P336: predictive scores for relapse and progression disease in pediatric acute lymphoblastic leukemia patients based on optical genome mapping technology

Topic: 1. Acute lymphoblastic leukemia - Biology & Translational Research Background: Acute childhood lymphoblastic leukaemia (cALL) is the most common childhood malignancy accounting for 30% of cancers in the paediatric population. Despite improvements in the combination of chemotherapy, haematopoietic cell transplantation and new immunotherapies, a percentage of 10-20% of patients succumb to the disease due to the treatment resistance or relapse. New useful biological markers are needed to redefine stratification criteria into risk groups and to predict the clinical outcome of current treatment protocols. This is particularly important in the group of patients classified as “intermediate risk” that comprises the most abundant number of patients and lack specific genetic characteristics. Aims: To create a predictive prognostic score for ALL stratification and relapse based on global cytogenomic findings of whole genome mapping. Methods: Cellular samples from diagnostic bone marrow aspirates of 63 patients with ALL diagnosed and treated under the SEHOP-PETHEMA 2013 protocol were processed for cytogenetic studies by Optical Genome Mapping (OGM) following a specific protocol from Bionano Genomics. Only patients classified as intermediate risk at the time of diagnosis were chosen for this study. In brief, cALL samples’ ultra-high molecular weight gDNA was extracted, labeled, and loaded into chips for use on the Saphyr platform. The collected data was then analyzed using the Bionano Access 1.6 software. Results: We first generated a binary matrix of genomic variations (presence/ absence) along the whole genome. Multi-Class Discriminant Analysis using Binary Predictors (binda 1.0.4 R package) was used to find relevant cytogenomic features that distinguished patients who upgraded their risk group classification during therapy (score 1), or that eventually relapsed (score 2). A set of chromosomal alterations was chosen using the binda.ranking function. Score 1 was build with 11 chromosomal alterations, variables with highest score were a deletion in 9p21, an insertion in 6p22, a intratranslocation in chromosome 10 t(10q11-10q23) and small gain in 7p13. Score 1 showed an accuracy in our cohort of 80%. Annotation of chromosomal alterations in the score identified CDKN2A as the most relevant gene involved in risk-group progression. Score 2 was built with 21 chromosomal alterations, the five variables with highest score were a deletion in 21q22, gain of one copy of entire chromosome 21, gain in 21q11, small inserion in 15q26 and a deletion in 22q11. Score 2 showed an accuracy of 98% in our cohort. Annotations of the most relevant chromosomal alterations showed genes such as mTOR. A validation process is ongoing to clarify if these scores can be useful in clinical decisions. Summary/Conclusion: We generated two scores to predict risk-group upgrading and leukemic relapse based only on global cytogenomic alterations found at diagnosis. A validation process is ongoing to clarify if these scores could be of help in clinical decisions and improve the current cytogenetics criteria for risk-group classification. Keywords: Pediatric, Acute lymphoblastic leukemia, Chromosomal abnormality, Childhood