• 
• 

Three Part Question

In  [patients presenting with symptoms compatible with cerebral venous thrombosis], can a negative [d-dimer] be used to exclude [cerebral venous thrombosis]?

Clinical Scenario

A 22-year-old female patient arrives to your emergency department with severe left-sided non-pulsating headache that began 3 days ago. She had been discharged from your ED a few hours earlier with a diagnosis of new onset migraines after partial resolution of her symptoms with analgesia and rest. She complains of progressive unilateral headache and mild nausea, which appears to be different from the occasional headaches she experienced in the past. She is healthy and takes only an oral contraceptive pill without any previous thromboembolic complications. Her neurological exam is once again unremarkable and you decide to order a non-contrast head CT which is completely normal. Once again, her symptoms partially resolve and you consider migraine headache to be the most likely diagnosis. However, the persistence and severity of her symptoms makes you wonder if cerebral venous thrombosis could be the cause of her headaches. Instead of repeating an imaging exam with venous contrast (magnetic resonance or computed tomography), you wonder if D-dimers, a laboratory assay frequently used in the exclusion of venous thromboembolism, could safely exclude cerebral venous thrombosis in this case.

Search Strategy

A. No completed BETs on the subject were found on the BestBETs database. Two similar BETs questions were submitted in 2010 and 2011 without completion.

B. The website clinicaltrials.gov was searched for an ongoing trial on the topic: one potentially relevant ongoing trial was found (NCT00924859). The trial is still incomplete and started in 2009.

C. MEDLINE (PUBMED): (((((intracranial venous thrombosis) OR intracranial vein thrombosis) OR (((((central vein thrombosis) OR cerebral vein thrombosis) OR sinus thrombosis) OR cerebral venous thrombosis) OR central venous thrombosis))) AND ((diagnosis) OR sensitivity)) AND ((((ddimers) OR d-dimers) OR ddimer) OR d-dimer): 159 papers
- After abstract/title review = 2 relevant papers.

D. EMBASE :

('diagnosis' OR 'sensitivity') AND ('d dimer' OR 'ddimer' OR 'ddimers' OR 'd dimers') AND ('intracranial venous thrombosis' OR 'intracranial vein thrombosis' OR 'intracranial veins thrombosis' OR 'cerebral veins thrombosis' OR 'cerebral vein thrombosis' OR 'cerebral venous thrombosis' OR 'central venous thrombosis' OR 'central vein thrombosis' OR 'central veins thrombosis' OR 'cerebral sinus thrombosis')

- After abstract review= 2 relevant papers (including 2 previous relevant papers and no new relevant papers).

E. Cochrane: There were no Cochrane review on the subject.

Search Outcome

Relevant papers about the subject = 2 meta-analysis including all relevant papers found.

Relevant Paper(s)

Author, date and country	Patient group	Study type (level of evidence)	Outcomes	Key results	Study Weaknesses
Dentali et al. 2012 Italy	1134 patients from 14 studies. 731 patients from prospective studies and 403 patients from retrospective studies. All patients included were either suspected of CVT of had confirmed CVT and underwent d-dimer testing and imaging (CTV, MRV, MRI, CT, Angiography, angio-CT).	Meta-analysis of cohort studies (prospective and retrospective)	Diagnostic accuracy in suspected cases (926 patients)	Sensitivity 93,9% Specificity 89,7% +LR 9,1 -LR 0,07 Factors associated with false negatives : isolated headache, young patients, longer duration of symptoms.	High prevalence of CVT diagnosis in “suspected cases” (17%). Wide range of included populations (duration of symptoms, presence of neurological symptoms, etc.) limiting the application of the numbers to a subset of lower risk patients. Different lab assays were used in the studies with different results and subgroup analysis of the performance for each assay were not conducted. Publication bias unclear.
			Sensitivity for confirmed cases	Sensitivity for confirmed cases Sensitivity 89,1%
Alons et al. 2015 Netherlands	692 patients with suspected of CVT (149 from their own retrospective data collection), presenting isolated headaches and normal head CT undergoing d-dimer testing and imaging.	Meta-analysis of cohort studies (prospective and retrospective)	Sensitivity, specificity	Sensitivity from all CVT cases 87.1 % Specificity (from 4 studies with complete data on no-CVT group) 84.9 %	Authors presented a sensitivity analysis that excluded 6 studies (56 patients) in which some information was missing on the characteristics of the patients in the no-CVT group. Sensitivity was superior in this subgroup (97,8%). Variable d-dimer assays were used in the included studies.

Comment(s)

Current guidelines issued by the Heart Association/Americain Stroke Association state that “A normal D-dimer level according to a sensitive immunoassay or rapid enzyme-linked immunosor- bent assay (ELISA) may be considered to help identify patients with low probability of CVT (Class IIb; Level of Evidence B). If there is a strong clinical suspicion of CVT, a normal D-dimer level should not preclude further evaluation.“ In two recently published meta-analysis, Dentali et al (2012) found a sensitivity of 89,1% to 93,9% among retrospectively and prospectively selected cases, while Alons et al (2015) found a sensitivity of 87,1% for all CVT cases among patients presenting with an isolated headache for which CVT was suspected. While relatively high pooled sensitivities confirms d-dimers as a potential adjunct in the investigation of CVT, missed rate up to 10% remains high for the exclusion of such a morbid and potentially deadly condition. This BET tends to demonstrate that the evidence supporting the exclusion of CVT with normal d-dimers remains limited and of poor quality. Of note, the variability in the degree of suspicion of CVT in patients included in the two systematic reviews, as well as the different d-dimer assays used make it difficult to conclude in the superiority of a diagnostic algorithm including d-dimers in the patient with a low pretest probability of CVT compared to liberal imaging based on the clinician’s gestalt. Alons et al tried to isolate this subgroup of « low probability » patients by including only isolated headaches with normal CT in their more recent meta-analysis, with similar results in terms of diagnostic accuracy. Contrary to other thromboembolic diseases, such as pulmonary embolism and deep venous thrombosis, no clinical tool is currently validated to determine the pretest probability of CVT and safely identify the patients in whom CVT could be safely excluded without imaging. Among important factors to consider as potential false-negative d-dimer causes, subacute symptoms (> 2 weeks) were identified in a significant number of missed cases.

Clinical Bottom Line

Based on current published evidence, d-dimers can identify most patients with CVT, but they present a relatively high missed rate and further research would need to identify if there is a low-risk subset of patients that could benifit from its inclusion in investigation algorithms.

References

1. Dentali F, Squizzato A, Marchesi C, Bonzini M, Ferro JM, Ageno W. D-dimer testing in the diagnosis of cerebral vein thrombosis: a systematic review and a meta-analysis of the literature. J Thromb Haemost. 2012;10:582–9
2. Alons IME, Jellema K, Wermer MJH, Algra A. D-dimer for the exclusion of cerebral venous thrombosis: a meta-analysis of low risk patients with isolated headache. BMC Neurology. 2015;15:118. doi:10.1186/s12883-015-0389-y.