Neutrophil Dynamics in Acute Coronary Syndrome

Aims: Clinical evidence indicates that innate immune cells may contribute to the onset and outcome of acute coronary syndrome (ACS). Our prospective study aimed at analysing neutrophil phenotypes in ACS and their role in predicting 1-year major cardiovascular events. 
Methods: Blood neutrophil phenotypes were analysed by flow cytometry. Differential blood cell count and plasma levels of soluble markers were recorded at admission and at 6-month follow-up. 
Results: 108 patients categorized in chronic stable coronary artery disease (n=37), unstable angina (UA) (n=19), Non-ST-Elevation Myocardial Infarction (NSTEMI) (n=25), and ST-Elevation Myocardial Infarction (STEMI) (n=27) were included. STEMI and NSTEMI patients displayed higher neutrophil count and neutrophil-to-lymphocyte ratio (NLR) than stable and UA patients (P<0.0001), which normalized at 6-month after MI. STEMI patients were characterized by elevated percentages of band cells in low-density neutrophils (P=0.007) and in high-density neutrophils (P=0.019) compared to the other patients. Multivariable logistic regression analysis revealed that plasma levels of total MPO was associated with STEMI when compared to stable (OR: 1.434; 95% CI: 1.119-1.837; P<0.0001), UA (1.47; 1.146-1.886; P=0.002), and NSTEMI (1.213; 1.1-1.134; P=0.0001) patients, while increased neutrophil SSC signal intensity was associated with NSTEMI compared to stable patients (3.828; 1.033-14.184; P=0.045). Based on multivariable Cox regression analysis, elevated plasma levels of PCSK9 and low-density neutrophil percentage predicted 1-year outcome independently of cardiovascular risk factors (c-index: 0.915; IQR: 0.908-0.929). 
Conclusions: Changes in neutrophil phenotype are concomitant to ACS. These changes may differ between STEMI and NSTEMI. They may also contribute to ACS risk and patient outcome.


INTRODUCTION
Increasing clinical evidence supports the existence of a link between low-grade systemic inflammation and cardiovascular (CV) risk 1 . Plasma level of high-sensitive C-reactive protein (hs-CRP) at the time of acute coronary syndrome (ACS) and its rise after the incident disease have been associated with a higher risk of major CV event 2 . CRP is produced by the liver in response to elevation of pro-inflammatory cytokines such as IL-6 and IL-1, two well-recognized players in atherosclerosis progression. In the CANTOS trial 3 , dampening inflammation by targeting the IL-1 pathway at the time of ACS lowered the risk of recurrent major CV events in patients who experienced a previous ACS. Hence, inflammation is currently considered as a modifiable CV risk factor. Among other inflammation-related risk factors, several studies pointed to the differential white blood cell count, with a particular focus on neutrophils, lymphocytes and monocytes. Neutrophil count by itself could predict both acute and chronic CVD 4,5 . Interestingly, an elevated neutrophil-to-lymphocyte ratio (NLR) has not only been associated with CVD but also with short-term adverse CVD outcomes, including mortality, coronary artery disease (CAD), stroke and heart failure 6 .
Changes in NLR are subsequent to a modification of the balance between innate (neutrophils) and adaptive (lymhocytes) immunity. Also, the association between elevated NLR and poor prognosis in cancer has been linked to cancer-elicited release of immune suppressive neutrophils from the bone marrow [7][8][9] . In view of recent reports on the existence of common risk factors in CVD and cancer 10 , it is therefore possible that CVD might similarly be associated with a mobilisation of neutrophils with altered phenotype, which might contribute to ACS risk. In chronic inflammatory diseases, as in cancer, peripheral blood neutrophils display diversity in their phenotype 8 . To date, neutrophil phenotypes in coronary artery disease patients and their association with ACS risk and outcome remain unexplored. Our study aimed at providing a first description of neutrophil dynamics in prospectively recruited coronary artery disease patients. We investigated the association of neutrophil markers with ACS, their evolution after 6 months, as well as their capability to predict patient outcome during 1-year follow-up.

Blood samples, sera and platelet poor plasma (PPP) preparation, laboratory measurements
Blood samples from patients were collected from radial or femoral site of arterial catheter and processed within 1 hour. Hematological parameters were determined in EDTA-anticoagulated blood on a Cell-Dyn 3700 (Abbott Laboratories). Platelet poor plasma was obtained after two consecutive centrifugations at 1750g during 15 min at room temperature. Sera were obtained following one centrifugation at 1750g during 15 min at RT. Plasma and sera were stored at -80°C until analysis. IL-6 and S100A9 were measured on EDTA-plasma using a Luminex assay kit (R&D Systems) according to the manufacturer's instructions. Nucleosome levels were quantified on EDTA-plasma with the Cell Death Detection ELISA plus kit (Sigma-Aldrich). Myeloperoxydase quantity and activity were analysed on EDTA-plasma by a method adapted from Franck et al. 11 using a human MPO Quantikine ELISA kit (R&D Systems). High sensitivity (hs)-CRP, CK-MB (Abbott Laboratories) and hs-TnT (Roche) were measured in sera by routine immunoassays.

Statistics
Qualitative variables are summarized as numbers and percentages. Normality of the distribution of quantitative variables was tested by Shapiro-Wilk test. Quantitative variables with normal distribution were presented as mean ± SD, while median (interquartile range [p25-p75]) were reported for quantitative variables with skewed distribution. Qualitative characteristics of patient groups were compared by Chi-square or Fisher exact tests. ANOVA or Kruskal-Wallis tests were applied for quantitative variables as appropriate, with Tukey or DSCF post-hoc test, respectively. Spearman correlation analyses were performed to assess the association between LDN percentage and other quantitative variables. Multinomial logistic regression (MLR) was applied to model association between CAD categories and demographic, clinical and biological variables. Significant variables with P <0.1 after univariate MLR analyses were considered into multivariable MLR analysis 13 . Results of the final MLR model were presented using odds ratios and corresponding 95% confidence interval. Cox proportional hazard models were used to investigate the occurrence of composite endpoint at 1-year follow-up (cardiovascular death, stroke, myocardial infarction or major bleeding) and to evaluate the prognostic value of changes in neutrophil markers from baseline to 6-month follow-up. Multivariable Cox model for composite endpoint at 1-year follow-up used the same variable selection method as for MLR. All hazard ratios (HR) were calculated with appropriate unit and corresponding 95% confidence interval. Concordance index (c-index) of the final model was calculated following Uno's method and is presented as c-index with interquartile range (IQR). All tests were performed 2-sided and P < 0.05 was considered significant except specified. Statistical analyses were performed using SAS 9.4 (SAS Institute, Cary NC).
Patients from the four categories did not differ in terms of age, sex and conventional CVD risk factors. However, differences were observed in regard with aspirin (P=0.0005) and lipidlowering therapy (P=0.001). Triglyceride levels were more elevated in STEMI patients than in UA (P=0.009) and NSTEMI (P=0.006). Regarding systemic inflammatory markers, the four patient groups displayed different levels of IL-6 (P=0.003) ( Table 1). As expected, differences of differential white blood cell counts were observed, mostly related to changes in neutrophil count. NSTEMI and STEMI patients had higher neutrophil counts than stable and UA patients (P<0.0001; NSTEMI vs. UA P=0.012). Among ACS, NSTEMI patients showed higher monocyte count than stable patients (P<0.0001). In contrast, lymphocyte counts did not differ between patient categories. Yet, changes in NLR likely reflected the increase of neutrophil count. Plasma levels of S100A9, total and active MPO, and of nucleosomes, wellknown circulating markers of neutrophil activation and NET (neutrophil extracellular trap) release that have previously been associated with CVD risk and ACS 14-17 , were higher in STEMI patients than in stable (S100A9: P=0.013; total MPO: P<0.0001; active MPO: P<0.0001; nucleosomes: P=0.032) and UA (S100A9: P=0.018; total MPO: P<0.0001; active MPO: P<0.0001; nucleosomes: P=0.007) patients (Fig. 1A). Despite similar increase of neutrophil count in NSTEMI and STEMI patients, the levels of these neutrophil markers were not significantly more elevated in NSTEMI patients than in stable and UA, suggesting that neutrophil phenotype, in addition to absolute cell numbers, differed between STEMI and other ACS conditions. Accordingly, total and active MPO levels were more elevated in STEMI than in NSTEMI patients (P<0.0001) (Fig. 1A).

Neutrophil phenotypes in ACS
A detailed phenotypic analysis of neutrophils was performed for the four patient groups. We  Table 1).
HDNs from NSTEMI patients displayed higher complexity than those from stable (P=0.003) and UA (P=0.031) patients, as determined by SSC (side scatter) signal intensity (Table 1) ( Fig. 1B). There were no significant differences in CD11b, CD16 or CD10 expression levels on the surface of HDNs or LDNs from the four patient groups. Altogether, these data indicate that the increase of neutrophil count in ACS likely reflects elevated numbers of circulating neutrophils of high density. Distinct phenotypic neutrophil features were identified in STEMI and NSTEMI ACS. Elevated numbers of band cells were detected in the blood from STEMI patients, while NSTEMI patients were characterized by the presence of high-complexity HDNs.

Neutrophil markers in ACS
The association of these neutrophil markers with the four patient groups was assessed. After adjustment with confounding variables, i.e., age, gender, daily low-dose aspirin, and lipidlowering therapy, plasma levels of total MPO (P<0.0001), HDN SSC (P=0.026) and hs-cTnT (P=0.015) were independently associated with the four patient groups (

Prognostic value of neutrophil markers
During 1-year follow-up, 12 patients (4 stable, 2 NSTEMI, 6 STEMI) out of 101 experienced an event. Among stable patients, we recorded 1 stroke, 2 cardiovascular deaths (cardiac arrest, recurrent MI) and 1 bleeding event (BARC 2). In the STEMI group, cardiovascular death (BARC 3b bleeding) occurred in 1 patient, 2 patients had recurrent MI, and 3 patients experienced major bleeding (2 BARC 2 and 1 BARC 3a). A total of 2 NSTEMI patients had a recurrent MI, which led to death for one of them. We then sought to determine whether neutrophil markers measured at admission could help predicting these major events. Based on univariable Cox regression analysis, elevated neutrophil count, plasma levels of S100A9, PCSK9 and LDN percentage in PBMC predicted 1-year outcome after adjustment for CV risk factors (Table 3). Multivariable analysis indicated that elevated levels of PCSK9 and LDN percentage were independent predictors of outcome (c-index: 0.918; IQR: 0.908, 0.929) ( Table 3). Of note, we did not detect any differences of PCSK9 levels between stable, UA, NSTEMI and STEMI patients.

Evolution of neutrophil markers during follow-up
To assess time-dependent changes in neutrophil markers and to determine whether changes concomitant to myocardial injury were maintained at distance of ACS, some of these markers were measured after 6 months post-admission. Data could be obtained for 51 patients from the four categories (18 stable, 11 UA, 11 STEMI, 11 NSTEMI). Baseline characteristics of patients from this subgroup are shown in supplementary Table S2. Although neutrophil count (P<0.0001), NLR (P=0.0004) and MPO levels (P=0.0009) and activity (P=0.0002) differed between ACS and stable or UA patients at admission, these markers no longer differed between these patient groups at follow-up (Table S2). Likewise, stable patients who had a previous myocardial infarction (11 patients out of 37) did not differ in any neutrophil markers, i.e., NLR, MPO levels, LDN and band cell percentages, HDN SSC (Table S3).
These data indicate that mobilization of neutrophils harbouring a specific phenotype likely occurs at the time of myocardial infarction, which globally resolves over time. We then

DISCUSSION
Despite advances in understanding the atherosclerotic processes, progression of CAD to acute events remains poorly understood and hardly possible to predict. Several studies indicate that neutrophils are involved in plaque progression [18][19][20][21] , and preclinical data in mice suggest that low-grade systemic inflammation could promote atherosclerosis progression through neutrophil reprogramming 22 . Our study provides first clinical evidence that phenotypic modifications of circulating neutrophils occur around the time of ACS. Moreover, we found that these neutrophil phenotypes normalize with time. In addition, differences were found between STEMI and NSTEMI patients. Importantly, we also identified neutrophil markers and neutrophil subpopulations (LDN) that may contribute to CAD outcome independently of known CV risk factors (Fig. S3).

Neutrophil phenotypes in ACS
A series of retrospective and prospective studies concur with a strong capability of NLR to predict CVD patient outcome. Since NLR also predicts outcome in cancer patients, it has been hypothesized that the NLR, as in cancer, could reflect the release of immune suppressor cells (e.g., granulocytic myeloid suppressor cells, G-MDSC) from the bone marrow. So, the NLR would not only be a marker of inflammation but could also reflect a biological process underlying ACS onset. In cancer and in various chronic diseases, G-MDSC can be isolated from the LDN neutrophil subset 23,24 . We identified variable percentages of LDN in the PBMC fraction of CAD patients' blood, irrespective of disease stage. Also, we showed that neutrophils with altered phenotype circulate in the blood of STEMI and NSTEMI patients.
We identified immature band cells in HDN and LDN from STEMI patient blood, while HDN with high complexity were observed in the blood from NSTEMI patients. Increased neutrophil complexity depicts higher granularity and changes in internal structures, which generally features early innate immune responses. Circulating neutrophils from NSTEMI patients may therefore have distinct functionality and granular content. Band cells are neutrophils at intermediary stage of maturation, characterized by a curved nucleus. The appearance of these immature cells in the blood results from signals to the bone marrow for increased neutrophil production, as in the case of infection or inflammation. In STEMI, this probably indicates a severe inflammatory response with a high demand for neutrophils, but not only. Importantly, in our whole cohort, increased LDN percentage contributed to predict major CV event at 1-year follow-up after comprehensive multivariable adjustment. Our observation that the LDN percentage moderately correlates with NLR could suggest an important biological role of these cells in CVD outcome that might be independent of inflammation and conventional CV risk factors. In chronic inflammatory diseases such as lupus erythematous or psoriasis, low-density granulocytes that are more prone to produce NETs are thought to contribute to higher CVD risk 25,26 . For instance, the frequency of these cells has been associated with non-calcified coronary plaque burden in patients with psoriasis 27 . Whether LDNs from CAD patients possess immune suppressive or pro-inflammatory function or whether they are more susceptible to release NETs remains to be determined.
Nevertheless, the fact that we did not observe any correlation between LDN percentage and the levels of circulating nucleosomes or inflammatory markers while it was moderately correlated with NLR might be in favour of an immune suppressive function.

Soluble neutrophil markers in ACS
We found elevated levels of soluble neutrophil activation markers, including MPO, S100A9 and nucleosomes in STEMI but not in NSTEMI patients and this, despite similar rise in neutrophil count and NLR. Total MPO levels appeared as a strong independent marker of STEMI. These distinct features between NSTEMI and STEMI might reflect different pathophysiological mechanisms leading to these ACS entities, which could not be picked up based on neutrophil count or NLR only. Noteworthy, it has recently been recognized that beside differences in ECG and troponin levels, plaque composition as well as mechanisms of atherogenesis influence CAD patient outcome [28][29][30] . MPO, stored in neutrophil azurophilic granules, is released upon neutrophil activation. Increased circulating levels of this protein have already been shown to predict the outcome of patients with ACS 15,16 . MPO would also be a marker of plaque vulnerability 16 . Interestingly, it has been reported that higher circulating MPO levels were found in ACS patients with culprit plaque erosion than in those with culprit plaque rupture 31 . In addition, luminal thrombi overlying eroded plaques contained higher density of MPO-positive cells, likely neutrophils, than thrombi formed on ruptured plaques. Although there is no clear relationship with culprit lesion morphology and STEMI or NSTEMI entities, our study indicates that MPO levels may reflect different acute complications in CAD, thereby revealing the existence of distinct mechanisms of ACS. It is noteworthy that in agreement with the study by Ferrante et al. 31 , we did not find any correlation between MPO levels and markers of systemic inflammation, CRP or IL-6, while correlations were observed with other neutrophil-derived markers, like S100A9 and nucleosomes. Hence, neutrophil markers are likely to carry additional information compared to classical inflammatory markers since they may underlie biological mechanisms leading to ACS.

Neutrophil dynamics after ACS
In the study by Kim et al. 6 , the shorter median time from ACS improved the strength of the association of NLR with a subsequent adverse event. Therefore, it has been suggested that the NLR would increase over time in CVD patients and reach a maximum around the time of event. In our study, in a subset of patients from each category, we could evaluate differential blood cell count, soluble neutrophil markers and inflammation at 6-months post-MI. We showed that neutrophil count, NLR and soluble neutrophil markers returned to levels of stable patients during the follow-up period, therefore confirming that these markers reach a maximum around the time of ACS. Notably, we found that the levels of changes between the baseline measurements of neutrophil count and NLR and the 6-month follow-up measurements might predict the occurrence of CV events.

Clinical implications
Our study highlights potential roles for neutrophil phenotypes as novel diagnostic and prognostic biomarkers in CAD. Advances in the cellular biology of atherosclerosis and its pathogenesis have led to major advances in its treatment. We provide first evidence that neutrophils could be a key player in the disease progression to ACS through changes in their phenotype and secretory profiles. Targeting pro-inflammatory or immune regulatory neutrophils (specific neutrophil subsets) and/or their downstream mediators that promote adverse effects (e.g., NET, pro-inflammatory cytokines, reactive nitrogen/oxygen species, neutrophil serine proteases, danger-associated molecular pattern molecules) may thus represent new therapeutic opportunities. Deoxyribonuclease (DNAse), an enzyme that breaks down NET, has been shown in a clinical study of STEMI patients to accelerate lysis of coronary thrombi 32 . Colchicine, an anti-inflammatory agent, has been shown to acutely suppress caspase-1 activity and inhibit inflammasome related cytokines in monocytes from ACS patients 33 . The use of colchicine in patients with recent myocardial infarction was also effective in reducing CV events 34 . In patients with a history of myocardial infarction and elevated hs-CRP, canakinumab, targeting IL-1, represented another drug effective at preventing adverse cardiac events 3 . Serine proteases liberated from activated neutrophils, e.g., cathepsin G, elastase, proteinase-3, are potent regulators of at least six IL-1 cytokine family members, including IL-1 35 . One might consider inhibiting these extracellular proteases since this strategy will not affect neutrophil antimicrobial capability 35 . Neutrophils may indeed represent a major source of IL-1 in ACS. In cancer, IL-1 and S100A9 have been involved in the mobilization of pro-tumoral immature neutrophils with immune suppressive functions 36,37 . These molecules might also contribute to the production of immune suppressive neutrophils at the time of ACS. Hence, our study may pave the way toward new therapeutic avenues in ACS, which could be based on available immunotherapy approaches.

Limitations
The number of patients examined may be perceived as low. Nevertheless, our study of neutrophil profiles is unique, as it required good logistics coordination, involving both the laboratory and the clinics. Neutrophils need to be processed in a highly standardized manner within 1-hour after blood collection. Blood samples were only obtained during working days.
The neutrophil profile during follow-up was obtained in only half of the patients. Given the distinct function of each neutrophil subpopulation, it is likely that each subset participates in the inflammatory pathophysiology of CAD to varying degrees. The relationship between the neutrophil phenotypes and the types of cytokines, reactive oxygen/nitrogen species, serine proteases released has not been studied. In addition, our study does not bring out a causal link between the neutrophilic profile and the destabilization of the coronary disease. It must be considered hypothesis-generating, even if the secretory and morphological dynamics of neutrophils are suggestive. Future larger studies evaluating the prognostic role of neutrophils in ACS are needed.

Conclusion
Our study indicates that in patients with CAD changes in neutrophil phenotypes occur concomitantly to ACS. These changes may reflect different acute complications of CAD and reveal new biological processes leading to ACS. Importantly, the presence of LDN in the blood of CAD patients may represent a novel CVD risk factor.

Disclosures
The authors have nothing to disclose.