A Negative D-Dimer Assay Does Not Rule Out Cerebral Venous Thrombosis
A Series of Seventy-Three Patients
Background and Purpose— Cerebral venous thrombosis (CVT) is an infrequent variety of cerebrovascular disease with a wide spectrum of clinical presentations and a notoriously difficult diagnosis. Previous reports have emphasized the potential clinical utility of D-dimer assay in CVT diagnosis.
Methods— A rapid sensitive D-dimer assay was performed at entry in 73 patients with CVT <30 days duration, examined in our institution between 1999 and 2003.
Results— The mean value of D-dimer levels was 1521 ng/mL; 7 patients (10% of all patients and 26% of those presenting with isolated headache) had values <500 ng/mL. In a multivariate analysis, isolated headache was the only variable associated with a negative D-dimer assay.
Conclusion— A negative D-Dimer assay does not confidently rule out CVT, particularly in the setting of recent isolated headache.
Cerebral venous thrombosis (CVT) is an infrequent variety of cerebrovascular disease with a remarkable spectrum of clinical presentations and modes of onset leading to great diagnostic difficulties. The most frequent symptom is headache, which can sometimes be isolated but is usually associated with other signs such as focal deficits, seizures, disorders of consciousness, and papilledema, which occur in highly variable combinations.1,2 Because of this great variability, CVT is frequently clinically suspected but confirmed only rarely by neuroimaging investigations such as MRI/MR angiography, which are presently considered the gold standard.1,2 It would, therefore, be of great practical interest to have a test that would be easy to perform in emergency care and would confidently rule out CVT.
The D-dimer measurement is a good candidate, particularly on a emergency basis, because of its well-established usefulness in the diagnostic approach of deep venous thrombosis (DVT). Indeed whereas high values are very sensitive but poorly specific markers for DVT, low values (<500 ng/mL) have a high negative predictive value.3
Four studies including, respectively 5, 12, 23, and 35, patients have so far been devoted to D-dimers in CVT.4–7 Eight patients had D-dimer levels <500 ng/mL, but 6 of them had chronic symptoms. Altogether, among the 68 patients with a recent CVT, only 2 (3%) had D-dimers <500 ng/mL, which led to the conclusion that a negative D-dimer assay makes the diagnosis of CVT very unlikely5,7 and is particularly useful in the emergency care of patients with headache.6
We present the results of D-dimer measurement using a rapid sensitive assay performed at entry between 1999 and 2003 in 73 patients with CVT and symptoms of less than 1-month duration.
Subjects and Methods
Between 1999 and 2003, 126 patients were hospitalized in our department for CVT. The diagnosis was based on the classical neuroimaging signs using MRI/MR angiography, CT venography/angiography, or conventional angiography. We included in the present study the 73 patients with CVT diagnosed in our institution (neurology department and emergency headache center), within 30 days of symptoms onset and who had D-dimer measured at entry, before starting heparin. The 53 other patients were excluded because they were referred to our department for CVT after the diagnosis of CVT (n=14), had symptoms of >1-month duration (n=8), were treated with heparin before D-dimer measurement (n=22), or did not have a D-dimer assay (n=9).
Patients were classified into 3 groups according to the clinical presentation: isolated headache, isolated intracranial hypertension (headache and papilledema), and encephalic signs (focal neurological deficits, seizures, or disorders of consciousness). The presence of parenchymal lesions (edema/infarction and/or hemorrhage) was assessed on the initial neuroimaging. Causes and risk factors were classified into 2 groups: (1) no cause or isolated oral contraceptive (OC) use and (2) well-identified causes.
D-Dimer levels were measured at entry using Liatest (Diagnostica Stago), a rapid and sensitive assay.3,8–10 This test has been validated for exclusion of DVT and pulmonary embolism and an exclusion cut off of 500 ng/mL (similar to that of the Elisa gold standard) was observed.8,9,11,12 This usual cut off was applied in previous CVT studies and in the present study. However, because it has not been validated specifically for CVT, we also looked at a cut off of 400 ng/mL.
D-Dimer levels were presented as means (SD) in univariate analysis. The relationships between D-dimer levels and patient characteristics were investigated by ANOVA. Statistical testing was performed after taking the square root of D-dimer level to normalize its distribution. We also studied the proportion of patients with levels<500 ng/mL according to their clinical and radiological characteristics. All significance levels reported were 2-sided and P<0.05 was considered to indicate statistical significance. All analyses were performed using SAS software (version 8.0 for Windows; SAS Institute).
The main characteristics of the 73 patients are shown in Table 1. The usual preponderance of CVT in young women was observed. All patients but 2 had headache as the first symptom, 19 (26%) had isolated headache, 18 (25%) intracranial hypertension, and 36 (49%) encephalic signs. Lateral sinus was most frequently involved (74%) followed by superior sagittal sinus (56%), and, in 68% of cases, 2 sinuses or more were involved.
The mean value of D-dimer levels was 1521 ng/mL, and the interquartile range was 850 to 2060 (median=1300). Seven patients (10%) had levels<500 ng/mL, and 4 patients (5%) had levels<400 ng/mL (Table 2). Among these 7 patients, 5 had isolated headache and 2 had intracranial hypertension1 or encephalic signs.1 D-Dimers were found at <500 ng/mL in 5 of 19 (26%) with isolated headache and 2 of 54 (4%) with other presentations. With a cut off of 400 ng/mL, the figures were 3 of 19 (16%) and 1 of 54 (2%).
Table 3 shows the mean D-dimer values and the percentage of patients with values <500 ng/mL according to patient characteristics. The main characteristics associated with values <500 ng/mL were young age, clinical presentation as isolated headache, no parenchymal lesion, and no cause but OC use. In a multivariate logistic regression model taking into account all variables associated with D-dimer levels (age, clinical presentation, parenchymal lesion, and potential causes for CVT), the only remaining variable associated with a negative D-dimer assay was isolated headache at presentation (odds ratio, 9.3; 95% CI, 1.6 to 53.1; P=0.012).
In this series of 73 patients with a recent CVT, 7 (10%) had D-dimer levels<500 ng/mL according to Liatest. Although nonvalidated for CVT, this cut off of 500 ng/mL was chosen because it is the usual one in DVT3 and the one that was used in the other studies devoted to CVT.4–7 Had a lower cut off of 400 ng/mL been chosen, there would still have been 4 patients (5%) with a negative assay. Liatest was chosen because it is feasible on an emergency basis and has a sensitivity comparable to that of ELISA.2
This 10% rate of negative assay is more than the average 3% observed with a cut off of 500 ng/mL in recent CVT in the 4 previous studies.4–7 Indeed, in the 3 largest published studies,5–7 none of the patients with a recent CVT (<3- to 4-week duration) had D-dimer<500 ng/mL, and only 2 of the 5 patients reported by Talbot et al4 had a recent CVT and a normal D-dimer. Because the duration of symptoms has been shown to be negatively correlated with the D-dimer levels in both DVT9 and CVT,7 a shorter duration of symptoms might have explained our higher rate of negative assays, but this seems unlikely because, first, we took into account only recent cases from the published series to calculate the 3% rate of negative D-dimer, and, second, we found no significant association in our series between the percentage of negative assays and the time elapsed between the onset of symptoms and the test.
In the present study, the only variable significantly associated with a negative D-dimer assay was isolated headache at presentation. In patients presenting with other symptoms, only 2 (4%) had low values, in keeping with other published results,4–7 whereas among patients presenting with isolated headache, 26% had low values. There is no similar series because, in the 3 main series,5–7 patients were recruited when they had headache “suggestive” of CVT, ie, when unusually severe, resistant, or associated with focal signs, seizures, or signs of raised intracranial pressure. In these 3 studies combined, 6 patients had D-dimer levels<500 ng/mL, but they all had chronic headache of >3- to 4-week duration.
Our series thus differs from all previous ones by the unusually high frequency of both CVT presenting with isolated headache (26%) and negative D-dimer assay (26.3%) among those. The frequency of headache as the only symptom of CVT was related mainly to the presence in our institution of an emergency headache center that recruits nearly 10 000 headache patients per year.13 The frequency of negative assay raises the question of the neuroimaging diagnosis of CVT, particularly because some of the patients with isolated headaches had only 1 sinus (usually 1 lateral sinus) involved. However, it seems unlikely that CVT was overdiagnosed because we used various magnetic resonance techniques to carefully rule out low flow, hypoplasia, and stenosis, which may be difficult to differentiate from lateral sinus thrombosis.1,2,14 It is of interest that the 2 patients reported by Talbot et al with acute CVT and normal D-dimer levels also had unilateral lateral sinus thrombosis, but, in contrast to our patients, they presented with encephalic signs.4
Because there is a correlation between isolated headache and both a limited sinus involvement and a negative D-dimer assay, it is tempting to speculate that the most important factor is the extension of thrombosis. In DVT, D-dimer levels are significantly higher in patients with proximal thrombosis than in those with thrombosis below the knee,10 and they are correlated with clot volume and surface as shown using thrombus MRI.15 In CVT, Kosinski et al also found an association between elevated D-dimer values and the extent of thrombosis.7 Our study differs in that it concentrated on negative D-dimer assays but found no significant association between a limited sinus involvement and a negative assay.
Another factor that might have played a role is the etiology of CVT because D-dimers are elevated in many of the conditions that predispose to venous thrombosis, such as pregnancy, systemic diseases, malignancies, or infections. Such causes were found in 62% of our patients, with the others having no cause or only OC use as a cause. We found a nearly significant trend in favor of a correlation between a negative assay and the absence of cause or predisposing factor other than OC use.
In summary, our results thus show that in recent CVT, when, clinically, there is a relatively high index of suspicion because of headache with encephalic signs or papilledema, a negative D-dimer assay is extremely rare (4%), which is in keeping with previous reports on CVT. In this setting, a negative D-dimer assay is of great clinical utility because it makes the presence of CVT very unlikely. By contrast, when CVT presents with a recent isolated headache, 26% of patients have D-dimer levels<500 ng/mL and 16%, <400 ng/mL. Therefore, when the degree of clinical suspicion for CVT is low because of isolated headache, particularly in the absence of cause other than OC use, a negative D-dimer does not confidently rule out CVT. This contrasts sharply with DVT, where a combination of low clinical probability and a negative D-dimer assay safely excludes the diagnosis.16
The fact that 53 patients seen in the same period of time were not included in the present study could have modified the percentage of negative assays, but even when taking into account all of the 45 who had a recent CVT, and supposing that they all had elevated D-dimers, there would still have been 7 of 118 (6%) patients with a negative assay. Therefore, this does not affect the conclusion that D-dimer may be negative in CVT with headaches of <1-month duration. This stresses the need for appropriate neuroimaging investigations in recent unexplained headache, whatever the D-dimer levels. It also suggests that studies should be performed to assess whether cut-off D-dimer values specific for exclusion of CVT would be different from those used for DVT and/or would differ according to the various clinical patterns of presentation, modes of onset, sites, and extent of thrombosis.
- Received March 22, 2005.
- Revision received April 25, 2005.
- Accepted May 25, 2005.
Bousser MG, Russell RR. Cerebral Venous Thrombosis. Vol. 1. London: WB Saunders; 1997.
Lalive PH, de Moerloose P, Lovblad K, Sarasin FP, Mermillod B, Sztajzel R. Is measurement of D-dimer useful in the diagnosis of cerebral venous thrombosis? Neurology. 2003; 61: 1057–1060.
Kosinski CM, Mull M, Schwarz M, Koch B, Biniek R, Schlafer J, Milkereit E, Willmes K, Schiefer J. Do normal D-dimer levels reliably exclude cerebral sinus thrombosis. Stroke. 2004; 35: 2820–2825.
Escoffre-Barbe M, Oger E, Leroyer C, Grimaux M, Le Moigne E, Nonent M, Bressollette L, Abgrall JF, Soria C, Amiral J, Ill P, Clavier J, Mottier D. Evaluation of a new rapid D-dimer assay for clinically suspected deep venous thrombosis (Liatest D-dimer). Am J Clin Pathol. 1998; 109: 748–753.
Cumurciuc R, Crassard I, Sarov M, Valade D, Bousser MG. Headache as the only neurological sign of cerebral venous thrombosis: a series of 17 cases. J Neurol Neurosurg Psychiatry. 2005. In press.
Mas JL, Meder JF, Meary E, Bousser MG. Magnetic resonance imaging in lateral sinus hypoplasia and thrombosis. Stroke. 1990; 21: 1350–1356.
Fraser DG, Moody AR, Martel A, Morgan P. Determinants of D-dimer level in patients presenting with deep venous thrombosis: assessment using magnetic resonance thrombus imaging. Abstract 513. Abstracts from the European Haematology Association 5th Congress. June 27, Birmingham, Ala. 2000.
Fancher TL, White RH, Kravitz RL. Combined use of rapid D-dimer testing and estimation of clinical probability in the diagnosis of deep vein thrombosis: systematic review. BMJ. 2004; 329: 821–824.