Genome-wide Single Nucleotide Polymorphism (SNP) Array-based Approach to Predict Chemoresponse and Survival in Patients With Acute Lymphoblastic Leukemia

Acute lymphoblastic leukemia (ALL) is not a single disease, but a composite of heterogeneous subgroup. Accordingly, more sophisticated classification in ALL is essential to achieve further improvement of treatment outcomes. However, only a few genetic markers are revealed to have significant prognostic implications in ALL patients. The current study is designed to stratify the ALL patients according to their prognosis and to predict their outcomes by a pharmacogenetic approach. A predictive model will be generated from 130 genotypes in adult ALL patients diagnosed at the Samsung Medical Center (SMC), Sungkyunkwan University School of Medicine, Seoul,Korea between 1994 and 2008. The validation of the predictive model will be performed using an independent external cohort of ALL patients. 1. Definition of the cohort: two hundred ALL patients from the SMC as a test set, another 100 patients from the SMC as a first validation set, and another 150 independent external patients as second external validation set. DNAs will be extracted and stored from patients' samples collected at the time of diagnosis. 2. In the test set, genotypes will be determined using a MALDI-TOF based platform (Sequenom, San Diego, CA, USA) for 130 SNPs of the candidate genes involved in DNA repair pathway, drug metabolism/transport pathway and folate metabolism pathway. 3. Bioinformatic analyses will be performed to identify around 13 genotypes (10%) having strongest predictive significance out of these 130 SNPs in terms of their treatment outcomes, drug toxicity and prognosis in the test set. 4. These 13 genotypes will be validated in the first cohort of 100 ALL patients using a multivariate Cox's proportional hazard model. 5. The predictive model will be built up based on Cox's proportional hazard model derived from 13 candidate genotypes and clinical risk factors. 6. The predictive model based on pharmacogenetic information will be validated again in the second, independent external cohort of