Authors: Donya Ghazinader1 | Thomas McLarnon1 | Frank McCarroll2 | Steven Watterson1 | Taranjit Singh Rai1*

Affiliations: 1 Personalised Medicine Centre, School of Medicine, Ulster University, Derry~Londonderry, Northern Ireland, BT48 7JL, UK. 2 Western Health and Social Care Trust, Altnagelvin Area Hospital, Derry~Londonderry, Northern Ireland, BT47 6SB, UK

Background/ Introduction: Acute kidney injury (AKI) and chronic kidney disease (CKD) are closely interconnected, yet the molecular mechanisms driving AKI-to-CKD progression remain incompletely understood. Incomplete renal repair after AKI may lead to persistent injury, maladaptive remodelling and irreversible CKD. Early identification of patients at risk is therefore essential, but reliable prognostic approaches at the time of AKI remain limited. Cellular senescence may contribute to this transition by promoting inflammation, fibrosis and impaired tissue recovery.

Material & Methods: Plasma proteomic data from AKI patients were analysed to identify biomarkers associated with subsequent CKD development. Differential protein expression analysis was performed to compare AKI-only patients with AKI-to-CKD progressors. Pathway enrichment analysis was used to explore biological mechanisms linked to progression. Feature selection and machine-learning modelling were then applied to identify the strongest predictive protein signature and evaluate its classification performance.

Results: The analysis identified a five-protein signature that distinguished AKI patients who progressed to CKD from those who did not. Although feature selection was unbiased, all five selected proteins were associated with cellular senescence, supporting a potential role for senescence-related biology in AKI-to-CKD transition. The model achieved 90.9% accuracy, 90.0% sensitivity and 91.7% specificity. Enrichment analysis highlighted pathways relevant to inflammation, stress signalling and tissue remodelling, including Wnt, cytokine-cytokine receptor interaction, NF-κB, TNF and MAPK signalling.

Conclusion: Senescence-associated plasma proteomic signatures may help identify AKI patients at increased risk of CKD progression. These findings support the potential of proteomics and machine learning for data-driven risk prediction and provide mechanistic insight into maladaptive renal repair following AKI.