CIRCULATING EXTRACELLULAR VESICLE DYSFUNCTION AND microRNA-BASED RISK PREDICTION FOR KIDNEY AND CARDIOVASCULAR OUTCOMES IN CKD PATIENTS

Adverse organ crosstalk among the kidney, cardiovascular system, and metabolic pathways has been conceptualized as the “cardiovascular–kidney–metabolic (CKM) syndrome”. Its molecular mechanisms remain largely unknown, and validated prognostic markers are lacking. We previously identified a set of micro (mi) RNAs encapsulated in circulating small extracellular vesicles (cEVs) that are physiologically protective against vascular calcification and are depleted in blood from chronic kidney disease (CKD) model rodents. In this study, we aimed to develop a longitudinal risk prediction model for renal and cardiovascular disease based on miRNA expression within CKD patient-derived cEVs.

We performed cEV-derived miRNAs transcriptome profiling in a derivation cohort of 36 CKD patients not yet receiving kidney replacement therapy (KRT). K-means clustering was used to identify co-expression modules, and a Lasso–Cox procedure was applied to derive a prognostic model. Model coefficients were re-estimated in an independent validation cohort of 234 CKD patients using quantitative PCR data to ensure external calibration. The primary outcome was a 30% decline in eGFR (serum creatinine-based) or initiation of KRT. The secondary outcome comprised a composite of all-cause mortality, initiation of KRT, and hospitalization for major adverse cardiovascular or cerebrovascular events (MACEs).

In the derivation cohort, a cluster enriched for CKD stage G5 showed depletion of 23 cEV-derived miRNAs. KEGG pathway analysis of predicted miRNA targets highlighted cellular senescence as the most significantly enriched pathway. Lasso–Cox selection reduced the variable set to three miRNAs (hsa‑let‑7d‑5p, hsa‑miR‑24‑3p, and hsa‑miR‑126‑3p), forming the “M3 prediction equation”, which demonstrated the strongest discrimination for the primary renal outcome (HR 30.7, 95% CI 3.43–275). In the validation cohort, repeating Lasso–Cox modeling and incorporating serum cystatin C and urine protein produced an optimized model (“M3V2 prediction equation”). Patients with high M3V2 scores showed markedly increased risk for the primary outcome (HR 7.8, 95% CI 4.75–13.0 versus low‑score patients) and for the composite secondary outcome (HR 10.3, 95% CI 4.86–22.0). The M3V2 prediction model outperformed conventional single‑variable predictors in discrimination (C‑index) and calibration; stratified analyses demonstrated consistent performance irrespective of baseline cardiovascular disease or CKD etiology, achieving predictive accuracy comparable to or exceeding CKM syndrome staging.

We characterized a distinct cEV-derived miRNA expression signature in CKD patients and defined coordinated depletion of physiologically important miRNAs as “cEVs functional impairment”. Using these miRNAs, we developed and externally validated a robust longitudinal risk prediction model for CKD progression and related adverse outcomes. Further investigation of mechanisms underlying miRNA depletion and resultant signaling changes in cEVs recipient cells may yield insights into the pathophysiology of CKD and the broader CKM syndrome.

This abstract was also submitted for the ASN Kidney Week 2025 congress.