An extracellular vesicle epitope profile is associated with acute myocardial infarction

Abstract The current standard biomarker for myocardial infarction (MI) is high‐sensitive troponin. Although powerful in clinical setting, search for new markers is warranted as early diagnosis of MI is associated with improved outcomes. Extracellular vesicles (EVs) attracted considerable interest as new blood biomarkers. A training cohort used for diagnostic modelling included 30 patients with STEMI, 38 with stable angina (SA) and 30 matched‐controls. Extracellular vesicle concentration was assessed by nanoparticle tracking analysis. Extracellular vesicle surface‐epitopes were measured by flow cytometry. Diagnostic models were developed using machine learning algorithms and validated on an independent cohort of 80 patients. Serum EV concentration from STEMI patients was increased as compared to controls and SA. EV levels of CD62P, CD42a, CD41b, CD31 and CD40 increased in STEMI, and to a lesser extent in SA patients. An aggregate marker including EV concentration and CD62P/CD42a levels achieved non‐inferiority to troponin, discriminating STEMI from controls (AUC = 0.969). A random forest model based on EV biomarkers discriminated the two groups with 100% accuracy. EV markers and RF model confirmed high diagnostic performance at validation. In conclusion, patients with acute MI or SA exhibit characteristic EV biomarker profiles. EV biomarkers hold great potential as early markers for the management of patients with MI.


| INTRODUC TI ON
Coronary artery disease is the most frequent cause of death worldwide. 1 Among acute coronary syndromes, ST-segment elevation myocardial infarction (STEMI) is associated with a 12%-mortality within 6 months. 2 Early diagnosis of myocardial infarction (MI) is critical to preserve cardiac function and improve outcomes. 3,4 This diagnosis currently relies on chest pain suggestive of myocardial ischemia, specific electrocardiogram (ECG) changes, and detection of increased serum levels of high-sensitive (hs) troponin, 4 which is released within 2-3 hours after the onset of cardiac injury. Although extremely useful in a daily clinical setting, a search for novel biomarkers to identify patients with MI early after onset of symptoms, when primary percutaneous coronary intervention (PCI) has the greatest chance of influencing their outcomes, is warranted. 5,6 Emerging potential biomarkers for coronary artery disease (CAD) include secreted extracellular vesicles (EVs), which are found in all biological fluids including peripheral blood. [6][7][8] Their diagnostic power derives from the enrichment of potential protein markers which otherwise constitute only a very small proportion (<0.01%) of the total proteome of body fluids. 9 EVs can be subdivided into exosomes, which originate from the endosomal system, and microvesicles, which directly shed by plasma membranes. Traditionally, EVs were subdivided by size: small EVs (30 nm-150 nm in diameter) mainly consisting of exosomal EVs, and medium-large EVs (150 nm-1000 nm in diameter) mainly representing microvesicles. Some EV cargo proteins specify a cellular origin or change in amount in certain diseases, so that they can serve as biomarkers of diseases. Exosomal surface epitopes include the tetraspanins CD9, CD63, and CD81. Intra-vesicular exosomal markers include tumour susceptibility gene 101 (TSG101). 6,10 Most circulating EVs in healthy individuals are derived from platelets and erythrocytes; however, vesicles also are released from leucocytes, endothelial cells, monocytes, neutrophils and lymphocytes. 7,11 Previous studies have shown increased numbers of platelet, endothelial cell-and leucocyte-derived EVs in patients with acute coronary syndromes (ACS) or chronic CAD. [12][13][14][15][16][17][18][19][20][21][22][23] However, partially divergent results have been reported, in part due to methodological differences in EV isolation. 24,25 Moreover, previous analyses focused on a few surface epitopes of interest, rather than on the EV epitope surface profile.
The main potential advantage of EV profiling in the management of patients with CAD is represented by the very early increase of circulating particles released by suffering but still alive cardiomyocytes, during heart ischemia. 6,22 Indeed, the increase in EV-derived biomarkers may even anticipate the raise of hs-troponin, which need cardiomyocytes death. However, EV profiling needs for protocol standardization and technology implementation, and that this approach remains to date relatively time-consuming and expensive when compared to troponin. 6 Potential advantages and disadvantages of EV profiling as compared to hs-troponin in the management of patients with CAD are summarized in Table S1.
In the present study, we analysed a comprehensive panel of 37 EV surface epitopes, representative of a variety of cells of origin, in serum samples from patients with acute MI and patients with chronic CAD presenting with stable angina (SA), compared to age-and sexmatched healthy controls, using a validated multiplex flow cytometry (FC) assay. 25,27 A random forest (RF) model based on EV surface epitopes accurately discriminated STEMI patients and controls.

| ME THODS
A detailed description of patient data, EV extraction and characterization protocols, statistical analyses and diagnostic modelling is provided in the Appendix S1.

| Participants and blood sampling
We analysed peripheral venous blood samples collected from individuals recruited at the Fondazione Cardiocentro Ticino, Lugano (Switzerland), as a training cohort for diagnostic modelling purposes.

| Sample processing
Blood was collected in 7 mL heparin-and EDTA-free polypropylene tubes. The first blood tube was discarded. Blood was centrifuged at 1600 g for 15 minutes at 4°C, and supernatant was centrifuged at 3000 g for 20 minutes, 10 000 g for 15 minutes and 20 000 g for 30 minutes to remove intact cells, cellular debris and larger EVs (apoptotic bodies and EV aggregates; see also Extended Methods and Figure S1A). After centrifugation steps, supernatant was divided into 0.1 mL aliquots and stored at −80°C until analysis. Pre-analytical factors for blood collection and storage complied with guideline for EV biomarkers. 6

| Nanoparticle tracking analysis (NTA)
To measure serum particle concentration and diameter, we used NTA with NanoSight LM10 (Malvern Instruments) equipped with a 405 nm laser and NTA 2.3 analytic software. EV concentration is shown as particle number/mL (median value and interquartile range).

| EV surface epitope analysis
Serum samples underwent bead-based EV immunocapture and were analysed by FC using MACSPlex human Exosome Kit (Miltenyi; Bergisch Gladbach, Germany), as detailed in the Appendix S1 (see also Figure S1A). Briefly, serum supernatant was incubated with 37 fluorescently labelled capture bead populations (Table S2), each coated with a specific antibody binding the respective surface epitope, and 2 control bead populations, followed by the EV detection reagent (ie fluorescently labelled antibodies for CD9/CD63/CD81). Median fluorescence intensity (MFI) was measured on a MACSQuant-Analyzer-10 flow cytometer (Miltenyi) according to previous validation studies. 26,27 All markers were analysed simultaneously. Surface epitope levels were referenced to EV-specific epitopes by subtracting the respective fluorescence values of blank control from MFI values for individual surface epitopes, and by normalizing them for CD9/ CD63/CD81 MFI, reflecting EV concentration. To rule out confounding effects of the protocol used, this method was compared with alternate protocols in a small subset of patients. Specifically, the effect of EV isolation by ultracentrifugation or size-exclusion chromatography (SEC) was assessed ( Figure S2A,B). Moreover, EV markers were determined in both serum and plasma samples from the same patients ( Figure S2C and S3). A shortened incubation time (1 hour) of serum supernatant with capture beads was tested as compared to overnight incubation ( Figure S2D). Finally, technical reproducibility of the assay was confirmed by analysing twice the same sample ( Figure S4).

| Clinical and biochemical characteristics of the study cohorts
We analysed 238 serum samples collected from 178 participants from a training cohort (n = 98) and an independent validation cohort (n = 80). The training cohort was divided into 3 groups (STEMI: Dyslipidemia (48.0%) and hypertension (43.9%) were highly prevalent in the study population, followed by smoking (14.3%), diabetes (12.2%) and chronic kidney disease (5.1%). Study groups did not significantly differ from one another with respect to age, sex,  (Table   S4). Based on the results in the training cohort, EV markers were validated for the diagnosis of STEMI, in an independent cohort.
Clinical and biochemical characteristics of patients in the validation cohort did not significantly differ from the training cohort (Table S5). Except for cardiac-specific hs-troponin I, WBC counts and LVEF, patients' characteristics were not significantly different in STEMI patients and controls in the validation cohort (Table   S6).

| EV number and size
Serum samples were pre-cleared by serial centrifugation steps in order to remove intact cells, cellular debris and larger EVs, while enriching in smaller particles. EV number and diameter decreased after each centrifugation step, showing a trend to depletion of particles larger than 250 nm (P > .05; Figure S1B). Before performing EV immunocapture, particle size and concentration were determined by NTA. In the training cohort, EV concentration measured by NTA was significantly increased in serum samples from STEMI patients on presentation to the emergency department (Tables S7 and S8;  EV concentration was also found between SA patients and controls (P < .05). Average EV size was similarly increased in STEMI patients, and to a lesser extent in SA patients ( Figure 1B

| EV characterization and assay validation
The specificity of EV immuno-capture assay was assessed by Western blot and correlation analysis with NTA data. Presence of the exosomal markers TSG101 and CD81 after incubation with capture beads was demonstrated by Western blotting ( Figure   S1D,E). Apolipoproteins A1 and B48 were also detectable in these preparations, as expected, being reduced by up to 90% compared with the respective serum samples, suggesting a negligible lipoprotein contamination. It should be noted, however, that the assay used for FC analysis, selectively measured EV surface epitopes of interest labelled with EV-specific markers (CD9, CD63 and CD81, tetraspanins generally accepted as EV surface markers), providing for an additional level of EV selectivity ( Figure S1). In addition, the expression levels of tetraspanins at FC analysis were directly  Figure S1C). As expected, CD9/CD63/CD81 MFI as a measure of EV concentration was also increased in serum samples from STEMI patients compared with SA patients and controls (P < .001; Figure 1F). Separate experiments were performed to evaluate whether different protocols for EV isolation may affect expression EV surface antigens. After ultracentrifugation, we found a similar profile as compared with the standard immunocapture procedure described above, whereas the supernatant was relatively EV-depleted, as expected ( Figure S2A). Similarly, EV isolation by SEC did not substantially affect EV marker profiles ( Figure S2B). No difference was found comparing serum and plasma samples from the same patients ( Figure S2C). EV marker profiles measured using 1-hour incubation of serum supernatant  Table S9. Data are shown as median and interquartile range. *P < .05; **P < .01 with capture beads were similar to those measured using overnight incubation ( Figure S2D). The low internal variability of our protocol was confirmed by analysing twice the same samples ( Figure S4).

| EV surface epitopes
EV surface profiling was performed by FC analysis after bead-based immuno-capture according to the protocol shown above. EV surface epitope levels in STEMI patients, SA patients and controls are shown in Figure 2 and Tables S9 and S10. In each group, the two most highly expressed EV biomarkers were CD62P (P-selectin α-granule membrane protein) and CD42a (platelet membrane glycoprotein), followed by CD41b (platelet membrane glycoprotein II-b) and CD31 (Platelet-Endothelial Cell Adhesion Molecule-1; PECAM-1). These EV epitopes, along with CD40 (antigen-presenting cells co-stimulatory receptor), were increased in STEMI patients on presentation to the emergency department (P < .001; Figure 2A). CD62P, CD42a and CD41b levels also were increased in SA patients vs. controls (P < .01; Figure 2B), albeit to a lesser extent than in STEMI patients (NS). CD31 and CD40 levels were significantly increased in STEMI patients vs. SA patients (P < .01). A heat map of EV surface epitopes in individual participants showed a clear clustering in STEMI patients ( Figure 2C). Each of these markers (CD62P, CD42a, CD41b, CD31 and CD40) correlated directly to peak hs-troponin and inversely to LVEF (Figure 3), which reflect cardiac injury. To assess the potential impact of the biological material, a comparative analysis of EV epitopes using plasma and serum samples from the same subjects was performed in a subset of patients. Each of the five markers that were significantly increased in STEMI patients at pre-PCI evaluation compared with controls using serum samples, remained significantly increased in STEMI patients using plasma samples ( Figure S3) Regression lines and 95% confidence intervals are shown. *P < .05; **P < .01 department, whereas it was increased at 24 hours and 48 hours (P = .027 and P = .035, respectively; Figure S5).

| Performance of EV markers in STEMI discrimination
The diagnostic performance of EV concentration and EV surface markers in discriminating STEMI patients and controls was assessed by multivariate logistic regression and ROC curve analyses. EV concentration and levels of CD62P, CD42a, CD41b, CD40 and CD31 were independent significant predictors of STEMI (Table 2). In the training cohort, ROC curves indicated a high sensitivity for these markers. An aggregate marker including the three most highly discriminating parameters (EV concentration, CD62P levels and CD42a levels) achieved a sensitivity of 90.0% and a specificity of 93.3% ( Figure 4A-C). The evaluation of the AUC confirmed excellent diagnostic performances for these markers (EV concentration, CD62P, CD42a, CD41b, CD40 and CD31). EV concentration, CD62P, CD42a and the aggregate EV marker were not inferior to hs-troponin ( Figure 4C). Of note, the aggregate EV marker was increased above its cut-off value in 92.3% of STEMI patients presenting with minimally elevated hs-troponin < 50 ng/L (a cut-off equalling mean + 1.95*SD of Ctrl values).
In the validation cohort, EV concentration, CD62P, CD41b, CD42a and CD31 were not inferior to hs-troponin, with AUC ranging from 0.865 to 0.979, and sensitivity ranging from 85.0% to 97.5%.
The aggregate EV marker achieved higher AUC and diagnostic performance compared with hs-troponin (P = .036), with a sensitivity/specificity of 100.0%/87.5% ( Figure 4B-D). This marker was increased in all STEMI patients presenting with minimally elevated hs-troponin levels. The combination of the aggregate EV marker and hs-troponin showed a higher AUC compared with hs-troponin alone, with a sensitivity/specificity of 100.0%/95.0% (P = .021; Figure 4D).

| Machine learning algorithm model based on expression levels of EV biomarkers
The linear combination of all EV surface epitopes is shown in the canonical plot ( Figure 5A). The model distinguished STEMI patients, SA patients and controls. At first screening analysis, only 6 of 30 STEMI patients were misclassified (5 of them as SA and 1 as control), resulting in an accuracy of 81.6% ( Figure 5B). We then developed a diagnostic model for STEMI based on the 37 EV surface epitopes included in the multiplex using RF classification algorithms ( Figure 5C,D) Figure 5E,F).

| D ISCUSS I ON
We have measured multiple EV biomarkers representative of various cell types of origin in the blood of patients with acute MI or chronic CAD presenting with SA. STEMI patients on presentation to the emergency department and to a lesser extent SA patient showed an increase in circulating EV numbers, compared to healthy individuals.
Both STEMI patients and SA patients showed an increase in smallsize EVs (≤150 nm, the size traditionally associated with exosomes), whereas only STEMI patients exhibited augmented numbers of medium/large-sized EVs (151-500 nm, the size traditionally associated with microvesicles). These results are consistent with the notion that release of microvesicles is induced by cellular stress, 31 which dramatically increases as a result of acute coronary artery occlusion leading to cardiac injury and the inflammatory response to it. It should be noted, however, that exosomes and microvesicles cannot be precisely discriminated by size. 10 The two most highly enriched EV markers in STEMI patients, and to a lesser extent in SA patients, were CD62P, a marker of platelet activation, and CD42a, another platelet-associated marker. 32 Two additional platelet-associated markers, CD41b and CD31 (the latter also being expressed by activated endothelial cells), along with CD40, which may reflect immune activation, were increased in STEMI patients. Each of the above markers correlated with STEMI diagnosis at multivariate regression analysis.
They also correlated positively to peak hs-troponin levels and inversely to LVEF, suggesting a direct correlation to injury sever- F I G U R E 4 Diagnostic performance of EV markers. Diagnostic performance of EV concentration (by NTA) and five EV markers, compared with hs-troponin (*P-values refer to the comparisons of the areas under the curves, AUCs) in the training cohort and in the validation cohort (n = 60 and n = 80, respectively; STEMI patients pre-PCI vs. Ctrl). A and B, ROC curve analysis for hs-troponin (black curve) compared to individual EV markers (dashed curves), an aggregate EV marker including EV concentration, CD62P MFI, and CD42a MFI; red curve), and to the combination of the aggregate EV marker with hs-troponin (blue curve). C and D, AUC (95% Confidence Interval; CI), best cut-off (according to Youden's index analysis), sensitivity (%) and specificity (%), and percentages of patients with minimally increased hs-troponin (troponin <50 ng/L) showing levels of EV markers higher than the respective cut-off. Bold characters indicate P-values of diagnostic performances showing non-inferiority to hs-troponin (AUC for EV markers < AUC for hs-troponin; P ≥ .05) or superiority to hs-troponin (AUC for EV markers > AUC for hs-troponin; P < .05) We then looked at the potential diagnostic performance of the above EV markers for STEMI. ROC curve analysis revealed a high performance for CD62P, CD42a, CD41b, CD40 and CD31, with CD62P and CD42a showing non-inferiority to hs-troponin. We also aggregated the three most discriminating markers (EV concentra- Regarding methodological issues, we confirmed the technical reproducibility of the flow cytometric assay used in the present study.
We also found no major differences between sample ultracentrifugation and SEC, and between 1-hour or overnight incubation of serum can be customized to optimize diagnostic accuracy, and the method is amenable to full automation.

ACK N OWLED G EM ENTS
Marco Moccetti, MD, Luigi Biasco, MD, and Giovanni Pedrazzini, MD, from the Interventional Cardiology Unit of the Cardiocentro Ticino, Lugano (Switzerland), participated in the collection of blood samples from STEMI patients.

CO N FLI C T O F I NTE R E S T
The authors confirm that there are no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data that support the findings of this study are available from the corresponding authors upon reasonable request.