Abstract
BACKGROUND AND AIMS: Cardiotoxic treatments like anthracyclines and heart-directed radiotherapy increase the risk of cardiac diseases (CDs) in childhood cancer survivors (CCSs), but individual differences in CD incidence are not fully understood. This study aims to identify transcriptomic biomarkers associated with CD occurrence after childhood cancer treatment. METHODS: A matched case-control study was conducted on a sample of 330 CCS: 165 cases with CD and 165 CD-free controls. The expression of 8557 genes was investigated to select those associated with CD and heart failure (HF), using three stabilization approaches derived for the conditional logistic regression with Lasso (Percentile lasso, Bolasso, and Sublasso). The intersection of the three selected gene sets formed the final selection. The interactions between cancer treatment doses and selected genes were investigated. RESULTS: One promising gene, NFE2L2, constituted the final selection, and its expression was lower in cases than in controls [CD: odds ratio (OR) .16, 95% confidence interval (CI) .09-.29; HF: OR .11, 95% CI .03-.37]. No interaction between treatment doses and NFE2L2 expression levels was found in our study. Incorporating NFE2L2 gene expression into prognostic models improved discrimination between cases and controls compared with models based solely on clinical and treatment variables [CD: area under the curve (AUC) .85 vs .66; HF: AUC .87 vs .77]. CONCLUSIONS: Using high-dimensional data selection methods has enabled the identification of the gene NFE2L2, associated with CD and HF in CCS. Further research is needed to validate this finding and achieve a better understanding of the biological mechanisms leading to cardiac toxicities and so to develop risk-adapted treatment and surveillance strategies.