Proteomic analysis of plasma extracellular vesicles for predicting acute kidney injury following laparoscopic radical nephrectomy

Abstract

Acute kidney injury (AKI) is a common complication following laparoscopic radical nephrectomy (LRN). Emerging evidence implicates extracellular vesicles (EVs) as promising candidates for diagnostic biomarkers in disease surveillance, therapeutic efficacy evaluation, and prognostic stratification. Currently, there is a lack of research on preoperative plasma EVs protein markers for predicting AKI after LRN. Plasma samples were obtained prior to the induction of anesthesia. Plasma-derived EVs were isolated using combinatorial chromatography and analyzed proteomically through off-label quantification (4D-DIA) from 5 patients with postoperative AKI and 5 patients without. Subsequently, plasma samples from an additional cohort comprising 28 AKI patients and 24 no_AKI patients were collected to validate differential protein expression in EVs. A total of 43 differential proteins were identified in EVs extracted from plasma samples, including 1 up-regulated protein and 42 down-regulated proteins. Subsequent validation demonstrated that the mean levels of Poly(rC)-binding protein 2 (PCBP2) and Profilin 1 (Pfn1) were significantly lower in AKI patients compared to no_AKI patients, yielding an AUC of 0.893 and 0.923 (95% CI: 0.776 –0.962; 0.814 –0.978), a sensitivity of 92.9% and 89.3%, and a specificity of 83.3% and 87.5%, respectively. This study employed 4D-DIA proteomic analysis and identified 43 differentially expressed proteins, among which PCBP2 and Pfn1 in plasma EVs were significantly downregulated in AKI patients and demonstrated good predictive ability for AKI. Moreover, preoperative risk stratification based on their optimal cutoff values showed high consistency with the postoperative KDIGO clinical diagnostic criteria.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-40480-z.

Introduction

Renal cell carcinoma is a common disease within the field of urology, with laparoscopic radical nephrectomy (LRN) being the preferred therapeutic intervention. Acute kidney injury (AKI) is a common complication in the postoperative period, attributable to factors such as the removal of the affected kidney, insufficient functional reserve of the contralateral kidney, ischemia-reperfusion injury, comorbidities, and the administration of nephrotoxic medications. AKI is characterized by a swift decline in estimated glomerular filtration rate (eGFR), reduced urine output, and elevated levels of serum creatinine (sCr) and blood urea nitrogen (BUN), which is associated with an increased risk of adverse outcomes, including admission to the intensive care unit, the necessity for renal replacement therapy, extended hospitalization, elevated healthcare costs, and higher 30-day readmission rates. The incidence of postoperative AKI following LRN is reported to be 50%~60%, with affected patients experiencing chronic renal insufficiency and elevated long-term mortality, ultimately leading to a poorer prognosis^1–4.

Although AKI is typically defined by alterations in sCr and urine output, these parameters have limitations, urine output exhibits low specificity and can be influenced by factors such as hypovolemia and the use of diuretics. sCr levels exhibit low sensitivity in individuals with normal pre-morbid renal function and a eGFR within the normal range, only becoming elevated after a loss of at least 50% of functional renal units. In patients with a low baseline eGFR, even minor alterations in renal function can result in a 0.3 mg/dL increase in sCr. Furthermore, sCr levels rise during a rapid decline in eGFR until a new equilibrium is established, with the magnitude of sCr increase being contingent upon the extent of eGFR reduction and the prolongation of sCr’s half-life. Under physiological conditions, the half-life of sCr elimination is approximately 4 h. It increases with decreasing eGFR and may take 24–36 h from the onset of renal injury to reach a new steady state, leading to a delay in recognizing AKI. The concentration of sCr is influenced by various factors, including dietary intake, medication use, muscle mass, gender, and race. Consequently, there is a pressing need for novel biomarkers to facilitate the early diagnosis of AKI development^5,6.

Extracellular vesicles (EVs), secreted by all somatic cells, including tumor cells, facilitate numerous biological processes such as intercellular communication, inflammation, immune response, and disease progression. The functional activity of EVs is mediated by proteins, nucleic acids, and lipids, and they have potential applications as diagnostic tools, efficacy predictors, and prognostic biomarkers^7–9. Research indicates that urine-derived EVs can reflect abnormal functions of renal endothelial cells, tubular cells, and podocytes, and the proteins, RNAs, and other molecules they carry potentially serving as early diagnostic markers for AKI¹⁰. However, it remains uncertain whether protein biomarkers present in plasma EVs prior to surgery can predict AKI following LRN. This study focused on LRN patients with normal preoperative renal function, collected preoperative plasma EVs from these individuals, and employed mass spectrometry to screen and validate differentially expressed proteins in EVs between patients who developed AKI and those who did not. The objective is to identify preoperative and more specific novel biomarkers that may effectively predict its occurrence.

Materials and methods

Collection of clinical specimens

This study is an observational, prospective cohort investigation, which has been approved by the Medical Ethics Committee of the Third Medical Center of PLA General Hospital (Approval No. KS2024-001) and registered with the Chinese Clinical Trial Registry (Registration No. ChiCTR2400084174), and informed consent was signed with the patients or their guardians. The study included 62 LRN patients treated at the third Medical Center of PLA General Hospital between May 2024 and July 2024.

The inclusion criteria were as follows: (1) Patients undergoing LRN under general anesthesia at the hospital. (2) Age 40–75. (3) Patients with American Society of Anesthesiologists (ASA) grades I-II. (4) Patients with normal preoperative cardiac, liver, and renal functions. The exclusion criteria were as follows: (1) Patients with a solitary kidney. (2) Patients who received nephrotoxic drugs preoperatively. (3) Patients with a history of previous kidney surgery. (4) Patients who eGFR < 60 mL/min/1.73m². (5) Patients who experienced serious complications during or after surgery.

For all enrolled patients undergoing LRN, a 5 mL peripheral venous blood sample was collected prior to the induction of anesthesia. The samples were immediately centrifuged at 3000 g for 10 min at 4 °C to obtain the supernatant, which was subsequently stored at -80 °C. At this point, none of the patients exhibited AKI. Postoperative AKI was evaluated according to the Kidney Disease: Improving Global Outcomes (KDIGO) criteria. Plasma-derived EVs were isolated from 5 patients who developed postoperative AKI and 5 patients who did not for the initial discovery-phase proteomic analysis. Subsequently, an independent validation cohort comprising 28 patients with postoperative AKI and 24 patients without AKI was enrolled. All subsequent statistical modeling and validation analyses were performed exclusively on this independent validation cohort. The sample size of the validation cohort was determined to ensure a significance level (α) of 0.05 and 80% power to detect an effect size (Cohen’s d = 0.8), based on prior pilot data.

Demographic and clinical variables including patient age, gender, BMI, preoperative comorbidities (hypertension, diabetes, and cardiovascular disease), healthy kidney volume, sCr before and 48 h after surgery, eGFR were estimated by applying the Cockcroft-Gault formula: Male eGFR (mL/min/1.73 m²)=(140 - age) × body weight (kg)/[0.818 × sCr (µmol/L)], female eGFR (mL/min/1.73 m²)=eGFR (male) × 0.85. Intraoperative details were also documented, including the volume of mean arterial pressure, volume of crystalloid infusion, urine output, blood volume, and surgery time. To quantify the expected decline in eGFR following LRN, the predicted new baseline glomerular filtration rate (NB-GFR) was estimated for each patient using a formula derived from previous studies: Predicted NB-GFR = 35.03 + 0.65 × preoperative eGFR − 18.19 × (if radical nephrectomy) − 0.25 × age + 2.83 × (if tumor size > 7 cm) − 2.09 × (if diabetes)¹¹.

According to the diagnostic criteria of KDIGO guidelines: AKI was diagnosed when sCr increased ≥ 26.5 µmol/L within 48 h or sCr increased to 1.5 times of the baseline value within 7 d, or when urine output was < 0.5 mL/(kg·h) for 6 h or more.

EVs purification and characterization

Plasma-derived EVs were isolated with a combined chromatography kit (EVTMC, EVbio, China), following the manufacturer’s instructions. Briefly, 100 µL plasma were loaded to the top of the column pre-packed with 500 µL of Sepharose CL-6B resins (Cytiva). The flow-through was discarded under gravity. Then 200 µL PBS was added to the top of the column. 200 µL filtrate was collected and added to the column containing a specific mixture of Fractogel EMD SO3-(M) (MilliporeSigma) and Capto Core 700 (Cytiva). After incubation, the columns were centrifuged at 500 x g for 30 s to separate the EVs. The isolated EVs were prepared for further downstream analysis¹².

Transmission electron microscopy (TEM)

TEM was utilized to examine the morphological features of the EVs. The EV samples were directly placed on the specimen mesh and subsequently stained with uranyl acetate solution. The samples were then observed under the microscope (JEOL, Japan).

Nanoparticle tracking analysis (NTA)

NTA was performed to detect the concentration and particle size distribution of exosomes. 5 µL aliquot of each fraction was taken and diluted 1:1000 in PBS, then particle counts and particle size distribution were determined using the Zetaview System (Particle Metrix, Diessen, Germany). The principle is to track and analyze the Brownian motion of each particle, calculate the hydrodynamic diameter and concentration of the nanoparticles in conjunction with the Stockes-Einstein equation. For NTA measurement, three cycles were performed by scanning 11 cell positions each and capturing 30 frames per position.

EVs protein extraction

300 µL of 8 M urea was added to the EV sample and the protease inhibitor was added at 10% of the lysate. After centrifuging at 14,100 g for 20 min, the supernatant was collected. The protein concentration was determined using Bradford method, and the rest was frozen to -80 °C.

EVs protein digestion and desalination

A 10 µg aliquot of extracted proteins from each sample was then subjected to reduction. An appropriate amount of protein was added to achieve a final concentration of 10 mM TCEP and 25 mM CAA. The mixture was vortexed and briefly centrifuged, followed by incubation at 37 °C for 30 min, after which it was allowed to return to room temperature. The sample was then diluted with 10 mM TEAB to a concentration suitable for trypsin digestion. Trypsin was added at a protein-to-trypsin ratio of 50:1, and the mixture was incubated overnight at 37 °C. The following day, formic acid was added to adjust the pH to less than 3 to terminate the digestion. AC18 desalting column (Beijing Qinglian Biotec Co., Ltd., QL-C18-C-10) was used for desalting. The column was activated with 100% acetonitrile and equilibrated with 0.1% formic acid. The sample was loaded onto the column, and impurities were washed away using 0.1% formic acid. Finally, the sample was eluted with 40% acetonitrile, and the flow-through liquid was collected and lyophilized.

LC-MS/MS analysis

Mobile phase A (100% water, 0.1% formic acid) and mobile phase B (80% acetonitrile, 0.1% formic acid) were prepared. The lyophilized powder was dissolved in 10 µL of mobile phase A, followed by centrifugation at 14,000 g for 20 min at 4 °C. A total of 200 ng of peptides were loaded onto a 25 cm column (100 μm inner diameter, packed with ReproSil-Pur C18-AQ 1.5-µm silica beads; QL-HPLC-100*15; Beijing Qinglian Biotech Co., Ltd, Beijing, China). Peptide separation was performed using a gradient starting from 4% to 25% B over 4 min, followed by an increase from 25% to 35% B in 1.8 min, and then from 35% to 99% B in 0.4 min, with a final hold at 99% B for 0.7 min. Mass spectrometry analysis was conducted using an Orbitrap Astral mass spectrometer (Thermo Fisher Scientific, USA) equipped with a Thermo ScientificTM NanoSpray FlexTM ion source. Mass spectra were acquired in Data-Independent Acquisition (DIA) mode, with a mass scan range of m/z 380–980. The resolution for the MS spectra was set to 240,000 (at m/z 200). The cumulative time for MS2 spectrometry parameters was set to 3 ms. The capillary voltage was maintained at 2.3 kV, and the total cycle time was set to 0.6 s¹³.

Data analysis

The identification and quantitation of protein

The MS data of the single-shot samples were searched against the spectral library using the DIANN software for final protein identification and quantitation. All searches were conducted against the Mus musculus UniProt reference proteome, which included canonical and isoform sequences. Carbamidomethylation was set as a fixed modification, while acetylation of the protein N-terminus and oxidation of methionines were specified as variable modifications. Default settings were applied for all other parameters. Briefly, a trypsin/P proteolytic cleavage rule was employed, allowing a maximum of two miscleavages and a peptide length range of 7–52 amino acids. A protein and precursor FDR of 1% was applied, and protein quantities were reported in samples only if the protein passed the filter¹⁴.

The functional analysis of protein and deps

Median normalization was applied to the raw intensity data to mitigate technical variations across samples and improve robustness to potential outliers. Data with more than 50% null values in the sample were filtered out. The filtered data were subjected to missing value imputation using the KNN algorithm (supplementary Table 1). Differentially expressed proteins (DEPs) were identified based on the following criteria: an average ratio-fold change > 1.2 and a p-value < 0.05.

Protein family and pathway analyses were conducted using the COG (Clusters of Orthologous Groups), KEGG (Kyoto Encyclopedia of Genes and Genomes), and Reactome (https://reactome.org/) databases. GO annotation of the proteome was obtained from the GO database (https://www.ebi.ac.uk/QuickGO/). Proteins were classified into three categories based on GO annotation: molecular function, biological process, and cellular component. Protein domain functional descriptions were annotated using Pfam, which analyzed the data through the Pfam domain database (https://pfam.xfam.org/) and protein sequence alignment. Pathway analysis was performed using the KEGG pathway database (http://www.kegg.jp/kegg/pathway.html). The KEGG database was utilized to identify enriched pathways, and the enrichment of differentially expressed proteins against all identified proteins was tested using the Hypergeometric distribution^15–17. Probable interacting partners were predicted using the STRING-db server (http://string.embl.de/) based on related species. STRING is a database of known and predicted protein-protein interactions.

Immune colloidal gold electron microscope assay for EVs proteins

The immunogold labelling experiment was performed according to previous studies¹⁸. EVs were captured by carbon supported membrane copper mesh, then primary antibody was added to bind to membrane proteins on the surface of EVs, and secondary antibody coupled to colloidal gold particles was added, which could recognise the bound primary antibody on EVs. The identification of EVs proteins (PCBP2, Profilin1, CD9) can be used by detecting the localisation of colloidal gold particles by transmission electron microscopy.

ELISA for EVs proteins

Randomly mix 30 EVs samples and prepare ELISA standard curve. According to the standard curve, take an appropriate amount of EV samples, dilute to 100 µL with the coating solution (0.1 M, pH = 9.6, carbonate solution), incubate at 4 °C overnight, and repeat three times in each well. After washing with PBST, 100 µL of 5% BSA blocking solution was added for 60 min at room temperature, the detection antibody was added and incubated at room temperature for 2 h. After washing with PBST, add Goat anti-Mouse Ig-G coupled with HRP (1:10,000; ThermoFisher, 32230), and react for 1 h at room temperature. After washing with PBST, 100 µL of TMB colour development solution was added to each well, incubated for 30 min at room temperature and protected from light, 100 µL of reaction termination solution was added to each well and the absorbance at 450 nm was measured¹².

Statistical analysis

A database was established using Excel 2021, and statistical analysis was performed using SPSS 22.0, GraphPad Prism 9.5, and MedCalc 19.5.6 software. Normally distributed measurement data were expressed as , while non-normally distributed measurement data were expressed as median and interquartile range M (P25, P75). Depending on whether the measurement data followed a normal distribution, either the independent samples t-test or the Mann-Whitney U test was used. Univariate and multivariate logistic regression analyses were employed to evaluate factors associated with postoperative AKI. The predictive value of extracellular vesicle proteins for postoperative AKI was assessed by plotting receiver operating characteristic (ROC) curves and calculating the area under the curve (AUC). The optimal cutoff value was determined using the Youden index, and the corresponding sensitivity and specificity were calculated. The consistency between the diagnosis based on the new biomarker and the diagnosis based on KDIGO criteria was evaluated using weighted Kappa statistics. Kappa values were interpreted as follows: < 0.20 indicated slight agreement; 0.21 –0.40, fair agreement; 0.41 –0.60, moderate agreement; 0.61 –0.80, substantial agreement; and > 0.80, almost perfect agreement. A P-value < 0.05 was considered statistically significant.

Results

Isolation and characterization of EVs

EVs were extracted from plasma samples of AKI and no_AKI patients by combinatorial chromatography (Fig. 1A), analysed by TEM and NTA. TEM analysis showed that the EVs were cup-shaped or rounded, which was consistent with the typical EVs results (Fig. 1B and C). The NTA of the AKI group showed that the diameter of the purified EVs was 134.4 ± 62.3 nm, and the main peak was 98.1 nm; the no_AKI group showed purified EVs with a diameter of 144 ± 69.1 nm and the main peak of 103.2 nm (Fig. 1D).

Fig. 1.

Isolation and characterization of EVs (A) Collection and extraction of EVs (created in BioRender. li, J. (2026) https://BioRender.com/q4w10lz); (B–C) Morphology maps of EVs of A4 (AKI) and B4 (no_AKI) obtained by TEM; (D) Particle size distribution of EVs of A4 (AKI) and B4 (no_AKI) obtained by NTA.

Global analysis of two groups of EVs proteins

A prerequisite for the use of EVs for the diagnosis of AKI is the identification of suitable EVs protein biomarkers. Therefore, in the present study, EVs proteins were first extracted from 5 AKI and 5 no_AKI male patients, and there was no statistical significance in age, BMI, tobacco and alcohol use, comorbidities (hypertension, diabetes, cardiovascular disease), surgery time, preoperative sCr and eGFR in the two groups (P > 0.05, Table 1). Raw data from LC-MS/MS were processed using Diann software, then compared and searched against Homo sapiens SP in the Uniprot library (downloaded on 2024.03.07). A total of 621 plasma EVs proteins were identified by non-labelled quantitative (4D-DIA) proteomic analysis. The identified proteins of the two groups were statistically analysed by Venn diagram (Fig. 2A), differences between the two sample groups could be observed by PLS-DA (Partial Least Squares Discriminant Analysis) analysis (Fig. 2B).

Table 1.

Clinical characteristics of patients with LRN.

Characteristics	AKI group (n = 5)	no_AKI group (n = 5)	P
Age (years)	60.40 ± 12.10	58.80 ± 7.86	0.81
BMI (kg/m2)	24.80 ± 2.29	26.25 ± 2.16	0.16
Smoking [n (%)]	3 (60%)	2 (40%)	0.53
Alcohol Consumption [n (%)]	1 (20%)	2 (40%)	0.49
Hypertension [n (%)]	2 (40%)	1 (20%)	0.49
Cardiovascular disease [n (%)]	1 (20%)	0 (40%)	0.29
Diabetes [n (%)]	0 (0%)	1 (20%)	0.29
Surgery time (min)	126.00 ± 37.15	194.4 ± 89.49	0.15
sCr (µmol/L)	64.20 ± 11.95	66.84 ± 13.76	0.75
eGFR (mL/min/1.73 m2)	92.54 ± 19.02	101.82 ± 11.02	0.37

The normally distributed data are presented as Mean ± SD. The enumeration data are presented as [n (%)]. BMI: body mass index; sCr: Serum creatinine; eGFR: Estimated Glomerular Filtration Rate.

Fig. 2.

Proteomic analysis of EVs (A) Venn shows the expression of EVs proteins in patients with AKI and no_AKI; (B) PLS-DA shows the difference in the expression of EVs proteins in patients with AKI and no_AKI, the greater distance between the two groups indicates the greater difference in protein expression between the two groups.

Screening and identification of differential proteins in EVs

T-test test was used for difference analysis, and a total of 43 differential proteins were screened in AKI and AKI groups, including 1 up-regulated protein and 42 down-regulated proteins (Table 2; Fig. 3A). Hierarchical clustering analysis was performed on the proteomics data of AKI and no_AKI groups and heatmaps were generated (Fig. 3B). PPI network analysis was performed on the differential proteins, and it was found that there might be direct or indirect interactions between the differential proteins (Fig. 3C).

Table 2.

Differential proteins in AKI and no_AKI groups.

Accession	Differential proteins	P	Sig
P01602	Immunoglobulin kappa variable 1–5	0.040	1
P19338	Nucleolin	0.001	-1
P30086	Phosphatidylethanolamine-binding protein 1	0.001	-1
P62899	Large ribosomal subunit protein eL31	0.002	-1
P04350	Tubulin beta-4 A chain	0.003	-1
P35268	Large ribosomal subunit protein eL22	0.003	-1
Q02878	Large ribosomal subunit protein eL6	0.010	-1
P25705	ATP synthase subunit alpha, mitochondrial	0.011	-1
P61978	Heterogeneous nuclear ribonucleoprotein K	0.011	-1
P07737	Profilin-1	0.013	-1
Q15366	Poly(rC)-binding protein 2	0.014	-1
P31943	Heterogeneous nuclear ribonucleoprotein H	0.014	-1
P05455	Lupus La protein	0.015	-1
P07910	Heterogeneous nuclear ribonucleoproteins C1/C2	0.015	-1
P78371	T-complex protein 1 subunit beta	0.018	-1
P62826	GTP-binding nuclear protein Ran	0.018	-1
P40227	T-complex protein 1 subunit zeta	0.019	-1
P68032	Actin, alpha cardiac muscle 1	0.019	-1
P60842	Eukaryotic initiation factor 4 A-I	0.021	-1
P02766	Transthyretin	0.021	-1
P37802	Transgelin-2	0.024	-1
P05388	Large ribosomal subunit protein uL10	0.025	-1
P62244	Small ribosomal subunit protein uS8	0.026	-1
P62937	Peptidyl-prolyl cis-trans isomerase A	0.026	-1
P36578	Large ribosomal subunit protein uL4	0.027	-1
P62841	Small ribosomal subunit protein uS19	0.029	-1
P02776	Platelet factor 4	0.029	-1
Q01081	Splicing factor U2AF 35 kDa subunit	0.033	-1
P10809	60 kDa heat shock protein, mitochondrial	0.034	-1
P62847	Small ribosomal subunit protein eS24	0.036	-1
Q9H1E1	Ribonuclease 7	0.037	-1
Q16610	Extracellular matrix protein 1	0.038	-1
P06576	ATP synthase subunit beta, mitochondrial	0.038	-1
P07195	L-lactate dehydrogenase B chain	0.041	-1
P62701	Small ribosomal subunit protein eS4, X isoform	0.041	-1
P29401	Transketolase	0.041	-1
P31942	Heterogeneous nuclear ribonucleoprotein H3	0.041	-1
P62081	Small ribosomal subunit protein eS7	0.045	-1
P40926	Malate dehydrogenase, mitochondrial	0.047	-1
P08195	Amino acid transporter heavy chain SLC3A2	0.048	-1
P28066	Proteasome subunit alpha type-5	0.048	-1
P68104	Elongation factor 1-alpha 1	0.049	-1
P11142	Heat shock cognate 71 kDa protein	0.049	-1

1: up-regulated protein; -1: down-regulated protein

Fig. 3.

Screening and identification of differential proteins in EVs (A) Red dot in the upper right corner of the volcano plot represent up-regulated protein, and green dots in the upper left corner represent down-regulated proteins; (B) In the heatmap, the horizontal coordinate represents each sample, the vertical coordinate is the differential proteins, the colours in the diagram represent the expression of the proteins, the redder the colour, the higher the protein expression, and the bluer the colour, the lower the expression of the proteins; (C) The PPI represents protein-protein interactions.

Functional enrichment analysis of EVs differential proteins

The functions of the differential proteins were interpreted by GO/KEGG enrichment pathway analysis. The GO results were mainly divided into three major categories: biological process, cellular component, and molecular function, biological processes mainly included translation, RNA processing, and protein folding, cellular processes mainly included cytoplasmic ribosomes, ribonucleoprotein complexes, and nucleus, molecular functions mainly included RNA binding, ATP binding and hydrolysis activity, and poly(U)RNA binding; the KEGG results mainly included oxidative phosphorylation, iron death, and ribosomes (Fig. 4). The clustering of each Term and differential proteins in the GO/KEGG analysis is demonstrated by the differential protein chord clustering diagram (Fig. 5).

Fig. 4.

GO and KEGG pathway analysis of EVs differential proteins (A) GO biological process histogram; (B) GO cellular component histogram; (C) GO molecular function histogram; (D) KEGG pathway histogram.

Fig. 5.

EVs differential protein chord clustering diagram (A) GO biological process chord clustering diagram; (B) GO cellular component chord clustering diagram; (C) GO molecular function chord clustering diagram; (D) KEGG pathway chord clustering diagram.

Immune colloidal gold electron microscope assay for EVs proteins

Based on bioinformatics analysis and literature review, we found that Poly(rC)-binding protein 2 (PCBP2), Profilin 1 (Pfn1) proteins may be associated with AKI. A review of the KEGG website revealed that PCBP2 is a down-regulated protein of the ferroptosis pathway, and Pfn1 is a down-regulated protein of regulation of actin cytoskeleton pathway (Fig. 6). Therefore, we chose PCBP2 and Pfn1 as the validation of the proteomic results. We recollected preoperative blood samples from 28 AKI and 24 no_AKI patients, and the basic information was not statistically significant as shown in Table 3 (P > 0.05). Additionally, to differentiate between the expected decline in eGFR due to reduced nephron mass and superimposed acute tubular injury, we calculated predicted postoperative eGFR using the NB-GFR formula. In the no_AKI group, the actual postoperative eGFR was significantly higher than the predicted value (P < 0.05), indicating that the actual compensatory capacity of the healthy kidney in no_AKI patients exceeded the predicted decline from nephron loss. In contrast, in the AKI group, the actual postoperative eGFR was significantly lower than the predicted value (P < 0.05), suggesting that the decline in renal function exceeded the predicted reduction due to nephron loss, implying the presence of additional pathological injury (Table 4).

Fig. 6.

KEGG pathway (A) PCBP2 in the ferroptosis pathway; (B) Pfn1 in the regulation of actin cytoskeleton pathway. KEGG pathway maps are copyrighted by Kanehisa Laboratories (https://www.kegg.jp/kegg/kegg1.html).

Table 3.

Clinical characteristics of patients with LRN.

Characteristics	AKI group (n = 28)	no_AKI group (n = 24)	P
Age (years)	58.71 ± 8.93	58.08 ± 8.69	0.80
BMI (kg/m2)	24.80 ± 3.41	25.05 ± 2.66	0.80
Gender (male/female)	(19/9)	(11/13)	0.11
Smoking [n (%)]	8 (28.6%)	5 (20.8%)	0.52
Alcohol Consumption [n (%)]	5 (17.9%)	8 (33.3%)	0.20
Hypertension [n (%)]	8 (28.6%)	4 (16.7%)	0.31
Cardiovascular disease [n (%)]	3 (10.7%)	5 (20.8%)	0.31
Diabetes [n (%)]	5 (17.9%)	4 (16.7%)	0.91
MAP (mmHg)	88.68 ± 10.89	87.21 ± 8.81	0.60
Crystalloid solution (mL)	1000 (850, 1500)	1050 (1000, 1500)	0.71
urine output (mL)	100 (100, 200)	100 (100, 200)	0.87
blood volume (mL)	50 (30, 50)	50 (20, 50)	0.50
Surgery time (min)	142 (101.5, 218)	165 (110.8, 195)	0.99
sCr (µmol/L)	70.73 ± 13.61	66.12 ± 12.43	0.21
Baseline GFR (mL/min/1.73 m2)	93.45 ± 11.37	97.32 ± 14.72	0.29
Healthy Kidney Volume (cm3)	147.5(121.6, 171.9)	144.1(121.0, 173.7)	0.88

The normally distributed data are presented as Mean ± SD. The non-normally distributed data are presented as median (P25, P75). The enumeration data are presented as [n (%)]. BMI: body mass index; sCr: Serum creatinine; eGFR: Estimated Glomerular Filtration Rate.

Table 4.

NB-GFR and actual eGFR of patients with LRN.

Group	NB-GFR	Actual eGFR	P
AKI group (n = 28)	66.6 (61.5, 70.3)	61.8 ± 10.13	0.004
no_AKI group (n = 24)	68.1 ± 9.87	74.1 ± 14.89	0.025
P	0.34	0.001

The normally distributed data are presented as Mean ± SD. The non-normally distributed data are presented as median (P25, P75). NB-GFR: new baseline Glomerular Filtration Rate.

EVs were extracted by combinatorial chromatography and detected by immunocolloidal gold lens, we chose Ig-G as a negative control and observed distinct gold nanoparticles around the EVs, labelled with anti-CD9, PCBP2, Pfn1 immunogold tags, which demonstrated the expression of CD9, PCBP2, Pfn1 on the EVs. CD9 is a specific protein for EVs, indicating that we may have successfully extracted EVs from the samples (Fig. 7).

Fig. 7.

Immune colloid gold labeling revealed, CD9, PCBP2, and Pfn1 on EV. Ig-G without primary antibody treatment were used as negative control. Scale bar: 200 nm.

ELISA for proteins in EVs

We performed an enzyme-linked immunosorbent assay (ELISA) to detect proteins in EVs. We used anti-PCBP2/Pfn 1 as a detection antibody to verify the success of the ELISA construction, and we found that PCBP2 and Pfn1 increased with the concentration of EVs, indicating that this ELISA method can effectively quantify EVs proteins, then detected PCBP2 and Pfn1 on EVs (Fig. 8A). The levels of PCBP2 and Pfn1 in AKI (n = 28) and no_AKI patients (n = 24) were measured by ELISA and plotted on a heat map (Fig. 8B), which showed that the mean levels of PCBP2 and Pfn1 in AKI patients were significantly lower than no_AKI. Scatter plots showed that there was a significant difference in the levels of PCBP2 and Pfn1 between AKI and no_AKI patients (Fig. 8C–D). Furthermore, Spearman correlation analysis indicated that the difference between actual eGFR and NB-GFR (ΔeGFR) was negatively correlated with PCBP2 (r=-0.30, P = 0.028) and Pfn1 (r=-0.37, P = 0.006), respectively. This suggests that patients with lower preoperative levels of PCBP2 and Pfn1 proteins experienced a more severe decline in postoperative eGFR beyond expected values.

Fig. 8.

Analysis of PCBP2 and Pfn1 in EVs (A) Standard curve of ELISA for PCBP2 and Pfn1 on EVs; (B) Heatmap of PCBP2 and Pfn1 levels on EVs in the AKI and no_AKI groups; (C–D) Scatterplot of PCBP2 and Pfn1 levels on EVs in the AKI and no_KI groups.

Univariate logistic regression analysis revealed that lower preoperative levels of PCBP2 and Pfn1 were risk factors for postoperative AKI (Table 5). However, Spearman correlation analysis indicated a significant positive correlation between the levels of PCBP2 and Pfn1 (r = 0.84, P < 0.001). To avoid multicollinearity in the multivariate model and to evaluate the independent predictive value of PCBP2 and Pfn1 proteins, we constructed separate multivariate logistic regression models by incorporating variables with P < 0.35 from the univariate analysis. In Model 1, lower preoperative PCBP2 levels and male gender were identified as independent risk factors for postoperative AKI. In Model 2, lower preoperative Pfn1 level was found to be an independent risk factor for postoperative AKI (Table 6).

Table 5.

Univariate logistic regression analysis of factors associated with postoperative AKI.

Characteristics	OR	95%CI	P
Age	1.00	0.95 –1.08	0.79
Gender
Female	1.00	(Ref)
Male	2.49	0.82 –7.06	0.11
BMI	0.97	0.81 –1.17	0.77
Hypertension
NO	1.00	(Ref)
YES	2.00	0.54 –8.50	0.31
Cardiovascular disease
NO	1.00	(Ref)
YES	0.46	0.08 –2.09	0.32
Diabetes
NO	1.00	(Ref)
YES	1.09	0.25 –4.92	0.91
MAP	1.02	0.96 –1.08	0.59
Crystalloid solution	0.99	0.98 –1.00	0.98
Blood volume	1.00	0.98 –1.03	0.62
Urine output	1.00	0.99 –1.01	0.75
Surgery time	1.00	0.99 –1.01	0.93
sCr	1.03	0.99 –1.08	0.21
Baseline GFR	0.98	0.94 –1.02	0.31
NB-GFR	0.96	0.89 –1.03	0.27
Healthy kidney volume	1.00	0.99 –1.01	0.93
PCBP2	1.54e-06	3.62e-10 –7.47e-04	< 0.001
Pfn1	2.39e-08	4.92e-13 –4.78e-05	< 0.001

BMI: body mass index, MAP: mean arterial pressure, sCr: serum creatinine, NB-GFR: new baseline Glomerular Filtration Rate.

Table 6.

Multivariate logistic regression analysis of factors associated with postoperative AKI.

Characteristics	Model 1			Model 2
	OR	95%CI	P	OR	95%CI	P
PCBP2	2.85e-07	1.10e-15 –3.75e-04	< 0.001
Pfn1				5.13e-09	5.84e-15 –2.66e-05	< 0.001
Gender
Female	1.00	(Ref)		1.00	(Ref)
Male	8.16	1.40 –73.54	0.03	4.53	0.67 –43.37	0.14
Hypertension
NO	1.00	(Ref)		1.00	(Ref)
YES	0.97	0.12 –7.08	0.92	2.68	0.24 –33.77	0.41
Cardiovascular disease
NO	1.00	(Ref)		1.00	(Ref)
YES	0.27	0.02 –2.95	0.27	0.56	0.04 –9.38	0.67
sCr	1.01	0.94 –1.08	0.80	0.99	0.91 –1.07	0.87
Baseline GFR	1.06	0.84 –1.34	0.62	1.22	0.93 –1.70	0.17
NB-GFR	0.89	0.61 –1.30	0.53	0.72	0.42 –1.11	0.16

sCr: serum creatinine, NB-GFR: new baseline Glomerular Filtration Rate.

To assess the diagnostic performance of the preoperative independent risk factors for AKI patients, we plotted ROC and calculated the AUC. Gender showed limited diagnostic performance (AUC = 0.61). Both PCBP2 and Pfn1 proteins demonstrated good diagnostic performance as single indicators, with AUCs of 0.893 and 0.923, sensitivities of 92.9% and 89.3%, and specificities of 83.3% and 87.5%, respectively. Furthermore, the combination of PCBP2 with gender significantly outperformed gender alone in diagnostic performance (P < 0.001), with the AUC increasing to 0.908, indicating that PCBP2 provides incremental predictive value to the clinical baseline information (gender) (Table 7).

Table 7.

Comparative analysis of different indicators for the diagnosis of AKI.

Characteristics	AUC	95% CI	Sensitivity	Specificity
Gender	0.61	0.465 –0.742	67.9	54.2
PCBP2	0.893	0.776 –0.962	92.9	83.3
Pfn1	0.923	0.814 –0.978	89.3	87.5
PCBP2 + Gender	0.908	0.795 –0.970	92.9	79.2

Consistency analysis of PCBP2 and Pfn1 versus sCr for diagnosing postoperative AKI

To further evaluate the potential of PCBP2 and Pfn1 as diagnostic tools, we plotted the ROC curves and determined their optimal cutoff values for diagnosing postoperative AKI, which were 0.78 (PCBP2) and 0.59 (Pfn1), respectively. The diagnostic results based on these cutoff values showed substantial agreement with the KDIGO criteria (PCBP2: Kappa = 0.77; Pfn1: Kappa = 0.73).

Discussion

AKI is a prevalent complication following LPN and is significantly associated with the occurrence of adverse outcomes. Therefore, early detection and prevention of AKI are crucial for enhancing patient prognosis. However, the current diagnostic methods for AKI, which rely on sCr levels and urine output, have inherent limitations that may delay AKI identification and increase the false-negative rate⁶. To overcome these challenges, novel biomarkers are essential for the early diagnosis of AKI.

In this study, we collected plasma samples from five patients with AKI and five without AKI for proteomic analysis of plasma EVs. We characterized the morphology and size of EVs using TEM and NTA. Through 4D-DIA proteomic analysis, we identified a total of 621 plasma EVs proteins, including 43 differential proteins (1 up-regulated and 42 down-regulated). Our preliminary findings suggest that the proteins PCBP2 and Pfn1 may be associated with AKI, which warrants further validation. Subsequent validation revealed that PCBP2 and Pfn1 are present on EVs, and their preoperative levels were significantly lower in patients with AKI compared to those without AKI, suggesting that PCBP2 and Pfn1 may be involved in the pathogenesis of AKI through plasma EVs.

Our study focused on a population undergoing LRN, in which postoperative sCr elevation is associated with factors such as nephron loss, ischemia, and inflammation. To distinguish between functional decline due purely to nephron reduction and superimposed tubular injury, we calculated the predicted postoperative NB-GFR based on preoperative data, the actual postoperative eGFR, and the ΔeGFR representing the decline beyond the expected extent. We found that in patients with AKI, the actual eGFR decrease significantly exceeded the predicted NB-GFR, whereas in no-AKI patients, renal function was better than the NB-GFR. Further analysis revealed that preoperative levels of PCBP2 and Pfn1 showed a significant negative correlation with ΔeGFR, suggesting that patients who developed AKI might experience superimposed pathological renal injury on top of nephron loss. However, the above inference relies on indirect calculations and comparisons. Since we did not directly measure split renal function using gold-standard techniques such as renal dynamic imaging, we could not precisely quantify the true compensatory potential and baseline status of the contralateral kidney. Although we minimized obvious bias at the group level by excluding patients with preoperative eGFR < 60 ml/min/1.73 m² and confirming no differences in preoperative contralateral renal volume, total eGFR, or NB-GFR between the two groups, heterogeneity in individual compensatory capacity may still exist. Therefore, future studies should incorporate functional imaging assessments of preoperative split renal function to more accurately evaluate whether PCBP2 and Pfn1 can predict genuine pathological tubular injury beyond individual expectations, ultimately clarifying their role as injury-specific biomarkers.

This study found that preoperative PCBP2 and Pfn1 levels were significantly associated with the risk of postoperative AKI. PCBP2, a member of the polycytosine-binding protein family, is involved in the regulation of RNA stabilization, translational silencing, and enhancement. It has been demonstrated to play a regulatory role in various diseases, including hepatitis C virus infection, neuronal degeneration, acute spinal cord injury, and chronic obstructive pulmonary disease. Furthermore, PCBP2 is critically implicated in cancer-related processes such as proliferation, apoptosis, invasion, migration, and epithelial-mesenchymal transition^19,20. Within the iron metabolism pathway, Fe³⁺ is reduced to Fe²⁺, with a portion stored as Fe²⁺ in the lysosomal lumen and another portion transferred to the cytoplasm via DMT1 or TRPML1 on the lysosomal membrane. PCBP2 functions as an iron chaperone by directly binding to DMT1 to accept Fe²⁺, forming the Fe²⁺/PCBP2 complex. This complex facilitates iron removal through ferritin depots or ferritin transporters. The accumulation of Fe²⁺ leads to the generation of a substantial amount of toxic hydroxyl radicals, which exacerbate oxidative stress and cause severe lipid peroxidation, ultimately resulting in ferroptosis²¹. Ferroptosis serves as a crucial mediator of ischemia-reperfusion-induced renal tubular cell death. Ferritin facilitates the conversion of Fe²⁺ to Fe³⁺ through its ferric oxidase activity. During extracorporeal circulation surgery, low levels of iron-binding proteins can result in elevated Fe²⁺ levels and oxidative stress, thereby exacerbating kidney injury in AKI. The knockdown of PCBP2 significantly heightens susceptibility to ferroptosis, while the application of ferroptosis inhibitors mitigates ischemia-reperfusion injury, thus reducing the incidence of AKI^22–25. Our study identified PCBP2 as a down-regulated protein, with KEGG analysis indicating its enrichment in the ferroptosis pathway. Further validation revealed the presence of PCBP2 on EVs, with significantly reduced levels observed in AKI patients. Therefore, we speculate that PCBP2 may be involved in the occurrence of AKI by participating in the process of ferroptosis, although further experiments are still required to verify the causal relationship.

Pfn1 is an evolutionarily highly conserved actin-binding protein ubiquitously present across eukaryotic organisms. It plays a critical role in regulating the actin cytoskeleton through its interactions with various actin-binding proteins and proline-rich domains, which are vital for cellular viability. Pfn1 is expressed throughout all stages of embryonic development, and its absence results in embryonic lethality in mice. It is crucial for the coordination of actin and microtubule networks during axonal growth and regeneration following nerve injury. Additionally, Pfn1 is implicated in cancer and cardiovascular diseases; its over-expression inhibits the oncogenicity of cancer cells by promoting apoptosis via the classical intrinsic apoptotic pathway, functioning as a tumor suppressor^26–28. In renal tissues, Pfn1 is expressed at low levels in glomerular structures and at moderate levels in renal tubules. In both mouse and human podocytes, reduced Pfn1 expression leads to mitotic anomalies, compromising the integrity of the podocyte actin cytoskeleton, which is essential for maintaining the glomerular filtration barrier. Consequently, Pfn1 deficiency can result in proteinuria and renal failure²⁶. The over-expression of Pfn1 has been observed to enhance apoptosis induced by high glucose levels and to inhibit the proliferation of renal tubular epithelial cells²⁹. In this study, Pfn1 emerged as a down-regulated protein in GO enrichment analysis and was identified as a regulator of the actin cytoskeleton through KEGG analysis. Immune colloid gold labeling revealed the presence of Pfn1 on the membranes of EVs. Notably, preoperative plasma EVs from patients in the AKI group exhibited significantly lower levels of Pfn1 compared to those from patients without AKI, and these levels were predictive of postoperative AKI occurrence. Therefore, we hypothesised that reduced preoperative Pfn1 protein levels in plasma EVs may be associated with impaired glomerular filtration barrier and facilitating the development of postoperative AKI. However, this hypothesis is primarily derived from proteomics analysis, and the underlying mechanisms as well as their correlations need to be further verified by functional assays.

EVs serve as novel mediators of intercellular communication and biomarkers in AKI, demonstrating significant potential in its diagnosis, mechanistic research, and treatment. Studies indicate that urinary EVs are potential indicators of glomerular and tubular injury, with changes in their levels occurring earlier than the decline in renal function detected by traditional markers⁷. Oh et al. found that activating transcription factor 3 protein in urinary EVs increases significantly after ischemia-reperfusion-induced AKI, rising earlier than sCr and persisting for 24–48 h³⁰. EVs also provide more precise information about the origin of injury through their cargo, distinguishing between different pathological mechanisms such as ischemia-reperfusion injury and AKI secondary to traumatic brain injury³¹. Previous research has primarily focused on urinary EVs and models such as ischemia-reperfusion, with limited investigation into EVs biomarkers for AKI in the high-risk surgical context of LRN. Studies on novel AKI biomarkers, such as neutrophil gelatinase-associated lipocalin and kidney injury molecule-1, have mainly emphasized early diagnosis after injury occurrence. Their value lies in detecting existing renal damage earlier than traditional sCr, rather than predicting risk prior to injury. In contrast, our study primarily employs preoperative EVs proteomic analysis, aiming to identify biomarkers that can pre-indicate the risk of AKI occurrence, thereby facilitating the development of individualized perioperative management strategies and improving patient outcomes.

Through preliminary proteomic analysis, we identified that PCBP2 and Pfn1 may be associated with AKI. Subsequent validation revealed that their expression levels were significantly decreased in AKI patients, and both demonstrated excellent predictive performance for postoperative AKI (AUC = 0.893 and 0.923, respectively). To clarify their clinical value, we compared them with conventional clinical indicators. The predictive ability of gender was limited (AUC = 0.61), while PCBP2 or Pfn1 alone showed significantly better predictive performance than this characteristic. A combined model integrating PCBP2 with gender achieved an AUC of 0.908, indicating significantly better diagnostic performance than gender alone and confirming its clear incremental predictive value. Additionally, based on preoperative protein concentrations, we performed preoperative risk stratification for AKI patients using the optimal cutoff value from the ROC analysis and found high consistency with clinical diagnosis based on sCr changes. Preoperative sCr levels showed no statistical difference between the two groups in this study, suggesting that the predictive ability of PCBP2 or Pfn1 may precede traditional sCr.

This study is subject to several limitations, which include the following: (1) The ELISA detection system’s limited sensitivity may introduce errors despite identifying significant PCBP2 and Pfn1 differences between AKI and no_AKI patients. Future studies should use orthogonal validation methods like Western blot or targeted mass spectrometry. (2) The high sensitivity and specificity of PCBP2 and Pfn1 are promising but based on a single-center cohort of malignancy patients undergoing LRN, which may not apply to non-oncologic AKI populations. These biomarkers also cannot be used as a universal, fixed concentration threshold for independent clinical diagnosis. Future validation and calibration of diagnostic thresholds are required in multicenter, large-sample prospective cohorts. (3) This study did not analyze the dynamic changes in extracellular vesicle protein levels before and after surgery. Longitudinal monitoring of the dynamic changes in PCBP2 and Pfn1 from the preoperative to postoperative periods is needed to determine whether their trajectory holds greater diagnostic or prognostic value. (4) PCBP2 and Pfn1 are linked to ferroptosis/cytoskeleton pathways, but their direct causal role in AKI is speculative. Functional assays like CRISPR knockdown in renal epithelial cells are needed. (5) This study did not assess potential confounding factors such as medications used during the perioperative period and other unrecorded comorbidities that may influence the levels of PCBP2 and Pfn1 in plasma EVs. Although basic clinical characteristics of the two patient groups (e.g., age, BMI, underlying diseases) were analyzed and showed no statistically significant differences, unmeasured confounding factors may still exist. Future multicenter studies should collect and analyze these variables more comprehensively to strengthen the robustness of the conclusions.

Conclusion

In this study, 4D-DIA proteomic analysis revealed 43 differentially expressed proteins in preoperative plasma extracellular vesicles of patients who developed AKI after LRN. Independent validation confirmed that PCBP2 and Pfn1 were located on EVs and were significantly downregulated in the preoperative plasma vesicles of AKI patients, demonstrating excellent preoperative predictive value. Risk stratification based on the optimal cutoff values of PCBP2 and Pfn1 showed a high degree of consistency with postoperative clinical diagnosis according to KDIGO criteria (based on sCr changes). However, the clinical applicability of these biomarkers still requires validation through multicenter, large-sample prospective studies, as well as in-depth mechanistic investigations to clarify their specific roles in the occurrence of AKI.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

JXL: Data curation; Formal analysis; Methodology; Project administration; Writing-original draft; Resources. LHX: Methodology; Project administration; Software. YL: Data curation; Formal analysis; Validation. XJL: Data curation; Visualization; Methodology. XFL: Data curation; Formal analysis; YZL: Conceptualization; Formal analysis; Resources; Writing-review & editing. All authors reviewed the manuscript.

Data availability

All data supporting the results reported in this study are available from the corresponding author upon request.

Declarations

Competing interests

The authors declare no competing interests.

Ethical approval

This study was approved by the Medical Ethics Review Panel of the Third Medical Centers of the Chinese People’s Liberation Army (PLA) General Hospital (No. KS2024-001 Chinese Clinical Trial Registry number: ChiCTR2400084174). All experiments using clinical samples were conducted under relevant ethical guidelines, and written informed consent was obtained from all participants.