Development and evaluation of a dark kinase–based transcriptional score in acute myeloid leukemia.

Abstract

e22641 Background: Dark kinases regulate cellular stress responses and transcriptional plasticity but remain largely unexplored in acute myeloid leukemia (AML). We hypothesized that coordinated dark kinase overexpression defines a high-risk transcriptional state and that a composite Dark Kinase Stress Score (DKSS) provides prognostic information beyond global transcriptomic variation. Methods: RNA-sequencing expression data and clinical annotations were obtained from The Cancer Genome Atlas (TCGA) AML cohort. Patients were stratified by cytogenetic risk. Dark kinases associated with overall survival (OS) on univariable Cox analysis were retained. DKSS was calculated as the mean of z-score–standardized expression values. Associations between DKSS and OS were assessed using Cox proportional hazards models adjusted for age. DKSS was analyzed per standard deviation (SD) increase and by quartiles. Model discrimination was evaluated using Harrell’s concordance index (C-index). Robustness was assessed using leave-one-out analyses. Independence from proliferation and global transcriptional programs was evaluated by adjustment for MKI67 expression and the first principal component (PC1). Results: In the full AML cohort (n = 140), cytogenetic risk was strongly prognostic, with intermediate/normal risk (HR 2.88; p = 0.003) and poor risk (HR 4.81; p < 0.001) compared with favorable risk. DKSS showed evidence of effect modification, with amplified risk in poor-risk AML (HR 1.95, 95% CI 0.88–4.30; p = 0.098). Primary analyses focused on poor-risk AML, where 32 patients with complete data were included (23 deaths, 71.9%). In age-adjusted models, DKSS was strongly associated with inferior OS (HR 2.27, 95% CI 1.41–3.65; p < 0.001), while age remained independently prognostic (HR 1.01 per year, 95% CI 1.01–1.02; p = 0.032). Model discrimination was high (C-index 0.738). Compared with the lowest DKSS quartile, the highest quartile had a markedly increased risk of death (HR 12.18, 95% CI 2.82–52.62; p < 0.001), with progressive risk escalation across quartiles (C-index 0.745). DKSS remained independently prognostic after adjustment for proliferation (HR 2.27; p < 0.001), whereas MKI67 was not associated with survival (p = 0.953). Adjustment for PC1 yielded similar effect sizes (HR 2.20, 95% CI 0.92–5.29; p = 0.078). Leave-one-out analyses confirmed score stability. CDK11B was the strongest individual kinase (HR 2.74, 95% CI 1.66–4.52; p < 0.001). Conclusions: DKSS identifies a high-risk transcriptional state in AML that is independently associated with inferior survival, with preferential amplification in cytogenetically poor-risk disease. Coordinated dark kinase activation captures aggressive biology beyond proliferation and global transcriptional variation, supporting DKSS as a novel prognostic framework and a rationale for dark kinase–directed therapeutic strategies in high-risk AML.