Deciphering the circadian signature in acute myeloid leukemia: NPAS2 downregulation as a master predictive feature

Purpose Considering the role of circadian clock genes in leukemogenesis through the regulation of cell cycle progression and genomic stability, this study aimed to evaluate the expression of NPAS2, CIPC, BMAL1, and CLOCK in acute myeloid leukemia (AML) and to determine their individual and combined diagnostic performance. Methods In this study, bone marrow (BM) and peripheral blood (PB) samples were obtained from newly diagnosed AML patients and healthy controls under standardized circadian conditions. Gene expression was measured by TaqMan RT-qPCR and normalized to ABL using the 2ΔΔCt method. Diagnostic performance was evaluated by ROC analysis and multivariate logistic regression, and a random forest classifier with SHAP-based feature importance was applied to assess predictive contribution of each gene. Results PB showed comparable expression levels to paired BM samples, supporting its use as a surrogate tissue for circadian biomarker assessment. All transcripts were significantly downregulated in AML compared with controls (p < 0.001), with NPAS2 showing the most pronounced reduction. NPAS2 exhibited the highest diagnostic performance (AUC = 0.925). The random forest classifier achieved 74% sensitivity and 100% specificity, and SHAP analysis identified NPAS2 as the most influential predictive feature. Conclusion As such, we propose downregulation of NPAS2 as a potential biomarker for diagnosis of AML.