Extracellular vesicles: Emerging biomarkers and therapeutic targets in sepsis-associated acute kidney injury

Acute kidney injury (AKI) affects up to 50% of septic patients and carries high risks of mortality, chronic kidney disease (CKD), and long-term disability.[1] Despite advances in supportive care, early AKI diagnosis and mechanism-targeted therapies remain elusive. Current diagnostic markers—serum creatinine and urine output—are nonspecific and reflect dysfunction only after significant kidney damage has occurred. The 28th Acute Disease Quality Initiative Workgroup defines sepsis-associated AKI (SA-AKI) as kidney injury, according to kidney disease: improving global outcomes occurring within 7 days of sepsis onset, as defined by Sepsis-3 criteria.[1] Several novel biomarkers, such as the product of tissue inhibitor of metalloprotease-2 and insulin-like growth factor-binding protein-2 ([TIMP-2] * [IGFBP7]), neutrophil-gelatinase associated lipocalin (NGAL), and proenkephalin A, have shown promise, but none have been universally adopted in routine care. SA-AKI is pathophysiologically heterogeneous, notably microcirculatory dysfunction, dysregulated inflammation, complement activation, and mitochondrial injury. Broad-spectrum immunomodulatory medications to treat inflammatory pathways can be dangerous in sepsis, as they may disable essential host-defense mechanisms for fighting infection. This complexity highlights the need for precision medicine tools, such as biomarkers that capture early mechanistic signatures and therapies targeting specific pathways to halt disease progression and facilitate repair. Given the complexity, attention has turned to extracellular vesicles (EVs), which are lipid bilayer-bound nanoparticles released by nearly all cell types, and they carry proteins, nucleic acids, metabolites, and lipids that reflect the state of their parent cells. In SA-AKI, EVs may act as mediators, propagating inflammatory and injurious signals, and as waste disposal mechanisms, removing obsolete organelles and molecules from cells. EVs are detectable in all biofluids, with urine and blood being the most frequently examined in research. EV cargos may provide earlier and more mechanistically informative signals of kidney injury than serum creatinine or urine output. Recent advances in EV isolation and detection technologies further support this potential. Nanoflow cytometry enables high-resolution particle counting and single-EV fluorescence analysis, while microfluidic chip-based platforms allow rapid, low-input profiling of EV surface proteins and nucleic acids. These approaches complement EV omic (proteomic/transcriptomic/metabolomic) strategies and may facilitate scalable, standardized EV assays for clinical application. In a rat model of SA-AKI by liposaccharide injection, serum EV-derived miR-181a-5p and miR-23b-3p (i.e., microRNAs implicated in inflammatory signaling and immune regulation) were detected at increased levels 2 h before the earliest significant elevation in serum creatinine at 4 h post-injection.[2] In another cecal ligation puncture mouse model of sepsis, activating transcription factor (ATF3) expression, a stress-responsive transcription factor linked to cellular injury and repair, was detected in kidney tissue before creatinine rise, but urinary measurement was not possible due to anuria.[3] In contrast to the animal studies, clinical studies showed that urinary exosomal ATF3 increased at the time of AKI onset (rather than before) and discriminated SA-AKI from sepsis without AKI with an area under the receiver operating characteristic curve (AUROC) of 0.84, outperforming urinary NGAL (AUROC = 0.64).[3] Urinary EV (uEV)-derived sodium-hydrogen exchanger 3 was increased across several experimental AKI models in rats, pointing toward disturbed tubular ion transport and epithelial cell polarity, and was also elevated in a small cross-sectional cohort of sepsis-AKI patients compared with healthy volunteers.[4] However, these human data remain preliminary, and the temporal relationship with creatinine rise has not been established. Together, these findings suggest that EV cargo reflects early molecular responses to SA-AKI and holds promise for developing sensitive, mechanism-based biomarkers that may precede traditional functional indicators such as serum creatinine. Beyond RNA-based signals, EV metabolomic profiling has been demonstrated in drug-induced AKI models.[5] In a recent study, serum analyses revealed early metabolic perturbations in SA-AKI; although EVs were not investigated, the findings support the potential of EV-based metabolomics for future biomarker development.[6] Together, these approaches underscore the promise of EV transcriptomics, proteomics, and metabolomics to enrich the SA-AKI biomarker landscape. Translating these preclinical findings to human disease, recent clinical studies have begun to test EV-derived biomarkers in sepsis patients, illustrating both their promise and limitations. In a multicenter proteomic analysis, uEV CD35 discriminated SA-AKI from sepsis without AKI with high accuracy (AUROC = 0.89) and identified subclinical AKI (AUROC = 0.84). By comparison, serum creatinine at the same time point had no diagnostic value (AUROC is about 0.50).[7] Its prognostic performance was more modest for persistent AKI (AUROC = 0.77), mortality (AUROC = 0.70), and acute kidney disease progression (AUROC = 0.66).[7] Mortality prediction was limited by only 10 deaths, restricting statistical power. In the same study, CD35-uEV levels (AUROC = 0.89) outperformed the urinary [TIMP-2] * [IGFBP7] test for early AKI diagnosis (AUROC = 0.77; P = 0.019), and the combination of both reached AUROC of 0.87, although external multicenter confirmation was lacking. Similarly, in a prospective study of 97 sepsis patients, urinary miR-452 emerged as an early marker of AKI (AUROC = 0.90, sensitivity 87.2%, and specificity 78.0%), outperforming the established [TIMP-2] * [IGFBP7] test.[8] By contrast, another study found that urinary miR-376b was significantly decreased in septic AKI patients compared with sepsis controls, correlating with kidney dysfunction in both animal models and humans.[9] Collectively, these studies highlight both the diagnostic potential and current fragmentation of EV-based research—suggesting that while EV-derived markers may capture distinct molecular signatures of renal injury and surpass established biomarkers in early detection, their clinical translation is still constrained by small cohorts, methodological variability, and the absence of standardized validation frameworks. Blood EVs can be obtained even in oligoanuria but contain vesicles from all organs, making kidney-specific signals difficult to isolate. By contrast, uEVs are enriched for kidney-derived vesicles, providing stronger signals for kidney metabolism and function.[10] Indeed, nephron mass is associated with the excretion rate of uEVs, highlighting the dynamic relationship between kidney function and uEV release.[10] Beyond their role as biomarkers, EVs also hold promise as therapeutic agents. EVs contribute to the propagation of inflammation in SA-AKI by increasing the expression of proinflammatory cytokines (miR-181a-5p and miR-23b-3p)[2] and promoting M1 macrophage activation through miR-19b-3p.[11] These novel pathways could be targeted to prevent further kidney damage in SA-AKI. The EV-associated cargos implicated in these processes include the immunoregulatory protein CD5L, miRNAs such as miR-19b-3p, and EV-associated lipids, which together may contribute to the immunomodulatory and reparative potential of EVs. Preclinical studies support this therapeutic potential. In a murine sepsis model, fibroblastic reticular cell-derived exosomes targeted injured kidneys, promoted PTEN-induced putative kinase 1 (PINK1)/Parkin-mediated mitophagy, and suppressed NOD-like receptor thermal protein domain associated protein 3 (NLRP3) inflammasome activation.[12] These effects were largely mediated by the immunoregulatory protein CD5L and were enhanced when exosomes were engineered with kidney injury molecule-1-targeting peptides to improve renal specificity. Such approaches remain experimental, and their translation will require scalable Good Manufacturing Practice (GMP)-grade manufacturing, rigorous preclinical safety testing, and regulatory approval pathways. For engineered EV-based therapies, several additional considerations are critical. Kidney-targeting ligands must be selected based on affinity, specificity, and ability to promote efficient internalization into renal tubular cells. The choice of source cells influences immunogenicity: Autologous EVs may reduce immune risks, whereas allogeneic or cell line-derived EVs offer greater scalability but necessitate stringent safety evaluation. Scalable manufacturing also remains a key challenge, as serum-free culture systems, standardized purification methods, and GMP-compliant production workflows will strongly determine the feasibility of clinical translation. Regulatory agencies, such as the US Food and Drug Administration (USFDA) and the European Medicines Agency (EMA), also emphasize the need for consistent production, quality control, and first-in-human safety data before kidney-targeted EV therapeutics can advance to clinical trials. The integration of EV-based diagnostics and therapeutics aligns with the Acute Dialysis Quality Initiative (ADQI)-endorsed vision of phenotyping SA-AKI into mechanistically distinct endotypes to guide targeted interventions. In this context, EV profiling can achieve the following: 1. Enable early detection by capturing molecular injury signatures before overt changes in kidney function. 2. Guide risk stratification by identifying patients at risk of maladaptive repair following AKI. 3. Support therapeutic monitoring by tracking dynamic biological responses to interventions. 4. Deliver targeted therapies by using engineered EVs to modulate injury pathways, such as mitochondrial dysfunction or inflammasome activation. Notably, EV research in SA-AKI must address translational hurdles before clinical application. As highlighted by the International Society for Extracellular Vesicles (ISEV) Urine Task Force, preanalytical variability, including urine collection timing, handling, storage, and normalization, can significantly affect EV yield and biomarker signal[13] [Figure 1]. Standardization and reproducibility across centers are essential, and the most promising biomarkers require large, multicenter validation with clinically actionable thresholds. A combined plasma-uEV approach may therefore deliver a more comprehensive picture of SA-AKI pathophysiology and enable multicompartmental phenotyping.Figure 1.: Minimum reporting standards from ISEV’s urine task force. ISEV, International Society for Extracellular Vesicles; uEVs, urinary extracellular vesicles. CONCLUSION SA-AKI remains a major unmet challenge in critical care nephrology. Despite advances in critical care, clinicians still rely on late functional markers, such as serum creatinine and urine output, to diagnose SA-AKI, highlighting the need for molecular tools that enable earlier and more precise intervention. EVs, by virtue of their dual role as biomarkers and potential therapeutic vectors, offer a promising pathway for both early detection and mechanism-directed intervention. For the intensivist, EV profiling could refine clinical decision making by enabling earlier recognition of renal injury, stratifying risk, and guiding monitoring. Realizing this potential will require standardized protocols, harmonized validation, integration into clinical biomarker panels, and regulatory-ready assays. With coordinated translational efforts and multicenter validation, EV-based diagnostics and therapeutics could redefine the clinical approach to SA-AKI, moving the field forward toward precision phenotyping and targeted intervention.