Patients

This cohort study evaluated 130 consecutive patients with CAD who did not achieve the therapeutic goals of hypercholesterolemia therapy. This group was described in our previous work.¹³ In brief, from January 2015 to February 2016, we enrolled 130 patients with prior ST‑segment elevation MI or non–ST‑segment elevation MI, coronary artery bypass grafting surgery or stenosis of more than 50% in at least 2 epicardial coronary arteries on coronary angiography.⁸ All patients were initially treated with statins (simvastatin 20 mg/d, atorvastatin 10 mg/d or 20 mg/d, or rosuvastatin 10 mg/d). The eligible patients had low‑density lipoprotein cholesterol (LDL‑C) level above 1.8 mmol/l or below 50% reduction in LDL‑C, if its baseline level was between 1.8 and 3.5 mmol/l, as recommended in the 2016 European guidelines¹⁴. The exclusion criteria were: AMI or percutaneous coronary intervention within preceding 6 months, end‑stage renal disease, active cancer, signs of acute infection, chronic inflammatory disease, and treatment with other lipid‑lowering agents, for example, ezetimibe or fibrates.¹⁵

Intervention

All patients were switched to high‑dose statin therapy (atorvastatin 80 mg/d or rosuvastatin 40 mg/d at the discretion of the investigators) and were treated for a minimum of 6 months (up to 12 months). The study protocol was approved by the Jagiellonian University Medical College Ethical Committee (1072.6120.59.2018), and the participants provided their informed written consent in accordance with the Declaration of Helsinki.

Laboratory investigations

At enrollment and after 6‑month therapy with high‑dose statins, blood was drawn after the overnight fast from the antecubital vein into either citrated tubes and centrifuged at 2500 g at 20 °C for 20 minutes or into serum tubes and centrifuged at 1600 g at 4 °C for 10 minutes. Serum total cholesterol (TC), LDL‑C, high‑density lipoprotein cholesterol (HDL‑C), and triglycerides were assayed using a biochemical analyzer Cobas 6000 (Roche Diagnostics GmbH, Mannheim, Germany). Non–HDL‑C was calculated as TC minus HDL‑C. C‑reactive protein (CRP) was measured by immunoturbidimetry (Roche Diagnostics GmbH) and D‑dimer with the Innovance D‑dimer assay (Siemens Healthcare, Marburg, Germany). Fibrinogen was determined using the von Clauss method.

Neutrophil extracellular trap–related proteins

Immunoenzymatic assays were used to determine serum levels of H3cit, MPO, and NE according to the manufacturer’s protocols (Cayman Chemical, Ann Arbor, Michigan, United States). The reference ranges for H3cit and MPO in our laboratory were 0.5–1.7 ng/ml and 0.13–0.43 ng/ml, respectively.¹⁶

Fibrin clot analysis

Fibrin clot permeability was measured as described previously.¹⁷ Briefly, citrated plasma was mixed with 20 mmol/l calcium chloride and 1 U/ml human thrombin (Sigma‑Aldrich, St. Louis, Missouri, United States). The buffer volume flowing through the clots was measured within 60 minutes. The K_s, which indicates the average pore size in fibrin networks (lower values indicate more compact fibrin structure), was calculated. The reference range of K_s established in our laboratory is 7.7–9.7 × 10^-9 cm².

Fibrinolysis

Clot lysis time (CLT) was measured as previously described.¹⁷ Briefly, citrated plasma was mixed with calcium chloride (15 mM), human tissue factor (0.6 pM; Innovin, Siemens Healthcare), phospholipid vesicles (12 µM), and recombinant tissue type plasminogen activator (rtPA, 60 ng/ml; Boehringer Ingelheim, Ingelheim, Germany). CLT was defined as the time from the midpoint of the clear‑to‑maximum‑turbid transition, which represents a clot formation, to the midpoint of the maximum‑turbid‑to‑clear transition (representing the clot lysis). The reference range of CLT established in our laboratory for 50 apparently healthy subjects is 61–80 minutes.

Moreover, thrombin activatable fibrinolysis inhibitor (TAFI) activity (Hyphen‑Biomed, Neuville‑Sur‑Oise, France) and α2‑antiplasmin (Berichrom, Siemens Healthcare Diagnostics, Marburg, Germany) were measured.¹⁷

Thrombin generation potential

Plasma thrombogenic potential was assessed using calibrated automated thrombography (CAT; Thrombinoscope BV, Maastricht, the Netherlands), using dedicated reagents (Stago, Asnieres sur Seine, France) in a 96‑well plate fluorometer (Ascent Reader, Thermolab Systems OY, Helsinki, Finland).¹⁷ The area under the curve represented the endogenous thrombin potential (ETP).¹⁶

Statistical analysis

Based on the previous data,^9-11 this study was powered to have an 80% chance to detect a 20% reduction in H3cit level at a P value of 0.05. To show such a difference or greater, at least 64 patients were required.

Depending on the variable type, they were presented as numbers and percentages or median and interquartile range (IQR). Normality was assessed by the Shapiro–Wilk test. Differences between the groups were compared using the t test for normally distributed variables or the Mann–Whitney test for non‑normally distributed variables. Categorical variables were compared with the Pearson χ² test or the Fisher exact test. Associations between nonparametric variables were assessed by the Spearman rank correlation coefficient. All independent variables potentially associated with the change of NET‑related protein level or with achievement of final LDL‑C concentration following statin therapy intensification were included in the multivariable linear or logistic regression model, respectively, to identify their independent determinants. Comorbidities, pharmacological treatment, D‑dimer, ETP, K_s, CLT as well as platelet and neutrophil levels were selected as potentially important confounding factors. A 2‑sided P value below 0.05 was considered significant. All statistical analyses were performed using Statistica software, version 13.3 (StatSoft, Kraków, Poland) or IBM SPSS Statistics, version 26.0 (IBM Corp, Armonk, New York, United States).

Baseline

As compared with our reference ranges, 120 CAD patients (92.3%) had elevated H3cit with a median (IQR) concentration of 2.63 ng/ml (2.05–3.31). H3cit level positively correlated with MPO (R = 0.543; P <⁠0.001) but not with NE. The concentration of NETosis‑related proteins did not differ in relation to demographics, cardiovascular risk factors, the type and daily dose of statins, lipid profile, or neutrophil count. Moreover, the levels of the NETosis‑related proteins were similar in patients with LDL‑C above 3 mmol/l and those with its lower concentrations. H3cit was positively associated with CRP (R = 0.54; P <⁠0.001), but not with fibrinogen.

An analysis of fibrin clot variables showed that all 3 NETosis‑related proteins inversely and weakly correlated with K_s (H3cit, R = –0.251; P = 0.004; MPO, R = –0.492; P <⁠0.001; NE, R = –0.214; P = 0.014), but not with CLT, thrombin generation, or fibrinolysis inhibitors.

High‑dose statin treatment

After high‑dose statin treatment, we observed a 14% reduction in TC, a 25% decrease in LDL‑C, along with a decrease in CRP (–42.9%; P <⁠0.001), D‑dimer (–22.6%; P <⁠0.001), and fibrinogen levels (–6.7%; P = 0.014) (Table 1). High‑dose statin therapy resulted in by 12.9% higher K_s (P <⁠0.001) and improved susceptibility to fibrinolysis, reflected by 11.1% shorter CLT (P <⁠0.001) and 8% lower TAFI activity (P <⁠0.001).

Intensification of statin treatment markedly reduced the level of H3cit (30.4%), MPO (28.1%), and NE (25.5%), as compared with baseline (Figure 1), while neutrophil count remained unchanged (Table 1). Of note, in 53 patients (40.8%) the on‑treatment H3cit concentrations were within the reference range. H3cit and MPO levels correlated with each other (R = 0.526; P <⁠0.001), similarly as at baseline, and showed associations with residual plasma CRP (R = 0.68; P <⁠0.001 and R = 0.253; P = 0.004, respectively). The ΔH3cit and ΔMPO were associated with each other (R = 0.413; P <⁠0.001), but neither of them was associated with ΔNE.

ΔLDL‑C, ΔTC, and Δnon–HDL‑C did not correlate with the magnitude of changes in NETosis‑related proteins. However, H3cit level was lower by 16.5% (P = 0.004) in the patients who achieved the target LDL‑C below 1.8 mmol/l (n = 50; 38.5%), and the reduction was even greater (by 25.5%; P = 0.001) in the patients with LDL‑C below 1.4 mmol/l (n = 19; 14.6%), as compared with the remaining participants (Figure 2A). No similar observations were made for MPO (Figure 2B), while NE was slightly, though significantly, higher in the patients with LDL‑C below 1.8 mmol/l (Figure 2C). Baseline levels of H3cit, MPO, and NE did not identify the individuals who achieved LDL‑C below 1.4 mmol/l or below 1.8 mmol/l after therapy intensification. The best discriminative value in the prediction of LDL‑C below 1.4 mmol/l was provided by H3cit concentrations measured at 6‑month high‑dose statin therapy, with sensitivity of 55%, specificity of 80%, and Youden index of 0.444 (Figure 3).

ΔH3cit was strongly associated with ΔCRP and weakly with ΔMPO (Figure 4A and 4B). Interestingly, ΔH3cit and ΔMPO, in contrast to ΔNE, positively correlated with ΔTAFI (Figure 4C and 4D), but not with ΔCLT, and inversely correlated with an increase in K_s (Figure 4E and 4F).

Multivariable analyses

The best multivariable model among all that incorporated NET‑related proteins was created for H3cit (R² = 0.75; P <⁠0.001; Table 2). We found that a decrease in H3cit level on high‑dose statins was independently associated with ΔCRP (β = 0.771; P <⁠0.001), ΔTAFI (β = 0.125; P = 0.013), and Δfibrinogen (β = 0.106; P = 0.034) but not with ΔLDL‑C (Table 2).

Moreover, in the multivariable logistic regression analysis, male sex, diabetes mellitus, smoking, lower baseline CRP level, and lower H3cit level at 6 months of intensive statin therapy were independently associated with the final LDL‑C below 1.4 mmol/l (R² = 0.463; Table 3).

Limitations

Our study has several limitations. First, even though it was adequately powered, the number of enrolled CAD patients was limited. Moreover, CAD progression evaluated using invasive and noninvasive diagnostic methods was not assessed during follow‑up; however, clinically evident events were not reported.⁴⁰ Second, we did not include a control group treated with other statin doses or presenting a different lipid profile, but it is highly unlikely that factors other than implementation of high‑dose statins were responsible for the changes reported at 6 months. Third, it should be recognized that H3cit is not entirely specific to NETosis, since other stimuli or cell types can also lead to histone citrullination.⁴ MPO‑DNA would be a more specific NETosis marker, however, there are no standardized diagnostic tools to detect such complexes, and the results of the in‑house methods are poorly comparable.¹⁶ The associations between NET‑related proteins and prothrombotic markers do not necessarily mean the cause‑effect relationship. In order to understand the mechanism underlying the statin‑induced NET‑related protein reduction and its relationship with coagulation along with fibrinolysis, in vitro studies and animal models should be conducted,¹⁹ but our hypothesis‑generating study indicates that such research is worth efforts. Finally, we analyzed only high‑risk CAD patients, and it is unknown whether our observations could be extrapolated to the entire CAD population with less advanced disease, and those treated with other lipid‑lowering agents with anti‑inflammatory properties.⁴¹

Conclusions

We demonstrated that high‑dose statin therapy in CAD patients results in decreased levels of circulating NETosis‑related proteins linked with a reduction of CRP and favorable modification of fibrin clot properties. Our study provides evidence for a novel effect of statins, which might contribute to clinical efficacy of cholesterol‑lowering therapy in CAD patients. It remains to be established in large cohort studies whether NETosis reduction can contribute to beneficial effects of high‑dose statins in terms of clinical outcomes in atherosclerotic vascular disease.