Background: Diabetic cardiomyopathy (DCM) is a serious cardiovascular complication specific to diabetes mellitus, with rising global prevalence. Ferroptosis, an iron-dependent form of regulated cell death driven by lethal lipid peroxidation, has been implicated in the pathogenesis of DCM. However, the key regulatory genes remain poorly characterized. This study aimed to identify and validate ferroptosis-related signature genes in DCM. Methods: Three murine transcriptomic datasets (GSE123975, GSE155377, and GSE210611) were retrieved from GEO and merged after batch correction. Differentially expressed genes were intersected with weighted gene co-expression network analysis disease-associated module genes and FerrDb ferroptosis annotations to define the ferroptosis-related differentially expressed gene candidate pool. LASSO regression and random forest selection then prioritized hub genes, defined operationally as candidates consistently prioritized by both machine-learning algorithms rather than by network-topological centrality. Classification performance was evaluated by ROC analysis and validated in two independent cohorts (GSE161931 and GSE274500). mMCPcounter estimated immune and stromal infiltration. ScRNA-seq (GSE290095) and spatial transcriptomic (GSE290094) profiling characterized cellular distribution, predicted cardiomyocyte network perturbations and tissue-level expression patterns. High-fat diet/streptozotocin (HFD/STZ)-induced DCM rat models provided experimental validation. Results: Acot1 and Txnip were identified as hub genes, with strong discriminatory performance in the discovery cohort (AUC = 1.000 and 0.988; in-sample estimates, n = 26) and independent external validation (AUC = 0.951 and 0.988). Immune profiling linked both genes inversely with vessel scores, and Txnip was also linked with eosinophils. Single-cell analysis localized Acot1 enrichment to cardiomyocytes and endothelial cells, while Txnip was broadly expressed across multiple cell types, with elevated levels in DCM. In silico knockout analysis predicted distinct cardiomyocyte network perturbation profiles for Acot1 and Txnip, and spatial transcriptomics revealed modest but disease-specific spatial associations between hub gene expression and ferroptosis driver scores (Acot1: rho = 0.123; Txnip: rho = 0.154). Both genes were significantly upregulated at mRNA and protein levels in HFD/STZ-induced DCM rats, with concurrent GPX4 depletion, ACSL4 accumulation, and FTH1 reduction consistent with ferroptosis activation. Conclusions: This study identifies Acot1 and Txnip as ferroptosis-related molecular signatures in DCM and provides multistep prioritization and validation spanning bulk transcriptomics, single-cell and spatial transcriptomics, and in vivo experimental verification, offering potential targets for ferroptosis-targeted therapeutic intervention.

Ferritin heavy Chain 1 (FTH1) is a key protective factor against ferroptosis in the heart, yet its protein abundance can diverge from transcript levels in cardiac ferroptosis contexts. Since resistance to ferroptosis depends on FTH1 protein abundance rather than transcript levels and its ferroxidase activity cannot be readily compensated by other ferritin subunits, this mismatch is functionally consequential. This review organizes the evidence into three regulatory layers acting downstream of transcription. Iron regulatory protein/iron‐responsive element (IRP/IRE) signaling gates ribosome recruitment at the 5 ^′ ‐untranslated region before translation begins. Epitranscriptomic marks and RNA‐binding proteins (RBPs) then route the transcript toward decay or translation, with the N ⁶ ‐methyladenosine (m ⁶ A) reader YTHDF2 acting in opposite directions across cardiac contexts. Finally, nuclear receptor Coactivator 4 (NCOA4) delivers assembled ferritin for lysosomal degradation, depleting FTH1 protein independently of transcript level. Together, these mechanisms provide an integrated framework for understanding the mRNA–protein discordance of FTH1 and its contribution to cardiac ferroptosis susceptibility. We also discuss how this framework may inform layer‐specific therapeutic strategies and identify the measurements that are missing (labile iron pool, ferritin iron loading, and ferritinophagic flux) before mechanistic insight can translate into clinical cardioprotection.