Circulating Markers of Neutrophil Extracellular Traps for Long‐Term Prognosis in Patients With Acute Chest Pain

Abstract

Background

Neutrophil extracellular traps are released from activated neutrophils and are involved in the pathogenesis of atherosclerotic lesions, atherothrombosis, and myocardial injury. We investigated the prognostic value of circulating neutrophil extracellular trap biomarkers in patients with suspected acute coronary syndrome (ACS).

Methods

A total of 1482 patients admitted with suspected non–ST‐segment elevation ACS were included and followed for a median of 4.2 years. The primary end point was a composite of death from any cause, incident myocardial infarction and hospitalization for heart failure. Secondary end points were all‐cause mortality, cardiovascular death, incident myocardial infarction, hospitalization for heart failure, and new‐onset atrial fibrillation. Admission blood samples were analyzed for the neutrophil extracellular trap biomarkers double‐stranded DNA (dsDNA), CitH3 (citrullinated histone H3), and myeloperoxidase‐DNA.

Results

A doubling of dsDNA concentration was associated with a hazard ratio (HR) of 3.11 (95% CI, 1.61–5.98, P<0.001) for the primary end point after adjusting for traditional risk factors, cardiac troponin T and N‐terminal pro‐B‐type natriuretic peptide. DsDNA served as a prognostic marker both in patients with (adjusted HR, 5.33 [95% CI, 1.67–17.06], P=0.005) and without ACS (adjusted HR, 2.86 [95% CI, 1.30–6.28], P=0.009). In contrast, CitH3 and myeloperoxidase‐DNA showed no significant prognostic value.

Conclusions

In patients with suspected ACS, dsDNA emerged as a long‐term prognostic marker for a composite outcome of death, incident myocardial infarction, or heart failure hospitalization, independent of conventional risk factors. DsDNA can independently from established risk factors identify high‐risk patients with and without ACS who may benefit from risk reduction.

Nonstandard Abbreviations and Acronyms

CitH3

citrullinated histone H3

dsDNA

double‐stranded DNA

GDF‐15

growth differentiation factor 15

MPO‐DNA

myeloperoxidase‐DNA

NETs

neutrophil extracellular traps

Clinical Perspective.

What Is New?

• By reflecting a distinct pathophysiological mechanism, elevated double‐stranded DNA concentrations predict death, myocardial infarction, and heart failure independently of cardiac troponins and NT‐proBNP (NT‐terminal pro‐B‐type natriuretic peptide).

• These associations persist in patients without acute coronary syndromes or myocardial injury, enabling identification of high‐risk individuals among the majority presenting with acute chest pain who are ultimately not diagnosed with acute coronary syndromes.

What Are the Clinical Implications?

• Measurement of circulating double‐stranded DNA may help identify high‐risk patients who are overlooked by conventional cardiac biomarkers; future studies should investigate if identifying high‐risk patients with elevated double‐stranded DNA may help guide selection for anti‐inflammatory or neutrophil extracellular trap‐targeted interventions.

Inflammation is an important driver of atherosclerosis, atherothrombosis, and cardiac injury. ¹ Neutrophils constitute the first line defense of the innate immune response and are involved in cardiovascular pathophysiological processes through a range of mechanisms. Neutrophils extensively infiltrate myocardial tissue during the first hours of myocardial infarction (MI), ² , ³ participate in ischemia–reperfusion injury, ⁴ and modulate the coagulation cascade and thrombus formation. ⁵ Neutrophils are also present in atherosclerotic lesions, ⁶ where they contribute to plaque progression and have been implicated in plaque instability. ⁷ , ⁸

Neutrophils exacerbate inflammation through the formation of extracellular traps (NETs), in which activated neutrophils release chromatin and granule proteins that form web‐like structures in a process called NETosis. ⁹ Core components of NETs comprise double‐stranded DNA (dsDNA), histones, and proteolytic enzymes such as MPO (myeloperoxidase). Studies have identified increased concentrations of NETs at the culprit lesion site in ST‐segment–elevation MI (STEMI), ¹⁰ , ¹¹ which have been associated with poor prognosis. ¹² , ¹³ The same observation has been made in patients with chronic coronary syndromes, ¹⁴ and dsDNA has in particular been linked to worse clinical outcomes. However, the prognostic role of NETs in unselected patients with chest pain remains unknown. Circulating levels of NET markers could provide valuable prognostic information in patients presenting with acute chest pain, including patients without acute coronary syndromes (ACS) who may still be at increased risk of cardiovascular events. Several studies have shown that targeting inflammatory pathways can reduce cardiovascular risk. ¹⁵ , ¹⁶ Given that NETs can be influenced by commonly used medications such as statins and metformin, ¹⁷ , ¹⁸ identifying a subgroup of patients with elevated NETs concentrations and heightened cardiovascular risk could define a population that may benefit from more intensive medical treatment.

Our study aimed to investigate whether circulating biomarkers of NETs, including dsDNA, CitH3 (citrullinated histone H3), and MPO‐DNA, can be used for long‐term risk prediction in patients presenting with acute chest pain (both with and without ACS), and compare their performance with that of established biomarkers. We also aimed to identify predictors of circulating dsDNA concentrations and assess their association with myocardial injury and coronary atherosclerosis reflected by the calcium score.

Methods

Study Design and Population

The study included 1482 patients from the prospective observational WESTCOR (Aiming Towards Evidence Based Interpretation of Cardiac Biomarkers in Patients Presenting With Chest Pain) study (ClinicalTrials.gov number: NCT02620202), which enrolled patients ≥18 years admitted to the emergency department with suspected non–ST‐segment elevation‐ACS at Haukeland University Hospital in Bergen and Stavanger University Hospital, Norway from September 2015 to March 2020. Details of the WESTCOR Study have been published previously ¹⁹ , ²⁰ and are provided in Data S1. The study was approved by the Regional Ethics Committee (REC number 2014/1365) and the study conformed to the Declaration of Helsinki. Written informed consent for participation in the study was provided by all patients.

Data Availability Statement

The data supporting the findings of this study are available from the corresponding author upon reasonable request, subject to institutional and ethical regulations.

End Points

Events after admission were collected through the Norwegian Patient Register and Norwegian Cause of Death Register. Patients were followed for a median of 1540 (25th to 75th percentile: 1087–1776) days until a primary end point consisting of all‐cause mortality, incident MI, and hospitalization for heart failure. Secondary end points included all‐cause mortality, cardiovascular death, incident MI, heart failure hospitalization, and new‐onset atrial fibrillation.

Biochemical Analyses

Circulating biomarkers of NETs were measured in peripheral blood samples obtained at admission. The concentration of dsDNA (ng/mL) was determined using Quant‐iT PicoGreen, a commercially available fluorescent dye that binds to nucleic acid. PicoGreen was added to diluted EDTA plasma, and fluorescence was measured at excitation/emission wavelengths of 485/538 nm using Fluoroskan Ascent fluorometer (Thermo Fisher Scientific Oy, Vantaa, Finland). MPO‐DNA complexes were measured in undiluted serum using an ELISA method previously described by Kessenbrock. ²¹ In summary, plates were coated with an anti‐MPO capture antibody (Bio‐Rad, Hercules) and incubated overnight at 4 °C. Following blocking with bovine serum albumin, patient samples, and a peroxidase‐labeled anti‐DNA antibody (Cell Death Detection Kit, Roche Diagnostics GmbH) were added and incubated for 2 hours. A peroxidase substrate was then added, and absorbance was measured and expressed in optical density units. CitH3 was analyzed in serum at a 1:2 dilution using a commercial ELISA kit (Cayman Chemical) according to the manufacturer’s protocol. The interassay coefficients of variation were 4.8% for dsDNA, 5.3% for MPO‐DNA, and 11.9% for CitH3. Other circulating biomarkers included hs‐cTnT (high‐sensitivity cardiac troponin T), CRP (C‐reactive protein), GDF‐15 (growth differentiation factor 15) and NT‐proBNP (N‐terminal pro‐B‐type natriuretic peptide), all from Roche diagnostics. Assay information on hs‐cTnT, CRP, GDF‐15, and NT‐proBNP has been published previously ²² and is also provided in Data S1.

Statistical Analysis

After excluding samples with hemolysis, dsDNA concentrations were obtained in 1481 patients, and MPO‐DNA and CitH3 were obtained in 1482 patients. Baseline characteristics were reported as numbers and percentages for categorical values, and continuous variables were reported as median (25th to 75th percentile). Differences among groups were tested using the Pearson χ ² test or Fisher’s exact test for categorical variables, and the Student’s t test or Mann–Whitney U test for continuous variables. Distribution was examined using the Shapiro–Wilk test and by visual assessment of Q‐Q plots. Right‐skewed variables were log2 transformed before being used in the logistic regression and in the Cox regression analysis.

Kaplan–Meier curves were generated to assess the ability of each biomarker, stratified by quartiles, to predict the primary end point. Differences between groups were compared using the log‐rank test.

Cox proportional hazard regression analyses were performed to calculate hazard ratios (HRs) for the association between biomarker concentrations (log2 transformed) and study end points. The Cox analysis was performed using an unadjusted model; model 1 adjusting for sex and age; model 2 adjusting for sex, age, diabetes, hypercholesterolemia, smoking, previous MI, hypertension, and estimated glomerular filtration rate; and model 3 adjusted for hs‐cTnT and NT‐proBNP in addition to variables in model 2. We also fitted a generalized additive model smoothed spline using a Cox model to explore the association between continuous biomarker concentrations and unadjusted hazard ratios. Recognizing the differences in pathophysiology, the prognostic value of dsDNA was also examined for patients with and without non–ST‐segment elevation‐ACS during admission, and for patients with and without myocardial injury (hs‐cTnT <99th percentile [<14 ng/L]). Given potential sex‐specific differences in the neutrophil response, ²³ we also assessed sex‐based variations in NETs concentrations and prognostic value. The effect of sex on the prognostic value of dsDNA for the primary end point was evaluated by including an interaction term between sex and dsDNA in the Cox regression model. The proportional hazard assumption was evaluated by plotting the residuals of covariates and the ranking of the failure times, and with P values >0.05 for the overall models, suggesting fulfilled criteria of assumptions. Martingale residual plots were created to assess potential nonlinearity in the relationship between continuous covariates and the log hazard. The plots indicated approximately linear relationships for most variables, although age, estimated glomerular filtration rate, log₂‐dsDNA, log₂‐NT‐proBNP, and log₂‐hs‐cTnT showed mild deviations from linearity. These findings were formally evaluated using likelihood ratio tests comparing linear and spline models. Only log₂‐hs‐cTnT demonstrated significant nonlinearity (P<0.001) and was therefore modeled using restricted cubic splines in all Cox regression analysis to allow for potential nonlinear relationships.

Discrimination was assessed by C‐statistics and receiver operating characteristic curves, in which C‐statistics were estimated for the different biomarker concentrations’ ability to predict the primary end point and compared using DeLong test. To evaluate the incremental value of adding dsDNA to existing biomarkers, predictor variables were incorporated into a binary logistic regression model, and the resulting predicted probabilities were used to calculate the area under the curve.

Binary logistic regression was used to determine the relationship between dsDNA concentrations above the median (dependent variable) and predictor variables (age, sex, hypercholesterolemia, current smoking, previous myocardial infarction, diabetes, hypertension, and log2‐transformed hs‐cTnT, NT‐proBNP, CRP, and GDF‐15). Correlation analyses were conducted using Spearman’s rank correlation (Spearman’s rho).

IBM SPSS Statistics for Windows, V.26.0, MedCalc statistical software V.17.6 and R V.4.2.3 were used for statistical analysis. For all statistical testing, 2‐sided P values were reported.

Results

Patient Characteristics

Patients with dsDNA > median concentration (400.1 ng/mL) were older, with higher rates of current smoking, hypertension, chronic kidney disease, peripheral arterial disease, and known heart failure (Table 1). They also had higher concentrations of hs‐cTnT, NT‐proBNP, CRP, and GDF‐15. Additionally, non–STEMI during the index hospitalization was more frequently observed in this group.

Table 1.

Baseline Characteristics of 1481 Patients Admitted With Acute Chest Pain Stratified by Median dsDNA Concentration

	All patients	< Median dsDNA concentration	> Median dsDNA concentration	P value*
	No.=1481	No.=741 (50%)	No.=740 (50%)
Age,y, median (interquartile range)	62 (52–73)	60 (50–70)	65 (53–75)	<0.001
Female sex, n (%)	588 (40)	311 (42)	277 (37)	0.074
Cardiovascular risk factors, n (%)
Obesity*	188 (24.6)	101 (27.1)	87 (24.2)	0.279
Current smoker	281 (18.9)	119 (16.0)	162 (21.9)	0.004
Hyperlipidemia†	573 (38.7)	270 (36.4)	303 (40.9)	0.075
Diabetes	180 (12.2)	81 (10.9)	99 (13.4)	0.150
Hypertension	609 (41.1)	268 (36.2)	341 (46.1)	<0.001
Medical history, n (%)
Previous MI	283 (19.1)	129 (17.4)	154 (20.8)	0.096
Atrial fibrillation	82 (5.5)	43 (5.8)	39 (5.3)	0.654
Previous stroke	42 (2.8)	20 (2.7)	22 (3.0)	0.575
Family history of coronary artery disease	271 (18.3)	136 (18.4)	135 (18.2)	0.956
Chronic kidney disease‡	183 (12.4)	71 (9.6)	112 (15.1)	<0.001
Peripheral arterial disease	29 (2.0)	4 (0.5)	25 (3.4)	<0.001
Known heart failure	51 (3.4)	13 (1.8)	38 (5.1)	<0.001
Laboratory parameters on admission, median (25th to 75th percentile)
Total cholesterol (mmol/L)	4.8 (3.9–5.8)	4.8 (3.9–5.8)	4.8 (3.9–5.7)	0.292
Low‐density lipoprotein cholesterol (mmol/L)	2.7 (1.9–4.4)	2.8 (2.0–3.7)	2.6 (1.9–3.5)	0.054
High‐density lipoprotein cholesterol (mmol/L)	1.4 (1.1–1.7)	1.4 (1.1–1.7)	1.3 (1.1–1.7)	0.143
Estimated glomerular filtration rate (mL/min/1.73 m2)	86 (72–97)	88 (74–97)	84 (69–96)	<0.001
High‐sensitivity cardiac troponin T (ng/L)	7 (3–16)	6 (3–13)	8 (4–20)	<0.001
N‐terminal pro‐B‐type natriuretic peptide (ng/L)	85 (35–251)	75 (32–180)	102 (39–387)	<0.001
Growth differentiation factor‐15 (pg/mL)	924 (629–1454)	826 (593–1242)	1060 (693–1721)	<0.001
C‐reactive protein (mg/L)	2.0 (0.7–4.0)	1.0 (0.6–2.3)	2.0 (0.9–6.0)	<0.001
Calcium score§	5 (0–109)	6 (0–118)	4 (0–101)	0.768
Adjudicated diagnosis during hospitalization, n (%)
Acute coronary syndrome	369 (24.9)	179 (24.2)	190 (25.7)	0.499
Non–ST‐segment–elevation MI	172 (11.6)	72 (9.8)	100 (13.5)	0.023
Unstable angina pectoris	197 (13.3)	107 (14.4)	90 (12.2)	0.197
Stable angina pectoris	14 (0.9%)	9 (1.2)	5 (0.7)	0.284
Noncoronary cardiac disease	95 (6.4)	39 (5.3)	56 (7.6)	0.114
Noncardiac chest pain	1003 (67.7)	514 (69.4)	489 (66.1)	0.262
Respiratory tract infection	11 (0.7)	3 (0.4)	8 (1.1)	0.130
Myocarditis, pericarditis, or pleuritis	19 (1.3)	5 (0.7)	14 (1.9)	0.037
Nonspecific chest pain	738 (49.8)	385 (52.0)	353 (47.7)	0.102
Medication at admission, n (%)
Aspirin	488 (32.9)	218 (29.4)	270 (36.5)	0.004
Other antiplatelet	111 (7.5)	55 (7.4)	56 (7.6)	0.916
Oral anticoagulants	160 (10.8)	70 (9.4)	90 (12.2)	0.092
Warfarin	75 (5.1)	29 (3.9)	46 (6.2)	0.043
Statin	564 (38.1)	264 (35.6)	300 (40.5)	0.052
Angiotensin‐converting enzyme inhibitor	499 (33.7)	231 (31.2)	268 (36.2)	0.040
Beta blocker	456 (30.8)	200 (27.0)	256 (34.6)	0.002
Diuretic	263 (17.8)	103 (13.9)	160 (21.6)	<0.001

dsDNA indicates double‐stranded DNA; and MI, myocardial infarction.

^*Body mass index >30, n=737 patients are included for this variable.

^† Use of lipid‐lowering agents before admission.

^‡ Estimated glomerulation filtration rate <60 mL/min/1.73 m².

^§ Calcium score: 603 patients were included for this variable.

Clinical Outcomes

Over a median follow‐up period of 1546 days (25–75 percentile: 871–1842), 200 patients (13.5%) reached the primary end point. In total, 143 patients died from any cause, out of whom 41 from cardiovascular causes, 60 suffered a MI, 46 were hospitalized for heart failure, and 46 were diagnosed with new‐onset atrial fibrillation.

Patients who met the primary end point had significantly higher dsDNA concentrations (median: 434 ng/mL [interquartile range, 391–480]) compared with those who did not (median: 396 ng/mL [interquartile range, 361–468]), P<0.001. In contrast, CitH3 concentrations were lower in patients who reached the primary end point (median: 5.8 ng/mL, 25th–75th percentile: 2.8–10.6) compared with those who did not (median: 7.0 ng/mL, 25th–75th percentile: 3.3–11.5; P=0.042). No significant difference was observed in MPO‐DNA concentrations between the 2 groups (P=0.340). Increasing quartiles of dsDNA concentrations were associated with an increased risk of the primary end point (Figures 1 and 2). A doubling of dsDNA concentration was associated with an HR of 3.11 (95% CI, 1.61–5.98), P<0.001 in the fully adjusted model (Table 2). Increasing CitH3 concentrations were associated with lower risk of the primary end point (Figures 1 and 2), but the association was not significant in the adjusted analysis (Table 2). There was no significant interaction between sex and dsDNA in predicting the primary end point (P=0.108).

Figure 1. Kaplan–Meier curves demonstrating the risk for the primary end point stratified by quartiles of biomarker concentrations (Q1: blue, Q2: green, Q3: yellow, Q4: purple) for double‐stranded DNA, citrullinated histone H3, myeloperoxidase‐DNA complexes.

CitH3 indicates citrullinated histone H3; dsDNA, double‐stranded DNA; and MPO‐DNA, myeloperoxidase‐DNA.

Figure 2. Generalized additive model curves demonstrating hazard ratios (unadjusted) for the primary end point along the y axis and biomarker concentrations (double‐stranded DNA, citrullinated histone H3, myeloperoxidase‐DNA complexes) along the x axis.

The 95% CIs are indicated by blue shade and the distribution of biomarker concentrations is indicated by gray shade along the x axis. CitH3 indicates citrullinated histone H3; dsDNA, double‐stranded DNA; MPO‐DNA, myeloperoxidase‐DNA; and OD, optical density.

Table 2.

HR (95% CI) in Proportional Hazards Analyses Were Calculated to Evaluate the Associations Between log2‐Transformed Biomarker Concentrations and the Primary Composite End Point

Variable	Unadjusted	Model 1	Model 2	Model 3
log2 DsDNA	7.29 (4.28–12.42) P<0.001	5.29 (2.73–10.26) P<0.001	4.14 (2.17–7.92) P<0.001	3.11 (1.61–5.98) P<0.001
log2 CitH3	0.87 (0.78–0.97) P=0.010	0.96 (0.87–1.07) P=0.492	0.98 (0.88–1.09) P=0.982	0.99 (0.89–1.10) P=0.794
Log2 MPO‐DNA	0.99 (0.78–1.28) P=0.987	0.94 (0.73–1.22) P=0.650	0.94 (0.73–1.21) P=0.615	0.92 (0.71–1.19) P=0.528

Model 1 adjusted for sex and age. Model 2 adjusted for sex, age, diabetes, hypercholesterolemia, smoking, previous myocardial infarction, hypertension, and estimated glomerular filtration rate. Model 3 adjusted for high‐sensitivity cardiac troponin T and N‐terminal pro‐B‐type natriuretic peptide in addition to the variables in model 2.

CitH3 indicates citrullinated histone H3; dsDNA, double‐stranded DNA; HR, hazard ratio; and MPO‐DNA, myeloperoxidase–DNA complexes.

Secondary End Points

Higher concentrations of dsDNA were significantly associated with increased risks of all‐cause mortality and cardiovascular death, with the associations remaining robust even after multivariable adjustments (Models 1 and 2). Increasing concentrations of dsDNA were associated with incident MI and hospitalization for heart failure in unadjusted analysis and model 1 but not model 2. No significant association was observed between dsDNA levels and the risk of new‐onset atrial fibrillation (Table 3).

Table 3.

HR (95% CI) in Proportional Hazards Analyses Were Calculated to Evaluate the Associations Between log2‐Transformed Biomarker Concentrations (Double‐Stranded DNA) and Secondary End Points (All‐Cause Mortality, Cardiovascular Death, Incident Myocardial Infarction, and Hospitalization for Heart Failure)

Outcome	Hazard ratios (95% CI)
	Variable	Unadjusted	Model 1	Model 2
All‐cause mortality (n=145)	log2 dsDNA	7.85 (4.16–14‐82) P<0.001	5.50 (2.34–12.69) P=0.001	4.12 (1.80–9.40) P=0.004
Cardiovascular death (n=41)	log2 dsDNA	11.51 (3.77–35.19) P<0.001	15.79 (4.90–50.87) P<0.001	7.86 (1.85–31.73) P=0.004
Heart failure hospitalization (n=47)	log2 dsDNA	7.20 (2.34–22.17) P<0.001	4.92 (1.26–19.18) P=0.022	3.39 (0.86–13.38) P=0.082
Myocardial infarction (N=60)	log2 dsDNA	5.78 (2.12–15.77) P<0.001	3.89 (1.22–12.25) P=0.017	3.01 (0.97–9.34) P=0.057
New‐onset atrial fibrillation (N=46)	log2 dsDNA	1.50 (0.43–5.24) P=0.527	0.88 (0.30–2.55) P=0.875	0.85 (0.33–2.13) P=0.722

Model 1 adjusted for sex and age. Model 2 adjusted for sex, age, diabetes, hypercholesterolemia, smoking, previous myocardial infarction, hypertension and estimated glomerular filtration rate. dsDNA indicates double‐stranded DNA; and HR, hazard ratio.

Prognostic Value of dsDNA Stratified by ACS and Myocardial Injury During Admission

DsDNA was an independent predictor of the primary end point in patients with and without ACS during index hospitalization (Table 4). DsDNA was also an independent predictor of the primary end point in patients without myocardial injury (Table 4). For patients without myocardial injury, having a dsDNA concentration above the median was associated with a HR of 1.87 (95% CI, 1.11–3.14, P=0.018) for the primary end point, after adjusting for traditional risk factors, hs‐cTnT and NT‐proBNP (Table 4).

Table 4.

Cox Proportional Hazard Regression Analysis for the Association Between Double‐Stranded DNA (log2 Transformed and dsDNA > Median) and the Primary End Point Stratified by Acute Coronary Syndrome During Hospitalization and by Myocardial Injury Defined as hs‐cTnT >14 ng/L (Measured in Admission Samples)

	Unadjusted	Model 1	Model 2	Model 3
Without ACS
log2 dsDNA	5.98 (3.07–11.64) P<0.001	4.97 (2.19–11.25) P=0.001	4.23 (1.88–9.53) P<0.001	2.86 (1.30–6.28) P=0.009
With ACS
log2 dsDNA	18.31 (5.89–56.86) P<0.001	8.02 (2.33–27.52) P<0.001	6.71 (2.12–21.21) P<0.001	5.33 (1.67–17.06) P=0.005
Without myocardial injury
log2 dsDNA	7.27 (2.76–19.11) P<0.001	6.38 (2.16–18.86) P<0.001	5.15 (1.77–14.90) P=0.003	3.62 (1.29–10.01) P=0.014
dsDNA >median	2.19 (1.34–3.59) P=0.002	1.94 (1.18–3.19) P=0.009	1.93 (1.18–3.18) P=0.009	1.87 (1.11–3.14) P=0.018
With myocardial injury
log2 dsDNA	3.75 (1.95–7.22) P<0.001	3.51 (1.55–7.96) P=0.003	2.78 (1.25–6.19) P=0.028	2.00 (0.88–4.56) P=0.098

Model 1 adjusted for sex and age. Model 2 adjusted for sex, age, diabetes, hypercholesterolemia, smoking, previous myocardial infarction, hypertension, and estimated glomerular filtration rate. Model 3 adjusted for hs‐cTnT and N‐terminal pro‐B‐type natriuretic peptide in addition to the variables in model 2. ACS indicates acute coronary syndrome; dsDNA, double‐stranded DNA; hs‐cTnT, high‐sensitivity cardiac troponin T.

Discriminative Performance of dsDNA Compared With Established Biomarkers

C‐statistic for dsDNA in predicting the primary end point was estimated at 0.68 (95% CI, 0.64–0.72), which was significantly lower than that of hs‐cTnT, NT‐proBNP, and GDF‐15 but higher than for CRP (Figure 3). Adding dsDNA to CRP led to an increase in C‐statistic from 0.63 (95% CI, 0.59–0.67) to 0.68 (95% CI, 0.64–0.72), P=0.025 for prediction of the primary end point. Adding dsDNA to hs‐cTnT led to a decrease in C‐statistic (Table S2). No differences were seen in C‐statistic by adding dsDNA to NT‐proBNP or GDF‐15 (Table S2).

Figure 3. Receiver operating characteristic area under the curve was calculated for double‐stranded DNA, citrullinated histone H3, myeloperoxidase‐DNA complexes, growth differentiation factor 15, N‐terminal pro‐B‐type natriuretic peptide, high‐sensitivity cardiac troponin T, and C‐reactive protein for prediction of the primary end point.

Differences in C‐statistics were compared using the DeLong test, with dsDNA as the reference. AUC indicates area under the curve; CitH3, citrullinated histone H3; CRP, C‐reactive protein; dsDNA, double‐stranded DNA; GDF‐15, growth differentiation factor 15; hs‐cTnT, high‐sensitivity cardiac troponin T; MPO‐DNA, myeloperoxidase‐DNA; NT‐proBNP, N‐terminal pro‐B‐type natriuretic peptide; and ROC, receiver operating characteristic.

Associations Between dsDNA, Clinical Variables, and Other Biomarkers

DsDNA was most strongly associated with CRP and GDF‐15 (Table S1). Weaker correlations were observed for estimated glomerular filtration rate, NT‐proBNP, cTnT, and age. Among the other NET markers, MPO‐DNA, and CitH3 correlated weakly with dsDNA. Notably, the correlation between MPO‐DNA and CitH3 was slightly stronger, at r _s=0.30 (P<0.001). There was no association between calcium score and dsDNA. The multivariable logistic regression analysis (Table 5) identified male sex, current smoking, hypertension, CRP, and GDF‐15 as independent predictors of having dsDNA concentration > median. R2 of this model was 0.13.

Table 5.

Association Between Predictor Variables and dsDNA Concentrations Above Versus Below the Median

Predictors	Odds ratio (95% CI) and P value (univariate)	Odds ratio (95% CI) and P value (multivariate)
Age (per y)	1.02 (1.01–1.02), P<0.001	1.00 (0.99–1.01), P=0.445
Female sex (vs male sex)	0.83 (0.67–1.02), P=0.074	0.74 (0.58–0.93), P=0.012
Hypercholesterolemia	1.14 (0.88–1.46), P=0.326	1.02 (0.77–1.35), P=0.871
Current smoking	1.47 (1.13–1.90), P=0.004	1.48 (1.12–1.95), P=0.005
Previous myocardial infarction	1.25 (0.96–1.62), P=0.096	0.98 (0.72–1.31), P=0.852
Diabetes	1.24 (0.91–1.70), P=0.175	0.87 (0.61–1.25), P=0.449
Hypertension	1.52 (1.24–1.87), P<0.001	1.29 (1.01–1.63), P=0.039
Estimated glomerular filtration rate (per mL/min/1.73 m2)	0.99 (0.98–0.99), P<0.001	1.00 (0.99–1.01), P=0.915
log2 high‐sensitivity cardiac troponin T (per doubling)	1.15 (1.08–1.22), P<0.001	0.97 (0.89–1.06), P=0.468
log2 N‐terminal pro‐B‐type natriuretic peptide (per doubling)	1.12 (1.07–1.17), P<0.001	1.02 (0.95–1.09), P=0.652
log2 C‐reactive protein (per doubling)	1.36 (1.28–1.45), P<0.001	1.31 (1.23–1.40), P<0.001
log2 growth differentiation factor 15 (per doubling)	1.58 (1.40–1.78), P<0.001	1.26 (1.06–1.51), P=0.010

DISCUSSION

We demonstrated that increasing dsDNA concentrations predicted a composite end point of death, MI, and heart failure hospitalization independently of traditional risk factors, hs‐cTnT and NT‐proBNP in patients presenting with acute chest pain. The prognostic value of dsDNA for death and cardiovascular events persisted in patients without ACS or detectable myocardial injury, highlighting its potential to identify a high‐risk group that may otherwise be overlooked using conventional clinical markers. DsDNA levels were modestly associated with systemic inflammation markers like CRP and GDF‐15 but weaker correlations with cardiovascular risk factors, myocardial injury markers and specific NETs markers (citH3 and MPO‐DNA) suggest that dsDNA primarily reflects broader inflammatory and pathogenic processes rather than NETosis. Circulating concentrations of CitH3 and MPO‐DNA showed no prognostic value.

The Prognostic Role of dsDNA and NETs in Cardiovascular Risk Assessment

In recent decades, there has been an increasing interest in the use of inflammatory markers for prognostic purposes in patients with suspected or established cardiovascular disease. Elevated concentrations of inflammatory markers such CRP, IL‐6 (interleukin‐6), and GDF‐15 are consistently associated with increased cardiovascular risk in patients with acute chest pain, in patients with MI, and also in the general population. ¹⁶ , ²⁰ , ²² , ²⁴ , ²⁵ Inflammatory markers such as CRP can be used to inform treatment strategies ²⁶ and have made their way into clinical guidelines and are currently recommended by the American Heart Association for risk stratification in the primary preventive setting ²⁷ and by the European Society of Cardiology in patients with chronic coronary syndromes. ²⁸ Cardiovascular inflammation is not necessarily mirrored by CRP, emphasizing the need for additional biomarkers to capture other inflammatory pathways. Previous studies have identified NETs as markers of cardiovascular risk in patients with STEMI ¹⁰ , ¹² and chronic coronary syndromes. ¹⁴ Our study group previously reported that in a cohort of 956 patients with STEMI, elevated dsDNA concentrations were associated with increased mortality over a 4.6‐year follow‐up period. ¹³ Similarly, in 1001 patients with stable angina, having dsDNA in the upper 3 quartiles was associated with an adjusted odds ratio of 2.01 (95% CI, 1.12–3.58, P=0.019) for experiencing death or cardiovascular events during a 2‐year follow‐up period. ¹⁴ Notably, MPO‐DNA showed no association with clinical outcomes in this study. Consistent with these findings, our study demonstrates that in a cohort of unselected chest pain patients, higher dsDNA concentrations independently predict death, MI, and heart failure hospitalization, even after adjusting for traditional risk factors, hs‐cTnT, and NT‐proBNP. The increased risk associated with higher dsDNA concentrations persisted in patients without ACS or myocardial injury on admission. In patients without myocardial injury, having a dsDNA concentration above the median was associated with an HR 1.87 (95% CI, 1.11–3.14, P=0.018) for the primary end point, after adjusting for traditional risk factors, hs‐cTnT and NT‐proBNP. Although NETs have been implicated in the pathogenesis of atrial fibrillation, this is to our knowledge the first study investigating whether elevated NETs concentrations can predict future atrial fibrillation, in which we found no association with increasing concentrations of ds‐DNA with the development of atrial fibrillation. Although associated with NETosis, findings with regard to the prognostic value of CitH3 and MPO‐DNA have been conflicting. ¹³ , ¹⁴ In our study, neither CitH3 or MPO‐DNA showed prognostic value for the primary or the secondary end points after covariate adjustments. Although CitH3 and dsDNA levels are elevated at the culprit lesion in patients with ACS, their concentrations in peripheral blood are substantially lower. ²⁹ Consequently, although involved in coronary pathophysiological process, they may be less reliable when measured peripherally.

NETs as Targets for Risk Stratification and Therapeutic Interventions

An exaggerated NETs response may cause unnecessary tissue damage and NETs are currently being explored as a therapeutic target, based on evidence suggesting that commonly used cardiovascular medications can modulate NETosis including ticagrelor ³⁰ and statins that have been shown to exert parts of its pleiotropic effects through NET modulation. ¹⁷ Investigators have also found that metformin significantly reduces circulating concentrations of NETs in diabetic patients compared with placebo. ¹⁸ Since commonly prescribed medications can modulate NET formation, ¹⁷ , ¹⁸ , ³⁰ , ³¹ identifying patients with elevated NET concentrations and increased cardiovascular risk could help guide targeted therapeutic strategies. However, this must be assessed in randomized clinical trials. We have also recently showed that patients with STEMI randomized to anti‐inflammatory therapy with the IL‐6 receptor inhibitor tocilizumab had lower NETs concentrations and less myocardial injury. ³¹ Identifying high‐risk individuals based on NET concentrations could be particularly valuable for patients without ACS, who are not typically prescribed statins or other risk reduction strategies, unless other indications such as high cholesterol exist.

Factors Influencing Circulating dsDNA Concentrations

DsDNA concentrations correlated modestly with inflammatory markers such as CRP and GDF‐15, suggesting that systemic inflammation and cellular stress are determinants of the amount of circulating dsDNA. Weaker correlations were observed with cardiovascular risk factors such as smoking, hypertension, as well as with cardiac‐specific markers like NT‐proBNP and cTnT. Additionally, dsDNA correlated weakly with specific NET markers such as CitH3 and MPO‐DNA, neither of which showed prognostic value, supporting the notion that dsDNA only partially reflects NETosis. This is in keeping with previous studies. ¹⁴

Furthermore, current smoking and hypertension were associated with higher odds of having dsDNA concentrations above the median (Table 5) and can imply that the cardiovascular inflammation caused by smoking and hypertension is mediated by NETs. However, the R ² value of 13.1% from the multivariable regression model indicates that the identified predictors account for only a modest proportion of the variability in dsDNA concentrations, pointing to the influence of additional unidentified factors. Interestingly, there was no association between calcium score and dsDNA, indicating that the atherosclerotic burden as assessed by calcium score has no impact on circulating dsDNA concentrations. These findings contrast those of Borissoff et al., who, in a smaller cohort (n=282 patients) of patients with stable chest pain referred for computed tomography coronary angiography found that circulating NET markers, including DNA, were independently associated with coronary atherosclerosis and predicted the number of atherosclerotic coronary arteries. ³² This discrepancy underscores the need to develop and validate circulating markers that are more specific to NETosis in cardiovascular disease. However, one of the inherent challenges is that inflammatory markers often lack specificity.

Sex Differences in the NETs Response and Prognostic Implications

Several studies suggest sex‐specific differences in the neutrophil response and signaling. ²³ Experimental studies in rodents have demonstrated that the higher mortality observed in male mice after MI was linked to increased neutrophil infiltration at the ischemic border zone, elevated protease secretion leading to infarct expansion, and a greater incidence of cardiac rupture. ³³ , ³⁴ In our study we found that female sex was associated with 26% lower odds of having dsDNA concentrations above the median (P=0.012) in multivariable analysis and after adjustment for biomarker estimated infarction size (cTn), supporting the notion of a greater NETs response in men. These findings align with existing evidence that men display a more powerful immune response in certain clinical scenarios, and our study also suggest a stronger NETs response in men. Despite these differences, the lack of a significant interaction between sex and dsDNA concentrations (P=0.108) in prediction of the primary end point, suggests that the increased odds of having high dsDNA concentrations in men does not translate to worse prognosis, and dsDNA can be used as a prognostic marker regardless of sex.

Strengths and Limitations

This is to our knowledge the first study to investigate the prognostic value of NET markers in a cohort of unselected patients with chest pain, providing argument for the use of dsDNA as a prognostic marker, also in patients without ACS.

The study used a large, well‐characterized cohort, measuring circulating NET markers alongside established biomarkers and comparing them with standard inflammatory and myocardial injury markers. Clinical end points were obtained from a national registry that covers all Norwegian health care contacts, ensuring comprehensive event capture.

Given the observational nature of our findings, causality cannot be inferred. NETs were only measured in baseline samples, and serial sampling could have provided a more comprehensive understanding given the temporal dynamics of the NETs response during MI. ¹² , ³¹ , ³⁵ The higher analytical variation (coefficient of variation) observed for the CitH3 assay (11.9%) compared with dsDNA (4.8%) and MPO‐DNA (5.3%) may have introduced greater measurement variability, potentially reducing the ability to detect subtle differences, which may affected the results for CitH3.

Another challenge is the specificity of the NET markers. DsDNA may have limited specificity as it also reflects DNA released into the circulation from necrotic or apoptotic cells, potentially including cardiomyocytes. Both MPO‐DNA and CitH3 are considered more specific indicators of NET formation. CitH3 arises from the citrullination of histones, a hallmark of NETosis, whereas MPO‐DNA combines MPO, a neutrophil‐specific enzyme, with extracellular DNA, and is therefore more directly reflecting NETosis. ³⁶ However, these markers are not exclusively specific for NETs and may also reflect broader processes of cell death and inflammation. Furthermore, the NETs proteome may vary across different pathological conditions, potentially influencing both their biological effects and detectability. ³⁷ Although our study cannot define the exact source of circulating dsDNA, the observed correlations and multivariable logistic analysis suggest that it mainly reflects broader inflammatory and pathogenic processes rather than specifically reflecting NETosis and myocardial injury. The cohort predominantly included White individuals with a median age >60 years, which may limit generalizability to more diverse populations.

Conclusions

In patients presenting with acute chest pain, circulating dsDNA emerged as a prognostic marker of death from all causes, cardiovascular death, incident MI, and heart failure hospitalization during long‐term follow‐up. DsDNA retained its prognostic significance in patients both with and without ACS and can identify patients with increased cardiovascular risk that may otherwise be overlooked using conventional clinical markers. In contrast, CitH3 and MPO‐DNA showed no prognostic value in this study.

Sources of Funding

The study was financed by a grant from the Western Norway Regional Health Authority; grant number: 912265. Gard Mikael Sæle Myrmel has had a part‐time research grant from Trond Mohn Foundation and currently has a PhD grant from the Western Norway Regional Health Authority (grant number: F‐12589). Ragnhild Helseth was supported by postdoctoral fellowship grant from the Norwegian Health Association (#43600) and by a grant from Marie Stenbergs Legat.

Disclosures

Kristin Moberg Aakre has served on advisory board for Roche Diagnostics, Abbott Diagnostics, Siemens Healthineers, and SpinChip, consultant honoraria form CardiNor, lecturing honorarium from Siemens Healthineers, Roche Diagnostics, Mindray, Wondf and Snibe Diagnostics and research grants from Siemens Healthineers and Roche Diagnostics, she is associate editor of Clinical Biochemistry and chair of the IFCC Committee of Clinical Application of Cardiac Bio‐markers. Torbjørn Omland has received honoraria from Abbott Diagnostics, CardiNor, Novo Nordisk, Roche Diagnostics, and SpinChip Diagnostics and has received research support from Abbott Diagnostics, ChromaDex, Novartis, Roche Diagnostics, via Akershus University Hospital. The remaining authors have no disclosures to report.

Supporting information

Data S1.