytogenetics to do this is complex and expensive, so the St. Jude researchers turned to microarrays to learn if this technology could be used instead.
The team, lead by St. Jude pathology department chair James Downing, first developed algorithms with microarray data derived from 215 bone marrow samples. His team used 12,600-probe Affymetrix chips.
Data allowed the researchers to come up with predictive gene-expression profiles for each of the six ALL subtypes. Activating only one gene was enough to identify two of the subtypes, the found. The four others required data from between seven and 20 genes in order to accurately classify the disease.
The team then tested their algorithms with 112 test samples and found that gene-expression data could identify the six ALL subtypes with an accuracy ranging from 96 to 100 percent--an improvement over conventional methods.
The researchers' report is published in the March issue of Cancer Cell.
Identifying the tumor subtype allows an oncologist to prescribe an appropriate level of chemotherapy, explains Downing. Similarly, patients identified early on as being at risk of relapse could be slated for more powerful therapy.
For two of those ALL subtypes, the team was also able to identify an expression profile that could predict relapse rates with 97 to 100 percent accuracy. It also identified a set of 20 genes that identify the patients that later develop chemotherapy-related leukemia.
Downing added that the approach needs further prospective testing before it could be used for prognosis, estimating that clinically useful diagnostic microarrays may still be about five years away.