Author + information
- Received July 8, 2009
- Revision received November 30, 2009
- Accepted December 7, 2009
- Published online May 18, 2010.
- ↵⁎Reprint requests and correspondence:
Dr. Tsuneaki Sadanaga, Department of Cardiology, Ueki Hospital, Iwano 285-29, Ueki, Kumamoto 861-0136, Japan
Objectives The aim of the present study was to evaluate whether elevated D-dimer levels can predict subsequent thromboembolic and cardiovascular events in patients with atrial fibrillation during oral anticoagulant therapy.
Background Atrial fibrillation is associated with hemostatic abnormalities even during oral anticoagulant therapy. D-dimer levels reflect a pro-thrombogenic state and thus might serve as a marker of thromboembolic and cardiovascular events.
Methods This was a single-center, prospective, observational study. Patients with atrial fibrillation (269 patients, age 74 ± 9 years, 160 paroxysmal atrial fibrillation) treated with warfarin (target prothrombin time–international normalized ratio: 1.5 to 3.0) were included. D-dimer levels were measured to assess the relationship of this parameter with subsequent thromboembolic and cardiovascular events. End points were thromboembolic events and combined cardiovascular events (thromboembolic events, cerebral hemorrhage, myocardial infarction, cardiovascular death).
Results D-dimer levels were elevated (≥0.5 μg/ml) in 63 (23%) patients. During an average follow-up period of 756 ± 221 days, 10 (1.8%/year) thromboembolic events (8 ischemic strokes, 1 transient ischemic attack, and 1 peripheral embolism) and 27 (4.8%/year) combined cardiovascular events (10 thromboembolisms, 9 deaths from heart failure, 3 sudden deaths, 2 myocardial infarctions, and 3 cerebral hemorrhages) occurred. Patients with elevated D-dimer levels experienced higher thromboembolic and combined cardiovascular events. Cox proportional hazard model revealed that elevated D-dimer levels were associated with both thromboembolic (p < 0.01, hazard ratio: 15.8; 95% confidence interval: 3.33 to 75.5) and combined cardiovascular (p < 0.01, hazard ratio: 7.64; 95% confidence interval: 3.42 to 17.1) events.
Conclusions D-dimer might be a useful marker of both thromboembolic and cardiovascular events in patients with atrial fibrillation during oral anticoagulant therapy.
Atrial fibrillation (AF) is associated with hemostatic abnormalities and increased risk of thromboembolic events. Effectiveness of oral anticoagulant therapy in reducing thromboembolic events is well documented (1–3); however, levels of coagulation markers are still elevated even during oral anticoagulant therapy in some patients (4–7). D-dimer originates from the formation and lysis of cross-linked fibrin and reflects activation of coagulation and fibrinolysis; thus, it can be used as an “in vivo” coagulation marker in addition to the standard “in vitro” marker of prothrombin time–international normalized ratio (PT-INR). The aim of the present study was to evaluate whether elevated D-dimer levels can predict subsequent thromboembolic and cardiovascular events in patients with AF during oral anticoagulant therapy.
This was a single-center, prospective, observational study. Patients eligible for our study were those undergoing AF treatment with warfarin. During entry periods (January 2006 to April 2007), 301 patients were initially screened, and finally 269 patients—152 (57%) male, 109 (41%) chronic AF (AF lasting more than 2 regular office visits), and 160 paroxysmal (or persistent) AF—were enrolled and followed until December 2008 (Fig. 1).The study protocol was approved by the institutional ethics committee, and informed written consent was obtained from all the patients.
Blood sampling and assays
All assays were performed in the laboratory of our institution within 2 h of blood sampling. For the quantitative measurement of D-dimers, a latex-enhanced photometric immunoassay (LPIA, Mitsubishi Chemical Medience Corporation, Tokyo, Japan) was used with automatic analyzer (LPIA-S500). The detection limit of this assay was 0.3 μg/ml. Intra-assay and interassay variability (coefficient of variability) was 2.2% and 4.4%, respectively. D-dimer was measured at the time of enrollment and repeated 2 or 3 times during entry period in some patients. The PT-INR was measured every 1 to 2 month throughout the observational periods. Target PT-INR was set at 1.5 to 3.0.
The end points were thromboembolic events (ischemic stroke, transient ischemic attack, peripheral embolism) and combined cardiovascular events, which included thromboembolic events, cerebral hemorrhage, myocardial infarction, and cardiovascular deaths. Major bleeding was defined as gastrointestinal bleeding requiring transfusion and intracranial bleeding, including cerebral hemorrhage and subdural hematoma.
Data were presented as the mean ± SD or percentages, as appropriate. Event frequencies were compared with the chi-square test. Other comparisons between 2 groups of data were made with unpaired Student ttest or Mann-Whitney Utest, as appropriate. Correlation analysis was performed by Spearman's rank correlation. The optimal D-dimer cutoff point was evaluated by receiver operator characteristic curve. Logistic regression models were applied to assess the determinants of high D-dimer levels. The outcomes were displayed with Kaplan-Meier event-free curve and compared with the use of log-rank tests. The prognostic values of D-dimer and clinical variables were analyzed with Cox-proportional hazard models. A value of p < 0.05 was accepted as statistically significant. The statistical software package SPSS (version 11.0, SPSS, Inc., Chicago, Illinois) and StatView (version 5.0 StatView, Berkeley, California) were used for analyses.
Baseline clinical characteristics
Baseline heart disease included 44 (16%) hypertensive heart disease, 37 (14%) coronary artery disease, 32 (12%) valvular heart disease, 17 (6.3%) hypertrophic cardiomyopathy, and 10 (3.7%) dilated cardiomyopathy. No apparent heart diseases were found in 131 (49%) patients. Other clinical risk factors are shown in Table 1.
At the time of enrollment, PT-INR levels were 1.93 ± 0.53 (1.2 to 6.9, median 1.82). The PT-INR levels were less optimal (<1.5) in 32 (12%), exceeded (>3.0) in 7 (2.6%), and within therapeutic range (1.5 ≤ PT-INR ≤3.0) in the remaining 230 (86%) patients (Fig. 2).Linear regression analysis revealed no relationship between D-dimer levels and PT-INR (p = 0.17, ρ = 0.084 by Spearman's rank correlation), consistent with the previous report (5). At the end of follow-up, PT-INR levels were 1.94 ± 0.60 (1.0 to 7.0, median 1.86) and were within therapeutic range in 82% of the patients.
Reproducibility of the D-dimer levels
Serial measurements of D-dimer levels to assess reproducibility with similar PT-INR levels (1.86 ± 038, median 1.82, vs. 1.97 ± 0.42, median 1.90, p = 0.055) were available in 105 patients. There was a high correlation between the 2 D-dimer measurements (p < 0.01, ρ = 0.86, by Spearman's rank correlation). The difference between these 2 D-dimer measurements was small (0.21 ± 0.42 μg/ml, median 0.01 μg/ml).
End points and major bleedings
Approximately 80% of the patients were followed in our institution, and the rest were followed by attending physicians who referred the patients to us. During a follow-up time of 756 ± 221 (1 to 1,091; median 740) days, 4 patients discontinued taking warfarin, and 6 were lost to follow-up after 1 year. These 10 patients were included in the final analysis (Fig. 1). There were 10 thromboembolic events (1.8%/year): 8 ischemic strokes, 1 transient ischemic attack, and 1 peripheral embolism. There were 27 cardiovascular events (4.8%/year): 10 thromboembolisms, 9 deaths from heart failure, 3 sudden deaths, 2 myocardial infarctions, and 3 cerebral hemorrhages. There were 9 (1.6%/year) major bleedings: 3 cerebral hemorrhages, 1 subdural hematoma, and 5 gastrointestinal bleedings requiring transfusion (Table 1).
Cutoff values and determinants of D-dimer levels
In the current study, the upper quartile of the D-dimer value was 0.48 μg/ml. With the receiver operator characteristic curve analysis (Fig. 3),the optimum cut point, identified as the point closest to upper left corner, yielded optimal cutoff value of approximately 0.5 μg/ml to detect both thromboembolic events and cardiovascular events. Previously, most studies used D-dimer cutoff values of 0.5 to 1.0 μg/ml for excluding pulmonary embolism (8) and acute aortic dissection (9). On the basis of these findings, the cutoff value of D-dimer level for the current study was set at 0.5 μg/ml. Median values in patients with high (≥0.5 μg/ml) and low (<0. 5 μg/ml) D-dimer values were 0.86 and 0 μg/ml, respectively. The PT-INR levels in patients with high and low D-dimer levels were not different (1.99 ± 0.74, median 1.82, vs. 1.92 ± 0.45, median 1.82, p = 0.83 by Mann-Whitney Utest). Table 1shows characteristics of patients with high (≥0.5 μg/ml) D-dimer levels. Patients with high D-dimer levels had higher prevalence of congestive heart failure (CHF), older age (≥75 years), and history of stroke. Table 2shows determinants of high D-dimer levels by logistic regression analysis. Univariate analysis showed that CHF, age ≥75 years, and history of stroke were the significant determinants of high D-dimer levels. Multivariate analysis including these 3 parameters, and PT-INR values revealed that these 3 parameters were the independent predictors of high D-dimer levels.
High D-dimer levels to predict thromboembolic and cardiovascular events
There were 8 thromboembolic events observed among patients with baseline D-dimer levels ≥0.5 μg/ml as compared with 2 such events in those with baseline D-dimer levels <0.5 μg/ml. Kaplan-Meier thromboembolic event-free curves for D-dimer level showed that elevated D-dimer level was the significant predictor of subsequent thromboembolic events (Fig. 4).Statistical significance of separation between the 2 groups was achieved at 600 days. With regard to PT-INR levels at the time of events, 5 (50%) of the 10 thromboembolic events occurred when the PT-INR levels were <1.5, and only 1 thromboembolic event took place when the PT-INR level exceeded 2.0. Table 3shows risk factors for thromboembolic events by Cox proportional hazard analysis. Univariate analysis showed that high D-dimer levels, CHF, and history of stroke were associated with increased thromboembolic events. High D-dimer level was the significant determinant after adjustment of baseline PT-INR (p < 0.01, hazard ratio: 15.8; 95% confidence interval [CI]: 3.33 to 75.5). Fully-adjusted models were not possible, given the low number of events.
There were 18 combined cardiovascular events observed among patients with baseline D-dimer levels ≥0.5 μg/ml, as compared with 9 such events in those with low levels. Kaplan-Meier curves showed that high D-dimer level was the significant predictor of subsequent cardiovascular events (Fig. 5).Statistical significance of separation between the 2 groups was achieved at 600 days. Table 4shows risk factors for combined cardiovascular events by Cox proportional hazard analysis. High D-dimer level, CHF, age ≥75 years, diabetes mellitus, and history of stroke were associated with increased cardiovascular events. High D-dimer level was also a significant determinant after adjustment of baseline PT-INR (p < 0.01, hazard ratio: 7.64; 95% CI: 3.42 to 17.1). Fully-adjusted models were not performed, given the low number of combined cardiovascular events.
With regard to bleeding complications, 4 (44%) of the 9 events occurred when the PT-INR levels were >3.0, but 3 (33%) events occurred when the PT-INR levels were <2.0 (i.e., 1.25, 1.55, 1.86). There were 8 bleeding complications observed among patients with baseline D-dimer levels ≥0.5 μg/ml, as compared with 1 such event in those with low levels. Cox proportional analysis revealed that high D-dimer levels were significant correlates of bleeding complications (p < 0.01, hazard ratio: 29.3; 95% CI: 3.65 to 235) after adjustment of baseline PT-INR.
Paroxysmal versus chronic AF
Prevalence of CHF (52% vs. 36%, p < 0.01) and history of stroke (17% vs. 8%, p = 0.021) were higher in chronic AF than those in paroxysmal AF. However, prevalence of high D-dimer levels (28% vs. 21%, p = 0.19) were comparable between these 2 groups.
It is well known that D-dimer levels are increased in patients with AF, and introduction of warfarin decreases D-dimer levels (4). However, D-dimer levels are still elevated even during oral anticoagulant therapy in some patients (4–7). We have found that they were the high-risk patients who subsequently suffer thromboembolic and cardiovascular events.
Clinical significance of D-dimer measurement
Circulating concentration of fibrin D-dimer levels reflect the extent of fibrin turnover. Highly elevated D-dimer values occur in various disorders in which the coagulation system is excessively activated, such as acute venous thromboembolism (8) and acute aortic dissection (9). It has been suggested that modestly elevated circulating D-dimer values reflect minor increases in blood coagulation, thrombin formation, and turnover of cross-linked intravascular fibrin. These increases might be related to thromboembolic (6) and cardiovascular (7) events. This study was consistent with these results. Interestingly, elevated D-dimer levels also predicted bleeding complications in this study. The reasons were not known, but it can be speculated that the fibrinolytic activation of which D-dimer is a marker might itself predispose to bleeding.
First, this is a single-center study, and selection bias is a major concern. We have tried to include every patient with AF who visited our office; however, it is difficult to start anticoagulation therapy in elderly patients with dementia, for fear of overdoses of warfarin, frequent falls, and subsequent bleeding complications. Age ≥75 years was not a risk factor for thromboembolic events partly due to the exclusion of these high-risk patients. Second, target PT-INR in this study was set below the recommended guideline of Western countries (PT-INR: 2.0 to 3.0). The Japanese Secondary Prevention Trial (10) suggested that there might be a racial difference in the optimal anticoagulation intensity for the prevention of ischemic stroke in patients with AF. Japanese guidelines (11) recommended target PT-INR should be 1.6 to 2.6 when the patients were older than 70 years. Anticoagulant therapies should always be considered along with the balance between antithrombotic effect and risk of bleeding. The annual incidence of thromboembolic events (1.8%/year) and bleeding complications (1.6%/year) were similar in this study, suggesting that target PT-INR was probably a reasonable range for the Japanese population. However, it might not be applicable to Western populations. Third, it is speculated that D-dimer levels might be increased at the time of events, but it was measured only at the entry periods, mainly due to budget constraints. Therefore, D-dimer levels at the time of events were unknown. However, reproducibility of D-dimer levels was acceptable in our study as well as in a previous study (5); thus we believe that D-dimer levels at baseline, if anticoagulated in a proper PT-INR range, are an important marker of thromboembolic and cardiovascular events. Finally and most importantly, the number of patients and events in this study was small; therefore further larger-scale, multicenter studies are needed to confirm these findings.
Elevation of D-dimer levels despite proper anticoagulant therapy can predict thromboembolic and cardiovascular events in patients with AF. Whether thromboembolic and cardiovascular events in patients with AF can be prevented by increasing anticoagulation intensity with the help of D-dimer levels remains to be elucidated.
For the acknowledgments, please see the online version of this article.
Evidence That D-Dimer Levels Predict Subsequent Thromboembolic and Cardiovascular Events in Patients With Atrial Fibrillation During Oral Anticoagulant Therapy
Continuing medical education (CME) is available for this article.
- Abbreviations and Acronyms
- atrial fibrillation
- congestive heart failure
- confidence interval
- prothrombin time–international normalized ratio
- Received July 8, 2009.
- Revision received November 30, 2009.
- Accepted December 7, 2009.
- American College of Cardiology Foundation
- Stroke Prevention in Atrial Fibrillation Investigators
- Lip G.Y.H.,
- Lowe G.D.O.,
- Rumley A.,
- Dunn F.G.
- Sodeck G.,
- Domanovits H.,
- Schillinger M.,
- et al.
- Yamaguchi T.,
- Japanese Nonvalvular Atrial Fibrillation-Embolism Secondary Prevention Cooperative Study Group
- Ogawa S.,
- Aizawa F.,
- Atarashi H.,
- et al.