Hypercoagulability, D-dimer and atrial fibrillation: an overview of biological and clinical evidence

Abstract

Atrial fibrillation (AF) is the most common among the severe cardiac arrhythmias and carries a significant risk of mortality and morbidity in the general population. The most important complication is represented by development of one or more thrombi in the left atrium of the dyskinetic heart, and their successive cerebral and peripheral embolization. The pathophysiological basis of the thromboembolic complications in AF entails the presence of a hypercoagulable state, which is mirrored by increased concentrations of a variety of prothrombotic markers. D-dimer is universally considered the gold standard among the various biomarkers that reflect activation of coagulation, fibrinolysis, or both, and several studies have assessed its diagnostic and prognostic role in AF. With a few exceptions and despite a broad heterogeneity in the study designs, published data seem to demonstrate that D-dimer values may be associated with the presence of atrial thrombosis, may be predictive of primary adverse outcomes and death, may be correlated with cerebral infarction volume, and may also be a useful parameter for assessing the degree of hypercoagulability of AF patients after cardioversion. If larger prospective studies confirm these findings, D-dimer assessment may hence become an integral part of the clinical decision-making in patients with AF.

Key words::

• Atrial fibrillation
• D-dimer
• hypercoagulability
• stroke
• thrombosis

Key messages

• The hypercoagulable state of atrial fibrillation is mirrored by increased concentrations of a variety of prothrombotic markers, among which D-dimer is the gold standard.

• D-dimer values are associated with the presence of atrial thrombosis, are predictive of primary adverse outcomes and death, are correlated with cerebral infarction volume, and may be a useful parameter for assessing the degree of hypercoagulability after cardioversion.

• D-dimer assessment represents an appealing perspective for clinical decision-making in patients with atrial fibrillation.

Introduction

Atrial fibrillation (AF) is universally considered the most common among severe cardiac arrhythmias, although the estimation of real prevalence of this disorder is challenging for a variety of reasons (1). Some published data, which describe an estimated prevalence of 1% to 2% in the general population and a predicted doubling in the next 50 years due to increasing life expectancy (especially in the Western world), are substantially underestimating the large burden of clinically silent disease (as discussed in the following parts of this article), which would leave AF undiagnosed for a long time (the so-called silent AF) (2). The prevalence of disease also varies broadly according to the age of the population, the presence of underlying heart disease, and the diagnostic approach used for its identification (e.g. ECG, dual chamber pacemaker, or implantable cardioverter defibrillator). According to the most recent statistics of the American Heart Association (AHA), the prevalence of non-valvular AF in the United States ranged from approximately 2.7 to 6.1 million in 2010, with an expected growth to between 5.6 and 12 million by 2050 (1). This prevalence is markedly heterogeneous among different age groups, however, being 0.5% in patients aged from 50 to 59 years, 1.8% in those aged from 60 to 69 years, 4.8% in those aged from 70 to 79 years, increasing to 8.8% in subjects aged 80 or older. It is also noteworthy that compared to whites, blacks (odds ratio (OR) 0.49; 95% CI 0.47–0.52), Asians (OR 0.68; 95% CI 0.64–0.72), and Hispanics (OR 0.58; 95% CI 0.55–0.61) have a lower adjusted prevalence (1).

The most powerful predictors of increased risk of new-onset AF include a familial history of AF (hazard ratio (HR) 1.85; 95% CI 1.12–3.06), aging, Caucasian ancestry, increased body mass index, presence of electrocardiographic abnormalities (left ventricular hypertrophy, left atrial enlargement), diabetes, hypertension, presence of cardiovascular disease (coronary heart disease, heart failure), clinical and subclinical hyperthyroidism, along with heavy alcohol consumption (1). It has also been recently acknowledged that AF has a strong genetic background, especially attributable to polymorphisms in genes encoding sodium and potassium channels, gap junction and signaling proteins (1). In particular, genome-wide association studies have found significant interactions with PITX2 (paired-like homeodomain transcription factor 2), a homeobox transcription factor involved in left–right signaling during embryogenesis (3); ZFHX3 (zinc finger homeobox 3), a transcription factor with multiple homeodomains and zinc finger motifs which modulates myogenic and neuronal differentiation (4); KCNN3 (potassium intermediate/small conductance calcium-activated channel, subfamily N, member), a potassium channel protein involved in atrial repolarization (5); along with other putative polymorphisms in genes encoding for PRRX1, CAV1, C9orf3, SYNPO2L, SYNE2, and HCN4 (6).

Classification

AF is typically defined as an irregular and often accelerated heart rate, which is hence mirrored by heart contractions between 50 and 180 times per min (usually between 140–180 times per min), with irregular pattern and varying force. According to the currently available guidelines of the European Society of Cardiology (ESC) and European Heart Rhythm Association (EHRA) (7), AF can be clinically classified in five main subtypes, based on timing of presentation and duration of arrhythmia.

Irrespective of the original subtype, an inexorable progression of AF from short and sporadic episodes to persistent or permanent forms occurs in the vast majority of cases, so that several patients will develop sustained forms of AF over time, and only a minority of patients will suffer from paroxysmal AF over several decades (8–10). As regards the clinical picture, AF is typically accompanied by a variety of symptoms that can be attributed to tachycardia, and thereby include palpitations, intolerance to exercise, and occasionally chest discomfort, congestive symptoms, shortness of breath, or even edema.

Leading complications of atrial fibrillation

It is now well established that AF carries a significant risk of mortality and morbidity in the general population. The AHA has recently estimated that AF is associated with an increased risk of mortality in both men (OR 1.5; 95% CI 1.2–1.8) and women (OR 1.9; 95% CI 1.5–2.2), with an age- and sex-adjusted mortality of 11% and 25% at 30 days and 1 year, respectively (1). Although there is still debate about the potential mechanisms that may contribute to increase the risk of death in patients with AF, it seems reasonable to conclude that the greater part of the risk is conferred by the onset of severe cardiac and extra-cardiac complications, which mainly include stroke, myocardial infarction (MI), heart failure (HF), and impaired renal function.

AF is indeed one of the stronger risk factors for stroke, as a result of cerebral embolization of one or more thrombi generated into the left-atrial appendage of the dyskinetic heart. Specifically, the presence of AF independently increases the risk of cerebrovascular accidents by 2.6- to 4.5-fold throughout all ages, although this may be substantially underestimated due to the frequently asymptomatic nature of AF. Interestingly, the frequency of stroke attributable to an underlying AF increases from 1.5% in patients aged 50 to 59 years, to over 20% in those aged 80 years or more (1). Healey et al. recently performed a prospective study including 2580 patients aged 65 years or older and with no history of AF, in whom pacemakers or defibrillators were implanted. Patients were monitored for 3 months to identify subclinical atrial tachyarrhythmias and were then followed up for a mean period of 2.5 years (8). The onset of subclinical atrial tachyarrhythmias was associated with a significantly increased risk of clinical atrial fibrillation (HR 5.6; 95% CI 3.8–8.2) and especially of ischemic stroke or systemic embolism (HR 2.5; 95% CI 1.3–4.8), even after adjustment for major predictors of stroke.

Another important aspect, that has only recently been emphasized, is the potential relationship between AF and incident MI. In a prospective study including 23,928 subjects without coronary heart disease at baseline who were followed up for a mean period of 4.5 years, Soliman et al. reported that AF was associated with a nearly 2-fold increased risk of MI (hazard ratio (HR) 1.70; 95% CI 1.26–2.30) in a fully adjusted model (11). A reasonable explanation has been brought in support of this association, that is the fact that atrial arrhythmias may trigger a systemic inflammatory response and, consequently, AF-mediated inflammation may lead to plaque rupture in susceptible patients (12).

There is a well-established relationship between HF and AF. Although the two disorders frequently coexist and share a number of predisposing factors (i.e. hypertension, diabetes, coronary artery disease, and valvular heart disease), this association is now regarded as mutually reinforcing, in that HF may promote AF, as well as AF may beget HF (13). It has also been reported that incident HF in patients with AF represents an independent risk factor for death in both men (HR 2.7; 95% CI 1.9–3.7) and women (HR 3.1; 95% CI 2.2–4.2) (14), and also confers an additive risk for development of stroke and thromboembolism (15).

Another challenging association is that existing between AF and impaired renal function. Takahashi et al. studied 386 patients undergoing catheter ablation of AF who were followed up for 1 year (16), and observed that the renal function (i.e. the estimated glomerular filtration rate) slightly improved in patients free from arrhythmia, whereas the estimated glomerular filtration rate significantly decreased in those with recurrent AF. This interesting finding has recently been confirmed by Bansal et al. (17), who demonstrated that incident AF was independently associated with a significant increase in the rate of end-stage renal disease (HR 1.67; 95% CI 1.46–1.91) in 206,229 adults.

All that said, the identification of biomarkers that may be predictive for developing severe complications, especially of AF- associated stroke risk, is an appealing perspective for management of patients with AF. It has been recently shown that cardiospecific troponins and natriuretic peptides are effective to improve risk stratification in addition to current clinical risk stratification models (18). Nevertheless, these cardiac biomarkers can only reflect a limited part (i.e. myocardiocyte injury and stress) of the complex biological pathway that leads to the pathogenesis of thrombogenesis and embolization in AF, but fail to provide useful information about the prothrombotic state of this condition, an aspect that can now be considered another important piece of the challenging puzzle of hypercoagulability, as described in the following parts of this article. Therefore, the aim of this narrative review is to provide an overview about the hypercoagulable state in AF and the potential usefulness of D-dimer assessment for predicting thromboembolic risk and therapeutic management of patients with AF.

The prothrombotic state in atrial fibrillation

The association between AF and hypercoagulability has long been recognized (19). The most frequent clinical manifestations include atrial thrombogenesis and further embolization, which often leads to the onset of stroke or other major thromboembolic complications. The former represents a major clinical problem because, as previously discussed, the risk of thromboembolism-related stroke is approximately 2% per year in AF patients. Although the pathophysiology of thromboembolism in AF is indeed multifactorial, there is increasing evidence that the so-called Virchow's triad (i.e. abnormalities of blood flow, blood constituents, and vessel wall) represents the underlying basis of thrombogenesis in AF (15) as well as in other thromboembolic conditions such as venous thromboembolism (20). A number of abnormalities referable to Virchow's triad are clearly present in AF.

Abnormal blood flow is represented in AF by increased left atrial stasis due to the loss of atrial systole, and is reflected by increased spontaneous echo contrast on transesophageal echocardiography (21) and blood flow velocities in the left-atrial appendage (LAA) (22), which is incidentally the major site of thrombus formation in AF patients. Flow abnormalities in AF may also be exacerbated by the close association between AF and HF, as previously discussed, or by the presence of non-valvular atrial fibrillation, which seems to promote progressive left atrial dilatation (23). In the presence of mitral stenosis, left atrial dilatation is increased and leads to further stasis and propensity to thrombosis (24). The contribution of left atrial dilatation to thrombogenesis is also supported by the finding that atrial size corrected for body surface area is an independent risk factor for stroke (25).

Vessel wall abnormalities in AF are typically described as endothelial/endocardial damage or dysfunction, and can be visualized by scanning electron microscopy, especially within the appendages (26). Goldsmith and colleagues described more severe endocardial changes in patients with AF compared with those in sinus rhythm. These structural changes were more frequently found in LAA, which is anatomically more predisposed to blood stasis than the right-atrial appendages (27). Changes in dimensions of left atrium and LAA often occur in AF and are associated with subsequent thromboembolism. However, not all structural changes in AF are of cardiac origin. Complex aortic plaque, indicating the presence of significant atherosclerotic disease and identified by transesophageal echocardiography (TEE), occurs in up to 57% of patients with AF (28) and is an independent predictor for subsequent thromboembolism (29).

The presence of abnormalities of blood constituents, that is the third component of Virchow's triad, is reflected by abnormal values of platelets, coagulation and fibrinolysis factors, which contribute to hypercoagulable state in AF patients. These abnormalities basically include increased fibrin turnover, enhanced release of platelet microparticles (i.e. a strong prothrombotic factor), as well as increased values of β-thromboglobulin, von Willebrand factor (VWF), prothrombotic biomarkers such as D-dimer, prothrombin fragments 1 and 2, and thrombin-antithrombin complexes (26).

Although several mechanisms support the prothrombotic state of AF, some studies have also focused on the potential role of inflammation (30,31), abnormalities in growth factor release (32), structural remodeling of atria (matrix metalloproteinase system) (33), and platelet activation (34,35). Moreover, genes influencing activation of hemostasis have been recently proposed as important components of the overall thrombotic tendency in AF, including FGA T331A (rs6050), FGB g.4577G> A (rs1800790), F13A1 V34L (rs5985), F2 g.25404A> G (rs3136517), F5 R506Q (rs6025), F7 g.4727_4728ins10 (rs5742910), and ITGA2 g.67214C> T (rs1126643) polymorphisms (36).

Risk assessment in atrial fibrillation

The strategies for identifying patients at risk for thromboembolism in AF are commonly based on clinical variables, e.g. congestive HF, hypertension, age, diabetes mellitus, and prior stroke or transient ischemic attack (TIA), as in the widely used CHADS₂ risk score (37). Recently, vascular disease, age ≥ 65 years, and female gender were also added, to give the variables in the CHA₂DS₂-VASc score, which assigns 2 points for a history of stroke or TIA, or age ≥ 75; and 1 point for age 65–74 years, a history of hypertension, diabetes, recent cardiac failure, vascular disease, and female gender (38). The latter stroke risk stratification schema provides some improvement in predictive value for thromboembolism over the CHADS₂, with low event rates in low-risk subjects and classification of only a small proportion of subjects into the intermediate-risk category. Nevertheless, the improvement in terms of discriminating value for individual patients remains modest, rising from a c-statistic of 0.58 with the CHADS₂ score to a maximum of 0.64 with the CHA₂DS₂-VASc (where 1.0 is perfect discrimination and 0.5 is no better than random chance) (38).

D-dimer as index of thrombogenesis and thrombus turnover in atrial fibrillation

D-dimer is now universally considered the gold standard among the various biomarkers that reflect activation of coagulation, fibrinolysis, or both (39). D-dimer is a product of the cross-linked fibrin degradation, and it is hence a circulating marker indicative of both thrombogenesis and thrombus turnover (40). Nevertheless, D-dimer is a generic term that is in conventional use for defining a heterogeneous species of fibrin degradation products, which are typically characterized by the presence of a covalent link between two adjacent D units against which monoclonal antibodies of the commercial immunoassays are directed.

Several studies have shown that D-dimer levels are higher in AF patients compared with matched controls in sinus rhythm, and for this reason this biomarker may be potentially useful for identification of subjects at higher risk of developing thromboembolic complications (41–43). In particular, Lip et al. showed that patients with paroxysmal AF have intermediate levels of median fibrin D-dimer compared with patients with chronic AF and controls in sinus rhythm (both P < 0.001), a finding that is consistent with the intermediate risk of thromboembolism in such patients (41). On the other hand, patients affected by chronic AF have higher risk of thromboembolic events, and the high levels of fibrin D-dimer observed in these patients are independent of the underlying cause (44).

Marin et al. observed that patients with both acute and chronic AF had higher concentrations of fibrin D-dimer than did healthy controls (P < 0.01). Moreover, mean D-dimer values in acute AF were higher when compared with chronic AF (acute AF: 2350 μg/L; chronic AF: 1120 μg/L), and these concentrations remained higher in patients with acute AF than in patients with chronic AF (P = 0.038) at day 30 after cardioversion (42). Additionally, in patients with known risk factors for embolism (i.e. history of embolism or TIA, congestive HF, hypertension, renal failure, surgery, liver impairment, acute or chronic infection, neoplastic disease, and diabetes mellitus), D-dimer levels were found to be higher than in those without risk factors (43). Interestingly, a positive correlation between levels of B-type natriuretic peptide (BNP) and D-dimer levels has recently been demonstrated, suggesting that these biomarkers may be used in combination for identifying high-risk patients (45).

The predictive value of D-dimer in AF

It is well known that a considerable variability exists both within and among quantitative methods used by laboratories to assay D-dimer. Laboratory practice still varies regarding both the type and the magnitude of units reported, so influencing the setting of the threshold for the exclusion of venous thromboembolism (VTE) (46). A growing body of evidence also demonstrates that age- and gender-adjusted cut-offs for D-dimer are more appropriate for risk stratification of patients with suspected thrombosis (47–52), so that the use of specific formulas designed to adjust the diagnostics thresholds for the age of the patient may be advisable (46).

It is conceivable that thromboembolism may occur when renewed atrial contraction displaces one or more thrombi that originated during a sufficiently prolonged episode of AF (usually lasting more than 48 h), or that the accentuation of atrial stasis immediately after restoration of sinus rhythm promotes the development of a new thrombus and its embolization (53). Some studies demonstrated that D-dimer levels are significantly increased in the presence of an atrial thrombus (54–57), so that this marker may be useful for risk assessment of thromboembolism in AF. In particular, it has been suggested that its assessment might be used to rule out the presence of atrial thrombi in AF patients, thus allowing safe cardioversion (58). The studies exploring such negative predictive value of D-dimer are scarce but have yielded promising results. The main findings are summarized in Table I.

Table I. Studies investigating the role of D-dimer in the exclusion of atrial thrombi.

First author	Type of AF	Patients, n	Primary end-point	Patients who experienced the end-point, %	D-dimer cut-off, ng/mL	Main results
Somloi et al. (56)	Patients with AF of > 48 hours duration or atrial flutter with documented history of AF	73	Atrial thrombi	12%	600	Sensitivity: 89% Specificity: 75% PPV: 33% NPV: 98% AUC: 0.78 (95% CI, 0.63–0.93; P = 0.007)
Habara et al. (57)	Non-valvular AF	925	LAA thrombi	22% of patients with higher D-dimer values and 3% of patients with lower D-dimer values	1150	Sensitivity: 76% Specificity: 73% PPV: 22% NPV: 97% AUC: 0.80 (95% CI 0.75–0.85; P< 0.001).
Yasaka et al. (54)	Valvular AF	63 (70% affected by AF)	Non-mobile atrial thrombi Mobile atrial thrombi	12.7% 15.9%	–	Sensitivity: 61% Specificity: 93% PPV: 79% NPV: 86% Sensitivity: 100% Specificity: 92% PPV: 71% NPV: 100%

AUC = area under ROC curve; LAA = left-atrial appendage; NPV = negative predictive value; PPV = positive predictive value.

According to the clinical evidence discussed in the previous paragraph, it seems reasonable to hypothesize that plasma D-dimer levels may be used to improve risk stratification for adverse events, especially thromboembolism, in patients with AF. To test this hypothesis, some authors studied the prognostic value of D-dimer for several outcomes including ischemic stroke, TIA, peripheral embolism, or even combined cardiovascular events such as thromboembolic episodes and MI, cardiovascular deaths and major bleeding. Results obtained in large cohorts of AF patients are summarized in Table I (59–66). These studies were single or multicenter prospective observational studies involving patients with non-valvular AF mostly on long-term anticoagulation therapy. Notably, only few authors reported the prognostic value of D-dimer by using regression analysis, and a minority of them performed multivariate analysis to test its role as independent predictor of adverse events. Moreover, the cut-off levels were rather heterogeneous across different studies. With a few exceptions and despite the broad heterogeneity in the study designs, published data seem to confirm that D-dimer may be predictive of primary thrombotic outcomes and death (Table II). In general, when the study population included patients both receiving and not receiving oral anticoagulants, three out of four studies reported that D-dimer values may help distinguish subjects at higher risk of thromboembolic complications or death. On the other hand, when only patients undergoing oral anticoagulant therapy were considered, the concentration of D-dimer did predict adverse cardiovascular outcomes or death in four out of five studies.

Table II. Studies reporting the prognostic value of D-dimer for cardiovascular events in AF.

First author	Type of AF	Patients, n	Patients’ treatment, n (type of therapy)	Mean follow-up, months	Primary end-point	Patients who experienced the end-point, n (%)	D-dimer cut-off, ng/mL	Main results^a
								All patients	Patients in OAT
Mahé I (59)	Paroxysmal or chronic	425	81 (APA), 285 (OAT), 59 (NO)	16	Combined CV events (MI, stroke or TIA, and arterial embolic events)	26 (6.1)	334	7.9% vs 1.7%, P = 0.01	6.3% vs 0.9%, P = 0.03
Mahè I (60)	Chronic	125	67 (OAT), 40 (ASA), 18 (NO)	23.2	Death	54/120 (45.0)	1000	44.4% vs 14.9%, P = 0.01 RR: 2.98 (1.56–5.67)
Krarup LH (61)	AF in combination with acute ischemic stroke	382	382 (ASA or LMWH)	12	Stroke progression and recurrence	90 (23.6)	100 unit increase	Stroke progression: OR: 0.99 (0.97–1.01), P = 0.34 ^b Stroke recurrence: OR: 0.98 (0.93–1.02), P = 0.26 ^b Stroke progression, recurrence, or death: OR: 0.99 (0.98–1.01), P = 0.24 ^b
Roldàn V (62)	Permanent/ paroxysmal	829	829 (OAT)	24	CV events (stroke, ACS, AHF), CV death	95 (11.6)	No cut-off found		CV events: HR: 1.00 (1.00–1.00), P = 0.989^c CV death: HR: 1.00 (0.99–1.00); P = 0.921^c
Sadanaga T (63)	Paroxysmal or chronic	269	269 (OAT)	25.2	TE (ischemic stroke, TIA, and peripheral embolism) and combined CV events (TE, cerebral hemorrhage, MI, and CV deaths)	37 (13.7)	500		HR: 15.7 (3.31–74.0) for TE, P < 0.01^c; HR: 7.57 (3.39–16.9) for combined CV events, P < 0.01^c
Sadanaga T (64)	Paroxysmal or chronic	245	245 (OAT)	25.2	TE events (ischemic stroke, TIA, and peripheral embolism)	9 (3.7)	500		HR: 14.3 (2.97–69.2), P < 0.01^c In the subgroup of patients with high CHADS2 score (n = 76) HR: 12.1 (1.44–101), P = 0.021 ^c
Nozawa T (65)	Permanent or paroxysmal	509	163 (APA), 263 (OAT) 83 (NO)	24	TE events (cerebral infarction, TIA or embolism of peripheral arteries)	31 (6.1)	150	3.8%/year vs 0.7%/year, P < 0.01	9.0% vs 0.4%, P < 0.01
Vene N (66)	Chronic	111	111 (OAT)	44.3	Combined CV event (stroke, MI, peripheral vascular occlusion or CV death)	22 (19.8)	Top vs lower two quartiles		HR 3.93 (1.49–10.38), P = 0.006^c Considering DD levels after stabilization of OAT, HR: 7.51 (2.39–23.66), P = 0.001^c; HR: 4.78, (1.39–16.41), P = 0.01^b

ACS = acute coronary syndrome; AHF = acute heart failure; APA = antiplatelet therapy; ASA = acetyl salicylic acid; CV = cardiovascular; DD = D-dimer; HR = hazard ratio; MI = myocardial infarction; NO = no treatment; OAT = oral anticoagulant therapy; OR = odds ratio; RR = relative risk; TE = thromboembolic; TIA = transient ischemic attack.

^a If not specified, the main results have been expressed as incidence of end-points for DD levels above and below the cut-off (comparison estimated with chi-square test); otherwise the risk of events has been reported as HR (95% CI), OR (95% CI), or RR (95% CI). In these latter cases, results are derived from ^b multivariate or ^c univariate regression analysis.

Data on the predictive impact of D-dimer over the classical clinical risk assessment emerged almost exclusively from two studies (64,65). Nozawa and co-authors (65) found that in AF patients with risk factors (i.e. age ≥ 75 years, cardiomyopathies, and prior ischemic stroke or transient ischemic attack), the annual incidence of thromboembolism was higher when the D-dimer level was ≥ 150 ng/mL than when it was below this threshold (6.5% versus 5.7%). Similarly, in patients with the above-mentioned clinical risk factors and treated with anticoagulant therapy the incidence of the primary end-points was 7.0% per year in those with low levels of D-dimer and 5.3% per year in those with high levels. Even more importantly, Sadanaga et al. (64) reported that when D-dimer level was added to the predictor model based on the CHADS₂ score, the c-statistic improved from 0.781 to 0.848.

The association between D-dimer and cerebral infarction volume is another aspect that has recently been highlighted. Matsumoto et al. enrolled 124 patients with ischemic stroke and non-valvular AF admitted within 48 h of symptom onset (67) and found that D-dimer levels were significantly associated with infarction volume (r = 0.309; P < 0.001) after adjusting for age, gender, current smoking status, diastolic blood pressure, and CHADS₂ score.

D-dimer in management of atrial fibrillation

It is now clearly acknowledged that cardioversion carries a high risk of thromboembolism, especially when AF persists for more than 48 h (53), although the precise underlying mechanisms have only been partially elucidated. Regardless of the pathophysiological basis, the results of several studies suggest that D-dimer may be a potentially useful parameter for assessing the degree of hypercoagulability of AF patients after cardioversion (68–70) and during anticoagulation therapy (44). It is also noteworthy that D-dimer levels were found to be reduced during anticoagulation with conventional warfarin therapy, but not with ultra-low-dose warfarin (1 mg) or aspirin 300 mg (71). In a recent study, Nakatani et al. (72) demonstrated that low therapeutic range, low percentage of international normalized ratio (INR) values in the range, and high coefficient of variation of INR values were significant contributors in increased D-dimer levels (P = 0.02; 0.03; and 0.02, respectively). Since the risk of atrial thrombus formation and systemic embolism persists after reversion to sinus rhythm following a successful electric cardioversion, it has also been emphasized that anticoagulation should be recommended for 3–4 weeks post-procedure also in a patient with a negative D-dimer result (73).

It has also been recently reported that the risk of hypercoagulability may increase during the early post-procedural period of AF ablation in patients inappropriately anticoagulated with warfarin (74). Sairaku et al. observed that the D-dimer level continued to increase over a period of 48 h after the ablation, and that the INR value measured just before the ablation had a significant inverse correlation with D-dimer levels assessed just before (r = –0.304), immediately afterwards (r = –0.440), 24 h after (r = –0.442), and 48 h after the ablation (r = –0.463) (74). Moreover, the reduction of D-dimer levels after initiation of oral anticoagulants was shown to be greater in patients treated with the novel oral anticoagulant dabigatran than in patients allocated with warfarin (75). This result is in accordance with the hypothesis that an anticoagulation strategy with dabigatran may be superior to one with warfarin in reducing the periprocedural risk of a hypercoagulable state.

The role of D-dimer in the management of patients with AF is expanding, and the potential of its clinical application as a test with high negative predictive value has been highlighted. In particular, Tayebjee et al. (76) proposed an interesting flow chart outlining the place of D-dimer measurement in AF management. They also emphasized that the finding of elevated fibrin D-dimer levels in patients who received anticoagulation therapy at the time of cardioversion may have relevant clinical implications. In the presence of remaining clot(s), cardioversion may hence be postponed and longer periods of anticoagulation might be advisable, which would ultimately lead to a decreased risk of stroke and thromboembolism after cardioversion.

Conclusions

It is now undeniable that hypercoagulability plays a major role in the pathogenesis of thromboembolism in patients with AF. The management of the prothrombotic state by anticoagulation also represents the mainstay of clinical management and prevention of thromboembolic complications in this life-threatening condition. Several lines of evidence suggest that the heterogeneous degree of hypercoagulability that accompanies AF is mirrored, and potentially monitored, by D-dimer levels. A clinical applicability of D-dimer in the treatment decision-making is also warranted, although evidence specifically addressing this aspect is still limited. Future studies would clarify the role of D-dimer in refining the thromboembolic risk predictable by clinical score and establish if such combined assessment may be used to select patients who will benefit from anticoagulation and guide clinicians about the intensity of anticoagulation needed to minimize thrombogenesis.

To date, most studies on D-dimer in the setting of AF have assessed its diagnostic and prognostic role. With a few exceptions and despite a broad heterogeneity in the study designs, published data seem to demonstrate that D-dimer values may be associated with the presence of atrial thrombosis, may be predictive of primary adverse outcomes and death in the majority of studies in patients with AF, may be correlated with cerebral infarction volume, and may also be a useful parameter for assessing the degree of hypercoagulability of AF patients after cardioversion. Further evidence suggests that D-dimer might have a potential role in ruling out atrial thrombus prior to attempting electrical or pharmacological cardioversion. If larger studies confirm these findings, D-dimer assessment may hence become an integral part of the clinical decision-making in patients with AF.

Declaration of interest: The authors report no conflicts of interest.