The current observational investigation is a substudy of the ASCET trial (Aspirin Nonresponsiveness and Clopidogrel Endpoint Trial) conducted at the Department of Cardiology Ullevål, Oslo University Hospital, Oslo, Norway. Between March 2003 and July 2010, patients with stable symptomatic CAD were enrolled in the ASCET trial (n = 1,000) (35). Eligibility screening was performed after coronary angiography had confirmed the presence of CAD. Patients with ACS and those treated with percutaneous coronary intervention (PCI) were eligible for enrolment after completing the guideline-recommended duration of dual antiplatelet therapy, as determined by the treating physician in accordance with the European Society of Cardiology (ESC) Guidelines, typically one to 12 months after initial screening. In the ASCET trial, all patients were on antiplatelet monotherapy with aspirin at the time of blood sampling. Patients using oral anticoagulants were excluded. Upon inclusion, they were randomized to either continue aspirin therapy or a switch to clopidogrel. At the two-year follow-up, the primary endpoint was defined as the first occurrence of a composite event consisting of myocardial infarction (MI), non-hemorrhagic stroke, or all-cause mortality, and was documented as a binary outcome (yes/no). This definition differed slightly from the original composite endpoint in the ASCET trial, as the less reliable diagnosis of unstable angina was excluded (35). The ASCET trial was conducted prior to the introduction of high-sensitivity troponin assays. All participants gave written informed consent. The ASCET study was approved by the regional committees for medical and healthcare research ethics (REK ID# 209/02), was registered at https://www.clinicaltrials.gov (ID# NCT00222261), complied with the Declaration of Helsinki and Good Clinical Practice (GCP) procedures, and the continued use of the biobank was granted (REK ID# 2017/652).

Venous blood samples were drawn in a fasting condition before morning medication and prior to inclusion. Whole blood was allowed to clot for up to 1 h at room temperature and centrifuged at 2,500×g for 10 min. Citrated blood (0.129 M; 1:10 dilution) was kept on ice and processed within 30 min with centrifugation at 4 °C, 3,000×g for 20 min. Aliquots of serum and citrated plasma were stored at −80 °C until analysis to preserve stability. CitH₃ was assessed in serum by a commercially available enzyme linked immunosorbent assay (ELISA) kit (Cayman Chemical, Ann Arbor, MI, USA, Cat nr 501620), according to the manufacturer's instructions. In brief, the assay employed a CitH₃-specific capture antibody and a horseradish peroxidase (HRP)-conjugated detection antibody. Color development was achieved with tetramethylbenzidine (TMB), and absorbance was measured at 450 nm on a microplate reader (EPOCH2TS, BioTek Instruments, supplied by AH Diagnostics, Oslo, Norway). The concentrations were calculated from a standard curve generated with known CitH₃ standards, and the inter-assay CV was 12.7%. The previously assessed biomarkers of NETs in the ASCET trial, i.e., dsDNA and MPO-DNA, were also analyzed in serum (32). In short, dsDNA was measured with use of the fluorescent nucleic acid stain Quant-iT PicoGreen (Invitrogen Ltd., Paisley, UK, Cat nr P7589) and detected by fluorometry (Fluoroskan Ascent, Thermo Fisher Scientific Oy, Vantaa, Finland). MPO-DNA complexes were identified using an ELISA technique, as described (36). Briefly, microtiter plates (Immulon 4HBX, flat bottom, Thermo Fisher Scientific, MA, USA, Cat nr 3855) were coated with anti-MPO monoclonal antibody (AbD Serotec, Hercules, CA, USA, Cat nr 0400-0002) and incubated overnight at 4 °C. After blocking with bovine serum albumin (BSA) 1%, patient serum and a peroxidase-labeled anti-DNA monoclonal antibody (Cell Death detection kit; Roche Diagnostics, Mannheim, Germany, Cat nr 11774425001) were added. Following incubation and washing, a peroxidase substrate was applied, and absorbance was measured after 40 min. Results were expressed as optical density units. The inter-assay CVs for dsDNA and MPO-DNA measurements were 7.0% and 10.5%, respectively. We have previously reported measurements of VWF, ADAMTS13 antigen (Ag), P-selectin, and high sensitivity (hs)-CRP (19, 35, 37). VWF, ADAMTS13 Ag, and P-selectin were analyzed in citrated plasma using the following commercially available ELISA kits: Asserachrom VWF Ag (Stago Diagnostica, Asnières, France, Cat nr 00492), ADAMTS13 Ag IMUBIND® (Sekisui Diagnostics GmbH, Pfungstadt, Germany, Cat nr 813) and P-selectin (R&D Systems Europe, Abingdon, Oxon, UK, Cat nr DPSE00). Hs-CRP was measured in serum (DRG Instruments, Marburg/Lahn, Germany, Cat nr EIA-3954). The inter-assay CVs for these assays were 6.3%, 8.7%, 7.2%, and 10.0%, respectively. Standard clinical chemistry analyses were conducted at the Department of Medical Biochemistry, Oslo University Hospital Ullevål using accredited methods.

Statistical analyses were performed with SPSS Inc., IL, USA version 29 and 30. Clinical characteristics were expressed as median (25th, 75th percentiles) or mean (±standard deviation) for continuous variables, according to normal or non-normal distribution-, and as numbers (percentages) for categorical variables. Group comparisons were performed with unpaired Student's t-test, Mann–Whitney U test and Chi-Square test, as appropriate. Groups were defined either by natural categories (e.g., sex) or by dichotomizing continuous variables at the median (above vs. below median) or between quartile Q1–Q3 and quartile Q4. Spearmans's rank correlation analyses were applied to measure the strength and direction of associations. Binary logistic regression analyses were used to estimate odds ratio (OR) with 95% confidence intervals (CIs) for the composite clinical endpoint. In the models, three combinations of thromboinflammatory markers including VWF and CitH₃ in the highest quartile (Q4) and ADAMTS13 Ag in the lowest quartile (Q1) (combination I), VWF and dsDNA in Q4 and ADAMTS13 Ag in Q1 (combination II), and VWF and MPO-DNA in Q4 and ADAMTS13 Ag in Q1 (combination III) were evaluated. Sensitivity analysis was performed with VWF and ADAMTS13 Ag. Diabetes mellitus, previous MI, hypertension and hs-CRP were considered potential confounders. Significant associations were adjusted for age and sex, in addition to confounders with a significant contribution by multivariate logistic regression analyses. A p-value <0.05 was considered statistically significant. The Bonferroni method was used for multiple comparisons in the correlation analyses (p = 0.05/15 = 0.0033). This observational cohort study was reported in accordance with the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines. A completed STROBE cohort checklist is provided in the Supplementary Material.

Baseline characteristics of the total stable CAD population (n = 1,000) are presented in Table 1, stratified by whether patients experienced a composite clinical endpoint (n = 73) or not (n = 927) at two-year follow-up. In brief, 21.8% were women, mean age was 62 years, and 96.8% were Caucasian. The average body mass index (BMI) was 27.4, 20% had diabetes, and more than half had hypertension (n = 556). Almost all patients used statins. Patients with the composite clinical endpoint registered at two-year follow-up, n = 73, more frequently had CVD, previous MI and diabetes. Of note, in the ASCET trial (35), the incidence of the composite clinical endpoint was similar between the aspirin and clopidogrel groups; therefore, the treatment arms were merged for this post-hoc analysis.

Baseline levels of CitH_3, and the previously measured NETs biomarkers MPO-DNA and dsDNA did not differ significantly between stable CAD patients experiencing the composite clinical endpoint or not (Table 1). As expected, CitH₃ was positively correlated with MPO-DNA and dsDNA, respectively (r = 0.291 and r = 0.146, p < 0.001, both), and correlated with neutrophil count (r = 0.221, p < 0.001).

We have previously reported significantly higher levels of VWF and VWF/ADAMTS13 Ag ratio in patients having a clinical endpoint (19), whereas neutrophil- and platelet count, as well as P-selectin and hs-CRP were similar in the two endpoint-groups (Table 1).

Patients younger than the mean age (<62 years) and those with BMI above the mean (>27.4 kg/m²) had significantly higher CitH₃ levels (p = 0.046 and p = 0.022, respectively). No other cardiovascular risk factors were associated with CitH₃ concentrations (Supplementary Table S1).

Among patients who experienced a clinical endpoint (n = 73; 36 MI, 28 non-hemorrhagic strokes, and 9 deaths), CitH₃ levels tended to be lower compared to those without an endpoint (Table 1). However, when CitH₃ was categorized into quartiles (Q1–3 ≤ 4.63 ng/mL vs. Q4 ≥ 4.64 ng/mL), univariate analysis showed no significant association with endpoint risk (OR: 0.70, 95% CI: 0.38, 1.28, p = 0.243; Supplementary Figure S1).

Overall correlations between circulating markers of thromboinflammation are summarized in Table 2. Initial analysis demonstrated significant correlations between dsDNA and both ADAMTS13 Ag and the VWF/ADAMTS13 Ag ratio, however, these associations did not remain significant after Bonferroni correction. Subgroup analyses [patients with a clinical endpoint (n = 73), and those with previous MI (n = 36)], revealed no significant correlations (data not shown).

We first evaluated the predictive value of combining Q4 of VWF with Q1 of ADAMTS13 Ag for the composite clinical endpoint (sensitivity analysis, Supplementary Table S2). Subsequently, we incorporated NETs biomarkers to create three combinations:

Combination I: VWF ≥1.33 IU/mL, ADAMTS13 Ag <461 ng/mL, and CitH₃ ≥ 4.64 ng/mL

Circulating levels of the measured NETs biomarkers (CitH₃, MPO-DNA and dsDNA) were not significantly associated with the VWF-ADAMTS13 axis in our relatively large cohort of stable CAD patients. To date, no studies have specifically examined this potential interplay. In contrast to our findings, a weak relationship among extracellular DNA, NETs biomarkers, and VWF was observed in patients with suspected CAD referred to cardiac computed tomography angiography (CCTA) (38). In a small study in patients with coronavirus disease (COVID 2019) vs. healthy controls, a strong correlation between NETs and VWF was found at high vs. low levels of cell-damage [measured with cell-free DNA (cfDNA)] (39). Despite guideline-directed medical therapy, patients with CAD continue to have a clinically relevant residual atherothrombotic risk driven by persistent platelet hyperreactivity, ongoing systemic inflammation, and existing atherosclerotic plaques (9). Therefore, one may hypothesize that the lack of association between circulating NETs biomarkers and VWF in stable CAD patients could reflect either the binding of NETs to VWF or sequestration of NETs components within microvascular thrombi (10, 40, 41). Such mechanisms would reduce their availability as circulating markers. Supporting this, a positive correlation between VWF and DNA has been demonstrated in arterial thrombi from patients with CAD and peripheral artery disease (42). Nevertheless, the lack of association in the current cohort of stable CAD patients may be ascribed to optimal treatment according to current guidelines. All patients were taking aspirin and 98% were using statins. Antiplatelet treatment has been shown to attenuate NETosis (43, 44).

When we explored CitH₃ as a single biomarker of NETs, it was based on the hypothesis that measuring NETosis at a single time point might reflect a state of chronically increased NETs release, however, it had no predictive value for the composite endpoint during the two-year follow-up. Limited data exists on the predictive role of CitH₃ and clinical outcomes in CAD. In the previously mentioned study in patients with suspected CAD (n = 282), dsDNA, MPO-DNA and nucleosomes predicted the severity of the disease and the occurrence of MACEs (n = 27, 9.7%) after a median of 545 days follow-up (38). Consistent with our findings, no association with citrullinated histone H4, another NET associated protein, was detected. The evidence for NETs biomarkers in predicting MACE and mortality in STEMI patients has been more established (45, 46). Recently, CitH₃-DNA measured in the acute phase was found to be elevated and independently associated with mortality after 30 days and MACE after one year, despite a markedly reduction in CitH₃-DNA concentration at six months (47). Its predictive value was enhanced when combined with CRP, reinforcing the concept of thromboinflammation. The type of NETs released have been suggested to depend on the activating stimuli, which may explain the varying results across clinical manifestations of CAD (48).

We have previously reported an association between circulating dsDNA and MACE in the present population (32). The contrasting results regarding specific NETs biomarkers, may also be attributable to the different methods applied, as well as their specificity for NETs. DsDNA also represent systemic inflammation and cell-death in general. Moreover, the lack of association between CitH₃ and MACE may be explained by a lower number of events in the current study, due to exclusion of unstable angina which was originally included in the composite endpoint definition.

Consistent with our observation of higher NETs levels in overweight stable CAD patients, NETs biomarkers were found to associate with BMI and hypertension in another CAD cohort (49). In patients with coronary stenosis compared to matched controls, higher levels of neutrophils were found and NETs biomarkers predicted the degree of stenosis (49). These findings are plausibly accounted for by the heightened state of chronic low-grade inflammation, endothelial dysfunction and atherogenesis associated with cardiovascular risk factors, driving the release of NETs (1).

Interestingly, in a subset of our stable CAD patients, we identified a distinctive biomarker profile characterized by low ADAMTS13 Ag and high VWF levels, in combination with high circulating levels of CitH₃ or dsDNA. These combined profiles independently predicted the incidence of new MI, non-hemorrhagic stroke, and all-cause mortality. To the best of our knowledge, this is the first study to demonstrate the utility of a combined thromboinflammatory biomarker profile in predicting adverse clinical outcomes in stable CAD. We speculate that such an imbalance in the VWF-NETs axis may cause a deleterious cycle with presence of ultralong VWF multimers, increased platelet- and neutrophil activation, NETosis, endothelial dysfunction and thrombosis. In accordance, we also found significantly higher levels of leukocytes and neutrophils in this subset of patients. In those with the biomarker profile low ADAMTS13 Ag, high VWF and high CitH₃, we found significantly higher concentration of the inflammatory marker hs-CRP and platelet count, but not markers of platelet activation. P-selectin has been shown to promote NETs in mice, and increased levels have been found in patients with cardiovascular diseases, and to be associated with an elevated risk of MI, stroke and cardiovascular death (50–52). The combinations were superior to the single NET biomarkers and may enlighten mechanisms behind progression of disease and the increased thrombotic risk. In patients with COVID 2019, thrombotic microangiopathy and heart failure, NETs are already well-known to drive disease progression (26, 53, 54). Based on our findings, we suggest that adding NETs to the biomarker profile enhance the predictive power for experiencing new clinical events in patients with stable CAD. Unlike with dsDNA and CitH₃, the biomarker profile with low ADAMTS13 Ag, high VWF and high MPO-DNA did not predict new clinical events in our cohort. This may be attributed to the heterogeneity of neutrophil subtypes and the diversity of NETs-inducing stimuli, which can influence the composition and functional properties of the NETs released (48).

Although CitH₃ is considered a specific biomarker of NETs, the detection of enzymatically citrullinated H₃ proteins has been impeded by variability in enzyme production and stability as well as poor antibody specificity in commercially available methods (55). Future research should focus on developing robust assays for NETs to enhance clinical utility. Assessing circulating levels of surrogate biomarkers may not reflect, and therefore limit, the understanding of vascular and local activity. This retrospective cohort study included small subgroups and limited endpoint numbers; therefore, the findings should be interpreted with caution and confirmed in a larger, prospective studies.

In stable CAD patients, combined extreme levels of thromboinflammatory biomarkers, i.e., low ADAMTS13 Ag, high VWF and high CitH₃ or dsDNA, respectively, predicted MI, non-hemorrhagic stroke, and all-cause mortality despite medication and treatment. The investigated combined biomarker profile may aid in identifying high-risk individuals, and targeting these pathways may become relevant to new treatment strategies. Our study refines risk stratification in stable CAD and creates potential for personalized antithrombotic therapy targeting thromboinflammation.