Secondary Logo

Journal Logo


Early matrix metalloproteinase-9 concentration in the first 48 h after aneurysmal subarachnoid haemorrhage predicts delayed cerebral ischaemia

An observational study

Triglia, Thibaut; Mezzapesa, Anna; Martin, Jean Charles; Verdier, Monique; Lagier, David; Dufour, Henry; Bruder, Nicolas; Alessi, Marie-Christine; Velly, Lionel J.

Author Information
European Journal of Anaesthesiology: September 2016 - Volume 33 - Issue 9 - p 662-669
doi: 10.1097/EJA.0000000000000494



Delayed cerebral ischaemia (DCI) related to cerebral vasospasm occurs in about 30% of patients with aneurysmal subarachnoid haemorrhage (SAH) within the first 2 weeks after the haemorrhage.1 DCI is responsible for increased morbidity and mortality.2 Identifying patients at risk of developing DCI after SAH is crucial in developing therapies for treating or preventing this potentially devastating complication. There is currently no established biomarker for predicting DCI. Although the pathophysiology of DCI is not fully understood, there is growing evidence that matrix metalloproteinases (MMPs) play a key role.3–5

MMP-9, a zinc endopeptidase, is produced by leukocytes6,7 and glial cells,8 and has collagenase activity. In-vivo experiments have linked MMP-9 to the pathophysiology of brain injury following SAH and to the development of vasospasm.9 After SAH, MMP-9 is up-regulated in rat cerebral arteries, and it co-localises with degraded type-IV collagen, a major structural component of the basal lamina of the blood–brain barrier.10 MMP-9 may be involved in early brain injury and cerebral vasospasm by reducing the anti-inflammatory properties of gelsolin11 and by converting the large endothelin-1 (ET-1) molecule into small, highly vasoactive fragments.12 MMP-9 knockout mice showed less brain oedema and improved neurological recovery after SAH compared with genetically unmodified animals.3

In a rat model, MMP-9 concentrations in microvessels increased as early as 3 hours after SAH10; MMP-9 concentrations peaked at postbleed day 3.13 The authors of that study noted a significant reduction in vasospasm following the administration of a MMP-9 inhibitor. In humans, a postmortem study after intracranial haemorrhage found that MMP-9 expression was up-regulated from day 2 post-SAH.14 The fact that MMP-9 was expressed shortly after brain haemorrhage indicated that it may be a potentially efficient biomarker for the early prediction of DCI after SAH. In this study, we tested the hypothesis that plasma or cerebrospinal fluid (CSF) concentrations of MMP-9 and ET-1 in the first 3 days post-SAH would be predictive of DCI. To that end, we conducted a prospective observational study on patients admitted to our Neuroscience Critical Care Unit (NCCU).


The study was designed and conducted following the Strengthening the Reporting of Observational Studies in Epidemiology guidelines.15 Ethical approval for this study (N° 2013-1316) was provided by the Ethical Committee of the Assistance Public Hôpitaux de Marseille, France (Chairperson Professor Jammes) on 16 May 2013. After obtaining written informed consent from next of kin, we enrolled 30 consecutive patients with spontaneous aneurysmal SAH who met the inclusion criteria. For inclusion, patients had to be over 18 years old admitted to our NCCU within 24-h postbleeding and their condition had to require clinically indicated external ventricular drainage. Patients were excluded if pregnant, if they had vasospasm on the first diagnostic angiogram or if their condition was so severe that they were unlikely to survive more than 48 h.

Details of patients were collected, including age, sex and individual scores on the Simplified Acute Physiology Score II,16 the World Federation of Neurological Surgeons (WFNS) grading system17 and the Hunt and Hess score.18 Radiographic characteristics of the initial haemorrhage were identified, including the localisation of the aneurysm, the modified Fisher grade19 and the volume of the intraventricular haemorrhage (IVH) evaluated with the IVH score.20 Radiographic grading scale scores were determined by two authors (L.V. and D.L.) who were blinded to the clinical outcome of study patients. Additionally, we collected data regarding duration of mechanical ventilation and external ventricular drainage placement, length of stay in NCCU, modified Rankin Scale score21 and Glasgow Outcome Scale22 at 1 month.

All patients were followed with transcranial Doppler sonography (TCD; Philips CX50 with a S3–1 transducer, Philips Healthcare, Suresnes, France) and received continuous intravenous nimodipine. Acceleration of TCD mean blood flow velocity above 120 cm s−1 in the middle or anterior cerebral artery or daily change in mean TCD velocities greater than 50 cm s−1 were suggestive of cerebral vasospasm. DCI was defined, based on the latest recommendations,23,24 as follows: a development of focal neurological signs; a reduction by at least 2 points on the Glasgow Coma Scale which lasts for at least 1 h and is associated with angiographic cerebral vasospasm, detected either with computed tomography (CT) angiography or digital subtraction angiography; or a new cerebral infarction detected by CT, either within 6 weeks after SAH or before discharge, after excluding a procedure-related infarction.

Blood and CSF were sampled in the first 24 h (T1) and between 48 and 72 h (T2) post-SAH. Samples were centrifuged immediately and stored at −80°C until analysis. CSF samples were collected from the external ventricular drain into sterile tubes which did not contain anticoagulant. Blood samples were collected in sterile tubes which contained EDTA.

MMP-9 antigen concentrations were measured in blood and CSF samples by blinded investigators using ELISA (Quantikine, R&D Systems, Minneapolis, Minnesota, USA). CSF and plasma samples were diluted 1 : 5 and 1 : 40, respectively. MMP-9 concentrations are expressed in ng ml−1. ET-1 was also measured in blood and CSF samples with ELISA. Concentrations are expressed in pg ml−1. CSF proteins were quantified with bicinchoninic acid and expressed in mg ml−1 (Bicinchoninic Acid Kit, Sigma Aldrich, Saint-Louis, Missouri, USA).

MMP-9 activity was assessed by gel zymography, as previously described.25 Briefly, 5 μl of diluted plasma (1 : 20) or 5 μg CSF proteins were added to nonreducing sample buffer (tris–glycine–SDS; Invitrogen, Carlsbad, Germany) to achieve a final volume of 20 μl. Samples were loaded onto a 10% tris–glycine gel with 0.1% gelatine as substrate (Invitrogen). Electrophoresis was performed at 75 V for 30 min, then at 100 V for 2 h at 4°C. Then, gels were incubated in the zymogram renaturing buffer (Invitrogen) for 30 min at room temperature under gentle agitation, rinsed once and incubated overnight in developing buffer at 37°C (Invitrogen). Developed gels were stained with Coomassie Blue (Sigma Aldrich). MMP enzymatic activities were detected as transparent bands on the blue background. Bands were quantified with Image J 1.42 software (National Institutes of Health, Bethesda, Maryland, USA) after image acquisition with a CCD camera. All assays were run in duplicate. Results are expressed in arbitrary units (AU) ml−1. The specific activity of MMP-9 was defined as the ratio of MMP-9 activity (AU ml−1) to the MMP-9 antigen concentration (ng ml−1). Specific activity ratios were calculated for both CSF and plasma and expressed in AU ng−1.

Blood cell counts and fibrinogen concentrations were measured in blood samples at the two time points with different autoanalysers: ADVIA 120 (Bayer Diagnostics, New York, New York, USA) was used for haemograms and STA-R (Stago, Asnières, France) was used for fibrinogen concentrations. These were determined using standard laboratory methods.

Statistical analysis

For the sample size calculation, we considered the null hypothesis (AUC = 0.5) and we assumed in this severe SAH population that the ratio of DCI+ to DCI− was equal to 1.26 The necessary sample size of 30 patients (15 in each group) was calculated to achieve at least 85% statistical power at a type 1 error probability of 5% to detect AUC of 0.9. The calculation was based on the estimate of AUC proposed by Hanley and McNeil27 that is a function of the receiver operating characteristic (ROC) area and the sample size only. This method is based on an exponential distribution approximation to the standard error of the ROC area which has been shown to provide a very good approximation for a variety of underlying distributions for continuous diagnostic characteristics.28

Categorical variables are expressed as absolute values and percentages. These variables were compared between groups with a χ2 test or Fisher's exact test. Continuous variables, expressed as mean ± SD or median (IQR), were compared using a Student's t-test if they were normally distributed, and Mann–Whitney U test if not. Associations between continuous variables were tested with the Spearman's coefficient (r) and 95% confidence intervals (95% CI). A ROC curve was plotted to determine the ROC area under the curve and the optimal cut-off value of MMP-9 that best predicted DCI.

Univariate analysis regarding the primary evaluation criterion, occurrence of DCI, was performed by creating contingency tables and Student's t-test for the following variables: WFNS grading system, modified Fisher grade, IVH, protein concentration in CSF, CSF and plasma MMP-9 activity and concentrations at T1 and ET-1 concentrations at T1. Multivariate analysis was performed using backward stepwise logistic regression to enter variables that yielded P values smaller than 0.2 in the univariate analysis to identify factors of DCI. The Hosmer–Lemeshow goodness-of-fit test and the area under the ROC curve were used to evaluate the overall fit of the final model.

For all tests, P < 0.05 was considered statistically significant. All analyses were performed with Prism 5.0 software (GraphPad, La Jolla, California, USA) except for the ROC curves which were constructed with MedCalc 9.2 software (MedCalc Software, Ostend, Belgium) and for the multivariate logistic regression analyses which were performed using JMP software 11.0 (SAS Institute, Cary, North Carolina, USA).


Among the 68 patients with SAH admitted to our unit between 2013 and 2014, 30 patients met the inclusion criteria (see Supplemental Digital Content, Figure S1,; 16 had DCI (group DCI+) and 14 had no DCI (group DCI−). The DCI+ diagnosis was based on clinical signs of deterioration and angiographic signs of cerebral vasospasm in 13 patients (six had new focal neurological signs and seven had a deterioration in level of consciousness), and, in three patients, on signs of brain infarction on CT. The characteristics of the patients are shown in Table 1 and the treatment of symptomatic cerebral vasospasm in Table S1, (see Supplemental Digital Content, Demographic variables were not significantly different between groups. At a 4-week follow-up, the median-modified Rankin Scale was not significantly different in patients with or without DCI (P = 0.08). Among the inflammatory parameters tested (see Supplemental Digital Content, Table S2,, leukocyte counts did not differ between the two groups. Fibrinogen concentrations were significantly higher at T2 than at T1, but only in patients with DCI (P = 0.046).

Table 1
Table 1:
Characteristics of patients with aneurysmal subarachnoid haemorrhage who developed (+) or did not develop (−) delayed cerebral infarction

Plasma and CSF MMP-9 antigen concentrations (Table 2) were significantly higher in patients with DCI compared with those without DCI at both time points. Zymography analyses confirmed that the concentration difference was consistent with a significant difference in the plasma and CSF MMP-9 activity concentrations between the groups (Table 2). These two methods of MMP-9 determination were significantly correlated in plasma at T1 (r = 0.70, 95% CI 0.45 to 0.86; P = 0.001) and T2 (r = 0.64, 95% CI 0.30 to 0.83; P = 0.004), and in CSF at T1 (r = 0.63, 95% CI 0.33 to 0.81; P = 0.003) and T2 (r = 0.70, 95% CI 0.38 to 0.87; P = 0.001).

Table 2
Table 2:
Matrix metalloproteinase-9 antigen content and activity levels and endothelin-1 concentrations in plasma and cerebrospinal fluid at two time points (T1 and T2) after aneurysmal subarachnoid haemorrhage in patients who developed (+) or did not develop (−) delayed cerebral infarction

CSF protein concentrations were significantly higher in patients with DCI than in those without DCI at both time points (Table 2). MMP-9 concentrations, both in plasma and CSF, were not correlated with the fibrinogen concentrations in plasma (data not shown). The MMP-9-specific activity ratio was slightly higher in CSF than in plasma at T1 (ratios in CSF to plasma 7.7 ± 10.2 and 4.2 ± 3.2 AU ng−1, respectively; P = 0.3) and significantly higher at T2 (9.6 ± 26.1 and 3.9 ± 2.7 AU ng−1, respectively; P = 0.01).

ET-1 concentrations in CSF, but not in plasma, were significantly higher in patients with DCI than in those without DCI at T1 and T2 (Table 2). CSF ET-1 activity concentrations were significantly correlated with MMP-9 activity concentrations at T1 (r = 0.78, 95% CI 0.57 to 0.89; P = 0.001) and T2 (r = 0.49, 95% CI 0.04 to 0.74; P = 0.02).

ROC curve analysis was used to determine the diagnostic performance of MMP-9 as a biomarker for DCI. The largest ROC area under the curve (0.91, 95% CI 0.75 to 0.98; P = 0.001) was obtained with CSF MMP-9 antigen concentrations at T1 (Table 3, Fig. 1). A threshold CSF MMP-9 concentration of 14.3 ng ml−1 was associated with the highest sensitivity (88%) and specificity (86%); the negative predictive value was 86% and the positive predictive value was 88% (Table 3). The predictive performance of MMP-9 was lower when measured in plasma. ET-l measurements were significantly less accurate than CSF MMP-9 antigen concentrations at T1 for predicting DCI (Table 3).

Table 3
Table 3:
Assessments of matrix metalloproteinase-9 antigen concentrations or activity and endothelin-1 antigen concentrations in cerebrospinal fluid or plasma for predicting delayed cerebral infarction after aneurysmal subarachnoid haemorrhage, based on receiver operator characteristics curves
Fig. 1
Fig. 1:
Receiver operating characteristic curves show MMP-9 sensitivity and specificity for the prediction of delayed cerebral ischaemia. (a, b) MMP-9 antigen concentrations and (c, d) MMP-9 activity levels were measured at (a, c) T1 and (b, d) T2 in CSF (solid lines) and plasma (dotted lines); black squares represent the highest accuracy on the plasma curve; black circles represent the highest accuracy on the CSF curve. CSF, cerebrospinal fluid; MMP-9, matrix metalloproteinase-9.

Logistic regression analysis

On univariate analysis, eight covariates – MMP-9 concentration (CSF and plasma), MMP-9 activity (CSF and plasma), Fisher grade, IVH volume, CSF ET1 concetration and CSF protein count – were identified as prognosis-predictive factors of DCI (P<0.2). Although the sample size of patients admitted to our study was small, we entered all patients into logistic regression analysis. As shown in Table 4, only MMP-9 concentration in CSF at T1 was significantly predictive of DCI.

Table 4
Table 4:
Multivariate analysis of risk factors for delayed cerebral infarction


The study demonstrated that, in the first 24 h after SAH, CSF MMP-9 concentrations are highly sensitive and specific for predicting subsequent DCI. Previous studies have investigated the value of MMP-9 expression as a biomarker after SAH, but they produced variable results. In 2002, McGirt et al.29 performed daily venous measurements of MMP-9, von Willebrand factor and vascular epidermal growth factor after SAH; they found that an increase in serum MMP-9 concentration could accurately predict the onset of vasospasm. A concentration higher than 700 ng ml−1 was associated with a 25-fold increase in the odds of subsequent vasospasm. However, the peak value was obtained 1 to 5 days before the onset of vasospasm, and thus required repeated measurements, which limited its clinical utility as a predictive biomarker. They did not find any difference between groups on the first day after SAH. Similarly, a recent study showed that patients with cerebral vasospasm had elevated MMP-9 serum concentrations compared with patients without vasospasm; however, the difference between groups was not significant within the first 4 days.7 Discrepancies between those results and our work may be explained by the use of serum samples rather than plasma. Indeed, MMP-9 concentrations were significantly higher in serum than in plasma and this may have reduced the discriminatory value of serum MMP-9. This hypothesis is consistent with previous results that compared serum and plasma MMP-9 in predicting gastric cancer development and progression.30

In the present study, we found that CSF MMP-9 concentrations showed better sensitivity and specificity than plasma concentrations at T1. Chou et al.6 found an association between blood and CSF MMP-9 concentrations and clinical outcome at 3 months after SAH, but they did not find an association with angiographic vasospasm. These major differences between study findings may be explained by different definitions of vasospasm. Vasospasm has been defined as a vessel narrowing detected on an angiograph6,29 or by means of transcranial Doppler sonography thresholds.7 However, less than 50% of patients with vasospasm develop DCI2; also, DCI can occur in patients without vasospasm.31,32 For this reason, we used a more recent, consensual definition of DCI which has been associated with long-term clinical outcome.23 We found strong MMP-9-specific activity in the CSF, which suggested that MMP-9 was highly active in the CSF. Accordingly, we expected that MMP-9 activity would provide better predictive value than MMP-9 antigen concentrations, a hypothesis that was not confirmed in our study. We speculate that a more sensitive assay may show a better predictive value of MMP-9 activity compared with MMP-9 antigen.

MMP-9 is also known to cleave the large ET-1 molecule, which is an inactive precursor, into potent vasoactive fragments.12 ET-1 has been shown to be involved in SAH-induced cerebral injury, oedema and vasospasm.33 In agreement with a previous study,34 we found a significant between-group difference in CSF ET-1 concentrations, but this difference had poor predictive value. The relationship between ET-1 concentrations and cerebral vasospasm or DCI remains elusive. Although some studies have found that the plasma or CSF content of ET-1 increased over time in patients who developed vasospasm,35–37 a recent study did not confirm that association.38 In that study, the ET-1 plasma content was significantly higher only on day 5 after SAH in patients who developed vasospasm. Again, this finding suggested that repeated measurements would be required, which would limit the practicability of ET-1 as a biomarker for the prediction of vasospasm.

The finding that MMP-9 antigen concentrations measured soon after SAH could predict whether brain ischaemia would occur several days later can be explained by the pathophysiology of SAH. In experimental models, cerebral vasospasm has been associated with the release of pro-inflammatory cytokines. Experimentally inhibiting the up-regulation of inflammatory cytokines prevented cerebral vasospasm.39 DCI occurs when brain inflammation gives rise to large vessel vasoconstriction, microcirculatory constriction, thrombosis or oxidative stress.40 Recent studies have demonstrated that the inflammatory pathway is activated as early as 2 h after SAH through neuronal release of high-mobility group box 1 proteins.41 Extracellular high-mobility group box 1 signalling up-regulates MMP-9 expression.42 Because inflammation is thought to play a major role in the development of DCI after SAH, this pathway could explain the link between increased CSF MMP-9 concentrations and DCI in our patients.

In addition, MMP-9 can cleave and inactivate plasma-type gelsolin, a scavenger of pro-inflammatory molecules.11 Two studies using brain microdialysis after SAH showed that the cerebral inflammatory response was more intense and occurred earlier than systemic inflammation.43,44 In one study, cerebral extracellular fluid but not serum IL-6 concentrations were predictive of the occurrence of secondary neurological deficits.43 These studies demonstrated a strong cerebral pro-inflammatory response after SAH that could be responsible for DCI, but no specific IL-6 threshold could be determined.43,44 MMP-9 may thus play an active role in the progression of DCI by enhancing brain inflammation and contributing to blood–brain barrier injury.7 These studies suggested a stronger relationship between extracellular fluid MMP-9 concentrations and outcome, rather than plasma concentrations.7 One clear advantage of CSF sampling from ventricular drains is that it is easier to obtain and technically much less complicated, and is more widely available than microdialysis.

The characteristics of MMP-9 make it a promising candidate biomarker for predicting DCI. Its CSF concentration was increased early after SAH, and it remained elevated 48 h after bleeding. The method for measuring MMP-9 is easy to standardise, with low variability between measures. In addition, the long delay (several days) between SAH onset and DCI occurrence facilitates obtaining a sufficiently early MMP-9 measurement, even in a busy clinical environment.

This trial had several limitations. First, our study was designed as a pilot study and included only a small number of patients, which is a potentially limiting factor. Second, we only included patients with severe SAH who required ventricular drainage. We did not investigate the performance of MMP-9 for predicting DCI in patients with less severe SAH or from causes other than aneurysmal rupture. However, our patients (WFNS 3 to 5) are the most severe SAH population with the highest risk of DCI. These patients are always referred to an ICU.45 Third, in our study we did not assess CSF leukocyte and neutrophil counts. Nevertheless, we estimated the degree of inflammation within the CSF compartment and its contribution to MMP-9 release in SAH with the CSF protein count. Finally, it is important to identify patients at high risk of a secondary neurological insult; however, it remains unknown whether this information can ultimately improve clinical outcome.

In conclusion, this study found that MMP-9 antigen concentrations and activity in both plasma and CSF, within the first 48 h after aneurysmal SAH, were highly predictive of DCI occurrence several days later. The highest predictive value was obtained with CSF MMP-9 concentrations. These findings must be validated in a larger patient population in a prospective multicentre trial to assess possible confounders before using MMP-9 as a clinical biomarker of impending DCI at the bedside.

Acknowledgements relating to this article

Assistance with the study: we thank Mrs J. Ansaldi and A. Vampiro for the biobanking and Mrs Lannelongue for the English correction.

Financial support and funding: this work was supported by the Assistance Publique des Hôpitaux de Marseille (AORC N° 2012-54).

Conflicts of interest: none.

Presentation: this study was presented in part at the Annual Congress of the ESA, Barcelona, June 2013.


1. Diringer MN, Bleck TP, Claude Hemphill J 3rd, et al. Critical care management of patients following aneurysmal subarachnoid hemorrhage: recommendations from the Neurocritical Care Society's Multidisciplinary Consensus Conference. Neurocrit Care 2011; 15:211–240.
2. Vergouwen MDI, Ilodigwe D, Macdonald RL. Cerebral infarction after subarachnoid hemorrhage contributes to poor outcome by vasospasm-dependent and -independent effects. Stroke J Cereb Circ 2011; 42:924–929.
3. Feiler S, Plesnila N, Thal SC, et al. Contribution of matrix metalloproteinase-9 to cerebral edema and functional outcome following experimental subarachnoid hemorrhage. Cerebrovasc Dis 2011; 32:289–295.
4. Wang Z, Meng CJ, Shen XM, et al. Potential contribution of hypoxia-inducible factor-1α, aquaporin-4, and matrix metalloproteinase-9 to blood-brain barrier disruption and brain edema after experimental subarachnoid hemorrhage. J Mol Neurosci 2012; 48:273–280.
5. Sarrafzadeh A, Copin JC, Bengualid DJ, et al. Matrix metalloproteinase-9 concentration in the cerebral extracellular fluid of patients during the acute phase of aneurysmal subarachnoid hemorrhage. Neurol Res 2012; 34:455–461.
6. Chou SH, Feske SK, Simmons SL, et al. Elevated peripheral neutrophils and matrix metalloproteinase 9 as biomarkers of functional outcome following subarachnoid hemorrhage. Transl Stroke Res 2011; 2:600–607.
7. Fischer M, Dietmann A, Beer R, et al. Differential regulation of matrix-metalloproteinases and their tissue inhibitors in patients with aneurysmal subarachnoid hemorrhage. Plos One 2013; 8:e59952.
8. Del Zoppo GJ. The neurovascular unit, matrix proteases, and innate inflammation. Ann N Y Acad Sci 2010; 1207:46–49.
9. Mehta V, Russin J, Spirtos A, et al. Matrix metalloproteinases in cerebral vasospasm following aneurysmal subarachnoid hemorrhage. Neurol Res Int 2013; 2013:943761.
10. Sehba FA, Mostafa G, Knopman J, et al. Acute alterations in microvascular basal lamina after subarachnoid hemorrhage. J Neurosurg 2004; 101:633–640.
11. Chou SH, Lee PS, Konigsberg RG, et al. Plasma-type gelsolin is decreased in human blood and cerebrospinal fluid after subarachnoid hemorrhage. Stroke J Cereb Circ 2011; 42:3624–3627.
12. Fernandez-Patron C, Zouki C, Whittal R, et al. Matrix metalloproteinases regulate neutrophil-endothelial cell adhesion through generation of endothelin-1[1-32]. FASEB J 2001; 15:2230–2240.
13. Wang Z, Fang Q, Dang BQ, et al. Potential contribution of matrix metalloproteinase-9 (mmp-9) to cerebral vasospasm after experimental subarachnoid hemorrhage in rats. Ann Clin Lab Sci 2012; 42:14–20.
14. Wu H, Zhang Z, Hu X, et al. Dynamic changes of inflammatory markers in brain after hemorrhagic stroke in humans: a postmortem study. Brain Res 2010; 1342:111–117.
15. Von Elm E, Altman DG, Egger M, et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Plos Med 2007; 4:e296.
16. Le Gall JR, Lemeshow S, Saulnier F. A new Simplified Acute Physiology Score (SAPS II) based on a European/North American multicenter study. JAMA 1993; 270:2957–2963.
17. Oshiro EM, Walter KA, Piantadosi S, et al. A new subarachnoid hemorrhage grading system based on the Glasgow Coma Scale: a comparison with the Hunt and Hess and World Federation of Neurological Surgeons Scales in a clinical series. Neurosurgery 1997; 41:140–147.
18. Hunt WE, Hess RM. Surgical risk as related to time of intervention in the repair of intracranial aneurysms. J Neurosurg 1968; 28:14–20.
19. Frontera JA, Claassen J, Schmidt JM, et al. Prediction of symptomatic vasospasm after subarachnoid hemorrhage: the modified fisher scale. Neurosurgery 2006; 59:21–27.
20. Hallevi H, Dar NS, Barreto AD, et al. The IVH score: a novel tool for estimating intraventricular hemorrhage volume: clinical and research implications. Crit Care Med 2009; 37:969–974.
21. Van Swieten JC, Koudstaal PJ, Visser MC, et al. Interobserver agreement for the assessment of handicap in stroke patients. Stroke J Cereb Circ 1988; 19:604–607.
22. Wilson JT, Pettigrew LE, Teasdale GM. Structured interviews for the Glasgow Outcome Scale and the extended Glasgow Outcome Scale: guidelines for their use. J Neurotrauma 1998; 15:573–585.
23. Vergouwen MDI, Vermeulen M, van Gijn J, et al. Definition of delayed cerebral ischemia after aneurysmal subarachnoid hemorrhage as an outcome event in clinical trials and observational studies proposal of a multidisciplinary research group. Stroke 2010; 41:2391–2395.
24. Vergouwen MDI. Participants in the International Multi-Disciplinary Consensus Conference on the Critical Care Management of Subarachnoid Hemorrhage. Vasospasm versus delayed cerebral ischemia as an outcome event in clinical trials and observational studies. Neurocrit Care 2011; 15:308–311.
25. Thuny F, Habib G, Le Dolley Y, et al. Circulating matrix metalloproteinases in infective endocarditis: a possible marker of the embolic risk. Plos One 2011; 6:e18830.
26. Crobeddu E, Mittal MK, Dupont S, et al. Predicting the lack of development of delayed cerebral ischemia after aneurysmal subarachnoid hemorrhage. Stroke 2012; 43:697–701.
27. Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982; 143:29–36.
28. Obuchowski NA. Sample size calculations in studies of test accuracy. Stat Methods Med Res 1998; 7:371–392.
29. McGirt MJ, Lynch JR, Blessing R, et al. Serum von Willebrand factor, matrix metalloproteinase-9, and vascular endothelial growth factor levels predict the onset of cerebral vasospasm after aneurysmal subarachnoid hemorrhage. Neurosurgery 2002; 51:1128–1134.
30. Wu CY, Wu MS, Chiang EP, et al. Plasma matrix metalloproteinase-9 level is better than serum matrix metalloproteinase-9 level to predict gastric cancer evolution. Clin Cancer Res 2007; 13:2054–2060.
31. Woitzik J, Dreier JP, Hecht N, et al. Delayed cerebral ischemia and spreading depolarization in absence of angiographic vasospasm after subarachnoid hemorrhage. J Cereb Blood Flow Metab 2012; 32:203–212.
32. Brown RJ, Kumar A, Dhar R, et al. The relationship between delayed infarcts and angiographic vasospasm after aneurysmal subarachnoid hemorrhage. Neurosurgery 2013; 72:702–707.
33. Yeung PK, Shen J, Chung SS, et al. Targeted over-expression of endothelin-1 in astrocytes leads to more severe brain damage and vasospasm after subarachnoid hemorrhage. BMC Neurosci 2013; 14:131.
34. Seifert V, Löffler BM, Zimmermann M, et al. Endothelin concentrations in patients with aneurysmal subarachnoid hemorrhage. Correlation with cerebral vasospasm, delayed ischemic neurological deficits, and volume of hematoma. J Neurosurg 1995; 82:55–62.
35. Jung CS, Lange B, Zimmermann M, et al. The CSF concentration of ADMA, but not of ET-1, is correlated with the occurrence and severity of cerebral vasospasm after subarachnoid hemorrhage. Neurosci Lett 2012; 524:20–24.
36. Suzuki K, Meguro K, Sakurai T, et al. Endothelin-1 concentration increases in the cerebrospinal fluid in cerebral vasospasm caused by subarachnoid hemorrhage. Surg Neurol 2000; 53:131–135.
37. Thampatty BP, Sherwood PR, Gallek MJ, et al. Role of endothelin-1 in human aneurysmal subarachnoid hemorrhage: associations with vasospasm and delayed cerebral ischemia. Neurocrit Care 2011; 15:19–27.
38. Bellapart J, Jones L, Bandeshe H, et al. Plasma endothelin-1 as screening marker for cerebral vasospasm after subarachnoid hemorrhage. Neurocrit Care 2014; 20:77–83.
39. Sasaki T, Kasuya H, Onda H, et al. Role of p38 mitogen-activated protein kinase on cerebral vasospasm after subarachnoid hemorrhage. Stroke J Cereb Circ 2004; 35:1466–1470.
40. Macdonald RL. Delayed neurological deterioration after subarachnoid haemorrhage. Nat Rev Neurol 2014; 10:44–58.
41. Sun Q, Wu W, Hu YC, et al. Early release of high-mobility group box 1 (HMGB1) from neurons in experimental subarachnoid hemorrhage in vivo and in vitro. J Neuroinflammation 2014; 11:106.
42. Qiu J, Xu J, Zheng Y, et al. High-mobility group box 1 promotes metalloproteinase-9 upregulation through Toll-like receptor 4 after cerebral ischemia. Stroke 2010; 41:2077–2082.
43. Sarrafzadeh A, Schlenk F, Gericke C, et al. Relevance of cerebral interleukin-6 after aneurysmal subarachnoid hemorrhage. Neurocrit Care 2010; 13:339–346.
44. Helbok R, Schiefecker AJ, Beer R, et al. Early brain injury after aneurysmal subarachnoid hemorrhage: a multimodal neuromonitoring study. Crit Care 2015; 19:75.
45. Velly LJ, Bilotta F, Fàbregas N, et al. Anaesthetic and ICU management of aneurysmal subarachnoid haemorrhage: a survey of European practice. Eur J Anaesthesiol 2015; 32:168–176.
© 2016 European Society of Anaesthesiology