Genetic and Epigenetic Biomarkers Associated with Early Relapse in Pediatric Acute Lymphoblastic Leukemia: A Focused Bioinformatics Study on DNA-Repair Genes

Abstract

Genomic instability is one of the main drivers of tumorigenesis and the development of hematological malignancies. Cancer cells can remedy chemotherapeutic-induced DNA damage by upregulating DNA-repair genes and ultimately inducing therapy resistance. Nevertheless, the association between the DNA-repair genes, drug resistance, and disease relapse has not been well characterized in acute lymphoblastic leukemia (ALL). This study aimed to explore the role of the DNA-repair machinery and the molecular mechanisms by which it is regulated in early- and late-relapsing pediatric ALL patients. We performed secondary data analysis on the Therapeutically Applicable Research to Generate Effective Treatments (TARGET)-ALL expansion phase II trial of 198 relapsed pediatric precursor B-cell ALL. Comprehensive genetic and epigenetic investigations of 147 DNA-repair genes were conducted in the study. Gene expression was assessed using Microarray and RNA-sequencing platforms. Genomic alternations, methylation status, and miRNA transcriptome were investigated for the candidate DNA-repair genes. We identified three DNA-repair genes, ALKBH3, NHEJ1, and PARP1, that were upregulated in early relapsers compared to late relapsers (p < 0.05). Such upregulation at diagnosis was significantly associated with disease-free survival and overall survival in precursor-B-ALL (p < 0.05). Moreover, PARP1 upregulation accompanied a significant downregulation of its targeting miRNA, miR-1301-3p (p = 0.0152), which was strongly linked with poorer disease-free and overall survivals. Upregulation of DNA-repair genes, PARP1 in particular, increases the likelihood of early relapse of precursor-B-ALL in children. The observation that PARP1 was upregulated in early relapsers relative to late relapsers might serve as a valid rationale for proposing alternative treatment approaches, such as using PARP inhibitors with chemotherapy.

Keywords: DNA repair; PARP1; acute lymphoblastic leukemia; childhood ALL; early relapse; late relapse; miRNA; precursor-B-ALL; prognostic biomarker.

Figure 1

Significantly upregulated DNA-repair genes in early-relapsing patients compared with late-relapsing patients. (A) 18 and 7 genes were found overexpressed in the early-relapse group relative to the late-relapse group using microarray and RNA seq, respectively. ALKBH3, NHEJ1, and PARP1 were found upregulated in the early-relapsing group (red) relative to the late-relapsing group (purple) in microarray (B) and RNA seq (C) datasets. Bold intermittent lines represent the mean values, while light intermittent lines represent 95% confidence interval values. Statistical analysis was computed using Student’s t-test.

Figure 2

Free survival curves for low-expression (blue) and high-expression patients (red) based on the expression of ALKBH3, NHEJ1, PARP1, and the three candidate genes combined (microarray). The shaded area represents the 95% confidence interval (CI) for each curve. Hazard ratio (HR) and p value were calculated using the log–rank test.

Figure 3

Free survival curves for low-expression (blue) and high-expression patients (red) based on the expression of ALKBH3, NHEJ1, PARP1, and the three candidate genes combined (RNA sequencing). The shaded area represents the 95% confidence interval (CI) for each curve. Hazard ratio (HR) and p value were calculated using a log–rank test.

Figure 4

Survival curves for low-score patients (blue) and high-score patients (red) based on the expression of miR-1301-3p. (A) miR-1301-3p expression levels among early (red) and late relapsers (purple). Bold intermittent lines represent the mean values, while light intermittent lines represent 95% confidence interval values. Statistical analysis was computed using Student’s t-test. (B) validated binding sites between PARP1 and miR-1301-3p as identified by [43]. (C) The shaded areas represent the 95% confidence interval (CI) for each curve. Hazard ratio (HR) and p-value were calculated using a log–rank test.