Transcriptomic analysis of childhood Acute Lymphoblastic Leukemia at diagnosis and its association with clinical evolution

RUIZ, MARÍA SOL; COTIGNOLA, JAVIER; RICCHERI, CECILIA; ABBATE, MERCEDES; VAZQUEZ, ELBA

Modalidad Virtual

Congreso; Translational Research E- Conference. ACUTE LYMPHOBLASTIC LEUKAEMIA; 2021

European School of Haematology

Patients diagnosed with Acute Lymphoblastic Leukemia (ALL) are stratified into risk groups based on biochemical parameters, cytogenetic and molecular signatures, and early response to therapy (determined by minimal residual disease at days 15 and 33). While considerable progress has been made on treatment efficacy and survival rates, the course of the disease is still unpredictable in many cases and there are patients that recur and die in all risk groups. In recent years, genome-wide and transcriptome analyses have allowed to identify ALL subgroups with distinctive molecular features. Examples of these molecular subgroups are the Phi-like and IKZF1plus ALL, which have poor outcome. Therefore, identification of new biomarkers or gene-expression profiles would help predicting the disease outcome, improving the response to treatment and reducing therapy-related toxicity. For this purpose, we collected bone marrow samples from 39 de-novo pediatric ALL samples at the time of diagnosis from three hospitals from Argentina. We isolated total RNA and performed paired-end transcriptome sequencing (RNAseq). Clinico-pathological characteristics and disease outcome were evaluated and recorded by trained oncohematologists. We performed differential gene expression analysis between early response to prednisone and occurrence of relapse/death (Event Free Survival, EFS). We considered that genes were differentially expressed if FDR-adjusted p-value≤0.05. Multivariate analyses included transcriptome batch, gender, and ALL risk groups as covariates. In both comparisons, we found that a high percentage of the Differentially Expressed Genes (DEG) corresponded to non-coding RNAs (ncRNA). We detected 161 significant DEG for EFS: 41 (25.5%) were protein-coding genes, 113 (70.2%) were ncRNA genes and 7 (4.3%) corresponded to TEC biotype (To be Experimentally Confirmed). The two predominant biotypes for ncRNAs were long intergenic ncRNA (36.3%) and pseudogenes (37.2%). One of the top dysregulated protein-coding genes was ZFPM2 (log2FC=-22, p-adj=1.5e-24), a transcription factor that modulates the expression of GATA genes (key regulators of hematopoiesis). The top dysregulated ncRNA was RNU4-2, a small nuclear RNA that regulates RNA transport and splicing (log2FC=-8, p-adj=6e-14). In the case of response to prednisone we found 40 DEG (75% protein-coding). We observed that ABCG2 was significantly upregulated (log2FC=3, p-adj=1.5e-3). This gene is an ATP Binding protein that functions as a xenobiotic efflux pump and might play a major role in multi-drug resistance. Among the ncRNAs we found that MIR99AHG (aka MONC) was significantly upregulated in patients with poor response to prednisone (log2FC=5, p-adj=8.4e-5). Interestingly, MIR99AHG was previously associated with myeloid leukemia and gastric cancer acting as oncogene. Overall, we detected significant dysregulation of protein-coding and ncRNAs that might identify ALL subgroups with different outcomes. These results suggest that the analysis of gene expression profiles at diagnosis might help to improve childhood ALL prognosis and identify new potential therapeutic targets.