Vol. 13 Issue 10
Efficacy of D-Dimer
Common questions regarding the accuracy and sensitivity of D-dimer assays are answered.
Venous thromboembolism (VTE) is a major health concern with significant morbidity and mortality when left untreated. VTE is a common medical condition with an annual incidence in the general population of about 0.1 percent.1 This incidence, however, rises exponentially with age to about 0.5 percent in persons at 80 years of age.1
Clinical manifestations are deep vein thrombosis (DVT) and pulmonary embolism (PE). Approximately one-third of patients with symptomatic VTE manifest PE, whereas two-thirds manifest DVT.
Mortality rates within one month of diagnosis are 6 percent and 12 percent in cases of DVT and PE, respectively.1 After anticoagulation, there is a risk of recurrence of 3 percent to 10 percent per year.2 About 10 percent of patients with DVT develop severe post-thrombotic syndrome, a chronic condition characterized by pain, swelling and skin ulceration of the leg.
DVT refers to an occluding thrombus in the venous circulation of the leg. This is less serious in distal calf veins and most serious in the proximal veins above the knee. Approximately 25 percent of untreated calf thrombi extend into proximal veins, mostly within a week of presentation.2 When the thrombus is dislodged, it may travel into the pulmonary circulation causing PE, the most common fatal aspect of VTE. Most fatal emboli probably arise from proximal-vein clots.
DVT and PE are manifestations of the same diseaseVTE. Asymptomatic PE is detected in about 50 percent of patients with documented DVT and asymptomatic DVT is found in about 70 percent of patients with confirmed symptomatic PE.2
VTE is a multifactorial disease and often results from a combination of risk factors, including environmental and genetic sources (Table 1).3 Depending on the type and combination of risk factors, prophylactic anticoagulant treatment is indicated. An idiopathic condition, without an identifiable risk factor, is present in 25 percent to 50 percent of first-time VTE cases.1 Once VTE is diagnosed, current practice is to provide an initial short treatment with intravenous unfractionated heparin or subcutaneous low-molecular-weight heparin, followed by an oral anticoagulant drug (e.g., warfarin) for several months to prevent embolization and recurrences.4 The optimum dosage and duration of oral anticoagulant therapy after a first event, however, is largely dictated by underlying risk factors.3,4
How is VTE Diagnosed?
Each stage of VTE–distal DVT, proximal DVT and PE–may or may not be associated with clinical symptoms such as pain and swelling of the affected leg and, in the case of PE, dyspnea, pleuritic pain and/or syncope. The diagnosis of VTE is not reliable when based on clinical symptoms alone and, therefore, clinical suspicion of VTE must be confirmed with objective tests.5
Venography6 or phlebography and pulmonary angiography7 are the gold standard methods to diagnose DVT and PE, respectively. However, these invasive methods are not universally available and require considerable expertise. Furthermore, they are costly and not without hazards. Alternative non-invasive diagnostic techniques have been introduced, including compression ultrasonography (CUS) for suspected DVT and ventilation-perfusion (V/Q) lung scintigraphy or spiral computed tomography (CT) for suspected PE.5
These non-invasive methods still have limitations in terms of availability, costs and effectiveness. CUS is widely used, but this method has a low sensitivity of isolated distal DVT. This means that suspected patients with an initial negative CUS at presentation have to be re-examined once or twice over a seven-day period.8 This is costly, time consuming and bears the risk of missing outpatients at follow up. V/Q lung scanning has a low diagnostic yield and is non-conclusive in 50 percent to 70 percent of cases.5
Furthermore, there has been a trend during the past two decades increasingly to refer patients to diagnostic centers in spite of a low clinical suspicion, and consequently the actual prevalence of VTE among suspected outpatients is only about 20 percent to 25 percent.5,9 This further underscores the importance of rapid and cost-efficient methods to safely exclude VTE in the majority of patients.
In recent years, clinical probability assessment and D-dimer testing have been evaluated as additional tools in sequential non-invasive diagnostic approaches to suspected DVT and PE.5,9 The assessment of pre-test probability (PTP), with categorization into low-, intermediate- and high-risk groups, is an essential initial step in the diagnostic management of patients with suspected VTE.10 PTP can be assessed empirically or by using decision rules or scoring systems. The best-known examples of the latter are the Wells scores for suspected DVT11 and PE12 and Geneva score for suspected PE.13
Several outcome studies have now shown that the approach of combining PTP with a modern D-dimer test can safely exclude disease in up to half of outpatients with suspected VTE, without the need for additional diagnostic investigations.14,15
What is D-dimer?
Fibrin, the main component of a thrombus, is formed by the activation of the coagulation system, whereas activation of the fibrinolytic system leads to the dissolution of the fibrin clot (Fig. 1). Under normal physiologic conditions there is a balance between these two opposing processes. Measurement of molecular markers of these two systems may reveal disturbances in the hemostatic balance and, therefore, may be an aid in the diagnosis of bleeding or thrombotic disorders.
Lysis of the fibrin clot results in the generation of soluble fibrin degradation products, which are a heterogeneous group of molecules characterized by the presence of multiple cross-linked D-domains (D-dimer; Fig. 1).16 D-dimer is a direct biomarker of fibrinolysis (plasmin generation) and an indirect marker of coagulation (thrombin generation). Increased levels of D-dimer occur in a variety of conditions where the coagulation system is activated, including surgery, trauma, infection, inflammation, pregnancy, disseminated intravascular coagulation and thrombosis.17 The half-life is approximately eight hours.
Why Consider D-dimer in the Diagnosis of VTE?
In patient management, an assay is clinically useful:
1. when it can confirm or rule in the diagnosis when the disease is present or
2. when it can rule out the diagnosis in suspected patients who do not have the disease.
Thrombi are large in symptomatic patients with VTE and the simultaneous activation of fibrinolysis results in raised levels of D-dimer. Activation of coagulation and subsequent fibrinolysis also occurs in a variety of other disorders. Consequently, D-dimer is not a specific marker for the presence of venous thrombi since its level also can be increased due to co-morbid conditions such as malignancy or recent surgery.18 D-dimer also is significantly elevated in the elderly19 and during pregnancy.20 Therefore, a positive D-dimer test will not confirm VTE.21
When assayed with a highly sensitive test, D-dimer is raised in virtually all patients presenting with acute VTE. A patient with a normal D-dimer concentration (i.e., below a predefined cut-off level) is highly unlikely to have DVT or PE. It has been well documented that the diagnostic utility of the D-dimer test lies in its ability to safely exclude the presence of VTE due to its high negative predictive value (NPV).22
How Should a D-dimer Assay Be Selected?
Since D-dimer is not a standardized analyte and the results depend on the assay being used, clinicians should know the diagnostic performance of the test used in their own institution.22
Digestion of a fibrin clot by plasmin results in a very heterogeneous population of fibrin degradation products with a variable number of D-dimer motifs.16 Hence, the size distribution of degradation products in plasma varies from person to person and from disease to disease. Furthermore, different monoclonals react differently with the available D-dimer epitopes. This, combined with the use of different types of calibrators in commercial kits and different units of expression, makes it impossible to compare test results between assays in absolute terms.23
Usually, D-dimer assays correlate, but results are not identical.24 This means that each D-dimer assay has its own typical normal range and needs its own validation before it can be introduced into clinical practice for the diagnostic approach of patients with suspected VTE.22
Validation of any procedure for the diagnosis of VTE includes an assessment of technical, operational and clinical criteria according to a three-step approach.5,25 How this works out for the various D-dimer assays is discussed below. The ideal characteristics of a D-dimer assay are listed in Table 2.
1. Technical and operational qualification
Three main detection principles can be distinguished for the measurement of D-dimer (Fig. 2): Elisa, latex particle agglutination and whole-blood agglutination.21 Depending on the method, the result is qualitative (above or below the detection limit), semi-quantitative (approximate concentration by comparison of color intensity with known controls or by testing dilutions) or fully quantitative (interpolation from a calibration curve). Methods can be manual or automated and differ in their time-to-result. Analytical and operational characteristics of the various commercial D-dimer assays are summarized in Table 3.
To exclude VTE in suspected patients, a test with a short turnaround time (TAT) is needed that reliably detects whether D-dimer is above or below the established cut-off. Preference should be given to quantitative observer-independent systems with the lowest coefficient of variation (CV) at the cut-off decision level.22 Combining analytical accuracy with ease of use and TAT, rapid automated quantitative assays such as the rapid Elisa (VIDAS®) or immunoturbidimetric tests are preferred. The quantitative immunoturbidimetric latex assays, however, may be subject to interfering substances26 and higher CVs.27
2. Accuracy study for selection of cut-off
The relation between a diagnostic test and its clinical usefulness usually is expressed in terms of its sensitivity (true positives) and specificity (true negatives). These indices are determined in a prospective accuracy study in a population of suspected VTE patients. The presence or absence of VTE is diagnosed with objective gold standard methods. True-positive fraction versus the false-positive fraction is plotted at various cut-off levels according to the ROC curve.28
The utility of D-dimer lies in its ability to exclude VTE; hence, the cut-off must be chosen to minimize the number of false negatives (target sensitivity is 100 percent) and maximize the NPV target of 100 percent.22 Lowering the cut-off will increase the sensitivity by providing fewer false negatives, but at the expense of a lower specificity or more false positives.
The safety of excluding VTE with a D-dimer test depends on the sensitivity of the test, whereas the clinical usefulness or efficacy is determined by the specificity. The specificity of highly sensitive D-dimer assays in outpatients usually is around 40 percent to 50 percent (i.e., 50 percent to 60 percent false positives). Specificity may be much lower, however, in populations such as the elderly, pregnant women, patients with cancer or hospitalized patients. Therefore, in these populations D-dimer has reduced clinical usefulness to exclude VTE.18
For example, in a population with only 10 percent negative D-dimer test results, the number of patients needed to test (NNT) to exclude one VTE is 1/0.1 = 10, whereas NNT is only 2 in a population with 50 percent negative test results. The decision to request a D-dimer assay in such populations is a matter of clinical judgment.
The NPV not only depends on the sensitivity of the test, but also on the prevalence of VTE.9 The higher the pre-test probability of VTE, the lower the NPV. This means that even a highly sensitive D-dimer assay may not safely rule out VTE in patients with a high clinical probability of disease. Conversely, in low-risk groups, a lower test sensitivity still may rule out the disease by achieving a high NPV.
The effect of disease prevalence, or PTP, can be approached by using the principles of Bayesian statistics.9 For example, a patient with a moderate probability on clinical assessment is thought to have a 17 percent chance of having DVT. This patient shows a negative result with a D-dimer test with a sensitivity of 96 percent and a specificity of 44 percent. Then, the following can be calculated:
pre-test odds = probability/(1-probability) = 0.17/(1-0.17)
likelihood ratio of a negative result, LR(neg) = (1-sens)/spec = (1-0.96)/0.44
post-test odds = pre-test odds x LR(neg) = 0.17/(1-0.17) x (1-0.96)/0.44 = 0.0186
post-test probability = odds/(1+odds) = 0.0186/(1+0.0186) = 0.018
This indicates that after the negative D-dimer test, the probability of DVT is 1.8 percent. However, if the probability of DVT on clinical assessment is thought to be 80 percent, it can be calculated that the post-test probability is 26.5 percent. This illustrates that this test with a LR(neg) of 0.09 can be used in populations with a low-moderate probability but not in very high-risk populations.
Applying Bayesian statistics, the effect of negative likelihood ratios on post-test probability of disease at various PTPs is shown in Table 4. Targeting a post-test probability of less than 3 percent, or NPV above 97 percent, one can see that only a test with a negative likelihood ratio of 0.10 or lower can be used to exclude VTE in low and medium-risk populations. Tests with a higher negative likelihood ratio, up till about 0.30, can only be used for exclusion in populations with a an expected low disease prevalence.
In outpatient populations with suspected VTE, the various D-dimer assays differ in sensitivity, specificity and negative likelihood ratio.29 Tests with a high sensitivity (> 90 percent; Elisa-based tests) usually have a moderate specificity (40 percent to 50 percent), whereas tests with lower sensitivity (80 percent to 90 percent; agglutination-based tests) tend to have a higher specificity (50 percent to 70 percent). A high sensitivity is required to obtain a low negative likelihood ratio. A recent meta-analysis indicates the following ranking in order of increasing negative likelihood ratio: rapid Elisa (VIDAS®) < membrane Elisa < immunoturbidimetric < SimpliRED® < manual latex.29
3. Prospective clinical management study
Once a D-dimer test has been selected based on its technical and operational merits (step 1) and accuracy criteria (step 2), its actual utility in terms of safety and cost-effectiveness needs to be demonstrated in real life. This requires a prospective clinical outcome study in which anticoagulation is withheld in patients with a diagnostic criterion ruling out VTE (step 3). A systematic three-month follow-up is required to allow detection of delayed events and establish the true diagnostic performance of the test. A three-month follow-up event rate of less than 3 percent (upper 95 percent confidence limit) generally is considered safe.5
Since 1999, several prospective clinical management studies have validated the safety and efficacy of sequential diagnostic strategies, including rapid D-dimer tests, for the non-invasive diagnosis of DVT or PE in suspected patients.5 For patients with suspected DVT, a strategy combining PTP and D-dimer (Fig. 3) has been shown to be a very safe and cost-effective approach, with a significant reduction in the number of ultrasound scans.4,30,31 For patients with suspected PE, a similar strategy based on initial PTP and D-dimer also is safe and cost-efficient, while limiting the number of V/Q and ultrasound scans (Fig. 4).31 A similar strategy also is possible with helical CT, replacing the ventilation-perfusion lung scan (Fig. 5).32
Since the vast majority of outpatients attending a hospital emergency room with clinical signs of VTE will not have the disease, there is a large emphasis on rapid and non-invasive diagnostic strategies for exclusion. During the past five years, evidence has accumulated confirming the accuracy and cost-effectiveness of incorporating a D-dimer assay into sequential diagnostic algorithms to diagnose DVT and PE.5
The Bayesian approach of combining a pre-test probability (PTP) model with a D-dimer test safely can exclude disease in approximately 30 percent of patients with suspected VTE.15 This will reduce greatly the need for time-consuming and expensive imaging procedures and avoid unnecessary treatment with anticoagulants. D-dimer and PTP now are highly recommended as the initial step in the investigation of patients with suspected VTE.33
D-dimer is not a standardized analyte and assays vary widely in analytical, operational and clinical characteristics.21,22 Clinicians and laboratory managers should be aware of these aspects before selecting a D-dimer assay as part of a diagnostic algorithm for VTE.
Dr. Houdijk is Market Manager, Sepsis-Thrombosis, Global Marketing and Strategic Development, bioMérieux.