D-dimer for risk stratification and antithrombotic treatment management in acute coronary syndrome patients: a systematic review and metanalysis

Background Data on the prognostic role of D-dimer in patients with acute coronary syndrome (ACS) are controversial. Our aim was to summarize current evidence on the association between D-dimer levels and short/long-term poor prognosis of ACS patients. We also investigated the association between D-dimer and no-reflow phenomenon. Methods Systematic review and metanalysis of observational studies including ACS patients and reporting data on D-dimer levels. PubMed and SCOPUS databases were searched. Data were combined with hazard ratio (HR) and metanalysed. The principal endpoint was a composite of cardiovascular events (CVEs) including myocardial infarction, all-cause and cardiovascular mortality. Results Overall, 32 studies included in the systematic review with 28,869 patients. Of them, 6 studies investigated in-hospital and 26 studies long-term outcomes. Overall, 23 studies showed positive association of high D-dimer levels with CVEs. D-dimer levels predicted poor prognosis in all studies reporting in-hospital outcomes. Five studies satisfied inclusion criteria and were included in the metanalysis, with a total of 8616 patients. Median follow-up was 13.2 months with 626 CVEs. The pooled HR for D-dimer levels and CVEs was 1.264 (95% CI 1.134–1.409). Five out of 7 studies (4195 STEMI patients) investigating the association between D-dimer levels and no-reflow showed a positive correlation of D-dimer levels with no-reflow. Conclusions In patients with ACS, D-dimer was associated with higher in-hospital and short/long-term complications. D-dimer was also higher in patients with no-reflow phenomenon. The use of D-dimer may help to identify patients with residual thrombotic risk after ACS. Trial registration The review protocol was registered in PROSPERO International Prospective Register of Systematic Reviews: CRD42021267233. Supplementary Information The online version contains supplementary material available at 10.1186/s12959-021-00354-y.


Introduction
Coronary artery disease (CAD) is the leading cause of death worldwide with acute myocardial infarction (MI) representing the clinical condition associated with the greatest morbidity and mortality, up to 30% of inhospital deaths [1,2]. Prompt reperfusion with percutaneous coronary intervention (PCI), advances in acute cardiovascular care and more effective medical therapy have improved the prognosis of patients with MI in the last years [3].
This increased survival revealed a previously unrecognized long-term risk of recurrent thrombotic events despite optimal medical therapy [2]. Indeed, almost one fifth of MI patients suffer from rehospitalization within 1 year and 10% from recurrent MI [3,4].
There is therefore a growing need of a more accurate risk stratification of MI patients to identify those at higher risk for adverse events and worse prognosis [5].
Current guidelines of the American Heart Association and the European Society of Cardiology highlighted the lack of useful biomarkers to predict in-hospital complications and short/long term prognosis of patients with MI, recommending for this purpose only to measure troponin and brain natriuretic peptide (BNP) serum levels, with the latter associated with a low level of evidence [6,7].
During last decades, several studies, mostly conducted in the 90's before the routine use of dual antiplatelet therapy, described the pivotal and prognostic role of coronary thrombosis and hypercoagulable state in the pathophysiology of MI, correlating increased coagulation activity with recurrent event and poor outcome after MI [8][9][10].
As a marker of rapid fibrin turnover and high thrombotic activation in both arterial and venous system, interest in D-dimer has grown over time, and its predictive role has been investigated in several acute and chronic cardiovascular care, [11]. In patients with MI, high circulating D-dimer levels have been correlated with recurrent MI and poor prognosis [12,13] and early reduction in plasma concentration of D-dimer by Ximelagatran administration was associated with decreased risk of new cardiovascular events [14].
In addition, there are some data indicating that Ddimer may be higher in MI patients experiencing noreflow phenomenon [15], a severe condition in which high thrombotic burden interfere with complete restoration of myocardial blood supply after PCI, known to be correlated with adverse outcomes and poor prognosis [16].
Despite many studies published, conclusive data on the real usefulness and additional value of D-dimer in the management of patients with MI are lacking.
The aim of this systematic review and meta-analysis is to summarize current evidence on the potential prognostic role of D-dimer levels in CAD patients and its association with no-reflow phenomenon.

Strategy search
We conducted a systematic review of literature searching MEDLINE via PubMed and Scopus databases using a combination of the following keywords: "D-dimer" and "myocardial infarction" and "coronary artery disease" and "acute coronary syndrome". There was no time restriction and the last search was run on May 1, 2021. We included only journal articles in English language with full text available. We excluded case-control studies, case reports, editorials/comments, letters, review and meta-analysis, and experimental studies. Supplementary  Fig. 1 report strategy searches which were performed according to the PRISMA guidelines. The systematic review was registered at PROSPERO (https://www.crd. york.ac.uk/PROSPERO) with registration number CRD42021267233.

Study selection and quality assessment
Two physicians (FGB, DP) independently screened the titles and abstracts of manuscripts identified through the database searches to identify studies potentially eligible for further assessment. A third physician (GT) reviewed eligible studies for appropriateness and completeness. The study selection was performed in multiple phases. In the first phase, potentially relevant studies were obtained by combined searches of electronic databases using the selected above-mentioned keywords and exclusion criteria. In the second phase, potentially eligible studies were reviewed to assess the appropriateness with the study question. Finally, the quality of pertinent studies included in the metanalysis was assessed by the Newcastle-Ottawa scale. Studies with a score ≥ 7 were considered of good quality. Quality assessment is reported in Supplementary Table 1.

Endpoints
The principal endpoint was a composite of cardiovascular events (CVEs) including MI, all-cause and cardiovascular mortality. The type of endpoint reported in each study is given in Table 1. Given the lack of high-quality studies, the association between D-dimer and no-reflow was not formally analysed but only systematically reviewed. Ethical approval was not required given the study type.

Statistical analysis
When available, hazard ratios (HRs) and odds ratios (ORs) were recorded from each study.
A separate meta-analysis was conducted for each endpoint, using a hierarchical Bayesian model which therefore would take into account heterogeneity among studies. This technique was chosen due to the limited number of studies included. Formally, each studyspecific effect was assumed to be distributed according to a log-Gaussian random variable, centered on a pooled effect. The variance on the log-scale was assumed to correspond to 1.25 times the squared standard error, as evaluated through confidence intervals reported in each study.
All analyses were conducted with R version 3.5.1, using the 'metafor and 'adaptMCMC' packages.

Characteristics of studies
The association between D-dimer levels and clinical outcomes has been investigated in populations enrolled using different definitions of ischemic heart disease, such as CAD, ST segment elevation myocardial (STEMI), non-ST segment elevation myocardial infarction (NSTE MI), acute MI and acute coronary syndrome (ACS). Characteristics of these studies have been separately described according to each definition (Table 1). Supplementary Table 2 and Supplementary Table 3 display the method used to dose D-dimer, cut-off levels and antithrombotic therapy across included studies.

Cardiovascular events
PRISMA flow diagram showing study search strategy about cardiovascular events is reported in Supplementary Fig. 1. After screening, 42 potentially eligible studies were identified and were considered for detailed analysis ( Supplementary Fig. 1); 32 studies were included in the systematic review with 28,869 patients. Of these, 9 were retrospective, 20 prospective, 1 ambispective, 1 cross sectional study and 1 post-hoc analysis of a randomized clinical trial.

ST segment elevation myocardial (STEMI)
Overall, a sample of 7729 patients was analysed. The mean age ranged from 41 to 70.2 years and the proportion of women ranged from 8.45 to 54.7%. Four studies investigated outcomes occurred during hospital stay, while 7 studies investigated long-term outcomes (Table 1).
Overall, 11 studies showed an association between high plasma D-dimer levels and CVEs in STEMI patients. In particular, D-dimer predicted worse prognosis in all studies considering in-hospital outcomes, and in 5 out of 7 studies with long-term follow-up (Table 1).

Non-ST segment elevation myocardial infarction (NSTEMI)
A total of 1949 NSTEMI patients, with a mean age that ranged from 57.2 to 67.4 years was included. The proportion of female patients ranged from 24.8 to 35.8%. All studies investigated long-term prognosis. Overall, 1 out of 3 studies showed a positive association between high plasma D-dimer levels and CVEs incidence. However, one study showing no association between Ddimer and outcomes showed an association between Ddimer and NSTEMI severity, as assessed by Global Registry of Acute Coronary Events (GRACE) and thrombolysis in myocardial infarction (TIMI) risk scores, and included only mortality as endpoint; the other study had a short follow-up (6 months) and included also heart failure among cardiovascular outcomes (Table 1).

Myocardial infarction
Seven studies included a total of 12,320 patients with MI, with a mean age ranging from 62.8 to 74.2 years. The proportion of women ranged from 20 to 33.6% and 2 studies investigated in-hospital outcomes. Overall, 4 out of 7 studies consistently showed that D-dimer associated with CVEs (Table 1). Two studies also investigated the existence of a sex-based difference in the prognostic role of D-dimer providing conflicting evidence. Indeed, the study by Wang et al. found that Ddimer predicted all-cause mortality in women but not men. In this study, women were significantly older than men (74 vs 61 years). The opposite was found in the THROMBO study, in which however the mean age was lower and the endpoint included a composited of nonfatal events and cardiovascular death (Table 1).

Acute coronary syndrome (ACS) or CAD
Nine studies included 6871 ACS/CAD patients with a mean age ranging from 57.2 to 71.1 years. The proportion of female patients ranged from 11.1 to 54.1%. Overall, 8 out of 9 studies found that D-dimer was a predictive risk factor for CVEs in patients with ACS or CAD. In one study [45], the association between Ddimer and mortality was not confirmed at multivariable analysis; however, this study included a very

Results of meta-analysis
Five studies satisfied inclusion criteria and were included in the metanalysis: 4 prospective and 1 retrospective studies (  (Fig. 1).

No-reflow phenomenon in STEMI patients
We also performed a systematic search using "D-dimer" and "no-reflow". After screening, 7 potentially eligible studies were identified (3 prospective, 3 retrospective and 1 ambispective) ( Supplementary Fig. 2). A total of 4195 STEMI patients were included, with mean age ranged from 52.6 to 64.0 mean years (female patients from 16 to 53.7%). Overall, 1327 patients (31.6%) had no-reflow phenomenon ( Table 2). Five out of 7 studies showed a positive correlation of D-dimer levels with noreflow phenomenon. Given the high variability in the methods of each study (cut-off of D-dimer used, measures of association) a formal analysis was not performed. In group 1, D-dimer levels at the end of the procedure were higher in patients with restenosis than in those without (p < 0.005). Increased D-dimer in patients with restenosis (61%) than those without (25%, p < 0.05). Study design: A ambispective, P prospective, R retrospective §including recurrent MI, required renal replacement therapy, stent thrombosis, bleeding and length of hospital stay, hospital costs, and mortality and long-term CVEs (mortality, stent restenosis, non-fatal MI and TVR) ACS acute coronary syndrome, AMI acute myocardial infarction, AUC area under curve, CABG coronary artery bypass grafting, CAD coronary artery disease, CHF congestive heart failure, CI-AKI Contrast-induced acute kidney injury, CV cardiovascular, CVEs cardiovascular events, ESRD end-stage renal disease, HF heart failure, HFpEF heart failure with preserved ejection fraction, HFrEF heart failure with reduced ejection fraction, HR hazard ratio, NSTEMI non-ST segment elevation myocardial infarction, NT-proBNP N-terminal pro-B-type natriuretic peptide, OR odds ratio, PCI percutaneous coronary intervention, STEMI ST segment elevation myocardial infarction, TIA transient ischemic attack, TIMI Thrombolysis in Myocardial Infarction, TVR target vessel revascularization, UA unstable angina

Discussion
Our study shows that most studies including patients with ischemic heart disease showed a consistent association between D-dimer levels and worse clinical outcomes. This association was particularly evident in studies investigating in-hospital adverse events. The rationale of investigating D-dimer levels in CAD patients relies on the persistent hypercoagulable state described in MI patients, which may lead to worse clinical outcomes in these patients through several mechanisms [14]. Clotting activation may contribute to thrombus growth and no-reflow phenomenon, along with impaired thrombolysis which has been frequently described in CAD patients [49], all factors contributing to an increased CAD severity and myocardial injury (Fig. 2).
Reperfusion fails to occur after thrombolytic therapy in approximate 40% of patients, as shown by a study including 427 STEMI patients undergo rescue PCI [50]. Indeed, high D-dimer levels have been shown to be significantly higher in lysis-resistant thrombi from STEMI patients [51].  The pathogenesis of no-reflow remains unclear, but it is likely to be multifactorial, including endothelial damage platelet and clotting activation leading to thrombus formation at the level of small vessels [52]. Few studies investigated the occurrence of no-reflow in STEMI patients undergoing PCI and its relationship with D-dimer [15,22,23,25,47,48]. In 229 consecutive STEMI patients, X. Gong et al. [47] found that a D-dimer value of 530 ng/mL (FEU) was associated with postprocedural no-reflow, with 85.7% of sensitivity and 67.7% of specificity (area under the curve [AUC] = 0.78; p = 0.049). Similarly, Sarli et al. found that optimal D-dimer cut-off value for predicting no-reflow was 549 μg/l (FEU) [25]. In 218 STEMI patients, Cheng et al. [48] found that noreflow patients were older, diabetics, with longer pain-to balloon time, lower blood pressure, higher platelet counts and higher levels of D-dimer and Cystatin C than patients without no-reflow.

Risk stratification using D-dimer
In the past decades, an increasing number of biomarkers have been tested to improve cardiovascular risk stratification in patients with ACS [6,53]. In particular, some studies suggested that the use of a combination of multiple biomarkers may improve risk stratification. This is the case of D-dimer, which if used alone is currently recommended to only exclude acute venous thromboembolism, but its addition to other biomarkers/risk scores gave promising results.
Indeed, the addition of D-dimer has been investigated to some biomarkers such as C-reactive protein (CRP), NT-proBNP and clinical scores, such as the Global Registry of Acute Coronary Events (GRACE) risk score has been investigated. Fu et al. included D-dimer levels ≥2.4 mg/L FEU in a risk score model together with CRP, left ventricular ejection fraction, age ≥ 65 years old and heart rate [32]. In ROC curve analysis, this model demonstrated a good power in predicting in-hospital mortality (AUC = 0.895, 95% CI 0.814-0.96; p < 0.001), better than the predictive power of the GRACE risk score alone (AUC = 0.754, 95% CI 0.641-0.868; p < 0.001). D-dimer has been tested combined to GRACE score also in the study by Lin et al. who found that the combination of NT-pro-BNP and D-dimer improved the predictive accuracy of GRACE score for all-cause death [29]. In 5923 ACS patients undergoing PCI, GRACE score combined with D-dimer achieved a better prognostic performance than GRACE score, and D-dimer could significantly improve the prognostic performance of GRACE score [35]. This result is in keeping with that described by Yu et al., in which [35] D-dimer levels significantly improved the prognostic performance of GRACE score.

Management and therapeutic implications
Altogether, current evidence suggest that D-dimer may identify patients at higher risk for a more severe CAD and worse prognosis. Of note, clotting activation is not lowered by antiplatelet therapy [54], as shown by previous studies reporting no effect of aspirin and DAPT in reducing D-dimer levels [55]. Conversely, oral and parenteral anticoagulation, in addition or not to DAPT, has been demonstrated to decrease D-dimer levels in patients with MI [41,56,57]. Thus, ATLAS ACS 2-TIMI 51 and COMPASS trials have recently proved the utility of adding low-dose anticoagulation to antiplatelet medications. In ATLAS ACS 2-TIMI 51, introduction of lowdose rivaroxaban therapy (2.5 mg or 5 mg twice daily, with 93% of patients on DAPT) has shown to be effective in reducing the risk of death from cardiovascular causes, myocardial infarction, or stroke [58], even if the use of rivaroxaban increased the risk of major bleeding and intracranial haemorrhages. Similarly, the COMPASS trial demonstrated that rivaroxaban plus aspirin was effective in decreasing the primary outcome of cardiovascular death, stroke or MI with a 22% reduction in the net clinical benefit endpoint (defined as the composite of primary outcome, fatal bleeding and symptomatic bleeding into a critical organ/area) [59]. Moreover, a subanalysis of the COMPASS trial has shown that patients with high thrombotic profile (i.e., peripheral arterial disease) may benefit more from adding anticoagulation therapy to reduce residual thrombotic risk.
However, these studies did not stratified patients according to D-dimer levels. Data from the literature let to hypothesize that the benefit conferred by an association therapy of oral anticoagulation and antiplatelet therapy may be even more evident in patients with the features of clotting activation (i.e. high D-dimer). This aspect would deserve a specific investigation.
Another group of patients who deserve particular attention is represented by patients presenting with noreflow. Preliminary data from the literature suggest an association between high-D-dimer levels and no-reflow. Moreover, in MI patients who cannot undergo cardiac catheterization, D-dimer concentrations may also be helpful to predict severity of coronary disease [60].
Finally, D-dimer levels might be helpful in identifying MI patients at high risk of complications and worse outcome. In Fig. 3, we propose a D-dimer guided strategy to tailored therapy in patients at higher thrombotic risk such as those with MI undergoing PCI. Firstly, MI patients with high D-dimer levels may benefit from specific pharmacological management and technical procedures to reduce the probability of no-reflow phenomenon. These include the use of glycoprotein IIb/IIIa inhibitors, higher heparin dose and direct stenting without predilation [61][62][63][64]. Those high-risk patients might also take advantage from prolonged parenteral anticoagulation after PCI during in-hospital staying to reduce early complications. At discharge, patients presenting with highrisk features, including presence of comorbidities, multivessel disease, age > 65 years, recurrent MI, multiple stent insertion, may benefit from adding a low dose of oral anticoagulant agent to conventional antiplatelet therapy to reduce the residual thrombotic risk (i.e., rivaroxaban 2.5 mg twice daily).
Study limitations. A formal analysis for D-dimer against no-reflow phenomenon was not performed given the high variability in the methods of each study (cut-off of D-dimer used, measures of association) was not performed.

Conclusions
In conclusion, given its association with CVEs, the use of D-dimer may help physicians to improve outcome and manage residual thrombotic risk in MI