Author + information
- Received May 17, 2018
- Revision received August 20, 2018
- Accepted September 11, 2018
- Published online February 18, 2019.
- Antoine Milhem, MDa,∗ (, )
- Pierre Ingrand, MD, PhDb,
- Frédéric Tréguer, MDc,
- Olivier Cesari, MDd,
- Antoine Da Costa, MDe,
- Dominique Pavin, MDf,
- Philippe Rivat, MDg,
- Nicolas Badenco, MDh,
- Sélim Abbey, MDi,
- Noura Zannad, MDj,
- Pierre François Winum, MDk,
- Jacques Mansourati, MDl,
- Philippe Maury, MDm,
- Hugues Bader, MDn,
- Arnaud Savouré, MDo,
- Frédéric Sacher, MDp,
- Marius Andronache, MDq,
- Caroline Allix-Béguec, PhDa,
- Christian De Chillou, MD, PhDq,r,
- Frédéric Anselme, MD, PhDo,
- for the ATE Study Group
- aGroupe Hospitalier de la Rochelle Ré Aunis, La Rochelle, France
- bEpidemiology and Biostatistics, INSERM CIC 1402 Université de Poitiers, CHU Poitiers, Poitiers, France
- cClinique St-Joseph, Trélazé, France
- dClinique Saint Gatien, Tours, France
- eCHU Saint-Etienne, Saint-Etienne, France
- fCHU Rennes, Rennes, France
- gPolyclinique Vauban, Valenciennes, France
- hAPHP, Hôpital Pitié Salpêtrière, Paris, France
- iNouvelles Cliniques Nantaises, Groupe Confluent, Nantes, France
- jCHR Metz-Thionville, Metz, France
- kCHU Nîmes, Nîmes, France
- lCHU Brest, Brest, France
- mCHU Toulouse, Toulouse, France
- nCH Pau, Pau, France
- oCHU Rouen, Rouen, France
- pCHU Bordeaux, Bordeaux, France
- qCHU Nancy, Vandœuvre lès-Nancy, France
- rINSERM-IADI U1254, Vandœuvre lès-Nancy, France
- ↵∗Address for correspondence:
Dr. Antoine Milhem, Service de Cardiologie, Groupe Hospitalier de la Rochelle Ré Aunis, rue du Dr. Schweitzer, 17019 La Rochelle, France.
Objectives This study hypothesized that the association of D-dimer blood level and several clinical items in a new risk score could predict the absence of atrial thrombus.
Background Symptomatic and drug resistant atrial fibrillation (AF) can be treated by catheter ablation. The procedure-related risk of thromboembolism is limited by the pre-operative use of transesophageal echocardiography (TEE) to detect atrial thrombi.
Methods Patients admitted for catheter ablation of AF (n = 2,494) were prospectively included in a multicenter study. TEE was systematically performed before the procedure to search for atrial thrombus (primary endpoint). D-dimer level, CHADS2 score, left ventricular ejection fraction, pre-operative anticoagulation regimen, and medical history were collected. A logistic regression model was used to identify factors associated with the presence of atrial thrombus (hypertension, history of stroke, heart failure, D-dimer level >270 ng/ml). These factors were aggregated in a new score called atrial thrombus exclusion (ATE).
Results The incidence of atrial thrombus was 1.92%. CHADS2 score and D-dimer level were significantly associated with atrial thrombus (p < 0.0001 and p < 0.0001, respectively). A zero CHADS2 score failed to exclude all atrial thrombi (5 false negatives; sensitivity: 89.58%, specificity: 52.2%). No false negative was found with a zero ATE score, which had a specificity of 37% and a higher sensitivity (100%) than the CHADS2 score (p < 0.031) to predict the absence of intra-atrial thrombi on TEE. Conversely, the positive predictive value was poor, and the ATE score should not be used to conclude a positive diagnosis of thrombus.
Conclusions An ATE score of zero was strongly associated with the absence of atrial thrombus. This new score could be useful to rule out a diagnosis of atrial thrombus before catheter ablation of AF.
Atrial fibrillation (AF) is the most common form of cardiac arrhythmia and is associated with increased morbidity and mortality mostly related to thromboembolic events. Catheter ablation is recommended as a curative treatment of drug-resistant and symptomatic AF. The presence of atrial thrombus, usually located in the left atrial appendage, is a contraindication to catheter ablation of AF. Transesophageal echocardiography (TEE) remains the gold standard for the diagnosis of left atrial thrombi (1). CHADS2 and CHA2DS2-VASc scores are widely used to predict the thromboembolic risk associated with AF (2,3).
D-dimers (plasma fibrin D-dimers) are degradation products of cross-linked fibrin by the endogenous fibrinolytic system, and their plasma levels are routinely measured to rule out diagnosis of deep venous thrombosis or pulmonary embolism. Likewise, we hypothesized that a low level of plasma D-dimers could contribute to ruling out a diagnosis of atrial thrombus.
This prospective multicenter study was designed to assess the ability of a new composite score, which combined plasma D-dimer levels and clinical data, to rule out a diagnosis of atrial thrombus before catheter ablation of AF.
This study enrolled consecutive patients hospitalized for catheter ablation of AF (paroxysmal or persistent) or left atrial tachycardia. It was conducted between August 2015 and December 2016 in 29 French electrophysiology facilities where TEE was regularly performed for thrombus formation screening before catheter ablation in the left atrium. The study complied with the Declaration of Helsinki. The patients received appropriate information by the physicians before inclusion in the trial. The study was approved by the institutional review board and was registered in the clinicaltrials.gov database (NCT02199080).
Plasma D-dimer level measurement
Plasma D-dimer level was systematically measured within 48 h before catheter ablation. Two different tests were used: LIATEST (Stago, Asnières sur Seine, France) and VIDAS (BioMerieux, Marcy l’Etoile, France). Both tests have the same cutoff value (500 ng/ml) to rule out a diagnosis of venous thromboembolism. The lowest measurable values were 270 and 45 ng/ml using the LIATEST and VIDAS tests, respectively. For the sake of data homogenization, results ranging from 45 to 270 ng/ml with the VIDAS test were all considered as being <270 ng/ml.
TEE was performed according to the guidelines of the European Association of Cardiovascular Imaging within 48 h before the ablation procedure (4). The detection of a hyperechoic mass revealed an intra-atrial thrombus. Its presence contraindicated the ablation procedure. Because the aim of the study was to propose an alternative method to TEE to eliminate an atrial thrombus, the findings that resulted from the TEE were not collected (sludge, left appendage velocity). Operators were blinded to the results of the D-dimer levels.
Clinical data including age, sex, hypertension, diabetes mellitus, heart failure, history of stroke, medical and surgical history, AF history, anticoagulation protocol, left ventricular ejection fraction (LVEF), plasma D-dimer level, and presence of atrial thrombus were collected. Effective anticoagulation was arbitrarily defined as the prescription of a direct oral anticoagulant agent or vitamin K antagonist therapy with a weekly international normalized ratio (INR) >2 during the 3 weeks preceding TEE. Paroxysmal AF was defined as recurrent AF that spontaneously terminated within 7 days. Persistent AF was defined as AF that lasted for at least 7 days or required cardioversion, and also included long-lasting AF (>1 year of AF).
The presence of left atrial thrombus was a contraindication to catheter ablation. The ablation procedure and patient monitoring were performed according to local practices.
The primary endpoint was the presence of atrial thrombus diagnosed by TEE.
Statistical significance was established with a p value <0.05. Continuous variables were presented as mean ± SD, and categorical variables were presented as counts and percentage. Relationships between the primary endpoint and continuous variables or categorical variables were explored using the Mann-Whitney test, chi-square analysis, and Fisher exact test, respectively. Continuous variables were converted into binary variables by determining a threshold value. Binary variables significantly associated with the primary endpoint were included in a multivariate logistic regression model using a stepwise selection method that used Wald's test significance criterion with entry and exit levels set both at p < 0.05. Independent factors associated with atrial thrombus were aggregated in a new score. The sensitivity, specificity, and negative predictive value of this new score were calculated. Sensitivities of the new score and CHADS2 score were compared using the likelihood ratio test. Calibration plots and the Hosmer-Lemeshow test were used to assess goodness of fit and model calibration.
Statistical analysis was performed using SAS V9.4 statistical software (SAS Institute, Cary, North Carolina).
Baseline demographic and clinical characteristics of participants
Among the 2,506 patients included over the 18-month study period, 12 were excluded from the analysis because of missing key data (D-dimer value). Of the remaining 2,494 patients, 166 were missing LVEF measurements. The study population consisted of 1,818 men (72.9%) and 676 women (27.1%), with a median age of 62 years. The patients experienced paroxysmal AF (53.5%), persistent AF (39.9%), and left atrial tachycardia (6.5%). Catheter ablation was a first attempt in 1,901 cases (76.2%). The median LVEF was 60%. Hypertension, diabetes mellitus, heart failure, and history of stroke were reported in 36.7%, 9.1%, 12.2%, and 4.6% of the cases, respectively (Table 1).
Forty-eight thrombi were documented by TEE (incidence of 1.92%; 95% confidence interval: 1.42% to 2.54%) before or at the beginning of the ablation procedure.
Plasma D-dimer threshold
Plasma D-dimer levels were measured with LIATEST and VIDAS in 1,173 and 1,321 patients, respectively.
The receiver-operating characteristic curve of plasma D-dimer level was built (area under the receiver-operating characteristic curve: 0.658) (ATE Score: Online Figure 2).
Among the tested continuous variables, age (p < 0.0001), CHADS2 score (p < 0.0001), LVEF (p < 0.0001), and D-dimer assay (p < 0.0001) were significantly associated with the presence of atrial thrombus.
The following categorical variables were significantly associated with the presence of atrial thrombus using univariate analysis: age older than 75 years (odds ratio [OR]: 3.1; p = 0.0086), hypertension (OR: 2.93; p = 0.0004), diabetes (OR: 2.35; p = 0.036), heart failure (OR: 4.97; p < 0.0001), history of stroke (OR: 3.06; p = 0.021), and D-dimer level >270 ng/ml (OR: 2.717; p = 0.0008). Persistent arrhythmia was also associated with the presence of atrial thrombus (p < 0.0001). The results are summarized in Table 2.
The center effect was tested and was not statistically significant (p = 0.98).
Using multivariate logistic regression analysis, hypertension (p = 0.0048), heart failure (p < 0.0001), history of stroke (p = 0.041), and a D-dimer plasma level >270 ng/ml (p = 0.007) were found to be independently associated with the existence of atrial thrombus (Table 3).
A new score was constructed by combining a history of stroke, congestive heart failure, high plasma D-dimer level, and hypertension. The distribution of the atrial thrombus exclusion (ATE) score in the study population is shown in Table 4. The distribution of thrombus number as a function of the ATE score is depicted in Figure 1.
This score was significantly associated with the presence of endocardial thrombus (p < 0.0001). The receiver-operating characteristic curve was constructed (area under the receiver-operating characteristic curve: 0.77).
An ATE score of zero stood for the absence of thrombus, with a sensitivity of 100%, a specificity of 37%, and a negative predictive value of 100% (no false negative). The sensitivity of the CHADS2 score was 89.9% (95% confidence interval: 77.34% to 99.65%) (5 false negatives). The sensitivities of the ATE score and the CHADS2 score were significantly different (p = 0.031). The Hosmer-Lemeshow test was significant (p = 0.0022) in relation with departures to log-linearity. This limit led to building the ATE score as an ordinal scale.
As stated in the 2017 HRS/EHRA/ECAS/APHRS/SOLAECE expert consensus statement on catheter ablation of AF, there is a wide disparity in the practice of experienced teams regarding indications for TEE. Although most teams systematically perform TEE before or at the beginning of ablation, some experts personalize their practice according to each patient’s AF type (paroxysmal or persistent), thromboembolic risk score, or previous effective anticoagulation. However, in the absence of previous anticoagulation in patients with AF who have episodes lasting >48 h, TEE is mandatory (1).
Incidence of thrombus
The incidence of intra-atrial thrombus of 1.92% (95% confidence interval: 1.42% to 2.54%) found in our study is in line with incidences reported in the published data (range 0.6% to 3.6%) (5–11). The incidence of atrial thrombus before a left atrial ablation is likely dependent on the characteristics of the study population, the quality of anticoagulation, and the diagnostic methods (Table 5).
Apparent effective anticoagulation before ablation does not definitively eliminate the risk of atrial thrombus formation. As reported in Table 5, almost all atrial thrombi were identified while patients were on effective anticoagulation treatment, as assessed by recurrent INRs within the therapeutic range weeks before AF ablation. To our knowledge, there are no data on thrombi incidence in patients on direct oral anticoagulant treatment in this setting. In a retrospective study of 672 patients with AF, Zylla et al. (12) reported an incidence of atrial thrombi that was significantly higher in patients treated with phenprocoumon (17.8%) than in those treated with dabigatran or rivaroxaban (3.9%).
In our study, 47 of 48 patients with atrial thrombus had previous anticoagulation before TEE. Among the 1,059 patients treated with warfarin, an atrial thrombus was found in 27 (2.5%), including 18 patients with a weekly INR >2 for >3 weeks. Among the 1,186 patients treated with direct oral anticoagulant agents, an atrial thrombus was found in 19 (1.6%) of them (dabigatran: 6 of 269; rivaroxaban: 10 of 596; apixaban: 3 of 321).
Plasma D-dimer level and atrial thrombus
Several studies have investigated the association between endocardial thrombus and plasma D-dimer level. Yasaka et al. (13) reported a significantly higher plasma D-dimer level in 18 of 63 mitral stenosis patients with an atrial thrombus visualized on transthoracic echocardiography compared with patients without atrial thrombus. Likewise, a higher D-dimer level was found in 19 of 109 patients with nonvalvular AF (14). In 73 AF patients who underwent systematic TEE before electrical cardioversion, a plasma D-dimer level <600 mg/ml had a negative predictive value of 98% to rule out the presence of thrombus (15). The authors pointed out that the only false negative was a patient with an intra-atrial ultrasound image that did not disappear when the patient received anticoagulant treatment, which suggested a possible false positive of echo. Recently, in a much larger study (925 patients with nonvalvular AF), in which TEE and plasma D-dimer measurement were systematically performed, a cutoff value of 1,115 ng/ml (determined by receiver-operating characteristic curve) had a negative predictive value of 97% to rule out a diagnosis of intra-atrial thrombus (16).
It is noteworthy, in these 2 studies, that the cutoff value was determined to obtain the best sensitivity/specificity ratio, but not to maximize the negative predictive value.
We reported here the results of the first multicenter study that evaluated plasma D-dimer level as a potential predictive factor to rule out the presence of intra-atrial thrombus in the setting of AF ablation. We found that a low plasma D-dimer value was significantly associated with absence of thrombus. However, it was not possible to determine a cutoff value to definitively exclude the presence of an intracardiac thrombus using this single parameter. Because of the measurement methods used (no value <270 ng/ml using LIATEST), the cutoff value of 270 ng/ml was chosen to favor the negative predictive value of the test.
Scores predicting the absence of thrombus
Several multicenter studies investigated the CHADS2 score or its constitutive variables as predictor of absence of atrial thrombus before isolation of the pulmonary veins (Table 5). Puwanant et al. (6), Scherr et al. (5), and Yamashita et al. (10) reported an increased incidence of intra-atrial thrombus as the CHADS2 score rose. However, among these 3 studies, only Scherr et al. (5) found that a CHADS2 score ≥2 was an independent factor associated with the presence of thrombus. In all of these studies, the incidence of thrombi was low when the CHADS2 score was equal to zero (0% to 1%). Based on this result, some authors suggested that TEE could be skipped in patients without clinical risk factors for thromboembolism. However, they warned that prospective multicenter studies were needed to confirm this hypothesis.
In our study, 1,282 patients had a CHADS2 score of zero, and atrial thrombus was found in 5 patients (0.4%). The plasma D-dimer levels of these 5 patients were all above the cutoff value of 270 ng/ml (411, 490, 640, 730, and 1,080 ng/ml). The new thromboembolic risk score, ATE, which combined clinical items and plasma D-dimer level, provided a negative predictive value of 100%. The diagnostic performance of this score was significantly better than that of the CHADS2 score. Among the patients with a zero CHADS2 score, 422 (including the 5 patients with intra-cardiac thrombus) had an ATE score of 1. Conversely, 51 patients with a CHADS2 score of 1 were considered at no risk of thrombus according to a zero ATE score.
Practical value of using the ATE score
This new score appears to be simple, combining common clinical data and a frequently used and inexpensive biological marker (plasma D-dimer level).
In our study, an ATE score of zero ruled out the presence of atrial thrombus with a sensitivity of 100%. Based on this result, 911 TEE examinations (36.5%) could have been avoided. Beyond the cost factor, scheduling TEE can sometimes be difficult in busy electrophysiology laboratories and may lengthen hospitalization time if performed the day before ablation. In that perspective, not having to schedule TEE could simplify the ablation procedure. Nevertheless, some physicians might perform TEE anyway, to guide transseptal punctures.
In the diagnosis of venous thromboembolic disease, the positive predictive value of D-dimer remains low. In this study, 930 patients without intra-atrial thrombi had D-dimer levels >270 ng/ml. A D-dimer level above this threshold cannot be considered as predictive of intra-atrial thrombus. In clinical practice, the D-dimer assay should be of interest only in the absence of the other risk factors (hypertension, stroke, heart failure).
We did not collect the history of vascular disease (peripheral artery disease, coronary artery disease). The CHA2DS2VASc score was therefore not considered. The CHA2DS2VASc score (3) is the standard score for thromboembolic risk assessment in nonvalvular AF (17). Nevertheless, it is known that arterial disease and female sex are moderate risk factors for thromboembolic events (18). The use of CHA2DS2VASc score would likely not have provided better results than the CHADS2 score to rule out intra-atrial thrombus.
Two methods were used for plasma D-dimer level measurement, but studies showed their equivalent effectiveness (19). The absence of the numerical value of <270 ng/ml might have limited the possibility of establishing a more precise threshold to eliminate the presence of endocardial thrombus. However, all patients with intra-atrial thrombus were found to have a plasma D-dimer level >411 ng/ml.
Echocardiography data were not exhaustive. Data for LVEF were missing in some cases. It was a likely relevant variable to assess the risk of endocardial thrombus (6). We decided not to include this parameter in the multivariate analysis in order to favor more accessible clinical data in the final model.
Measurements of the left atrium were not collected although some studies showed a relationship between the size of the left atrium and the presence of atrial thrombus (5,20).
Effective anticoagulation was systematically considered for patients treated with direct oral anticoagulant agents without a specific questionnaire on treatment adherence.
Because positive and negative predictive values varied according to the prevalence of the outcome, these values might change if the score is applied in a different cohort.
Regardless of previous anticoagulation and duration of AF episodes, the proposed ATE score, which combined clinical items and plasma D-dimer level, could have ruled out the presence of atrial thrombus in our prospective multicenter study. This result needs to be confirmed in other independent sets of patients before conclusions can be drawn.
COMPETENCY IN MEDICAL KNOWLEDGE: Pending confirmation by a large multicenter study, the use of this new score could prevent use of imaging examinations, including TEE, before left atrial ablation procedures. This concept could also be extended to other clinical situations when a thrombus is suspected.
TRANSLATIONAL OUTLOOK: These results could encourage the research and development of new biomarkers of thrombosis.
The authors are grateful to Nikki Sabourin-Gibbs for her help in editing the manuscript.
The authors are particularly grateful to Virginie Laurençon for her involvement in carrying out this research.
Collaborators and members of the ATE study group: Alain Al Arnaout, Walid Amara, Mathieu Amelot, Clément Bars, Lucile Becoulet, Paul Bru, Philippe Chevalier, Jean-Philippe Darmon, Jean-Claude Deharo, Antoine Dompnier, Cécile Duplantier-Duchene, Fabrice Extramiana, Jean-Paul Faugier, Charles Guenancia, Jérôme Horvilleur, François Jourda, Gabriel Laurent, Nicolas Lellouche, Isabelle Magnin Poull, Olivier Piot, Antoine Roux, Yannick Saludas, Julien Seitz, and Jérôme Taieb.
This study was supported by a grant from Medtronic. Dr. de Chillou has been a consultant for Biosense Webster, Abbott, Boston Scientific, and Stereotaxis. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
All authors attest they are in compliance with human studies committees and animal welfare regulations of the authors’ institutions and Food and Drug Administration guidelines, including patient consent where appropriate. For more information, visit the JACC: Clinical Electrophysiology author instructions page.
- Abbreviations and Acronyms
- atrial fibrillation
- atrial thrombus exclusion
- international normalized ratio
- left ventricular ejection fraction
- odds ratio
- transesophageal echocardiography
- Received May 17, 2018.
- Revision received August 20, 2018.
- Accepted September 11, 2018.
- 2019 American College of Cardiology Foundation
- Calkins H.,
- Hindricks G.,
- Cappato R.,
- et al.
- Puwanant S.,
- Varr B.C.,
- Shrestha K.,
- et al.
- Yamashita E.,
- Takamatsu H.,
- Tada H.,
- et al.
- Zylla M.M.,
- Pohlmeier M.,
- Hess A.,
- et al.
- Yasaka M.,
- Miyatake K.,
- Mitani M.,
- et al.
- Heppell R.M.,
- Berkin K.E.,
- McLenachan J.M.,
- Davies J.A.
- Kirchhof P.,
- Benussi S.,
- Kotecha D.,
- et al.