Identification of a Diagnostic Gene Signature Associated with Centrosome Amplification in Pressure Injuries: A Cross-Sectional Transcriptome and Machine Learning Study

Purpose: Centrosome amplification (CA) contributes to cancer but remains poorly characterized in non-neoplastic conditions such as pressure injuries (PI). This study investigated CA-related genes (CARGs) in PI to identify potential therapeutic targets. Patients and Methods: Transcriptomic data from 15 patients with PI and 15 healthy control blood samples were analyzed. Differentially expressed genes were intersected with CARGs, and key genes were identified using machine learning algorithms and receiver operating characteristic curve analysis. A nomogram was constructed, and underlying mechanisms were investigated using functional enrichment, immune infiltration, drug prediction, and quantitative reverse transcription polymerase chain reaction (RT-qPCR) analyses. Results: GADD45A, LFNG , and DUSP13 were identified as key PI-associated genes, each demonstrating strong diagnostic performance (area under the curve > 0.8). These genes were primarily enriched in the spliceosome pathway. Neutrophil infiltration correlated strongly with all three genes. Decitabine was predicted as a potential agent targeting these genes, with DUSP13 showing strong binding affinity (− 6.9 kcal/mol). RT-qPCR validation confirmed upregulation of GADD45A/DUSP13 and downregulation of LFNG in PI. Conclusion: This study preliminarily identifies GADD45A, LFNG , and DUSP13 as key CA-associated genes in PI. Their expression patterns may provide supplementary molecular evidence to support the early identification and dynamic risk monitoring of high-risk patients. However, the clinical translational potential of these findings requires further validation through large-scale, multicenter prospective studies. Keywords: pressure injuries, centrosome amplification, machine learning, bioinformatics