Seminars in Vascular Medicine 2005; 05(4): 311-314
DOI: 10.1055/s-2005-922475

Copyright © 2005 by Thieme Medical Publishers, Inc., 333 Seventh Avenue, New York, NY 10001, USA.

Fibrin D-Dimer Testing for Venous and Arterial Thrombotic Disease

Jan Jacques Michiels1 , 2  Editor in Chief , Gualtiero Palareti3 , 4  Guest Editor , Philippe de Moerloose3 , 4  Guest Editor 
  • 1Goodheart Institute & Foundation, Hematology Hemostasis Thrombosis Science Center, Rotterdam, The Netherlands
  • 2Department of Hematology, University Hospital Antwerp, Antwerp, Belgium
  • 3Department of Angiology & Blood Coagulation “Marino Golinelli,” University Hospital S. Orsola-Malpighi, Bologna, Italy
  • 4Hemostasis Unit, University Hospital of Geneva, Geneva, Switzerland
Further Information

Publication History

Publication Date:
22 November 2005 (online)

The present issue of Seminars in Vascular Medicine assembles a series of articles that nicely elucidate the current knowledge on the use of fibrin D-dimer testing for venous and arterial thrombotic disease in vascular medicine.

In the first article Dempfle provides relevant background information on D-dimer assays used for thrombosis exclusion. D-dimer antigen assays are heterogeneous concerning epitope specificity and calibration. Dimerized D-domains (D-dimers), D-trimers, and D-tetramers are needed for the branching of fibrin fibrils to cross-linked fibrin during clot formation. Fragment D-dimer is a specific indicator for plasmin proteolysis of cross-linked fibrin as the essential part of a blood clot. Proteolysis of cross-linked fibrin generates various fragments, which contain dimerized D-domains. Proteolytic fragment of cross-linked fibrin also include fibrin fragments formed by elastase or other proteolytic enzymes. The monoclonal antibodies used in D-dimer antigen assays react with epitopes of fibrin compounds containing dimerized D-domains. Three main types of immune assays are available for the measurement of D-dimers. Qualitative latex agglutination assays are insufficiently sensitive for use in the exclusion of venous thrombosis. The whole blood red cell agglutination uses bivalent antibodies against D-dimer antigen and red blood cells. In the presence of sufficient D-dimer antigen the red cells agglutinate. Automated latex-enhanced light-scattering assays use latex particles coated with monoclonal antibodies against D-dimer antigen and are performed on various automatic laboratory analyzers with photometric detection systems. An automated enzyme-linked immunoassay (ELISA) has been developed for the VIDAS analyzer. Fibrin derivatives detected by D-dimer antigen assays are highly heterogeneous. Manufacturers of D-dimer assays use either purified D-dimer, cross-linked fibrin degradation products, or pooled plasma for calibration. The concentration label is based on either the amount of D-dimer present in the solution or the amount of fibrinogen used for the preparation of fibrin degradation products and expressed as D-dimer units (concentration) or fibrinogen equivalent nuts (FEU). Standardization and harmonization of D-dimer assays using standardized preparation for D-dimer concentration are warranted.

In the second article Meijer and Kluft discuss the current status of standardization and harmonization of the available quantitative D-dimer methods. The different quantitative D-dimer tests can differ significantly, which is mainly caused by the variety of fibrin degradation products in plasma, the specificity of antibodies against D-dimer antigen, and the calibrator used in the D-dimer assay. Depending on the stage of degradation of cross linked fibrin in various diseases like diffuse intravascular coagulation or venous thrombosis, the patient samples contain various amounts of large and smaller fibrin degradation and D-dimer products. Depending on the specificity of antibodies used in different D-dimer assays, the mixtures of smaller and larger fibrin D-dimer degradation products are recognized by the different methods with different reactivity. These variations hamper the exchange and comparability of text results of the various quantitative D-dimer methods. Within the context of the scientific standardization committee (SSC) on fibrinogen of the International Society on Thrombosis and Haemostasis (ISTH), standardization of quantitative D-dimer assays including latex immune assays (LIAs) and ELISA was not possible, but harmonization of D-dimer assays using a pooled patient sample seems to be feasible. The principle of harmonization of D-dimer assays is based on the use of a method-specific conversion factor, which is the ratio between the consensus value of all methods included and the value of a particular method for pooled plasma. The authors developed a harmonization model for conversion of the D-dimer results over a wide range by transformation of the measured method-specific D-dimer result to the harmonized D-dimer concentration using the slope and intercept of the method-specific regression line and the reference line regression line. One limitation of the harmonization model is the fact that the reference regression line is based on the results of the commonly used D-dimer methods in the study. Initiatives are ongoing to improve the harmonization of quantitative test results of different D-dimer methods.

In the next article Freyburg and Labrouche compare the different D-dimer assays for optimization of their proper use for thrombosis exclusion despite the differences caused by the lack of harmonization of D-dimer results. The ELISA VIDAS D-dimer, the SimpliRed, and the more recent turbidimetric LIA D-dimer assays are validated in prospective outcome studies of patients after a 3-month follow-up as end point. Systematic reviews or meta-analyses pooling the results of D-dimer methods for thrombosis exclusion are very confusing and terminated in rather discordant conclusions. The performance of each D-dimer antigen assay depends on the technical principle; whether ELISA, quantitative turbidimetric LIAs, or qualitative agglutination assays; the antibody specificity; and the calibration used. Objective quantitative measurements are obviously preferable to qualitative ones, which are subject to interobserver variability. The quantitative latex D-dimer assays, which have been selected by many manufacturers for their convenience in routine use, exhibit variable performances. At the exclusion threshold range, confidence variable values are always higher for the turbidimetric latex D-dimer assays than those observed with the ELISA D-dimer assays. The anti-D-dimer specificity of the antibodies used is probably not a major factor for the differences observed between D-dimer assay performances. In the ELISA method, the use of a second tagged sandwiching antibody greatly enhances sensitivity, reproducibility, and accuracy.

Two major methods exist to prepare a D-dimer calibrator: (1) a clot lysed by plasmin that gives FEUs and (2) purified D-dimer fragments that give D-dimer units. In spite of this unit heterogeneity, most manufacturers define the thrombosis threshold around a value corresponding with 500 ug/mL when measured with a reference ELISA (Asserachrom® from Stago).

The only objective criterion for thrombosis exclusion is a negative predictive value (NPV) of > 99% to near 100%. The specificity is much less important and critically dependent on the sensitivity of the D-dimer test and prevalence of venous thromboembolism (VTE) in the patients population studied. Both ELISA and turbidimetric D-dimer assays provide good results for thrombosis exclusion. ELISA D-dimer assays in general dominate in terms of sensitivity, the comparative ranking among the quantitative D-dimer assays, due to the lack of precision of the turbidimetric latex D-dimer assays at the exclusion threshold range. The rapid ELISA is most sensitive (close to 100%) at the cutoff point between normal and abnormal for clinically relevant VTE. The rapid ELISA VIDAS D-dimer has been approved by the U.S. Food and Drug Administration and may stand alone for thrombosis exclusion including both deep vein thrombosis (DVT) and pulmonary embolism (PE). The restricted use of low-sensitive latex assays (SimpliRed) to patients with a low clinical score is related to safety in terms of statistics. The low likelihood of a negative latex D-dimer (0.10) multiplied by low pretest odds in patients with a low clinical score (0.10) results statistically in very low posttest odds and probability (0.01).

In the fourth article, Wells presents the Canadian experiences and views on how clinical assessment, qualitative D-dimer, and noninvasive imaging test can be employed for the investigation of patients with suspected DVT or PE. After a negative compression ultrasonography (CUS) for proximal DVT the posttest probability of DVT during 3 months follow-up is 2 to 3%. A second CUS within 1 week (serial CUS) will reduce the posttest probability of DVT from 2 to 3% to < 1%, but this is not cost-effective. A strategy that employs a combination of clinical score assessment, D-dimer, and CUS is recommended. The clinical prediction rule of Wells incorporates signs, symptoms, and risk factors, which can be applied to categorize patients as low, moderate, or high probability for DVT. The use of the prediction rule is very helpful as over 80 to 90% referred for diagnostic imaging do not have DVT. After a negative CUS, the prevalence of DVT during 3 months follow-up is 2 to 3%. After a negative CUS, DVT is safely excluded in patients by a low clinical score or a negative SimpliRed test. The Canadian studies have shown that, when a qualitative D-dimer test (SimpliRed) is used, the pretest clinical probability should be rather low (less than 5%) to reliable exclude the diagnosis of DVT without the need of CUS. The sensitivity of the rapid ELISA D-dimer assay can exclude DVT without the need for diagnostic tests in patients with moderate or low clinical score, but those who have persistent symptoms suspicious for DVT should undergo CUS testing to differentiate between DVT and alternative diagnosis.

Accepted methods to diagnose PE are pulmonary angiography or spiral computed tomography (SCT). Accepted methods to rule out PE are a normal perfusion scan, a normal pulmonary angiogram, or a normal SCT. The positive predictive value of a high-probability lung scan for PE is 85 to 90%, indicating that 10 to 15% of patients with a high-probability lung scan do not have PE. The prevalence of PE in patients with suspected PE and a nondiagnostic lung scan is 20 to 30%, indicating the need of further testing. A strategy that employs the combined use of clinical score assessment, D-dimer, and noninvasive imaging, preferentially SCT, is recommended. Clinical score assessment in patients with suspected PE is cumbersome but feasible when performed by or under the strict supervision of experienced clinicians. The original and the shortened clinical prediction rule of Wells incorporates signs, symptoms, and risk factors, which can be applied to categorize patients as low (score ≤ 2), moderate (score 2 to 6), or high (score ≥ 6) probability for PE. The Canadian studies have shown that when qualitative D-dimer test (SimpliRed) is used, the pretest clinical probability should be low (prevalence of PE < 5%) to reliably exclude the diagnosis of PE without the need of lung scanning or SCT. The shortened prediction rule of Wells can be used to score patients as “PE unlikely” (score ≤ 4) or “PE likely” (score > 4). In patients who are “PE unlikely” the physician can avoid the need for diagnostic imaging when the rapid ELISA VIDAS D-dimer is negative.

In the fifth article, Michiels et al review the European experiences and views on how to screen for DVT and PE in outpatients with suspected DVT or PE by the sequential use of the clinical score of Wells, D-dimer, and noninvasive imaging techniques. The requirement for a safe diagnostic strategy should be based on an overall posttest incidence of VTE of < 1% during 3-month follow-up. The prevalence of DVT in seven studies ranged from 3.2 to 14% in the low clinical score group of Wells. This wide variation is surely due to the “score of minus 2” for the subjective evaluation of an alternative diagnosis as more or less likely than DVT. Four clinical outcome studies showed that a negative SimpliRed test reduces the prevalence of DVT in the low clinical score group from 4.6 to 8% to 1.6 to 2%, consistent with a NPV of 98 to 98.6%, indicating the need of CUS testing. A negative first CUS (prevalence of DVT 2 to 3%) followed by a negative SimpliRed test or an ELISA VIDAS result of < 1000 ng/mL safely excluded DVT to < 1% (NPV > 99%) without the need to repeat CUS. As compared with venography for DVT and pulmonary angiography for PE, a normal rapid ELISA D-dimer is equally safe (sensitivity > 99 to 100%) to exclude VTE in outpatients with suspected VTE and a low to moderate clinical score (< 6 according to Wells) without the need of noninvasive imaging. However, clinicians should always be aware that no laboratory test is 100% sensitive. Therefore, every patient with a negative ELISA VIDAS D-dimer but with persistent or recurrent symptoms suspicious of VTE is a candidate for noninvasive imaging (CUS or SCT) to exclude or detect VTE or an alternative diagnosis.

The interobserver agreement of an angiographically documented subsegmental PE is only 60%. Consequently, pulmonary angiography remains the gold standard for segmental but not for subsegmental PE. Recent prospective outcome studies showed that a negative SCT will exclude PE with an NPV of 98.2 to > 99% during 3-month follow-up. The NPVs of a normal SCT critically depends on the experience and skill of the radiologist. Only a small number of patients with nondiagnostic SCT (5 to 10%) are candidates for pulmonary angiography or lung scanning. The sequential use of a rapid ELISA VIDAS D-dimer followed by CUS will reduce the number of SCTs by 40 to 50% and is therefore cost-effective.

In the next article, Cosmi and Palareti focus on risk factors and prevention of late recurrences after long-term treatment of DVT. Anticoagulant treatment with vitamin K antagonists (VKA) is recommended for at least 3 months after a first episode of proximal DVT secondary to a transient (reversible) risk factor and for at least 6 to 12 months for a first episode of idiopathic DVT. A normal D-dimer level has a high NPV for VTE recurrence when performed 1 month after discontinuation of VKA treatment. Increased D-dimer levels 1 month after VKA withdrawal is an independent risk factor for recurrent VTE, although residual venous thrombosis on CUS is of minor importance for risk prediction of recurrent DVT. Poor-quality VKA treatment after a first idiopathic DVT is associated with increased long-term risk recurrence after discontinuation of VKA. A prospective intervention study showed that increased D-dimer levels 1 month after discontinuation of VKA is associated with a significantly higher risk of recurrence and deserve prolonged VKA treatment.

In the seventh article, Samama et al address the question whether D-dimer levels are of clinical, biological, or prognostic value in patients with constitutional thrombophilia. In constitutional thrombophilia in the absence of comorbidity or of a venous thromboembolic episode, the D-dimer levels are usually normal. D-dimer testing for thrombosis exclusion has a high NPV for DVT and PE in hereditary thrombophilia similar to nonthrombophilic patients when used in combination with clinical score assessment according to Wells. A low concentration of D-dimer after discontinuation of anticoagulant treatment for thrombotic event has a strong NPV in the absence and presence of a hereditary thrombophilic factor. The role of inherited thrombophilia in potentiating the risk of VTE in patients undergoing surgery has not been well documented mainly because most patients will receive prophylactic treatment with a low-molecular-weight heparin. Anticoagulation with VKA is associated with a decrease of D-dimer levels to the same extent in patient with or without constitutional thrombophilia.

In the eighth article, Eichinger reports on D-dimer testing in pregnancy. Pregnancy is a major risk factor for venous thrombosis. Uncomplicated pregnancy is accompanied with a substantial hemostatic system activation as indicated by increased markers of coagulation activators like prothrombin fragment F1 + 2 and D-dimer. D-dimer has very little value for the diagnosis of venous thrombosis during pregnancy and the postpartum period. D-dimer levels are significantly higher throughout pregnancy in healthy women heterozygous for factor V Leiden as compared with pregnant noncarriers of the mutation. Because reference values of D-dimer during normal pregnancy are not validated, the sensitivity and specificity of D-dimer in diagnosing and predicting pregnancy-related complications is reduced.

In the ninth article, Squizzato and Ageno review the literature on D-dimer testing for the diagnosis of acute stroke and cerebral sinus and venous thrombosis (CSVT). D-dimer levels are elevated for up to 1 or 2 months after acute ischemic stroke and not significantly increased after a transient ischemic attack (TIA). D-dimer levels are significantly higher in the group of patients with cardioembolic stroke or TIA related to atrial fibrillation than in the group of atherothrombotic patients or those with a lacunar stroke. The accuracy of D-dimer testing is not high enough to diagnose or exclude stroke or TIA. When a diagnosis of stroke has been reached, the D-dimer level may be used for risk stratification of treatment. Conflicting data do not allow one to draw conclusions on the prognostic value of D-dimer in stroke patients. D-dimer is not an independent long-term prognostic factor for mortality and new cerebrovascular events in stroke patients. Exclusion of DVT after stroke at a certain cutoff level for fibrin D-dimer is cumbersome, because stroke itself is associated with increased D-dimer levels.

CSVT is rare, with high mortality rates between 5 and 30%, and has been associated with several causes and risk factors such as thrombophilia, oral contraceptives, pregnancy, and puerperium. A normal sensitive ELISA D-dimer test very likely will exclude CSVT with an NPV > 99%.

In the last article, Lowe extensively reviews the epidemiological studies on fibrin D-dimer and cardiovascular disease. Activated coagulation and in vivo fibrin formation and lysis play a role in arterial, intracardiac, and VTE. Epidemiological studies have observed an inverse correlation between regular physical activity and D-dimer, significant correlation of D-dimer with blood pressure, and no correlation with cholesterol, triglyceride, apolipoproteins, insulin, homocysteine, hematocrit, white cell count, and platelet count. D-dimer levels correlate with C-reactive protein, fibrinogen, and viscosity. D-dimer levels are higher in patients with stable cardiovascular disease, in persons with previous myocardial infarction, and in patients with unstable angina or myocardial infarction. Following myocardial infarction left ventricular dysfunction, aneurysm, and heart failure may develop, and each of these conditions is associated with increased D-dimer levels. Patients with atrial fibrillation and left atrial thrombus have higher D-dimer levels than patients without complications. The potential use of D-dimers to select high-risk patients with atrial fibrillation for warfarin or aspirin requires further studies. Increased D-dimer levels in persons with atrial fibrillation are associated with risk of stroke and cognitive decline. D-dimer levels are raised in patients with acute stroke with higher levels in cardiovascular or atherothrombotic stroke as compared with lacunar stroke and TIAs. A single D-dimer measurement for stroke severity at admission and a D-dimer level at day 9 will identify a substantial proportion of patients with proximal DVT. The potential use of D-dimer in the selection of stroke patients at high risk of progression and DVT for anticoagulant therapy requires further studies.

D-dimer levels are elevated in patients with peripheral artery disease and are correlated with clinical and angiographic severity. Higher levels are seen in patients with abdominal aneurysms.

Patients with previous VTE have increased D-dimer levels, when studied at least 6 months after the acute event following discontinuation of anticoagulation. Baseline D-dimer levels are associated with risk of postoperative DVT, but D-dimer was associated with other risk factors for DVT (age, obesity, varicose veins), and therefore D-dimer was not a significant predictor of postoperative DVT after adjustment for these clinical variables.

We sincerely wish to thank all the authors for their valuable contributions and their enthusiasm to critically evaluate the role of D-dimer testing for venous and arterial thrombotic disease. We very much hope that readers will find the content of the articles as informative and useful for implementation in daily clinical practice as we have intended.