Changes in circulating extracellular vesicle cargo are associated with cognitive decline after major surgery: an observational case-control study

Abstract

Background: Postoperative neurocognitive decline is a frequent complication triggered by unclear signalling mechanisms. This observational case-control study investigated the effects of hip or knee replacement surgery on the composition of circulating extracellular vesicles (EVs), potential periphery-to-brain messengers, and their association with neurocognitive outcomes.

Methods: We mapped the microRNAome and proteome of plasma-derived EVs from 12 patients (six with good and six with poor neurocognitive outcomes at 3 months after surgery) at preoperative and postoperative timepoints (4, 8, 24, and 48 h). Complement C3-EV association was confirmed by flow cytometry in plasma- and cerebrospinal fluid (CSF)-derived EVs, with total plasma and CSF C3 and C3a concentrations determined using enzyme-linked immunosorbent assay.

Results: Differential expression analysis found eight dysregulated EV microRNAs (miRNAs) exclusively in the poor neurocognitive outcomes group. Pathway analysis suggested potential downregulation of proliferative pathways and activation of extracellular matrix and inflammatory response pathways in EV target tissues. Proteome analysis revealed a time-dependent increase in immune-related EV proteins, including complement system proteins, notably EV surface-associated C3. Such upward kinetics was detected earlier in the poor neurocognitive outcomes group. Interestingly, CSF-derived EVs from the same group showed a drastic drop of C3 at 48 h with unchanged concentrations in the good neurocognitive outcomes group. Functionally, the complement system was activated in both patient groups in plasma, but only in the poor neurocognitive outcomes group in CSF.

Conclusions: Our findings highlight the impact of surgery on plasma- and CSF-derived EVs, particularly in patients with poor neurocognitive outcomes, indicating a potential role for EVs. The small sample size necessitates verification with a larger patient cohort.

Keywords: circulating extracellular vesicle; miRNA; orthopaedic surgery; postoperative cognitive decline; proteomics.

Fig 1

Study design and methodology. Hip or knee surgery was performed in patients under spinal anaesthesia supplemented by light sedation. Cerebrospinal fluid (CSF; 5 ml) and blood (20 ml) were collected preoperatively (PreOp) and at 4, 8, 24, and 48 h after skin incision. Preoperative (1–2 weeks before surgery) and postoperative (3 months after surgery) neurocognitive capacity was evaluated using the International Study of Postoperative Cognitive Dysfunction (ISPOCD) test battery (Supplementary Methods). Red thumbs up icon, patients with good neurocognitive function (n=6). Blue thumbs down icon, patients with poor neurocognitive function (n=6). The microRNAome and proteome of plasma-derived extracellular vesicles (EVs) isolated from 12 patients before and after the completion of orthopaedic surgery were mapped by RNA-sequencing and tandem mass spectrometry. EVs were characterised by standard methods and results were validated and analysed in both blood and CSF samples. DE, differentially expressed; ELISA, enzyme-linked immunosorbent assay; FC, fold change; miRNA, microRNA; TEM, transmission light microscopy; NTA, nanoparticle tracking analysis; WB, western blot; RT-qPCR, quantitative reverse transcription polymerase chain reaction.

Fig 2

Differential expression (DE) of microRNAs (miRNAs) in circulating extracellular vesicles. (a) Principal component analysis (PCA) of circulating extracellular vesicles (EVs) miRNAome. CPM (counts per million)-normalised miRNA read counts were filtered for low-expression miRNAs, log2-transformed, and used as a matrix table by Qlucore software (QLUCORE, Lund, Sweden) to generate PCA plots. PCA plots for all timepoints in the good (upper panel) and poor outcome groups (lower panel). (b) DE of miRNAs in plasma-derived EVs from patients from the combined, good, and poor outcome groups. DE is estimated relative to preoperative values and miRNAs with a false discovery rate (FDR) ≤0.1 were considered statistically significant and referred to as DE miRNAs. (c) Pathway analysis of DE miRNA target messenger RNAs (mRNAs) in the poor outcome group. mRNAs for each of upregulated or downregulated miRNAs were searched in the Reactome database for enriched pathways. The top significantly (FDR ≤0.05) enriched pathways are shown. FC, fold change; Yellow, upregulated pathways; blue, downregulated pathways.

Fig 3

Differential expression (DE) of proteins in circulating extracellular vesicles (EVs). (a) Principal component (PC) analysis of the EV proteome (normalised abundancies of 214 proteins consistently expressed in all samples). (b) Number of DE proteins in the combined, good, and poor outcome groups. (c) Hierarchical clustering analysis of the normalised abundances of the 66 DE proteins. Data are presented for each patient across all timepoints and separately for each cognitive outcome group. The proteins showing different protein abundances at the 8-h timepoint between groups (mostly complement system) are enclosed in red. Green arrow indicates the central complement component C3. Each protein is associated with its corresponding Gene Ontology (GO) term: biological function (BP; bottom). (d) Pathway enrichment analysis using the merged list of 66 DE proteins from the good and poor outcome groups at the 48-h timepoint. Presented are the results from the David-based search in the Reactome database visualizing the top significantly (false discovery rate ≤0.05) enriched pathways. The DE protein list from the 24-h timepoint generated a nearly identical bubble plot (results not shown). PreOp, preoperative.

Fig 4

Venn diagram and protein classification analysis of the 66 differentially expressed proteins. Among the 66 differentially expressed proteins, seven main classes of proteins were identified (see legend). The complement component was the class with the highest percentage of proteins (32%), followed by immunoglobulin-related proteins (15%), blood coagulation proteins (15%), apolipoproteins (12%), and acute-phase proteins (8%). Analysis of the set of proteins unique for the poor group showed relatively higher abundance of immunoglobulin-related proteins compared to the good group (32% vs 10%). The complement component was the most represented term within the unique and overlapping proteins in the two groups. Enclosed rectangles indicate the name and the vector of the regulation of the proteins included in the immunoglobulin (IG) and complement component terms.

Fig 5

Effect of surgery on total and extracellular vesicle (EV)-associated C3 in serum and cerebrospinal fluid (CSF). (a) Left panel: log₂ fold changes (log₂FC) of the relative C3 concentrations determined by mass spectrometry from plasma-derived EVs isolated from the good and poor neurocognitive outcome groups relative to corresponding preoperative (PreOp) values. Patients in the poor cognitive outcome group (poor) showed an early and persisting increase in the abundance of C3 in plasma-derived EVs (two-way analysis of variance with repeated measures: multiple comparison; uncorrected Fisher's least significant difference [LSD]): poor (#, P=0.0185, ns <0.1). Right panel: confirmation of the C3 association with plasma EVs. Latex beads were coated with anti-human CD9, incubated with Exo-EVs from plasma, and analysed for the presence of CD9, CD63, CD81, and C3 by flow cytometry for pooled EVs from three patients (two from the good cognitive outcome group and one from the poor cognitive outcome group) at the PreOp and 48-h timepoints. Mean Fluorescence Intensity (MFI) ratio is the ratio of mean fluorescence intensity of capture antibody against isotype control. (b) Enzyme-linked immunosorbent assay-based quantification of C3, C3a, and resulting C3a-to-C3 ratio in the serum and CSF in both cognitive outcome groups. Data are presented as log₂FC relative to PreOp values. Statistical differences were tested by two-way analysis of variance with repeated measures followed by multiple comparison analysis by uncorrected Fisher's LSD: good ∗P<0.05, ∗∗ P<0.01 (vs PreOp); poor #P<0.05, ##P<0.01 (vs PreOp); good vs poor, §P<0.05; ns, P<0.1. (c) Left panel: C3 log₂FC relative to PreOp values from the flow cytometry experiment on CSF-derived EVs from good and poor groups at 48 h after surgery. Two-way analysis of variance with repeated measures: time: time × outcome: ∗P=0.0156. Multiple comparison (uncorrected Fisher's LSD): poor (##P=0.0035) and good vs poor (§§§P=0.0005). Right panel: results from the flow cytometry analysis of CD81 and C3 from the CSF-derived EVs (EVs CSF) of patients from both outcome groups.