C-reactive protein and D-dimer in cerebral vein thrombosis: Relation to clinical and imaging characteristics as well as outcomes in a French cohort study

Introduction Cerebral venous sinus thrombosis (CVST) is a rare disease with highly variable clinical presentation and outcomes. Clinical studies suggest a role of inflammation and coagulation in CVST outcomes. The aim of this study was to investigate the association of inflammation and hypercoagulability biomarkers with CVST clinical manifestations and prognosis. Methods This prospective multicenter study was conducted from July 2011 to September 2016. Consecutive patients referred to 21 French stroke units and who had a diagnosis of symptomatic CVST were included. High-sensitivity C-reactive protein (hs-CRP), neutrophil-to-lymphocyte ratio (NLR), D-dimer, and thrombin generation using calibrated automated thrombogram system were measured at different time points until 1 month after anticoagulant therapy discontinuation. Results Two hundred thirty-one patients were included. Eight patients died, of whom 5 during hospitalization. The day 0 hs-CRP levels, NLR, and D-dimer were higher in patients with initial consciousness disturbance than in those without (hs-CRP: 10.2 mg/L [3.6-25.5] vs 23.7 mg/L [4.8-60.0], respectively; NLR: 3.51 [2.15-5.88] vs 4.78 [3.10-9.59], respectively; D-dimer: 950 μg/L [520-2075] vs 1220 μg/L [950-2445], respectively). Patients with ischemic parenchymal lesions (n = 31) had a higher endogenous thrombin potential5pM than those with hemorrhagic parenchymal lesions (n = 31): 2025 nM min (1646-2441) vs 1629 nM min (1371-2090), respectively (P = .0082). Using unadjusted logistic regression with values >75th percentile, day 0 hs-CRP levels of >29.7 mg/L (odds ratio, 10.76 [1.55-140.4]; P = .037) and day 5 D-dimer levels of >1060 mg/L (odds ratio, 14.63 [2.28-179.9]; P = .010) were associated with death occurrence. Conclusion Two widely available biomarkers measured upon admission, especially hs-CRP, could help predict bad prognosis in CVST in addition to patient characteristics. These results need to be validated in other cohorts.

Conclusion: Two widely available biomarkers measured upon admission, especially hs-CRP, could help predict bad prognosis in CVST in addition to patient characteristics.
These results need to be validated in other cohorts.

K E Y W O R D S
cerebral venous sinus thrombosis, D-dimer, high-sensitivity C-reactive protein, thrombin generation assay

| I N T R O D U C T I O N
Cerebral venous sinus thrombosis (CVST) is a rare thrombotic disorder with an estimated incidence of 1.3 to 1.57 per 100,000 personyears from studies of various European and Australian populations and 2.85 to 3.16 in the United States [1][2][3][4]. Several risk factors for CVST have been identified, including oral contraceptive use; pregnancy; thrombophilia; inflammatory diseases; infections; trauma; malignancy; medications, such as steroids; endocrine, and hematologic disorders [1,[3][4][5]. Delays in diagnosis of CVST are common and substantial, with the first symptoms occurring from less than 48 hours until beyond 30 days [5]. Four main patterns have been identified as CVST clinical manifestations: isolated intracranial hypertension, focal syndrome, diffuse encephalopathy, and cavernous sinus syndrome [6].
Clinical presentation of CVST is affected by the patient's age, the time between the onset of symptoms and the admission to hospital, location of CVST, and the presence of cerebral parenchymal lesions.
Only a few studies have focused on the potential interest of biomarkers in CVST diagnosis and prognosis. Although plasma Ddimer has been widely validated as an effective tool for excluding the diagnosis of deep venous thrombosis or pulmonary embolism, its potential usefulness in the management of patients with suspected CVST is debated [7,8]. Indeed, D-dimer testing is of little use for CVST diagnosis [9]. No study has yet evaluated the association of D-dimer with CVST clinical manifestations. Besides, thrombin generation (TG) has been proposed as a global assay to evidence hypocoagulable [10,11] and hypercoagulable states [12][13][14]: increased TG is an independent predictor of recurrent venous thrombosis [15]. Inflammation biomarkers, such as fibrinogen and high-sensitivity C-reactive protein (hs-CRP), have been recently described as possible predictors of poor prognosis in CVST [16]. The neutrophil-to-lymphocyte ratio (NLR) reflects subclinical inflammation, predicting mortality in myocardial infarction and cardiovascular risk [17].
We recently described the clinical and imaging features of patients presenting with CVST enrolled in a French prospective cohort [18]. The aim of the present study was to investigate, in the French prospective cohort, the association of inflammation biomarkers, Ddimer, and TG parameters with clinical or imaging characteristics in the patients with CVST in this cohort at diagnosis and during the follow-up months.

| Population study
The "French prospective cohort-cerebral vein thrombosis" (FPCCVT) multicenter study was prospectively conducted from July 2011 to September 2016 [18]. Briefly, symptomatic patients with CVST from 21 French stroke unit centers were included. This study was funded • Association between inflammation and CVST severity has been observed.
• D-dimer and C-reactive protein predicted bad prognosis in CVST.
• Elevated endogenous thrombin potential on admission was associated with ischemic parenchymal lesions. BILLOIR ET AL. (NCT02013635). Patients were included regardless of race/ethnicity, and fully covered by the French health system.

| Clinical and imaging characteristics
The diagnosis of CVST had to be confirmed by cerebral tomodensitometry venography and/or magnetic resonance imaging (MRI) combined with magnetic resonance venography and/or digital subtraction angiography following established diagnostic criteria [19]. Patients' data were recorded. Clinical status was evaluated at admission using the Glasgow Coma Scale and the National Institutes of Health Stroke Scale (NIHSS). Presenting syndromes were categorized as isolated intracranial hypertension and/or isolated headache, focal syndrome, cavernous sinus syndrome, and diffuse encephalopathy. Radiological evolution was evaluated at 3 months using preferably MRI with magnetic resonance venography. "Partial recanalization" was defined as recanalization of at least one previously thrombosed sinus or vein in an individual patient with CVST. "No recanalization" was defined as total absence of any type of recanalization in any previously throm-

| Laboratory parameters and assays
Laboratory tests, including whole blood count and plasma fibrinogen, were locally performed for each patient upon admission before any initiation of anticoagulant therapy (D 0 ). Additional blood was collected on EDTA, 0.109M citrate, and dry tubes to build a plasma and serum bank. In addition to D 0 sampling, blood samples were taken at different times, especially at day 5 (D 5 ) after the diagnosis.
Anemia was defined by decreased hemoglobin <12.0 g/dL in females and 13.0 g/dL in males.
Platelet-poor plasma (PPP) was prepared as recommended by the "Groupe Français d'études sur l'Hémostase et la Thrombose," ie, doublecentrifuged at 2000 g for 10 minutes, and aliquots were kept frozen at −80 • C (www.geht.org). Samples were transported on dry ice for central analysis. Plasma D-dimer was assessed using Liatest DDI+ (Diagnostica Stago) on a STAR analyzer and serum hs-CRP using Creactive protein (CRP)-Vario on a C8000 analyzer (Abbott) at Lariboisière Hospital (Paris). TG was performed on a calibrated automated thrombogram system using a microplate fluorometer (Fluorocan Ascent, Thermoscientific Labsystems) at Rouen University Hospital.
PPP samples were run using either PPP reagent (5pM tissue factor and 4μM phospholipid vesicles) or low PPP reagent (1pM tissue factor and 4μM phospholipid vesicles) (Diagnostica Stago). We recorded and analyzed the following parameters: lag time, time to peak, endogenous thrombin potential (ETP) (area under the thrombin time-concentration curve), maximal thrombin peak, and velocity. TG assay was not performed if patients were on anticoagulant therapy before D 0 sampling; activity was systematically measured to detect anti-Xa inhibitors.

| Statistical analysis
The Shapiro-Wilk test was used to test normality of continuous variables. Normally, distributed data were presented as mean ± standard deviations and non-Gaussian variables as median ± IQR. We performed analyses to test relationships between biomarkers, namely inflammation biomarkers, D-dimer, and TG parameters measured at different times and i/clinical presentation or imaging data upon admission; ii// clinical evolution evaluated by modified Rankin Score, NIHSS, and death occurrence; iii/clinical/imaging evolution on 3 months. Univariate subgroup comparisons were analyzed using the Mann-Whitney U-test.
Multiple group comparisons were performed using the Kruskal-Wallis test completed with the two-stage step-up method of Benjamini, Krieger, and Yekutieli. The correlation between clinical and biological parameters was evaluated with Spearman and Pearson correlation.
Univariate logistic regression analysis of clinical outcome (death, presence of a parenchymal lesion) was performed using the following variables as predictors: age >75th percentile, sex, body mass index, dyslipidemia, tobacco, diabetes, initial consciousness disturbances, Ddimer >75th percentile, hs-CRP >75th percentile, and maximal thrombin peak. P values of <.05 were considered significant. Data were analyzed using GraphPad Prism 9.4 and the R software, version 4.0.0.

| Clinical characteristics
Two hundred thirty-one patients with a diagnosis of CVST were included. The main demographic and clinical characteristics at baseline are shown in Table 1

| Laboratory parameters and clinical presentation characteristics
Laboratory results obtained at D 0 are shown in Supplementary Table 1.
They are displayed according to the time between the onset of symptoms and diagnosis and the different clinical presentations (Table 2).

| Association with the initial clinical presentation
We analyzed the association of biomarkers with symptoms at presentation categorized as isolated intracranial hypertension and/or T A B L E 1 Epidemiologic, demographic, and clinical characteristics of patients with cerebral venous sinus thrombosis.

| Laboratory parameters and imaging at presentation
Laboratory parameters on D 0 according to imaging data are displayed in

| NIHSS evolution from D0 to D5 and biomarkers
We calculated the NIHSS variation from D 0 to D 5 (Supplementary

| Death occurrences during hospitalization and biomarkers
Five patients died during hospitalization. No significant correlations were observed between D0 hs-CRP and D-dimer (Supplementary Table 3

| D I S C U S S I O N
In our prospective study including 231 patients with CVST, we identified several biomarkers associated with clinical presentation characteristics and/or predictors of evolution. We confirmed, in a limited number of events, that hemorrhage and elevation of inflammation biomarkers, particularly hs-CRP and D-dimer, were associated with early death.
CVST is a rare thrombotic disorder caused by systemic or local imbalances in hemostasis [9]. Prognosis is usually good, with more than 80% of the patients making a complete recovery [5,6] [5]. More recently, a practical clinical score was established to predict outcomes after CVST [20].
The clinical score associating parenchymal lesion size of >6 cm, bilateral Babinski signs, male sex, parenchymal hemorrhage, and level of consciousness are associated with unfavorable outcomes. However, no laboratory marker had been evaluated in these 2 studies. We confirmed that age >52 years and hemorrhagic parenchymal lesions were associated with unfavorable outcomes. We also found that dyslipidemia was associated with unfavorable outcomes; this abnormality has never been reported before. However, the confidence interval of the odds ratio is very wide, and this result deserves to be cautiously interpreted.
The association of inflammation biomarkers with diagnosis, severity, and poor prognosis of cerebrovascular events has already been reported, especially in patients with CVST [15,[20][21][22][23][24][25]. Recently, a multicenter prospective observational study was conducted in Portugal: the Pathophysiology of Venous Infarction -Prediction of Infarction and Recanalization in CVT (PRIORITy-CVT) study, including 62 patients, demonstrated that biomarkers associated with inflammation predicted poor prognosis in cerebral vein thrombosis (CVT) [15]. In this study, patients had more unfavorable outcomes than the FPCCVT cohort (21% vs 3% of deaths, respectively). Three biomarkers, namely NLR, CRP, and interleukin (IL)-6, were evaluated as predictors of evolution. As reported in our study, NRL was associated with initial parenchymal brain lesions [16]. A higher baseline NLR, CRP, and IL-6 levels were predictors of unfavorable functional outcome at 90 days [16,24]. We demonstrated in our cohort that baseline systemic inflammation biomarkers were also associated with poor outcomes during hospitalization. Finally, we reported a non-significant fibrinogen increase associated with parenchymal lesions and an increasing number of sinus veins with thrombosis, as well as a lower baseline NLR in patients with improvement in imaging findings at 3 months. Noteworthy, in the PRIORITy-CVT, higher NLR, CRP, and IL-6 levels at admission were not associated with the early evolution of brain lesions or early recanalization. In a recent study including 52 CVST, the hs-CRP level was significantly higher in patients with acute or subacute CVT than in patients with chronic CVT [25]. However, in our study, including a much higher number of patients, we did not find a significant difference.
In addition, we studied D-dimer levels. The D-dimer increase at diagnosis was associated with neurological deficit [22]. In our FPCCVT cohort, the initial D-dimer levels were associated with initial consciousness disturbance, early death occurrence, and the severity of parenchymal lesions. Together with hs-CRP, D-dimer measured at D 5 predicted early death. The association between inflammatory biomarkers and clinical outcomes has been reported in different studies [16,26]. Interestingly, a previous study demonstrated that the D-dimer decrease between D 0 and day 180 was associated with complete sinus recanalization in MRI findings at 6 months [21]. In a study evaluating fibrin clot properties, an increase in D-dimer was observed in patients who had developed recurrent CVST [27]. An altered fibrin clot and fibrinogen increase were observed in patients with recurrent CVST [28] and in patients with CVST-related pregnancy [29]. The combined activation of inflammatory and hemostatic responses after infection or tissue injury is a defense mechanism. This process is referred to as immunothrombosis: within this novel concept, inflammatory cells such as leukocytes and platelets directly drive the progression of venous thrombosis [30]. Under pathological conditions, an increase in tissue factor expression and increased levels of the plasminogen activation inhibitor 1, blocking fibrinolysis, amplify coagulation [31]. The elevation of D-dimer in our patients at baseline in parallel to that of other inflammation biomarkers shows that D-dimer is an acute phase reactant [32,33]. Once anticoagulant treatment had been initiated, the subsequent D-dimer decrease makes the interpretation of D-dimer more complex, although D-dimer measured on D 5 still predicted early death.
Finally, in our cohort, TG parameters measured on D 0 , namely, ETP and velocity, were predictive of the presence of ischemic parenchymal lesions and a greater number of thrombosed sinuses. In the literature, to our knowledge, no study had previously evaluated TG as a predictor of CVST evolution. TG assay is a global coagulation assay that may be altered in hypercoagulable states [12,33] or different types of endothelial injury [34][35][36]. In our study, lower ETP 5pM was associated only with hemorrhagic parenchymal lesion development.
The strengths of our study were a prospective study design, with a prolonged follow-up, together with clinical and imaging evaluation at similar time points, including the evolution of parenchymal brain lesions and functional outcomes. However, our study has some limitations. Whereas inflammation markers and D-dimer are routinely performed, TG assay is performed only in some laboratories, and the calibrated automated thrombogram system is poorly standardized.
We had a few severe clinical presentations. However, Coutinho et al. [37] showed an improvement over time in initial CVT presentations especially underlying comorbid conditions. Also, we could not assess race or ethnicity as a social determinant related to our findings for regulatory reasons. The power was limited in that we only had 5 deaths and therefore could not perform age-adjusted or multivariable analysis.
In conclusion, we reported here that 2 widely available biomarkers, namely hs-CRP and D-dimer, could help predict early death in CVST in addition to patient characteristics. These results need to be validated in external cohorts.

RELATIONSHIP DISCLOSURE
There are no competing interests to disclose.