ORIGINAL ARTICLE
In-hospital C-reactive protein predicts outcome after aneurysmal subarachnoid haemorrhage treated by endovascular coiling L. Z. Csajbok1, K. Nylén2, M. Öst1, H. Sonander1 and B. Nellgård1 1
Department of Anaesthesiology and Intensive Care, Institution of Clinical Sciences, Sahlgrenska Academy, Gothenburg University, Gothenburg, Sweden 2 Department of Neurology, Institution of Neurological Sciences, Sahlgrenska Academy, Gothenburg University, Gothenburg, Sweden
Correspondence L. Csajbok, Department of Anaesthesiology and Intensive Care, Sahlgrenska University Hospital, S-413 45 Göteborg, Sweden E-mail: [email protected] Conflicts of interest The authors confirm that there are no conflicts of interest. Funding This study was supported by grants from the Göteborg Medical Society, ALF Swedish government research grants, the Elsa and Gustav Lindh Foundation, and the John and Brit Wennerström Foundation. Submitted 9 September 2014; accepted 19 October 2014; submission 25 March 2014. Citation Csajbok LZ, Nylén K, Öst M, Sonander H, Nellgård B. In-hospital C-reactive protein predicts outcome after aneurysmal subarachnoid haemorrhage treated by endovascular coiling. Acta Anaesthesiologica Scandinavica 2015 doi: 10.1111/aas.12441
Background: This study aimed to examine prospectively whether the inflammatory marker C-reactive protein (CRP) increases in patients with aneurysmal subarachnoid haemorrhage (aSAH) treated by endovascular coiling and investigate whether CRP could be used as prognostic factor for long-term neurological outcome. Methods: This single-hospital study comprised 98 consecutive patients with confirmed aSAH treated by endovascular coiling. Admission status was classified according to the World Federation of Neurosurgical Societies (WFNS) Scale and initial cerebral computed tomography according to Fisher scale. CRP was analysed on days 0, 1, 2, 3, 4, 6 and 8 after the initial bleed. A neurological follow up was performed 1 year later according to the Extended Glasgow Outcome Scale (GOSE) for overall outcome and National Institute of Health Stroke Scale (NIHSS) for focal deficit. Results: CRP values increased from normal to peak at 53 mg/l at day 3–4 and then declined, without normalising, at day 8. Patients with a higher increase in CRP had a poorer neurological outcome after 1 year. CRP during the first week had a stronger correlation to outcome (r = 0.417) and NIHSS (r = 0.449) than initial clinical status (WFNS; r = 0.280 and 0.274) and radiology (Fisher scale; r = 0.137 and 0.158). CRP increase indicated a risk of poor outcome (GOSE) (P < 0.001) and permanent loss of neurological function (NIHSS) (P < 0.001). Logistic regression analysis suggested that elevated CRP already on day 2 is an independent prognostic marker for outcome. Conclusion: Early CRP values can perhaps be used as a prognostic factor for long-term neurological outcome prediction after endovascular treatment of aSAH.
Editorial comment: what this article tells us This cohort study found that the early increase in CRP is a strong prognostic factor for poor 1-year neurological outcome in aSAH patients undergone endovascular treatment. This observation may help clinicians to adjust optimal treatment and monitoring shortly after coiling of an aneurysm in SAH patients.
Acta Anaesthesiologica Scandinavica 59 (2015) 255–264 © 2015 The Acta Anaesthesiologica Scandinavica Foundation. Published by John Wiley & Sons Ltd
bs_bs_banner
255
L. Z. CSAJBOK ET AL.
Stroke is one of the leading causes of death and disability in modern society. Subarachnoid haemorrhage (SAH) accounts for up to 7% of all strokes1 and thus has an incidence of around 10/100,000 persons/year.1 The rupture of an intracerebral aneurysm is a devastating event, accounting for up to 25% of cerebrovascular deaths,1,2 rendering an overall poor outcome, dependence or death in 65–70% of the patients.2,3 However, with early surgical or endovascular intervention to secure a ruptured aneurysm, outcome has improved.4 This is achieved primarily by reducing the risk of re-bleeding, otherwise seen in 50% of patients within 6 months.5 Even when the aneurysm is secured, many patients develop late neurological complications. The underlying pathogenic mechanisms for these delayed neurological deficits are still obscure, but inflammation is one factor that has been suggested to play a role.6 The surgical clipping of aneurysms, like all surgical traumas, including neurosurgery, is followed by an inflammatory reaction, manifested by elevated levels of different inflammatory markers, including C-reactive protein (CRP).7 CRP is mainly synthesised in the liver, after induction by cytokines, particularly interleukin6. Following release into the circulation, CRP is evenly distributed in the vascular compartment. CRP responds rapidly after an inflammatory stimulus and can increase up to 1000-fold with a peak at approximately 48 h after the initial stimulus, after which it falls once the inflammatory stimulus is removed. The distribution in the vascular compartment could be explained by CRP’s putative role as a substance that neutralises or detoxifies harmful substances escaping from the site of inflammation into the circulation. The number of hepatocytes synthesising CRP increases as the intensity of the stimulus increases. Most of the CRP is mono-exponentially cleared from the circulation with a biological halflife of ∼19 h, so the only significant determinant of plasma levels of CRP is the rate of synthesis. This justifies the use of CRP to monitor the inflammatory activity. The surgical clipping of aneurysms has been shown to result in elevated CRP levels, which could be correlated to the 1-year outcome measured by the Glasgow Outcome Scale.8,9 Endovascular coiling has become more or less the
therapeutic choice for most aneurysms. We hypothesised that CRP levels may also be elevated in this group of patients, and, if so, it might correlate to the 1-year outcome. If the initial bleed causes an inflammatory reaction and this reaction is not clouded by a surgical intervention, the CRP elevation should be more moderate and should relate more closely to the effect of the haemorrhage than in operated patients. The aim of this study was to evaluate whether there is a CRP response after SAH in patients treated by endovascular coiling and if so, what relationship it has to the long-term outcome. Methods Patients The patients recruited to this study were admitted to the Neurosurgical Intensive Care Unit (NICU) at Sahlgrenska University Hospital during a 2-year period after subarachnoid blood was detected either on a computed tomography (CT) scan or at a lumbar puncture. The following inclusion criteria were met: the patient was admitted for acute aneurysmal SAH (aSAH), within 48 h after the initial haemorrhage, an aneurysm was visible on the digital subtraction selective angiography, and the treatment was determined to be endovascular coiling of the aneurysm. Furthermore, the patients had to be residing permanently in Sweden, that is, available for outcome investigation, and informed consent had to be obtained from patients or next of kin. The Gothenburg University Medical Ethics Committee (Forskningsetikkommiten, Avd. För allmänmedicin, Vasa Sjukhus, 411 33 Göteborg) approved the study with the approval number S 161-00 on 29 June 2000. All patients were treated according to a standardised protocol10 that included intravenous injections of tranexamic acid (Cyclokapron®, Pfizer AB, Sollentuna, Sweden) 1 g four times daily, a fibrinolytic agent to minimise the risk of re-bleeding.11 To reduce the risk of cerebrovascular vasospasm, a continuous infusion of nimodipine (Nimotop®, Bayer AB, Solna, Sweden), a selective calcium-receptor antagonist, was started at 2 mg/h at NICU admission and was then given for 10 days. However, a few patients received oral nimodipine (60 mg every 4 h) for the final 3 days. Acta Anaesthesiologica Scandinavica 59 (2015) 255–264
256
© 2015 The Acta Anaesthesiologica Scandinavica Foundation. Published by John Wiley & Sons Ltd
CRP AS OUTCOME PREDICTOR AFTER aSAH
Neurological assessment At hospital admission, a short neurological examination was performed by the admitting physician. The status was recorded and classified according to the World Federation of Neurosurgical Societies (WFNS) Scale.12 A CT of the brain was acquired and categorised according to the Fisher scale13 of SAH. Neurological status was registered in the NICU from four times daily to every 15 min, depending on the patient’s condition. One year (range 11.5–13 months) after the aSAH, one neurologist, blinded to all laboratory data, examined all the patients according to a standardised protocol. Outcome was measured using the Extended Glasgow Outcome (GOSE) score,14 an eight-level scale, where GOSE 1 is deceased, and GOSE 8 represents full recovery. This scale was subsequently dichotomised to calculate cut-off values for favourable (GOSE 5–8) and poor outcome (GOSE 1–4). The patients’ focal neurological deficit was assessed using the National Institute of Health Stroke Scale (NIHSS).15 Blood sampling procedures and laboratory assay Blood sampling for CRP analysis was started after admission and continued on days 1, 2, 3, 4, 6 and 8 after the initial haemorrhage (day 0). CRP was analysed with an immunochemical kinetic turbidimetry method16 within an hour of sampling in the hospitals accredited laboratory. The level of faulty margin is under 4% according to the ISO/ IEC 17 025 accreditation standard. Statistical analysis The Kolmogorov–Smirnov test showed that the CRP data were not normally distributed, and non-parametric statistics were therefore used. Statistical analyses were performed with the SPSS version 21 (IBM, Armonk, NY, USA) and SAS version 9.4 (SAS Institute Inc, Cary, NC, USA) statistics packages, where the correlation between variables was analysed with Spearman’s rank correlation test, while Wilcoxon’s signed rank test was used to compare dependent samples. All analyses were two tailed and performed at a 5% level of significance.
One way of looking at data related to variables changing over time, according to Matthews,17 is to analyse of the mean and maximum of the series. The CRPmean, and CRPmax during the first 8 days were therefore used for further statistical analysis. In search of an early prediction, we added the means of the CRP values on observation days CRPd1-8. We tested the parameters for associations with Spearman’s correlation analysis. The potential prognostic factors were tested with bivariate logistic regression, and odds ratio (OR), 95% confidence interval and P-values were calculated. To clarify the internal relationships between the prognostic factors, CRP was inserted into a multivariate regression model. On the strongest predictors, receiver operating characteristic curve (ROC) analysis18,19 was employed to choose a cut-off value where possible. Sensitivity, specificity and positive predictive values (PPVs) were calculated where the likelihood ratio (LH) of the test was highest. Results During the study period, 202 patients were initially admitted to the NICU with a diagnosis of aSAH, and of these, 98 met the inclusion criteria. Of the non-eligible patients, three were admitted later than day 2, in 33 patients no aneurysm was detected, angiography could not be performed in four patients, in nine cases no informed consent could be obtained or consent was withdrawn, three patients were lost to long-term follow up and in 39 patients surgical clipping was the treatment of choice. In 13 patients, consecutive data were missing. As a result, the study population consisted of 74 women (76%) and 24 men (24%) with a median age of 57 years (range 26–81). The patients’ demographics, their initial neurological evaluation, infectious status during the study period and CRP values are shown in Table 1. The CRP values showed a continuous rise from an initial median value on day 0 of 5 mg/l [interquartile range (IQR) 5–7] until its maximum at day 3–4 (median 53 mg/l, IQR 24–110); thereafter, it decreased. The increase from the initial value until its peak on day 3–4 was significant between each of the days as well as the decrease from day 4 to day 6 and 8. The CRP value did not return to baseline, as the day 8 value (median 24 mg/l, IQR 10–47) was still significantly higher than the initial value (Fig. 1).
Acta Anaesthesiologica Scandinavica 59 (2015) 255–264 © 2015 The Acta Anaesthesiologica Scandinavica Foundation. Published by John Wiley & Sons Ltd
257
L. Z. CSAJBOK ET AL.
Table 1 Patients’ characteristics at admission and during observation.
Variables Gender Female Male Age (years)
Total (n = 98)
Favourable outcome (GOSE 5–8) (n = 67)
74 (75.5%) 24 (24.5%) 56.5 (10.7) 57.5 (26.0; 81.0) n = 98 Neurological status at first attention (WFNS score) WFNS I 34 (34.7%) WFNS II 30 (30.6%) WFNS III 3 (3.1%) WFNS IV 20 (20.4%) WFNS V 11 (11.2%) Radiological classification at first available CT scan (Fisher score) Fisher II 3 (3.1%) Fisher III 31 (31.6%) Fisher IV 64 (65.3%) Infectious status during observation Non infected patients 41 (41.8%) Infected patients 57 (58.2%) CRP max (mg/l) 98.9 (73.9) 79.5 (6.0; 340.0) n = 98 CRP mean (mg/l) 51.5 (38.2) 41.4 (5.2; 167.5) n = 98 CRP day 0 (mg/l) 10.5 (19.7) 5.0 (5.0; 100.0) n = 23 CRP day 1 (mg/l) 17.9 (24.4) 9.0 (5.0; 180.0) n = 87 CRP day 2 (mg/l) 47.2 (40.5) 27.0 (5.0; 180.0) n = 98 CRP day 3 (mg/l) 71.6 (57.0) 62.5 (5.0; 260.0) n = 98 CRP day 4 (mg/l) 76.0 (66.2) 53.0 (5.0; 340.0) n = 98 CRP day 6 (mg/l) 59.7 (65.3) 34.0 (5.0; 260.0) n = 98 CRP day 8 (mg/l) 42.9 (52.2) 24.0 (5.0; 240.0) n = 98
52 (77.6%) 15 (22.4%) 56.1 (10.3) 58.0 (26.0; 81.0) n = 67
Poor outcome (GOSE 1–4) (n = 31)
22 (71.0%) 9 (29.0%) 57.4 (11.4) 57.0 (30.0; 78.0) n = 31
30 (44.8%) 18 (26.9%) 1 (1.5%) 12 (17.9%) 6 (9.0%)
4 (12.9%) 12 (38.7%) 2 (6.5%) 8 (25.8%) 5 (16.1%)
3 (4.5%) 23 (34.3%) 41 (61.2%)
0 (0.0%) 8 (25.8%) 23 (74.2%)
31 (46.3%) 36 (53.7%) 80.8 (68.4) 68.0 (6.0; 340.0) n = 67 41.7 (34.6) 31.4 (5.2; 167.5) n = 67 6.58 (3.45) 5.00 (5.00; 16.00) n = 12 14.6 (14.7) 8.0 (5.0; 78.0) n = 58 38.5 (36.1) 24.0 (5.0; 140.0) n = 67 59.7 (54.9) 37.0 (5.0; 260.0) n = 67 62.2 (61.4) 41.0 (5.0; 340.0) n = 67 47.6 (54.2) 23.0 (5.0; 250.0) n = 67 30.9 (33.5) 17.0 (5.0; 180.0) n = 67
10 (32.3%) 21 (67.7%) 138.1 (71.1) 140.0 (22.0; 280.0) n = 31 72.7 (37.5) 71.0 (14.9; 165.5) n = 31 14.8 (28.4) 5.0 (5.0; 100.0) n = 11 24.6 (36.3) 14.0 (5.0; 180.0) n = 29 66.0 (43.6) 62.0 (12.0; 180.0) n = 31 97.5 (53.4) 89.0 (18.0; 190.0) n = 31 105.8 (67.2) 87.0 (18.0; 280.0) n = 31 85.8 (79.5) 47.0 (10.0; 260.0) n = 31 68.8 (73.1) 34.0 (6.0; 240.0) n = 31
For categorical variables n (%) is presented. For continuous variables, mean (SD)/median (min; max)/n = is presented. CRP, C-reactive protein; CT, computed tomography; WFNS, World Federation of Neurosurgical Societies Scale.
Acta Anaesthesiologica Scandinavica 59 (2015) 255–264
258
© 2015 The Acta Anaesthesiologica Scandinavica Foundation. Published by John Wiley & Sons Ltd
CRP AS OUTCOME PREDICTOR AFTER aSAH
Fig. 1. Box plot of C-reactive protein (CRP) values from day 0 to day 8 with median, mean and interquartile range. Differences tested with Wilcoxon’s signed rank test.
Table 2 Patients’ neurological outcome after 1 year. All patients (n = 98) Glasgow Outcome Score Extended GOSE 1 (dead) GOSE 3 (severe disability lower) GOSE 4 (severe disability upper) GOSE 5 (moderate disability lower) GOSE 6 (moderate disability upper) GOSE 7 (favourable recovery lower) GOSE 8 (favourable recovery upper) National Institutes of Health Stroke Scale NIHSS 0 p NIHSS 1–10 p NIHSS 11–20 p NIHSS 21–30 p NIHSS not accessible
15 (15.3%) 16 (16.3%) 0 (0%) 20 (20.4%) 18 (18.4%) 7 (7.1%) 22 (22.4%) 59 (60.2%) 15 (15.3%) 3 (3.1%) 4 (4.1%) 17 (17.3%)
For categorical variables n (%) is presented.
The patients’ infectious status was evaluated during the observation period. It was considered positive if any bacterial culture proved positive and/or more than prophylactic antibiotics (3 doses) were administered. No association could be shown between the status of infection and neurological outcome (GOSE, NIHSS). The results of the 1-year neurological follow up are presented in Table 2 as outcome parameters. The time-dependent course of CRP levels in the
Fig. 2. Box plot of time-dependent course of C-reactive protein (CRP) levels in the favourable and unfavourable patient groups with median, mean and interquartile range.
favourable and unfavourable patient groups is depicted in Fig. 2. There was a significant correlation between the initial WFNS score, CRPmean and CRPmax during the first week (r = 0.393, P < 0.001 and r = 0.316, P = 0.002 respectively), as well as between the initial WFNS score and the GOSE (r = 0.441, P < 0.001). An even stronger correlation was found between CRPmean/CRPmax and the GOSE (r = 0.483, P < 0.001/0.444, P < 0.001), while the radiological Fisher scale at admission showed a weaker, but nevertheless significant association
Acta Anaesthesiologica Scandinavica 59 (2015) 255–264 © 2015 The Acta Anaesthesiologica Scandinavica Foundation. Published by John Wiley & Sons Ltd
259
L. Z. CSAJBOK ET AL.
Table 3 Spearman’s correlation coefficient for relationship between WFNS, Fisher, CRP variables and outcome parameters GOSE and NIHSS. Spearman correlation coefficients Prob > |r| under H0: rho = 0 GOSE WFNS Neurological status at first attention n = 98 Fisher Radiological Fisher scale at admission n = 98 CRPmax n = 98 CRPmean n = 98 CRPd1 n = 87 CRPd2 n = 98 CRPd3 n = 98 CRPd4 n = 98
NIHSS
−0.44140 0.27385 P < 0.0001 P = 0.0133 −0.25239 0.15813 P = 0.0121 P = 0.1585 −0.44440 P < 0.0001 −0.48320 P < 0.0001 −0.19317 P = 0.0730 −0.42370 P < 0.0001 −0.42669 P < 0.0001 −0.38964 P < 0.0001
0.45189 P < 0.0001 0.44894 P < 0.0001 0.03161 P = 0.7920 0.26881 P = 0.0152 0.38633 P = 0.0004 0.39018 P = 0.0003
CRP, C-reactive protein; GOSE, Glasgow Outcome Score Extended; NIHSS, National Institutes of Health Stroke Scale; WFNS, World Federation of Neurosurgical Societies Scale.
with the GOSE (Table 3). The association between the initial WFNS score and the NIHSS was weak (r = 0.274, P = 0.01) while it was much stronger between CRPmean/CRPmax and the NIHSS (r = 0.449, P < 0.001/r = 0.452, P < 0.001). The relationship between the Fisher scale and the NIHSS was non-significant (r = 0.158, P = n.s.). Furthermore, we tested whether the available parameters, age, gender, radiological Fisher scale, clinical WFNS scale, infectious status and CRP, were independently able to predict outcome in a univariate logistic regression analysis by 1-year outcome, defined by the dichotomised GOSE scale. It showed significant ORs only for the WFNS scale (OR = 1.44, P = 0.019) and CRP parameters (Table 4). We inserted WFNS, Fisher and the different CRP variables into a multivariate logistic regression model, where the various CRP values were the only variables, which showed, after adjustment for the other factors, a significant relationship with long-term outcome. The ORs showed that each increase of CRPmean of 50 mg/l
(P < 0.001) or an increase in CRPmax of 100 mg/l (P = 0.012) tripled the risk of poor outcome. The CRP increase as early as on day 2 showed significant risk escalation on 1-year outcome (OR = 1.17/10 units, P = 0.006). To find cut-off values for poor or favourable outcome prediction, ROC curves were used. The area under the ROC curve showed moderate accuracy for CRP (CRPmean 0.76, P < 0.001, CRPmax 0.74, P < 0.001) by Swets’ criteria,20 poor accuracy for the WFNS score (0.67, P = 0.008) and no association for the Fisher scale (0.57, P = n.s.), rendering the WFNS and Fisher scale inapplicable for outcome prediction. To obtain earlier prognosis, we added the ROC curves for CRPd2 and CRPd3. Surprisingly, CRPd2 (0.7, P = 0.002) already provides a better opportunity for prediction than initial neurology (WFNS) and the Fisher scale. In predicting poor outcome, the best LR is found for a CRPmean of 3.8 at a cut-off value of 70 mg/l. This results in a PPV of 64%, with a specificity of 87% and a sensitivity of 52%. CRPmax had an LR of 4.3 at a cut-off point of 175 mg/l, with a PPV of 67%, specificity 90% and sensitivity 39%. The WFNS with its highest LR at 1.8 is inadequate for outcome prediction. The probability of poor outcome for the best predictors, derived from logistic regression, is depicted in Fig. 3. The area under the ROC curve, LHs and PPVs are thus higher for CRP than initial neurology measured by the WFNS scale in predicting outcome. Discussion In this study, we have demonstrated that there is a CRP response during the first week after an aSAH, treated with percutaneous endovascular coiling, and it correlates to the 1-year neurological outcome, which can be used as a prognostic factor. A natural hypothesis would be that any kind of relevant neurochemical marker is elevated in deeply unconscious patients with a large intracranial haemorrhage after an aSAH. Ideally, the marker should decrease if the patient’s condition improves and similarly increase if the cerebral ischemia propagates and the patient deteriorates. If the marker can be analysed in blood, this is advantageous. Recent investigations have revealed that ischaemic stroke is connected to Acta Anaesthesiologica Scandinavica 59 (2015) 255–264
260
© 2015 The Acta Anaesthesiologica Scandinavica Foundation. Published by John Wiley & Sons Ltd
CRP AS OUTCOME PREDICTOR AFTER aSAH
Table 4 Logistic regression analysis – prediction of poor outcome.
Variable
Value
n (%) of poor outcome
Gender
Female Male 26 ≤ 50 50 ≤ 57 57 ≤ 64 64–81 WFNS I WFNS II WFNS III WFNS IV WFNS V Fisher II Fisher III Fisher IV Non infected patients Infected patients 6–31 31 ≤ 80 80 ≤ 140 140–340 5 ≤ 20 20 ≤ 40 40 ≤ 71 71–168 5 ≤ 20 20 ≤ 50 50 ≤ 90 90–180 0 ≤ 20 20 ≤ 50 50 ≤ 90 90–180 5 ≤ 20 20 ≤ 70 70 ≤ 110 110–260 5 ≤ 20 20 ≤ 72 72 ≤ 110 110–340
22 (29.7%) 9 (37.5%) 7 (30.4%) 7 (30.4%) 7 (28.0%) 10 (37.0%) 4 (11.8%) 12 (40.0%) 2 (66.7%) 8 (40.0%) 5 (45.5%) 0 (0.0%) 8 (25.8%) 23 (35.9%) 10 (24.4%) 21 (36.8%) 2 (8.7%) 7 (26.9%) 6 (28.6%) 16 (57.1%) 1 (4.3%) 7 (30.4%) 7 (25.9%) 16 (64.0%) 21 (72.4%) 4 (13.8%) 2 (6.9%) 2 (6.9%) 3 (9.7%) 10 (32.2%) 11 (35.5%) 7 (15,3%) 1 (5.6%) 9 (26.5%) 7 (22.6%) 14 (56.0%) 1 (5.0%) 10 (27.8%) 6 (40.0%) 14 (51.9%)
Age (years)
Neurological status at first attention (WFNS score)
Radiological classification at first available CT scan (Fisher score) Infectious status during observation CRP max (OR per 10 units)
CRP mean (OR per 10 units)
CRP day 1 (OR per 10 units)
CRP day 2 (OR per 10 units)
CRP day 3 (OR per 10 units)
CRP day 4 (OR per 10 units)
OR (95%CI)
P-value
Area under ROC curve (95%CI)
1.42 (0.54–3.72)
0.48
0.53 (0.44–0.63)
1.01 (0.97–1.05)
0.58
0.54 (0.41–0.66)
1.44 (1.06–1.94)
0.019
0.67 (0.56–0.77)
1.91 (0.80–4.57)
0.14
0.57 (0.47–0.67)
1.81 (0.74–4.42)
0.19
0.57 (0.47–0.67)
1.12 (1.05–1.19)
0.0009
0.74 (0.64–0.84)
1.25 (1.10–1.42)
0.0006
0.76 (0.66–0.86)
1.19 (0.96–1.48)
0.1115
0.62 (0.48–0.73)
1.18 (1.06–1.32)
0.0030
0.70 (0.60–0.80)
1.13 (1.04–1.22)
0.0037
0.72 (0.62–0.82)
1.11 (1.03–1.19)
0.0048
0.73 (0.63–0.83)
The tests are performed with bivariate logistic regression and ROC analysis. OR is the odds ratio for an increase of the predictor per one unit. CI, confidence interval; CRP, C-reactive protein; CT, computed tomography; OR, odds ratio; ROC, receiver operating characteristic curve; WFNS, World Federation of Neurosurgical Societies Scale.
increases in CRP, with values correlated to prognosis.21,22 In stroke, it is assumed that the CRP elevation can be explained by inflammatory mechanisms involved in the reparatory phase, which begins after the ischeamic event.23 A persistently elevated CRP in stroke may represent either an ongoing inflammatory process or an
extension of the cerebral ischaemia. A growing body of evidence from animal models and from human studies indicates that inflammatory mechanisms contribute to secondary neuronal injury after cerebral ischaemia.24,25 It can be assumed that these mechanisms may also be applicable to patients with aSAH.
Acta Anaesthesiologica Scandinavica 59 (2015) 255–264 © 2015 The Acta Anaesthesiologica Scandinavica Foundation. Published by John Wiley & Sons Ltd
261
L. Z. CSAJBOK ET AL.
Fig. 3. Probability of poor outcome expressed as a measurement of C-reactive protein (CRP) on day 2 and day 3, CRPmean and CRPmax during the first week after aSAH calculated from the logistic regression model.
During the last 15 years, a new treatment for aneurysmal closure has been introduced. Using an endovascular technique detachable titanium coils fill the aneurysm, inducing thrombi in the aneurysmal sac. The lack of major surgical trauma, in this group of patients, makes it more probable that a CRP elevation is secondary to the effect of the initial haemorrhage. The main result of this study is that this CRP response exists, and it is strongly correlated to both outcome assessed with the GOSE and focal neurological deficit measured by the NIHSS after 1 year. The explanation for the noted CRP increase is unclear. It is possible that the angiography and coiling could result in an inflammatory reaction with elevated CRP. In studies with angiography26 and stenting27 of coronary and renal vessels, CRP increased but to a far less pronounced degree than in the present study. CRP elevation from this kind of catheterisation would not correlate to longterm neurology. Vasospasm is a putative cause of CRP increase. However, there is no proof of a causal relationship between CRP and vasospasm. On the contrary, purified human CRP has been shown to have vasodilatory effects both in vitro28 and in vivo.29,30 Further, it would be remarkable if, with its high dynamic range, CRP were an important regulator of vascular tone, although there is evidence that inflammatory mechanisms play a role in vasospasm.6,31 Thus, it seems plausible that the initial bleed and its consequences are important reasons for the elevated CRP. In recent years, increasing evidence has been gath-
ered, showing the importance of inflammatory processes with CRP elevation9,32,33 and IL accumulation after aSAH.32,34 Some investigators were able to demonstrate an association between delayed cerebral ischaemia and inflammatory mediators32 while others failed to show a relationship.9 Infection is a usual cause of CRP elevation, but in the present study the status of infection showed no association with neurological outcome. We believe that early prognosis is important. With signs of an unfavourable prognosis it is possible, at an early stage, to intensify monitoring, plan for ICU availability and mobilise early rehabilitation resources. Because predicting favourable outcome is more precise than predicting poor outcome, LR being three times higher, this enables the physician to reassure both patients and relatives. Conclusions CRP values showed a significant transient elevation after aSAH in patients treated with endovascular coiling. This elevated CRP pattern during the first 8 days correlated significantly to patients’ 1-year neurological outcome. Acknowledgement We would like to express our gratitude to Ingrid Eiving for her dedicated technical assistance. References 1. Feigin VL, Lawes CM, Bennett DA, Anderson CS. Stroke epidemiology: a review of population-based studies of incidence, prevalence, and case-fatality in the late 20th century. Lancet Neurol 2003; 2: 43–53. 2. Feigin VL, Findlay M. Advances in subarachnoid hemorrhage. Stroke 2006; 37: 305–8. 3. Fountas KN, Tasiou A, Kapsalaki EZ, Paterakis KN, Grigorian AA, Lee GP, Robinson JS Jr. Serum and cerebrospinal fluid C-reactive protein levels as predictors of vasospasm in aneurysmal subarachnoid hemorrhage. Clinical article. Neurosurg Focus 2009; 26: E22. 4. Batjer HH. Timing of operation for ruptured aneurysms: early surgery, In: Ratcheson RAWF ed. Concepts in neurosurgery. Baltimore: Williams & Wilkins, 1994: 46–53. Acta Anaesthesiologica Scandinavica 59 (2015) 255–264
262
© 2015 The Acta Anaesthesiologica Scandinavica Foundation. Published by John Wiley & Sons Ltd
CRP AS OUTCOME PREDICTOR AFTER aSAH
5. Jane JA, Kassell NF, Torner JC, Winn HR. The natural history of aneurysms and arteriovenous malformations. J Neurosurg 1985; 62: 321–3. 6. Bhardwaj A. SAH-induced cerebral vasospasm: unraveling molecular mechanisms of a complex disease. Stroke 2003; 34: 427–33. 7. Mirzayan MJ, Gharabaghi A, Samii M, Tatagiba M, Krauss JK, Rosahl SK. Response of C-reactive protein after craniotomy for microsurgery of intracranial tumors. Neurosurgery 2007; 60: 621–5, discussion 25. 8. Rothoerl RD, Axmann C, Pina AL, Woertgen C, Brawanski A. Possible role of the C-reactive protein and white blood cell count in the pathogenesis of cerebral vasospasm following aneurysmal subarachnoid hemorrhage. J Neurosurg Anesthesiol 2006; 18: 68–72. 9. Juvela S, Kuhmonen J, Siironen J. C-reactive protein as predictor for poor outcome after aneurysmal subarachnoid haemorrhage. Acta Neurochir (Wien) 2012; 154: 397–404. 10. Öst M, Nylén K, Csajbok L, Blennow K, Rosengren L, Nellgård B. Apolipoprotein E polymorfism and gender difference in outcome after severe traumatic brain injury. Acta Anaesthesiol Scand 2008; 52: 1364–9. 11. Hillman J, Fridriksson S, Nilsson O, Yu Z, Saveland H, Jakobsson KE. Immediate administration of tranexamic acid and reduced incidence of early rebleeding after aneurysmal subarachnoid hemorrhage: a prospective randomized study. J Neurosurg 2002; 97: 771–8. 12. Teasdale GM, Drake CG, Hunt W, Kassell N, Sano K, Pertuiset B, De Villiers JC. A universal subarachnoid hemorrhage scale: report of a committee of the World Federation of Neurosurgical Societies. J Neurol Neurosurg Psychiatry 1988; 51: 1457. 13. Fisher CM, Kistler JP, Davis JM. Relation of cerebral vasospasm to subarachnoid hemorrhage visualized by computerized tomographic scanning. Neurosurgery 1980; 6: 1–9. 14. 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: 573–85. 15. Brott T, Adams HP Jr., Olinger CP, Marler JR, Barsan WG, Biller J, Spilker J, Holleran R, Eberle R, Hertzberg V, Walker M. Measurements of acute cerebral infarction: a clinical examination scale. Stroke 1989; 20: 864–70. 16. Roche-Diagnostic GmbH. BH/Hitachi 917, CRP 1929372. Available at: www.biodinamica
17.
18.
19.
20. 21.
22.
23.
24.
25.
26.
27.
-Itda.com.br/documentos/bulas_H-902/CRP.pdf. 2001. Matthews JN, Altman DG, Campbell MJ, Royston P. Analysis of serial measurements in medical research. BMJ 1990; 300: 230–5. Zweig MH, Campbell G. Receiver-operating characteristic (ROC) plots: a fundamental evaluation tool in clinical medicine. Clin Chem 1993; 39: 561–77. Hanley JAM, McNeil BJ. The meaning and use of the area under the Receiver Operating Characteristic (ROC) curve. Radiol 1982; 143: 29–36. Swets JA. Measuring the accuracy of diagnostic systems. Science 1988; 240: 1285–93. Di Napoli M, Papa F. Clinical use of C-reactive protein for prognostic stratification in ischemic stroke: has the time come for including it in the patient risk profile? Stroke 2003; 34: 375–6, author reply 75-6. Sattar N, Murray HM, McConnachie A, Blauw GJ, Bollen EL, Buckley BM, Cobbe SM, Ford I, Gaw A, Hyland M, Jukema JW, Kamper AM, Macfarlane PW, Murphy MB, Packard CJ, Perry IJ, Stott DJ, Sweeney BJ, Twomey C, Westendorp RG, Shepherd J. C-reactive protein and prediction of coronary heart disease and global vascular events in the Prospective Study of Pravastatin in the Elderly at Risk (PROSPER). Circulation 2007; 115: 981–9. Tuttolomondo A, Di Sciacca R, Di Raimondo D, Renda C, Pinto A, Licata G. Inflammation as a therapeutic target in acute ischemic stroke treatment. Curr Top Med Chem 2009; 9: 1240–60. Wen YD, Zhang HL, Qin ZH. Inflammatory mechanism in ischemic neuronal injury. Neurosci Bull 2006; 22: 171–82. Denes A, Ferenczi S, Kovacs KJ. Systemic inflammatory challenges compromise survival after experimental stroke via augmenting brain inflammation, blood-brain barrier damage and brain oedema independently of infarct size. J Neuroinflammation 2011; 8: 164. Alhadad A, Guron G, Fortuna-Nowakowska E, Saeed A, Mattiasson I, Jensen G, Lindblad B, Gottsater A, Herlitz H. Renal angioplasty causes a rapid transient increase in inflammatory biomarkers, but reduced levels of interleukin-6 and endothelin-1 1 month after intervention. J Hypertens 2007; 25: 1907–14. Almagor M, Keren A, Banai S. Increased C-reactive protein level after coronary stent implantation in
Acta Anaesthesiologica Scandinavica 59 (2015) 255–264 © 2015 The Acta Anaesthesiologica Scandinavica Foundation. Published by John Wiley & Sons Ltd
263
L. Z. CSAJBOK ET AL.
28.
29.
30.
31.
patients with stable coronary artery disease. Am Heart J 2003; 145: 248–53. Sternik L, Samee S, Schaff HV, Zehr KJ, Lerman LO, Holmes DR, Herrmann J, Lerman A. C-reactive protein relaxes human vessels in vitro. Arterioscler Thromb Vasc Biol 2002; 22: 1865–8. Pepys MB, Hawkins PN, Kahan MC, Tennent GA, Gallimore JR, Graham D, Sabin CA, Zychlinsky A, de Diego J. Proinflammatory effects of bacterial recombinant human C-reactive protein are caused by contamination with bacterial products, not by C-reactive protein itself. Circ Res 2005; 97: e97–103. Bisoendial R, Birjmohun R, Keller T, van Leuven S, Levels H, Levi M, Kastelein J, Stroes E. In vivo effects of C-reactive protein (CRP)-infusion into humans. Circ Res 2005; 97: e115–6. Peterson JW, Kwun BD, Hackett JD, Zervas NT. The role of inflammation in experimental cerebral vasospasm. J Neurosurg 1990; 72: 767–74.
32. McMahon CJ, Hopkins S, Vail A, King AT, Smith D, Illingworth KJ, Clark S, Rothwell NJ, Tyrrell PJ. Inflammation as a predictor for delayed cerebral ischemia after aneurysmal subarachnoid haemorrhage. J Neurointerv Surg 2013; 5: 512–7. 33. Rodling-Wahlstrom M, Olivecrona M, Koskinen LO, Naredi S, Hultin M. Subarachnoid haemorrhage induces an inflammatory response followed by a delayed persisting increase in asymmetric dimethylarginine. Scand J Clin Lab Invest 2012; 72: 484–9. 34. Muroi C, Mink S, Seule M, Bellut D, Fandino J, Keller E. Monitoring of the inflammatory response after aneurysmal subarachnoid haemorrhage in the clinical setting: review of literature and report of preliminary clinical experience. Acta Neurochir Suppl 2011; 110: 191–6.
Acta Anaesthesiologica Scandinavica 59 (2015) 255–264
264
© 2015 The Acta Anaesthesiologica Scandinavica Foundation. Published by John Wiley & Sons Ltd
Copyright of Acta Anaesthesiologica Scandinavica is the property of Wiley-Blackwell and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.
In-hospital C-reactive protein predicts outcome after aneurysmal subarachnoid haemorrhage treated by endovascular coiling L. Z. Csajbok1, K. Nylén2, M. Öst1, H. Sonander1 and B. Nellgård1 1
Department of Anaesthesiology and Intensive Care, Institution of Clinical Sciences, Sahlgrenska Academy, Gothenburg University, Gothenburg, Sweden 2 Department of Neurology, Institution of Neurological Sciences, Sahlgrenska Academy, Gothenburg University, Gothenburg, Sweden
Correspondence L. Csajbok, Department of Anaesthesiology and Intensive Care, Sahlgrenska University Hospital, S-413 45 Göteborg, Sweden E-mail: [email protected] Conflicts of interest The authors confirm that there are no conflicts of interest. Funding This study was supported by grants from the Göteborg Medical Society, ALF Swedish government research grants, the Elsa and Gustav Lindh Foundation, and the John and Brit Wennerström Foundation. Submitted 9 September 2014; accepted 19 October 2014; submission 25 March 2014. Citation Csajbok LZ, Nylén K, Öst M, Sonander H, Nellgård B. In-hospital C-reactive protein predicts outcome after aneurysmal subarachnoid haemorrhage treated by endovascular coiling. Acta Anaesthesiologica Scandinavica 2015 doi: 10.1111/aas.12441
Background: This study aimed to examine prospectively whether the inflammatory marker C-reactive protein (CRP) increases in patients with aneurysmal subarachnoid haemorrhage (aSAH) treated by endovascular coiling and investigate whether CRP could be used as prognostic factor for long-term neurological outcome. Methods: This single-hospital study comprised 98 consecutive patients with confirmed aSAH treated by endovascular coiling. Admission status was classified according to the World Federation of Neurosurgical Societies (WFNS) Scale and initial cerebral computed tomography according to Fisher scale. CRP was analysed on days 0, 1, 2, 3, 4, 6 and 8 after the initial bleed. A neurological follow up was performed 1 year later according to the Extended Glasgow Outcome Scale (GOSE) for overall outcome and National Institute of Health Stroke Scale (NIHSS) for focal deficit. Results: CRP values increased from normal to peak at 53 mg/l at day 3–4 and then declined, without normalising, at day 8. Patients with a higher increase in CRP had a poorer neurological outcome after 1 year. CRP during the first week had a stronger correlation to outcome (r = 0.417) and NIHSS (r = 0.449) than initial clinical status (WFNS; r = 0.280 and 0.274) and radiology (Fisher scale; r = 0.137 and 0.158). CRP increase indicated a risk of poor outcome (GOSE) (P < 0.001) and permanent loss of neurological function (NIHSS) (P < 0.001). Logistic regression analysis suggested that elevated CRP already on day 2 is an independent prognostic marker for outcome. Conclusion: Early CRP values can perhaps be used as a prognostic factor for long-term neurological outcome prediction after endovascular treatment of aSAH.
Editorial comment: what this article tells us This cohort study found that the early increase in CRP is a strong prognostic factor for poor 1-year neurological outcome in aSAH patients undergone endovascular treatment. This observation may help clinicians to adjust optimal treatment and monitoring shortly after coiling of an aneurysm in SAH patients.
Acta Anaesthesiologica Scandinavica 59 (2015) 255–264 © 2015 The Acta Anaesthesiologica Scandinavica Foundation. Published by John Wiley & Sons Ltd
bs_bs_banner
255
L. Z. CSAJBOK ET AL.
Stroke is one of the leading causes of death and disability in modern society. Subarachnoid haemorrhage (SAH) accounts for up to 7% of all strokes1 and thus has an incidence of around 10/100,000 persons/year.1 The rupture of an intracerebral aneurysm is a devastating event, accounting for up to 25% of cerebrovascular deaths,1,2 rendering an overall poor outcome, dependence or death in 65–70% of the patients.2,3 However, with early surgical or endovascular intervention to secure a ruptured aneurysm, outcome has improved.4 This is achieved primarily by reducing the risk of re-bleeding, otherwise seen in 50% of patients within 6 months.5 Even when the aneurysm is secured, many patients develop late neurological complications. The underlying pathogenic mechanisms for these delayed neurological deficits are still obscure, but inflammation is one factor that has been suggested to play a role.6 The surgical clipping of aneurysms, like all surgical traumas, including neurosurgery, is followed by an inflammatory reaction, manifested by elevated levels of different inflammatory markers, including C-reactive protein (CRP).7 CRP is mainly synthesised in the liver, after induction by cytokines, particularly interleukin6. Following release into the circulation, CRP is evenly distributed in the vascular compartment. CRP responds rapidly after an inflammatory stimulus and can increase up to 1000-fold with a peak at approximately 48 h after the initial stimulus, after which it falls once the inflammatory stimulus is removed. The distribution in the vascular compartment could be explained by CRP’s putative role as a substance that neutralises or detoxifies harmful substances escaping from the site of inflammation into the circulation. The number of hepatocytes synthesising CRP increases as the intensity of the stimulus increases. Most of the CRP is mono-exponentially cleared from the circulation with a biological halflife of ∼19 h, so the only significant determinant of plasma levels of CRP is the rate of synthesis. This justifies the use of CRP to monitor the inflammatory activity. The surgical clipping of aneurysms has been shown to result in elevated CRP levels, which could be correlated to the 1-year outcome measured by the Glasgow Outcome Scale.8,9 Endovascular coiling has become more or less the
therapeutic choice for most aneurysms. We hypothesised that CRP levels may also be elevated in this group of patients, and, if so, it might correlate to the 1-year outcome. If the initial bleed causes an inflammatory reaction and this reaction is not clouded by a surgical intervention, the CRP elevation should be more moderate and should relate more closely to the effect of the haemorrhage than in operated patients. The aim of this study was to evaluate whether there is a CRP response after SAH in patients treated by endovascular coiling and if so, what relationship it has to the long-term outcome. Methods Patients The patients recruited to this study were admitted to the Neurosurgical Intensive Care Unit (NICU) at Sahlgrenska University Hospital during a 2-year period after subarachnoid blood was detected either on a computed tomography (CT) scan or at a lumbar puncture. The following inclusion criteria were met: the patient was admitted for acute aneurysmal SAH (aSAH), within 48 h after the initial haemorrhage, an aneurysm was visible on the digital subtraction selective angiography, and the treatment was determined to be endovascular coiling of the aneurysm. Furthermore, the patients had to be residing permanently in Sweden, that is, available for outcome investigation, and informed consent had to be obtained from patients or next of kin. The Gothenburg University Medical Ethics Committee (Forskningsetikkommiten, Avd. För allmänmedicin, Vasa Sjukhus, 411 33 Göteborg) approved the study with the approval number S 161-00 on 29 June 2000. All patients were treated according to a standardised protocol10 that included intravenous injections of tranexamic acid (Cyclokapron®, Pfizer AB, Sollentuna, Sweden) 1 g four times daily, a fibrinolytic agent to minimise the risk of re-bleeding.11 To reduce the risk of cerebrovascular vasospasm, a continuous infusion of nimodipine (Nimotop®, Bayer AB, Solna, Sweden), a selective calcium-receptor antagonist, was started at 2 mg/h at NICU admission and was then given for 10 days. However, a few patients received oral nimodipine (60 mg every 4 h) for the final 3 days. Acta Anaesthesiologica Scandinavica 59 (2015) 255–264
256
© 2015 The Acta Anaesthesiologica Scandinavica Foundation. Published by John Wiley & Sons Ltd
CRP AS OUTCOME PREDICTOR AFTER aSAH
Neurological assessment At hospital admission, a short neurological examination was performed by the admitting physician. The status was recorded and classified according to the World Federation of Neurosurgical Societies (WFNS) Scale.12 A CT of the brain was acquired and categorised according to the Fisher scale13 of SAH. Neurological status was registered in the NICU from four times daily to every 15 min, depending on the patient’s condition. One year (range 11.5–13 months) after the aSAH, one neurologist, blinded to all laboratory data, examined all the patients according to a standardised protocol. Outcome was measured using the Extended Glasgow Outcome (GOSE) score,14 an eight-level scale, where GOSE 1 is deceased, and GOSE 8 represents full recovery. This scale was subsequently dichotomised to calculate cut-off values for favourable (GOSE 5–8) and poor outcome (GOSE 1–4). The patients’ focal neurological deficit was assessed using the National Institute of Health Stroke Scale (NIHSS).15 Blood sampling procedures and laboratory assay Blood sampling for CRP analysis was started after admission and continued on days 1, 2, 3, 4, 6 and 8 after the initial haemorrhage (day 0). CRP was analysed with an immunochemical kinetic turbidimetry method16 within an hour of sampling in the hospitals accredited laboratory. The level of faulty margin is under 4% according to the ISO/ IEC 17 025 accreditation standard. Statistical analysis The Kolmogorov–Smirnov test showed that the CRP data were not normally distributed, and non-parametric statistics were therefore used. Statistical analyses were performed with the SPSS version 21 (IBM, Armonk, NY, USA) and SAS version 9.4 (SAS Institute Inc, Cary, NC, USA) statistics packages, where the correlation between variables was analysed with Spearman’s rank correlation test, while Wilcoxon’s signed rank test was used to compare dependent samples. All analyses were two tailed and performed at a 5% level of significance.
One way of looking at data related to variables changing over time, according to Matthews,17 is to analyse of the mean and maximum of the series. The CRPmean, and CRPmax during the first 8 days were therefore used for further statistical analysis. In search of an early prediction, we added the means of the CRP values on observation days CRPd1-8. We tested the parameters for associations with Spearman’s correlation analysis. The potential prognostic factors were tested with bivariate logistic regression, and odds ratio (OR), 95% confidence interval and P-values were calculated. To clarify the internal relationships between the prognostic factors, CRP was inserted into a multivariate regression model. On the strongest predictors, receiver operating characteristic curve (ROC) analysis18,19 was employed to choose a cut-off value where possible. Sensitivity, specificity and positive predictive values (PPVs) were calculated where the likelihood ratio (LH) of the test was highest. Results During the study period, 202 patients were initially admitted to the NICU with a diagnosis of aSAH, and of these, 98 met the inclusion criteria. Of the non-eligible patients, three were admitted later than day 2, in 33 patients no aneurysm was detected, angiography could not be performed in four patients, in nine cases no informed consent could be obtained or consent was withdrawn, three patients were lost to long-term follow up and in 39 patients surgical clipping was the treatment of choice. In 13 patients, consecutive data were missing. As a result, the study population consisted of 74 women (76%) and 24 men (24%) with a median age of 57 years (range 26–81). The patients’ demographics, their initial neurological evaluation, infectious status during the study period and CRP values are shown in Table 1. The CRP values showed a continuous rise from an initial median value on day 0 of 5 mg/l [interquartile range (IQR) 5–7] until its maximum at day 3–4 (median 53 mg/l, IQR 24–110); thereafter, it decreased. The increase from the initial value until its peak on day 3–4 was significant between each of the days as well as the decrease from day 4 to day 6 and 8. The CRP value did not return to baseline, as the day 8 value (median 24 mg/l, IQR 10–47) was still significantly higher than the initial value (Fig. 1).
Acta Anaesthesiologica Scandinavica 59 (2015) 255–264 © 2015 The Acta Anaesthesiologica Scandinavica Foundation. Published by John Wiley & Sons Ltd
257
L. Z. CSAJBOK ET AL.
Table 1 Patients’ characteristics at admission and during observation.
Variables Gender Female Male Age (years)
Total (n = 98)
Favourable outcome (GOSE 5–8) (n = 67)
74 (75.5%) 24 (24.5%) 56.5 (10.7) 57.5 (26.0; 81.0) n = 98 Neurological status at first attention (WFNS score) WFNS I 34 (34.7%) WFNS II 30 (30.6%) WFNS III 3 (3.1%) WFNS IV 20 (20.4%) WFNS V 11 (11.2%) Radiological classification at first available CT scan (Fisher score) Fisher II 3 (3.1%) Fisher III 31 (31.6%) Fisher IV 64 (65.3%) Infectious status during observation Non infected patients 41 (41.8%) Infected patients 57 (58.2%) CRP max (mg/l) 98.9 (73.9) 79.5 (6.0; 340.0) n = 98 CRP mean (mg/l) 51.5 (38.2) 41.4 (5.2; 167.5) n = 98 CRP day 0 (mg/l) 10.5 (19.7) 5.0 (5.0; 100.0) n = 23 CRP day 1 (mg/l) 17.9 (24.4) 9.0 (5.0; 180.0) n = 87 CRP day 2 (mg/l) 47.2 (40.5) 27.0 (5.0; 180.0) n = 98 CRP day 3 (mg/l) 71.6 (57.0) 62.5 (5.0; 260.0) n = 98 CRP day 4 (mg/l) 76.0 (66.2) 53.0 (5.0; 340.0) n = 98 CRP day 6 (mg/l) 59.7 (65.3) 34.0 (5.0; 260.0) n = 98 CRP day 8 (mg/l) 42.9 (52.2) 24.0 (5.0; 240.0) n = 98
52 (77.6%) 15 (22.4%) 56.1 (10.3) 58.0 (26.0; 81.0) n = 67
Poor outcome (GOSE 1–4) (n = 31)
22 (71.0%) 9 (29.0%) 57.4 (11.4) 57.0 (30.0; 78.0) n = 31
30 (44.8%) 18 (26.9%) 1 (1.5%) 12 (17.9%) 6 (9.0%)
4 (12.9%) 12 (38.7%) 2 (6.5%) 8 (25.8%) 5 (16.1%)
3 (4.5%) 23 (34.3%) 41 (61.2%)
0 (0.0%) 8 (25.8%) 23 (74.2%)
31 (46.3%) 36 (53.7%) 80.8 (68.4) 68.0 (6.0; 340.0) n = 67 41.7 (34.6) 31.4 (5.2; 167.5) n = 67 6.58 (3.45) 5.00 (5.00; 16.00) n = 12 14.6 (14.7) 8.0 (5.0; 78.0) n = 58 38.5 (36.1) 24.0 (5.0; 140.0) n = 67 59.7 (54.9) 37.0 (5.0; 260.0) n = 67 62.2 (61.4) 41.0 (5.0; 340.0) n = 67 47.6 (54.2) 23.0 (5.0; 250.0) n = 67 30.9 (33.5) 17.0 (5.0; 180.0) n = 67
10 (32.3%) 21 (67.7%) 138.1 (71.1) 140.0 (22.0; 280.0) n = 31 72.7 (37.5) 71.0 (14.9; 165.5) n = 31 14.8 (28.4) 5.0 (5.0; 100.0) n = 11 24.6 (36.3) 14.0 (5.0; 180.0) n = 29 66.0 (43.6) 62.0 (12.0; 180.0) n = 31 97.5 (53.4) 89.0 (18.0; 190.0) n = 31 105.8 (67.2) 87.0 (18.0; 280.0) n = 31 85.8 (79.5) 47.0 (10.0; 260.0) n = 31 68.8 (73.1) 34.0 (6.0; 240.0) n = 31
For categorical variables n (%) is presented. For continuous variables, mean (SD)/median (min; max)/n = is presented. CRP, C-reactive protein; CT, computed tomography; WFNS, World Federation of Neurosurgical Societies Scale.
Acta Anaesthesiologica Scandinavica 59 (2015) 255–264
258
© 2015 The Acta Anaesthesiologica Scandinavica Foundation. Published by John Wiley & Sons Ltd
CRP AS OUTCOME PREDICTOR AFTER aSAH
Fig. 1. Box plot of C-reactive protein (CRP) values from day 0 to day 8 with median, mean and interquartile range. Differences tested with Wilcoxon’s signed rank test.
Table 2 Patients’ neurological outcome after 1 year. All patients (n = 98) Glasgow Outcome Score Extended GOSE 1 (dead) GOSE 3 (severe disability lower) GOSE 4 (severe disability upper) GOSE 5 (moderate disability lower) GOSE 6 (moderate disability upper) GOSE 7 (favourable recovery lower) GOSE 8 (favourable recovery upper) National Institutes of Health Stroke Scale NIHSS 0 p NIHSS 1–10 p NIHSS 11–20 p NIHSS 21–30 p NIHSS not accessible
15 (15.3%) 16 (16.3%) 0 (0%) 20 (20.4%) 18 (18.4%) 7 (7.1%) 22 (22.4%) 59 (60.2%) 15 (15.3%) 3 (3.1%) 4 (4.1%) 17 (17.3%)
For categorical variables n (%) is presented.
The patients’ infectious status was evaluated during the observation period. It was considered positive if any bacterial culture proved positive and/or more than prophylactic antibiotics (3 doses) were administered. No association could be shown between the status of infection and neurological outcome (GOSE, NIHSS). The results of the 1-year neurological follow up are presented in Table 2 as outcome parameters. The time-dependent course of CRP levels in the
Fig. 2. Box plot of time-dependent course of C-reactive protein (CRP) levels in the favourable and unfavourable patient groups with median, mean and interquartile range.
favourable and unfavourable patient groups is depicted in Fig. 2. There was a significant correlation between the initial WFNS score, CRPmean and CRPmax during the first week (r = 0.393, P < 0.001 and r = 0.316, P = 0.002 respectively), as well as between the initial WFNS score and the GOSE (r = 0.441, P < 0.001). An even stronger correlation was found between CRPmean/CRPmax and the GOSE (r = 0.483, P < 0.001/0.444, P < 0.001), while the radiological Fisher scale at admission showed a weaker, but nevertheless significant association
Acta Anaesthesiologica Scandinavica 59 (2015) 255–264 © 2015 The Acta Anaesthesiologica Scandinavica Foundation. Published by John Wiley & Sons Ltd
259
L. Z. CSAJBOK ET AL.
Table 3 Spearman’s correlation coefficient for relationship between WFNS, Fisher, CRP variables and outcome parameters GOSE and NIHSS. Spearman correlation coefficients Prob > |r| under H0: rho = 0 GOSE WFNS Neurological status at first attention n = 98 Fisher Radiological Fisher scale at admission n = 98 CRPmax n = 98 CRPmean n = 98 CRPd1 n = 87 CRPd2 n = 98 CRPd3 n = 98 CRPd4 n = 98
NIHSS
−0.44140 0.27385 P < 0.0001 P = 0.0133 −0.25239 0.15813 P = 0.0121 P = 0.1585 −0.44440 P < 0.0001 −0.48320 P < 0.0001 −0.19317 P = 0.0730 −0.42370 P < 0.0001 −0.42669 P < 0.0001 −0.38964 P < 0.0001
0.45189 P < 0.0001 0.44894 P < 0.0001 0.03161 P = 0.7920 0.26881 P = 0.0152 0.38633 P = 0.0004 0.39018 P = 0.0003
CRP, C-reactive protein; GOSE, Glasgow Outcome Score Extended; NIHSS, National Institutes of Health Stroke Scale; WFNS, World Federation of Neurosurgical Societies Scale.
with the GOSE (Table 3). The association between the initial WFNS score and the NIHSS was weak (r = 0.274, P = 0.01) while it was much stronger between CRPmean/CRPmax and the NIHSS (r = 0.449, P < 0.001/r = 0.452, P < 0.001). The relationship between the Fisher scale and the NIHSS was non-significant (r = 0.158, P = n.s.). Furthermore, we tested whether the available parameters, age, gender, radiological Fisher scale, clinical WFNS scale, infectious status and CRP, were independently able to predict outcome in a univariate logistic regression analysis by 1-year outcome, defined by the dichotomised GOSE scale. It showed significant ORs only for the WFNS scale (OR = 1.44, P = 0.019) and CRP parameters (Table 4). We inserted WFNS, Fisher and the different CRP variables into a multivariate logistic regression model, where the various CRP values were the only variables, which showed, after adjustment for the other factors, a significant relationship with long-term outcome. The ORs showed that each increase of CRPmean of 50 mg/l
(P < 0.001) or an increase in CRPmax of 100 mg/l (P = 0.012) tripled the risk of poor outcome. The CRP increase as early as on day 2 showed significant risk escalation on 1-year outcome (OR = 1.17/10 units, P = 0.006). To find cut-off values for poor or favourable outcome prediction, ROC curves were used. The area under the ROC curve showed moderate accuracy for CRP (CRPmean 0.76, P < 0.001, CRPmax 0.74, P < 0.001) by Swets’ criteria,20 poor accuracy for the WFNS score (0.67, P = 0.008) and no association for the Fisher scale (0.57, P = n.s.), rendering the WFNS and Fisher scale inapplicable for outcome prediction. To obtain earlier prognosis, we added the ROC curves for CRPd2 and CRPd3. Surprisingly, CRPd2 (0.7, P = 0.002) already provides a better opportunity for prediction than initial neurology (WFNS) and the Fisher scale. In predicting poor outcome, the best LR is found for a CRPmean of 3.8 at a cut-off value of 70 mg/l. This results in a PPV of 64%, with a specificity of 87% and a sensitivity of 52%. CRPmax had an LR of 4.3 at a cut-off point of 175 mg/l, with a PPV of 67%, specificity 90% and sensitivity 39%. The WFNS with its highest LR at 1.8 is inadequate for outcome prediction. The probability of poor outcome for the best predictors, derived from logistic regression, is depicted in Fig. 3. The area under the ROC curve, LHs and PPVs are thus higher for CRP than initial neurology measured by the WFNS scale in predicting outcome. Discussion In this study, we have demonstrated that there is a CRP response during the first week after an aSAH, treated with percutaneous endovascular coiling, and it correlates to the 1-year neurological outcome, which can be used as a prognostic factor. A natural hypothesis would be that any kind of relevant neurochemical marker is elevated in deeply unconscious patients with a large intracranial haemorrhage after an aSAH. Ideally, the marker should decrease if the patient’s condition improves and similarly increase if the cerebral ischemia propagates and the patient deteriorates. If the marker can be analysed in blood, this is advantageous. Recent investigations have revealed that ischaemic stroke is connected to Acta Anaesthesiologica Scandinavica 59 (2015) 255–264
260
© 2015 The Acta Anaesthesiologica Scandinavica Foundation. Published by John Wiley & Sons Ltd
CRP AS OUTCOME PREDICTOR AFTER aSAH
Table 4 Logistic regression analysis – prediction of poor outcome.
Variable
Value
n (%) of poor outcome
Gender
Female Male 26 ≤ 50 50 ≤ 57 57 ≤ 64 64–81 WFNS I WFNS II WFNS III WFNS IV WFNS V Fisher II Fisher III Fisher IV Non infected patients Infected patients 6–31 31 ≤ 80 80 ≤ 140 140–340 5 ≤ 20 20 ≤ 40 40 ≤ 71 71–168 5 ≤ 20 20 ≤ 50 50 ≤ 90 90–180 0 ≤ 20 20 ≤ 50 50 ≤ 90 90–180 5 ≤ 20 20 ≤ 70 70 ≤ 110 110–260 5 ≤ 20 20 ≤ 72 72 ≤ 110 110–340
22 (29.7%) 9 (37.5%) 7 (30.4%) 7 (30.4%) 7 (28.0%) 10 (37.0%) 4 (11.8%) 12 (40.0%) 2 (66.7%) 8 (40.0%) 5 (45.5%) 0 (0.0%) 8 (25.8%) 23 (35.9%) 10 (24.4%) 21 (36.8%) 2 (8.7%) 7 (26.9%) 6 (28.6%) 16 (57.1%) 1 (4.3%) 7 (30.4%) 7 (25.9%) 16 (64.0%) 21 (72.4%) 4 (13.8%) 2 (6.9%) 2 (6.9%) 3 (9.7%) 10 (32.2%) 11 (35.5%) 7 (15,3%) 1 (5.6%) 9 (26.5%) 7 (22.6%) 14 (56.0%) 1 (5.0%) 10 (27.8%) 6 (40.0%) 14 (51.9%)
Age (years)
Neurological status at first attention (WFNS score)
Radiological classification at first available CT scan (Fisher score) Infectious status during observation CRP max (OR per 10 units)
CRP mean (OR per 10 units)
CRP day 1 (OR per 10 units)
CRP day 2 (OR per 10 units)
CRP day 3 (OR per 10 units)
CRP day 4 (OR per 10 units)
OR (95%CI)
P-value
Area under ROC curve (95%CI)
1.42 (0.54–3.72)
0.48
0.53 (0.44–0.63)
1.01 (0.97–1.05)
0.58
0.54 (0.41–0.66)
1.44 (1.06–1.94)
0.019
0.67 (0.56–0.77)
1.91 (0.80–4.57)
0.14
0.57 (0.47–0.67)
1.81 (0.74–4.42)
0.19
0.57 (0.47–0.67)
1.12 (1.05–1.19)
0.0009
0.74 (0.64–0.84)
1.25 (1.10–1.42)
0.0006
0.76 (0.66–0.86)
1.19 (0.96–1.48)
0.1115
0.62 (0.48–0.73)
1.18 (1.06–1.32)
0.0030
0.70 (0.60–0.80)
1.13 (1.04–1.22)
0.0037
0.72 (0.62–0.82)
1.11 (1.03–1.19)
0.0048
0.73 (0.63–0.83)
The tests are performed with bivariate logistic regression and ROC analysis. OR is the odds ratio for an increase of the predictor per one unit. CI, confidence interval; CRP, C-reactive protein; CT, computed tomography; OR, odds ratio; ROC, receiver operating characteristic curve; WFNS, World Federation of Neurosurgical Societies Scale.
increases in CRP, with values correlated to prognosis.21,22 In stroke, it is assumed that the CRP elevation can be explained by inflammatory mechanisms involved in the reparatory phase, which begins after the ischeamic event.23 A persistently elevated CRP in stroke may represent either an ongoing inflammatory process or an
extension of the cerebral ischaemia. A growing body of evidence from animal models and from human studies indicates that inflammatory mechanisms contribute to secondary neuronal injury after cerebral ischaemia.24,25 It can be assumed that these mechanisms may also be applicable to patients with aSAH.
Acta Anaesthesiologica Scandinavica 59 (2015) 255–264 © 2015 The Acta Anaesthesiologica Scandinavica Foundation. Published by John Wiley & Sons Ltd
261
L. Z. CSAJBOK ET AL.
Fig. 3. Probability of poor outcome expressed as a measurement of C-reactive protein (CRP) on day 2 and day 3, CRPmean and CRPmax during the first week after aSAH calculated from the logistic regression model.
During the last 15 years, a new treatment for aneurysmal closure has been introduced. Using an endovascular technique detachable titanium coils fill the aneurysm, inducing thrombi in the aneurysmal sac. The lack of major surgical trauma, in this group of patients, makes it more probable that a CRP elevation is secondary to the effect of the initial haemorrhage. The main result of this study is that this CRP response exists, and it is strongly correlated to both outcome assessed with the GOSE and focal neurological deficit measured by the NIHSS after 1 year. The explanation for the noted CRP increase is unclear. It is possible that the angiography and coiling could result in an inflammatory reaction with elevated CRP. In studies with angiography26 and stenting27 of coronary and renal vessels, CRP increased but to a far less pronounced degree than in the present study. CRP elevation from this kind of catheterisation would not correlate to longterm neurology. Vasospasm is a putative cause of CRP increase. However, there is no proof of a causal relationship between CRP and vasospasm. On the contrary, purified human CRP has been shown to have vasodilatory effects both in vitro28 and in vivo.29,30 Further, it would be remarkable if, with its high dynamic range, CRP were an important regulator of vascular tone, although there is evidence that inflammatory mechanisms play a role in vasospasm.6,31 Thus, it seems plausible that the initial bleed and its consequences are important reasons for the elevated CRP. In recent years, increasing evidence has been gath-
ered, showing the importance of inflammatory processes with CRP elevation9,32,33 and IL accumulation after aSAH.32,34 Some investigators were able to demonstrate an association between delayed cerebral ischaemia and inflammatory mediators32 while others failed to show a relationship.9 Infection is a usual cause of CRP elevation, but in the present study the status of infection showed no association with neurological outcome. We believe that early prognosis is important. With signs of an unfavourable prognosis it is possible, at an early stage, to intensify monitoring, plan for ICU availability and mobilise early rehabilitation resources. Because predicting favourable outcome is more precise than predicting poor outcome, LR being three times higher, this enables the physician to reassure both patients and relatives. Conclusions CRP values showed a significant transient elevation after aSAH in patients treated with endovascular coiling. This elevated CRP pattern during the first 8 days correlated significantly to patients’ 1-year neurological outcome. Acknowledgement We would like to express our gratitude to Ingrid Eiving for her dedicated technical assistance. References 1. Feigin VL, Lawes CM, Bennett DA, Anderson CS. Stroke epidemiology: a review of population-based studies of incidence, prevalence, and case-fatality in the late 20th century. Lancet Neurol 2003; 2: 43–53. 2. Feigin VL, Findlay M. Advances in subarachnoid hemorrhage. Stroke 2006; 37: 305–8. 3. Fountas KN, Tasiou A, Kapsalaki EZ, Paterakis KN, Grigorian AA, Lee GP, Robinson JS Jr. Serum and cerebrospinal fluid C-reactive protein levels as predictors of vasospasm in aneurysmal subarachnoid hemorrhage. Clinical article. Neurosurg Focus 2009; 26: E22. 4. Batjer HH. Timing of operation for ruptured aneurysms: early surgery, In: Ratcheson RAWF ed. Concepts in neurosurgery. Baltimore: Williams & Wilkins, 1994: 46–53. Acta Anaesthesiologica Scandinavica 59 (2015) 255–264
262
© 2015 The Acta Anaesthesiologica Scandinavica Foundation. Published by John Wiley & Sons Ltd
CRP AS OUTCOME PREDICTOR AFTER aSAH
5. Jane JA, Kassell NF, Torner JC, Winn HR. The natural history of aneurysms and arteriovenous malformations. J Neurosurg 1985; 62: 321–3. 6. Bhardwaj A. SAH-induced cerebral vasospasm: unraveling molecular mechanisms of a complex disease. Stroke 2003; 34: 427–33. 7. Mirzayan MJ, Gharabaghi A, Samii M, Tatagiba M, Krauss JK, Rosahl SK. Response of C-reactive protein after craniotomy for microsurgery of intracranial tumors. Neurosurgery 2007; 60: 621–5, discussion 25. 8. Rothoerl RD, Axmann C, Pina AL, Woertgen C, Brawanski A. Possible role of the C-reactive protein and white blood cell count in the pathogenesis of cerebral vasospasm following aneurysmal subarachnoid hemorrhage. J Neurosurg Anesthesiol 2006; 18: 68–72. 9. Juvela S, Kuhmonen J, Siironen J. C-reactive protein as predictor for poor outcome after aneurysmal subarachnoid haemorrhage. Acta Neurochir (Wien) 2012; 154: 397–404. 10. Öst M, Nylén K, Csajbok L, Blennow K, Rosengren L, Nellgård B. Apolipoprotein E polymorfism and gender difference in outcome after severe traumatic brain injury. Acta Anaesthesiol Scand 2008; 52: 1364–9. 11. Hillman J, Fridriksson S, Nilsson O, Yu Z, Saveland H, Jakobsson KE. Immediate administration of tranexamic acid and reduced incidence of early rebleeding after aneurysmal subarachnoid hemorrhage: a prospective randomized study. J Neurosurg 2002; 97: 771–8. 12. Teasdale GM, Drake CG, Hunt W, Kassell N, Sano K, Pertuiset B, De Villiers JC. A universal subarachnoid hemorrhage scale: report of a committee of the World Federation of Neurosurgical Societies. J Neurol Neurosurg Psychiatry 1988; 51: 1457. 13. Fisher CM, Kistler JP, Davis JM. Relation of cerebral vasospasm to subarachnoid hemorrhage visualized by computerized tomographic scanning. Neurosurgery 1980; 6: 1–9. 14. 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: 573–85. 15. Brott T, Adams HP Jr., Olinger CP, Marler JR, Barsan WG, Biller J, Spilker J, Holleran R, Eberle R, Hertzberg V, Walker M. Measurements of acute cerebral infarction: a clinical examination scale. Stroke 1989; 20: 864–70. 16. Roche-Diagnostic GmbH. BH/Hitachi 917, CRP 1929372. Available at: www.biodinamica
17.
18.
19.
20. 21.
22.
23.
24.
25.
26.
27.
-Itda.com.br/documentos/bulas_H-902/CRP.pdf. 2001. Matthews JN, Altman DG, Campbell MJ, Royston P. Analysis of serial measurements in medical research. BMJ 1990; 300: 230–5. Zweig MH, Campbell G. Receiver-operating characteristic (ROC) plots: a fundamental evaluation tool in clinical medicine. Clin Chem 1993; 39: 561–77. Hanley JAM, McNeil BJ. The meaning and use of the area under the Receiver Operating Characteristic (ROC) curve. Radiol 1982; 143: 29–36. Swets JA. Measuring the accuracy of diagnostic systems. Science 1988; 240: 1285–93. Di Napoli M, Papa F. Clinical use of C-reactive protein for prognostic stratification in ischemic stroke: has the time come for including it in the patient risk profile? Stroke 2003; 34: 375–6, author reply 75-6. Sattar N, Murray HM, McConnachie A, Blauw GJ, Bollen EL, Buckley BM, Cobbe SM, Ford I, Gaw A, Hyland M, Jukema JW, Kamper AM, Macfarlane PW, Murphy MB, Packard CJ, Perry IJ, Stott DJ, Sweeney BJ, Twomey C, Westendorp RG, Shepherd J. C-reactive protein and prediction of coronary heart disease and global vascular events in the Prospective Study of Pravastatin in the Elderly at Risk (PROSPER). Circulation 2007; 115: 981–9. Tuttolomondo A, Di Sciacca R, Di Raimondo D, Renda C, Pinto A, Licata G. Inflammation as a therapeutic target in acute ischemic stroke treatment. Curr Top Med Chem 2009; 9: 1240–60. Wen YD, Zhang HL, Qin ZH. Inflammatory mechanism in ischemic neuronal injury. Neurosci Bull 2006; 22: 171–82. Denes A, Ferenczi S, Kovacs KJ. Systemic inflammatory challenges compromise survival after experimental stroke via augmenting brain inflammation, blood-brain barrier damage and brain oedema independently of infarct size. J Neuroinflammation 2011; 8: 164. Alhadad A, Guron G, Fortuna-Nowakowska E, Saeed A, Mattiasson I, Jensen G, Lindblad B, Gottsater A, Herlitz H. Renal angioplasty causes a rapid transient increase in inflammatory biomarkers, but reduced levels of interleukin-6 and endothelin-1 1 month after intervention. J Hypertens 2007; 25: 1907–14. Almagor M, Keren A, Banai S. Increased C-reactive protein level after coronary stent implantation in
Acta Anaesthesiologica Scandinavica 59 (2015) 255–264 © 2015 The Acta Anaesthesiologica Scandinavica Foundation. Published by John Wiley & Sons Ltd
263
L. Z. CSAJBOK ET AL.
28.
29.
30.
31.
patients with stable coronary artery disease. Am Heart J 2003; 145: 248–53. Sternik L, Samee S, Schaff HV, Zehr KJ, Lerman LO, Holmes DR, Herrmann J, Lerman A. C-reactive protein relaxes human vessels in vitro. Arterioscler Thromb Vasc Biol 2002; 22: 1865–8. Pepys MB, Hawkins PN, Kahan MC, Tennent GA, Gallimore JR, Graham D, Sabin CA, Zychlinsky A, de Diego J. Proinflammatory effects of bacterial recombinant human C-reactive protein are caused by contamination with bacterial products, not by C-reactive protein itself. Circ Res 2005; 97: e97–103. Bisoendial R, Birjmohun R, Keller T, van Leuven S, Levels H, Levi M, Kastelein J, Stroes E. In vivo effects of C-reactive protein (CRP)-infusion into humans. Circ Res 2005; 97: e115–6. Peterson JW, Kwun BD, Hackett JD, Zervas NT. The role of inflammation in experimental cerebral vasospasm. J Neurosurg 1990; 72: 767–74.
32. McMahon CJ, Hopkins S, Vail A, King AT, Smith D, Illingworth KJ, Clark S, Rothwell NJ, Tyrrell PJ. Inflammation as a predictor for delayed cerebral ischemia after aneurysmal subarachnoid haemorrhage. J Neurointerv Surg 2013; 5: 512–7. 33. Rodling-Wahlstrom M, Olivecrona M, Koskinen LO, Naredi S, Hultin M. Subarachnoid haemorrhage induces an inflammatory response followed by a delayed persisting increase in asymmetric dimethylarginine. Scand J Clin Lab Invest 2012; 72: 484–9. 34. Muroi C, Mink S, Seule M, Bellut D, Fandino J, Keller E. Monitoring of the inflammatory response after aneurysmal subarachnoid haemorrhage in the clinical setting: review of literature and report of preliminary clinical experience. Acta Neurochir Suppl 2011; 110: 191–6.
Acta Anaesthesiologica Scandinavica 59 (2015) 255–264
264
© 2015 The Acta Anaesthesiologica Scandinavica Foundation. Published by John Wiley & Sons Ltd
Copyright of Acta Anaesthesiologica Scandinavica is the property of Wiley-Blackwell and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.
