- Email: [email protected]
CSF levels of neurofilament is a valuable predictor of long-term outcome after cardiac arrest
Journal of the Neurological Sciences 221 (2004) 19 – 24 www.elsevier.com/locate/jns
CSF levels of neurofilament is a valuable predictor of long-term outcome after cardiac arrest H. Rose´n *, J.-E. Karlsson, L. Rosengren Institute of Clinical Neuroscience, Department of Neurology, Sahlgrens University Hospital, University of Go¨teborg, S-413 45 Go¨teborg, Sweden Received 23 December 2003; accepted 1 March 2004
Abstract Aims: Prognostication of brain damage after cardiac arrest mainly relies on clinical observations. Recently, it has been shown that biochemical markers of brain damage measured in serum aid in this process. In the present study, we wanted to test the usefulness of CSF determinations of a neuronal protein, the neurofilament protein (NFL). Methods and results: Lumbar punctures were performed during week 2 or 3 in 22 patients surviving cardiac arrests. CSF NFL concentrations were analysed using an ELISA. Levels were increased in cardiac arrest patients. Patients with poor outcome according to the Glasgow outcome scale (GOS), low performance at a mini mental state examination (MMSE) and dependent according to Katz at 1 year follow up had the highest NFL levels. The NFL levels correlated well with anoxia time and coma depth. High positive and negative predictive values, particularly for poor outcome according to GOS were observed. Conclusions: Levels of CSF NFL give a reliable measure of the brain damage following cardiac arrest and the levels are highly predictive of poor outcome. This observation urges the development of sensitive serum assays of this marker to be used in the clinical setting. D 2004 Elsevier B.V. All rights reserved. Keywords: Cardiac arrest; Prognosis; Lumbar puncture; Neurofilament protein (NFL)
1. Introduction In Sweden, between January 1990 and March 1995, a total of 10,966 persons had an out of hospital cardiac arrest where resuscitations efforts were attempted. Among these, 1692 patients were successfully resuscitated and brought to hospital. Only 544 were alive 1 month after admittance to hospital [1]. The initial high mortality in this patient group reaches a plateau after 2 weeks [2]. The global anoxia has neurological consequences to the majority of survivors. For proper prognostication, the magnitude of brain damage needs to be evaluated. The development of the clinical state is important and particularly persistent lack of motor response to painful stimuli is a reliable predictor of poor outcome [3]. Neuroimaging has reached increasing value and impaired separation between white and gray substance seems to forecast poor outcome [4]. The use of newer MRI techniques has given new opportunities to measure brain injury. Extensive pathology visible on diffusion weighted images (DWI) is indicative for a poor prognosis [5]. * Corresponding author. Tel.: +46-31-342-10-00; fax: +46-31-342-2467. E-mail address: [email protected] (H. Rose´n). 0022-510X/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.jns.2004.03.003
Biochemical methods to quantify brain damage have been sought. High levels of the brain specific proteins S-100 and NSE in serum have been shown to be related to clinical factors indicative of brain injury and can be used as predictors for poor short-term [6] and long-term outcome [7]. The neurofilament is a major structural element of neurons. The light subunit of the neurofilament protein (NFL) is the primary component of the neurofilament core. CSF-NFL have been analysed among patients with neurodegeneration and elevated levels were observed in patients with amyotrophic lateral sclerosis (ALS) [8], multiple system atrophy (MSA) [9] and normal pressure hydrocephalus (NPH) [10] but not after cardiac arrest. Several studies have analysed other brain damage markers, i.e. CK-BB and NSE in cerebrospinal fluid among cardiac arrest patients. Most studies have used early lumbar punctures and the results have been contradictory [11 – 13]. This work focuses on patients surviving the first two critical weeks after admission. The rationale of this study was to test a hypothesis that NFL is released into cerebrospinal fluid after cardiac arrest and can be measured at 2 to 3 weeks post arrest. Furthermore, we intended to verify that its determination would be correlated to the Glasgow Outcome Scale (GOS), activity of daily living (ADL) and mini mental state examination
20
H. Rose´n et al. / Journal of the Neurological Sciences 221 (2004) 19–24
(MMSE) as well as to compare the prognostic value of NFL with neurological signs of brain damage.
2. Methods 2.1. Study design The study was performed at Sahlgrens University Hospital in Go¨teborg, Sweden between November 1994 and April 1999. Patients with an out-of-hospital cardiac arrest brought to the emergency ward after successful resuscitation were enrolled in the study. Patients of >18 years with a return of spontaneous circulation for a duration of z 12 days were eligible. Exclusion criterias were acute neurological disease and trauma. After admission to hospital surviving patients were generally transferred to the intensive care unit and treated according to the general guidelines that applies for advanced cardiac life support and post resuscitation care. Patients whose clinical condition would not permit a lumbar puncture to be performed were excluded (e.g. raised intracranial pressure, brain death) as were patients with anticoagulant medication. The study was approved by the Medical Ethics Committee at the University of Go¨teborg. Informed consent was given by all patients or, in the cases of unconsciousness, by relatives. 2.2. Definitions The data of the arrest were collected from ambulance reports. A cardiac arrest was defined as the ceasing of cardiac mechanical activity and was verified by the absence of a palpable pulse, coma and apnoea. The aetiologies of the heart arrests were categorised as either cardiac or noncardiac. Bystander cardiopulmonary resuscitation (CPR) is an attempt to perform CPR by a person not included in an emergency rescue team. The anoxia time was defined as the interval between collapse and the return of spontaneous circulation (ROSC) [14]. Thus, it included both the no-flow (arrest time) and low-flow (CPR time) periods.
centrifuged and stored in aliquots at 80 jC until analysed. The NFL concentrations were measured using an enzyme-linked immunosorbent assay (ELISA) [8]. The detection limit was 125 ng/l. Cerebrospinal fluids from 22 age-matched volunteers were used as reference. These 22 persons were examined at our hospital, shown to be healthy and bore no sign of neurological and/or cardiologic disease. 2.4. Assessment of outcome The GOS was determined using the Pittsburgh cerebral performance modifications, at T2, T3, T4 and T5 [3,14]. The result was dichotomised as good (good recovery and moderate disability) or poor (severe disability, permanent vegetative state and death). We used the best-recorded result at any of the registrations above as outcome, as some patients awoke from the initial coma but died later for reasons other than anoxic brain damage, mostly cardiac. To determine the cognitive function of the patients we used the mini mental state examination (MMSE) at T3, T4 and T5 [17]. MMSE values were ranked between 0 and 30, with values < 28 regarded as low. The best value at any of the examinations was used as the MMSE-index in the outcome records. The overall functional performance of the patients was ranked by their dependence/independence in activity of daily living (ADL) according to Katz [18] (between 0 and 6). A value of < 4 was regarded as dependence. The best ranking at any of the examinations (T3, T4, T5) was used in the outcome records. 2.5. Statistical analysis Statistical analysis was performed using Stat View and SPSS. Nonparametric statistics was applied with Mann – Whitney U-test to compare groups. Correlations were tested using Spearman’s rank correlation test. A value of p < 0.05 was considered as significant. Cross-tables for diagnostic precision (sensitivity, specificity, positive and negative predictive values) were calculated at three levels of specificity, 100%, 90% and 80%.
2.3. Neurological investigations 3. Results The examinations of the patients were planned to occur at admission (T0), days 2 –4 (T1), days 12 – 14 (T2), day 45 (T3), 3 months (T4) and 1 year (T5). The coma level was assessed at T1 and T2 using the Swedish RLS 85 score [15]. The neurological examination was done by one of the authors (HR) at T1 and T2. The neurological status was indexed using the NIH stroke scale [16]. Computed Tomography (CT) scans and/or Magnetic Resonance Imaging (MRI) were done in the majority of the patients and in two cases at the follow up after 1 year. The intention was to lumbar puncture the patients 12 – 14 days post arrest. Routinely 12 ml of CSF was collected,
During the study period, 105 patients were successfully resuscitated and brought to our intensive care unit. Sixtytwo patients survived 12 days or more. Ten of these patients declined to participate in the study and eight were transferred to other hospitals. Lumbar puncture was contraindicated due to poor clinical status (n = 9) or warfarin treatment (n = 8) in 17 cases. Five patients were lost from this study due to practical reasons. Thus, in the present study 22 patients were lumbar punctured. The mean age of the study population (n = 22) was 59.6 F 3.4 years (S.E.M.) (range 23 to 81 years). Nineteen
H. Rose´n et al. / Journal of the Neurological Sciences 221 (2004) 19–24
21
signs of brain oedema, three white matter disease and one older brain infarctions. Twelve examinations were without evident pathology. The study population did not differ from the patients not lumbar punctured (n = 40) regarding anoxia time (11.9 min vs. 13.9), coma level (median coma level 3 vs. 2) or age (59.6 years vs. 66.5) on outcome ( p>0.3 in all cases). 3.1. Outcome variables
Fig. 1. Mean levels of neurofilament protein (NFL) in the cerebrospinal fluid after cardiac arrest according to outcome (S.E.M. within brackets). Outcome groups: GOS, good and poor; MMSE, normal (norm) and pathological result (path). ADL, independence (ind) and dependence (dep). ***: p V 0.001.
were men and three women. Eleven patients had a history of cardiovascular disease before the arrest and seven noncardiovascular. Of the latter patients two had several years earlier suffered from minor stroke, a third had previously been investigated due to vertigo (>1 year ago) and a fourth had multiple sclerosis, however with no bout during the last years. Only four patients were assumed to be in a healthy state pre-arrest. The arrest had a cardiac background in 21 of the patients. One case was caused by asphyxia. All but one case had been subjected to bystander CPR (cardiopulmonary resuscitation). Eighteen patients had ventricular fibrillation, one asystolia and one pulseless electrical activity (PEA) when seen by the emergency team. For two patients the initial rhythm could not be determined. The mean anoxia time was 11.9 min F 1.7 (S.E.M.) (range 1 to 25 min). The actual time points for the clinical examinations were T0; admission, T1; 3.8 days F 0.2 (S.E.M.), range 2 – 5, T2; 14.2 days F 0.7 (S.E.M.), range 10– 24, T3; 47.4 days F 2.4 (S.E.M.), range 32 – 57, T4; 99.5 days F 3.4 (S.E.M.), range 85 –113, T5; 402.5 days F 20.8 (S.E.M.), range 365– 502. The median of the RLS score at T1 was RLS 3. The median of the RLS score at T2 was RLS 2 (range 1 to 6). The median score of NIH scale at T1 was 13 and at T2 6. The mean CSF levels of NFL were 11,381 F 2393 (S.E.M.) ng/l (range 144 –37 231). Thirteen patients had elevated levels of serum CKMB (>15 Ag/l) and/or troponin T (>0.2 Ag/l) in blood as signs of acute myocardial infarctions. MRI and/or CT scans were done in 17 cases. One patient had
Two patients were dead (GOS 5), one in a permanent vegetative state (GOS 4), nine severely disabled (GOS 3), one moderately disabled (GOS 2) and nine had a good recovery (GOS 1). Thus, 12 patients were poor (GOS 3– 5) and 10 were good outcome (GOS 1 and 2). Thirteen patients had pathologic results according to the MMSE and eight were normal (>27). Of these five were scored 0 due to severe encephalopathy. One patient did not do the MMSE. Eleven patients were dependent and 11 independent according to ADL score. The sampling of the CSF was done at mean day 17.5 F 1.1 (S.E.M.) (range 12 –30). In a few cases a second (n = 4), a third and a fourth (n = 1) lumbar puncture was managed. The mean concentration of CSF-NFL was increased at the first lumbar puncture (11,381 ng/l F 2393 (S.E.M.)). The corresponding mean level of the agematched controls was 217 ng/l (S.E.M.) (range 125– 633). Levels of poor outcome, and dependent patients were higher as were those with a pathological MMSE ( p < 0.001) (Fig. 1). Correlations between the various NIH, coma level as well as NFL and the three outcome variables are given in Table 1. Very high correlations (r f 0.90) were observed between the scores of the NIH-scale as well as coma level at T2 and all outcome variables. The corresponding correlations for NFL were somewhat lower (r f 0.80). Outcome did not correlate with coma level at admission, age or CKMB-level (not shown). In Fig. 2, levels of NFL are plotted against the NIH score at T2. The predictive capacity of the various clinical variables as well as NFL with regard to the three outcome variables are given in Tables 2, 3 and 4. Sensitivity, positive and negative predictive values at various cut off levels rendering
Table 1 Correlations (Spearman) between outcome and NFL, coma level and NIH scoring
NFL RLS T2 NIH T2
GOS 1 – 5
MMSE
ADL
r = 0.79 p < 0.0003 0.90 0.0001 0.92 0.0001
r = 0.82 p < 0.0002 0.91 0.0001 0.87 0.0001
r = 0.79 p < 0.0003 0.86 0.0001 0.84 0.0002
NFL, CSF neurofilament protein concentration; RLS, level of coma according to the Swedish reaction level scale; NIH, score according to the NIH stroke scale; T2, see Methods.
H. Rose´n et al. / Journal of the Neurological Sciences 221 (2004) 19–24
22
Table 3 Cross-table ADL, predicting dependency at follow up at specificity levels 100%, 90% and 80%
Fig. 2. Correlations between levels of NFL protein (NFp) and scoring according to the NIH stroke scale at T2. Black dot indicates seven overlapping values.
an approximate specificity of 100%, 90% and 80%, respectively, were calculated.
4. Discussion If a patient survives the first 14 days after cardiac arrest intensive care generally can be discontinued. A decision regarding the future ward needs to be taken and rehabilitation can commence. Even though clinical signs are the main prognostic factors the rationale for future ward must be balanced between a high confidence that improvement can be achieved and that further treatment in many cases are unethical due to the extent of the brain damage. It is obvious
Cut off
NFL 18,668 (Ag/l)
Anoxia time 22 (min)
NIH T2 15
RLS T2 3
Specificity Sensitivity PPV NPV
100% 46% 100% 65%
100% 36% 100% 61%
100% 60% 100% 71%
100 46 100 65
Cut off
12,780 (Ag/l)
16 (min)
4
Specificity Sensitivity PPV NPV
91% 73% 89% 77%
91% 64% 88% 72%
91% 100% 92% 100%
* * * *
Cut off
9622 (Ag/l)
9 (min)
1
2
Specificity Sensitivity PPV NPV
82% 91% 83% 90%
82% 91% 83% 90%
82% 100% 85% 100%
82 100 85 100
PPV: Positive predictive value. NPV: Negative predictive value. *: Not possible to estimate the specificity at that respective level.
that adjunctive methods of measuring brain damage using biochemical means are needed. NFL is a biomarker previously used in a number of clinical situations. Chronic disorders with degeneration of white matter or myelinated tracts, such as vascular dementia, multiple sclerosis or amyotrophic lateral sclerosis were associated with moderately increased levels of NFL [8,19 – 21]. In acute conditions like neonatal asphyxia high concentrations of NFL were related with extensive ischemic brain damage and a poor prognosis [22]. Similarly, focal ischemia caused very high levels [8].
Table 2 Cross-table GOS, predicting poor outcome at specificity levels 100%, 90% and 80%
Table 4 Cross-table MMSE, predicting pathological result at MMSE at follow up at specificity levels 100%, 90% and 80%
Cut off
NFL 12,800 (Ag/l)
Anoxia time 22 (min)
NIH T2 4
RLS T23
Cut off
NFL 12,780 (Ag/l)
Anoxia time 9 (min)
NIH T2 4
RLS T2 2
Specificity Sensitivity PPV NPV
100% 75% 100% 77%
100% 33% 100% 55%
100% 100% 100% 100%
100% 42% 100% 59%
Specificity Sensitivity PPV NPV
100% 69% 100% 67%
100% 85% 100% 80%
100% 85% 100% 80%
100% 92% 100% 88%
Cut off
9622 (Ag/l)
16 (min)
1
2
Cut off
1326 (Ag/l)
8 (min)
1
Specificity Sensitivity PPV NPV
90% 92% 92% 90%
90% 58% 87% 64%
90% 100% 92% 100%
90% 100% 92% 100%
Specificity Sensitivity PPV NPV
88% 100% 93% 100%
88% 92% 93% 87%
88% 85% 92% 78%
* * * *
* * * *
* * * *
* * * *
* * * *
Cut off
5888 (Ag/l)
9 (min)
Specificity Sensitivity PPV NPV
80% 92% 85% 89%
80% 83% 83% 80%
PPV: Positive predictive value. NPV: Negative predictive value. *: Not possible to estimate.
Cut off
549 (Ag/l)
7 (min)
Specificity Sensitivity PPV NPV
75% 100% 87% 100%
75% 92% 86% 85%
PPV: Positive predictive value. NPV: Negative predictive value. *: Not possible to estimate.
H. Rose´n et al. / Journal of the Neurological Sciences 221 (2004) 19–24
One patient had the diagnosis of multiple sclerosis. However, it had been long since he had an active phase of the disease and according to Lycke et al. [21] it is only during the bout that MS patients have augmented levels of NFL. This patient combined a longstanding cardiac arrest with a following poor clinical status and deep coma. He carried extreme high levels of NFL (>30,000 ng/l) as signs of advanced brain damage. Three other patients had anamnesis of neurological disease (minor strokes and episodes of vertigo) but in all cases several years ago. NFL is abundant in myelinated axons and it has been suggested that degeneration of white matter causes the CSF increase [8]. The number of injured axons will rise with longer anoxia time and more extended brain damage. Since the neuronal compartment including axons accounts for a substantial part of the cerebral tissue volume, diffuse neuronal and axonal loss following the anoxic injury serves well to explain the extended increase of NFL in CSF after cardiac arrest. Focused neurological examinations and measurement of CSF NFL are different but complementary ways to evaluate consequences of brain damage. A very strong relationship between these two parameters was seen in the present study. However, it is obvious that a careful clinical examination is the most reliable prognostic instrument. The highest predictive values at 14 days were observed for the NIH classification. But, to correctly use the advantages of a clinical scale, the physician has to be neurologically trained. Although the predictive capacity of the NFL was somewhat lower the clinical benefit is at a comparable level. It is an important knowledge for the clinician that patients with the worst outcome according to GOS as well as functional outcome carried the highest NFL levels. Also, the results are not subjected to interrater variation and can be done in confused or sedated patients. Very high specificity and sensitivity for NFL predicting poor outcome according to the GOS at the cut off level about 10,000 Ag/l is shown in the present study. Thus, 9 out of 10 patients with levels exceeding this cut off have a poor outcome and 9 out of 10 patients with levels below have a good outcome according to GOS. Interestingly, already a modest increase of NFL was associated with worse performance at the MMSE but the cut off value for predicting dependence according to ADL was higher. This probably can be explained by different capacity of various brain regions to endure anoxia. In hippocampus neurons are selectively vulnerable even to relative mild anoxia and the patient is at risk of ensuing memory impairment. Dependency implies a more extensive destruction of neurons in larger regions of the brain as reflected by the high NFL values. However, a rather recent MRI study of Grubb and coworkers was at variance with this hypothesis. It was proposed that global cerebral atrophy rather than hippocampal lesions was correlated with memory impairment [23]. Relatively few other studies have been committed to investigate
23
radiological findings after cardiac arrest. Less than a decade earlier, Roine observed that patients surviving cardiac arrest often had deep cerebral infarctions in the white tissue. However, the findings did not correlate to the outcome but later studies have shown the benefit of modern MRI techniques [12,24]. Although our primary aim not was to perform a radiographic study we did CT scans or MRI among the majority of the patients. Their clinical values were minor. Most brain imagings were normal or with minor pathology. A single CT scan revealed signs of brain oedema in a deeply comatose patient. This is not the first study where brain specific proteins in the CSF are investigated after cardiac arrest, but no other group has previously analysed NFL. Already in the 1980s it was shown that CK-BB was augmented in CSF among victims of cardiac arrest [11]. The results were controversial but later it was shown that both S-100 and NSE (neurone specific enolase) are increased in CSF the first 48 h after circulatory arrest and levels correlate to the clinical course [13,25]. S-100 and NSE are easily soluble proteins and are also transiently found in serum during the first days after a cardiac arrest [7]. It has also been shown that these two brain damage markers, in contrast to NFL, rapidly disappear from the CSF after the acute stage of herpes encephalitis [26]. This dynamics was one of the main reasons why we chose the time of the lumbar tap to be at approximately 2 to 3 weeks post collapse. At this time CSF levels of NFL are approximately maximal, whereas NSE and S-100 have normalised. We did actually measure the CSF-levels of those markers as well as GFAp in this study. Although somewhat higher levels of GFAp could be observed in certain patients we found no correlation to our outcome variables. Regarding NSE and S-100 both were normal for all patients at this time point. Due to practical reasons the time points for the punctures varied between days 12 and 30, with the majority around day 18. However, it was possible for us to follow CSF NFL concentrations for a few months in some cases and the increased levels seemed to have reached a plateau between days 14 and 45. This indicates that the puncture can be performed at an optional time during the first 2 months. Contrary, a few occasional early samples measured at our laboratory have had substantially lower levels of CSF NFL but levels of the other markers were pathological. The dropout rate from the study was relatively high due to patients declining the puncture, poor clinical status or anticoagulant treatment. However, we managed to include 22 patients, which approximately represent every third patient eligible for the study surviving 12 days or more. The cases represent different times of anoxia as well as various stages of brain damage and no differences were observed between the study population and the dropout patients with respect to age, coma level or anoxia time. Thus, the dropout rate does not seem to affect the results or conclusions drawn from the study.
24
H. Rose´n et al. / Journal of the Neurological Sciences 221 (2004) 19–24
It is obvious that CSF NFL is a very accurate and sensitive biochemical marker of neuronal injury after a cardiac arrest. If levels are only moderately increased at 2 –3 weeks a hopeful prognosis with rehabilitation potential is possible. On the other hand, high levels presage a more pessimistic destiny with very low chances of improvement. Although lumbar puncture is fairly invasive it is a safe investigation when performed at correct indications. If the examination is preceded with a routine CT scan cases with increased intracranial pressure can be discovered and a routine blood sample reveals cases where bleeding might occur. Even so, in various clinical settings, it might not be feasible to lumbar puncture patients at the start of the rehabilitation period. However, the very precise results of our study are also interesting from a theoretical point of view. If levels of NFL can be measured in CSF, serum determination should be possible if more sensitive techniques were available. We have recently contributed to the development of very well characterised monoclonal antiNFL antibodies aiming at serum determination of NFL [27]. The present study urges this process.
[9]
[10]
[11]
[12]
[13]
[14]
[15]
Acknowledgements [16]
This study was supported by the John and Brit Wennerstro¨m Foundation, the Rune and Ulla Amlo¨v Foundation, the Hjalmar Svensson Foundation, the Laerdal foundation for Acute Medicine, the Heart– Lung Foundation, Go¨teborg foundation for Neurological research, the Edit Jacobsson Foundation and the Go¨teborg Medical Society. We also thank our statistician, Martin Gellerstedt for calculating the statistics and Shirley Fridlund for excellent technical assistance.
[17]
[18]
[19]
[20]
References [1] Holmberg M, Holmberg S, Herlitz J, Ga˚rdelo¨v B. Survival after cardiac arrest outside hospital in Sweden. Resuscitation 1998;36:29 – 36. [2] Snyder BD, Loewenson RB, Gumnit RJ, Hauser WA, Leppik IE, Ramirez-Lassepas M. Neurologic prognosis after cardiopulmonary arrest: II. Level of consciousness. Neurology 1980;30:52 – 8. [3] Edgren E, Hedstrand U, Kelsey S, Sutton-Tyrrell K, Safar P. BRCT I Study Group. Assessment of neurological prognosis in comatose survivors of cardiac arrest. Lancet 1994;343:1055 – 9. [4] Torbey MT, Selim M, Knorr J, Bigelow C, Recht L. Quantitative analysis of the loss of distinction between gray and white matter in comatose patients after cardiac arrest. Stroke 2000;31:2163 – 7. [5] Wijdicks EF, Campeau NG, Miller GM. MR imaging in comatose survivors of cardiac resuscitation. AJNR Am J Neuroradiol 2001;22: 1561 – 5. [6] Rose´n H, Rosengren L, Herlitz J, Blomstrand C. Increased serum levels of the S-100 protein are associated with hypoxic brain damage after cardiac arrest. Stroke 1998;29:473 – 7. [7] Rose´n H, Stibrant Sunnerhagen K, Herlitz J, Blomstrand C, Rosengren L. Serum levels of the brain-derived proteins S-100 and NSE predict long-term outcome after cardiac arrest. Resuscitation 2001;49:183 – 91. [8] Rosengren LE, Karlsson J-E, Karlsson J-O, Persson LI, Wikkelsø C.
[21]
[22]
[23]
[24]
[25]
[26]
[27]
Patients with amyotrophic lateral sclerosis and other neurodegenerative diseases have increased levels of neurofilament protein in CSF. J Neurochem 1996;67:2013 – 8. Holmberg B, Rosengren L, Karlsson J-E, Johnels B. Increased cerebrospinal fluid levels of neurofilament protein in progressive supranuclear palsy and multiple-system atrophy compared with Parkinson’s disease. Mov Disord 1998;13:70 – 7. Tullberg M, Rosengren L, Blomsterwall E, Karlsson J-E, Wikkelsø C. CSF neurofilament and glial fibrillary acidic protein in normal pressure hydrocephalus. Neurology 1998;50:1122 – 7. Kjekshus J, Vaagenes P, Hetland Ø. Assessment of cerebral injury with spinal fluid creatine kinase (CSF-CK) in patients after cardiac resuscitation. Scand J Clin Lab Invest 1980;40:437 – 44. Kjekshus J, Vaagenes P. Cerebrospinal fluid creatine kinase isoenzyme BB levels do not predict the clinical outcome in patients unconscious following cardiac resuscitation. Clin Cardiol 1988;11:286 – 364. Roine RO, Somer H, Kaste M, Viinikka L, Karonen S-L. Neurological outcome after out-of-hospital cardiac arrest. Prediction by cerebrospinal fluid enzyme analysis. Arch Neurol 1989;46:753 – 6. Cummins R, Chamberlain D, Abramson N, Allen M, Baskett P, Becker L, et al. Recommended guidelines for uniform reporting of data from out-of-hospital cardiac arrest: The Utstein Style. A statement for health professionals from a task force of the American Heart Association, the European Resuscitation Council, the Heart and Stroke Foundation of Canada, and the Australian Resuscitation Council. Circulation 1991;84:960 – 75. Starmark J-E, Sta˚lhammar D, Holmgren E. The reaction level scale (RLS 85): manual and guidelines. Acta Neurochir (Wien) 1988; 91:12 – 20. Brott T, Adams Jr HP, Olinger CP, Marler JR, Barsan WG, Biller J, Spilker J, et al. Measurements of acute cerebral infarction: a clinical examination scale. Stroke 1989;20:864 – 70. Folstein M, Folstein S, McHugh P. Mini mental state. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975;12:189 – 98. Katz S, Ford A, Moskowitz R, Jackson B, Jaffe M. Studies of illness in the aged. The index of ADL: a standardized measure of biological and psychosocial function. JAMA 1963;185:914 – 9. Rosengren LE, Karlsson J-E, Sjo¨gren M, Blennow K, Wallin A. Neurofilament protein levels in CSF are increased in dementia. Neurology 1999;52:1090 – 3. ˚ , Lind Sjo¨gren M, Blomberg M, Jonsson M, Wahlund L-O, Edman A K, et al. Neurofilament protein in cerebrospinal fluid: a marker of white matter changes. J Neurosci Res 2001;66:510 – 6. Lycke J, Karlsson J-E, Andersen O, Rosengren L. Neurofilament protein in cerebrospinal fluid: a potential marker of activity in multiple sclerosis. J Neurol Neurosurg Psychiatr 1998;64:402 – 4. Blennow M, Savman K, Ilves P, Thoresen M, Rosengren L. Brainspecific proteins in the cerebrospinal fluid of severely asphyxiated newborn infants. Acta Paediatr 2001;90:1171 – 5. Grubb NR, Fox KAA, Smith K, Best J, Blane A, Ebmeier KP, et al. Memory impairment in out-of-hospital cardiac arrest survivors is associated with global reduction in brain volume, not focal hippocampal injury. Stroke 2000;31:1509 – 14. Roine RO, Raininko R, Erkinjuntti T, Ylikoski A, Kaste M. Magnetic resonance imaging findings associated with cardiac arrest. Stroke 1993;24:1005 – 14. Martens P, Raabe A, Johnsson P. Serum S-100 and neuron-specific enolase for prediction of regaining consciousness after global cerebral ischemia. Stroke 1998;29:2363 – 6. Studahl M, Rosengren L, Gunther G, Hagberg L. Difference in pathogenesis between herpes simplex virus type 1 encephalitis and tickborne encephalitis demonstrated by means of cerebrospinal fluid markers of glial and neuronal destruction. J Neurol 2000;247:636 – 42. Norgren N, Karlsson J, Rosengren L, Stigbrand T. Monoclonal antibodies selective for low molecular weight neurofilaments. Hybrid Hybridomics 2002;21:53 – 9.
CSF levels of neurofilament is a valuable predictor of long-term outcome after cardiac arrest H. Rose´n *, J.-E. Karlsson, L. Rosengren Institute of Clinical Neuroscience, Department of Neurology, Sahlgrens University Hospital, University of Go¨teborg, S-413 45 Go¨teborg, Sweden Received 23 December 2003; accepted 1 March 2004
Abstract Aims: Prognostication of brain damage after cardiac arrest mainly relies on clinical observations. Recently, it has been shown that biochemical markers of brain damage measured in serum aid in this process. In the present study, we wanted to test the usefulness of CSF determinations of a neuronal protein, the neurofilament protein (NFL). Methods and results: Lumbar punctures were performed during week 2 or 3 in 22 patients surviving cardiac arrests. CSF NFL concentrations were analysed using an ELISA. Levels were increased in cardiac arrest patients. Patients with poor outcome according to the Glasgow outcome scale (GOS), low performance at a mini mental state examination (MMSE) and dependent according to Katz at 1 year follow up had the highest NFL levels. The NFL levels correlated well with anoxia time and coma depth. High positive and negative predictive values, particularly for poor outcome according to GOS were observed. Conclusions: Levels of CSF NFL give a reliable measure of the brain damage following cardiac arrest and the levels are highly predictive of poor outcome. This observation urges the development of sensitive serum assays of this marker to be used in the clinical setting. D 2004 Elsevier B.V. All rights reserved. Keywords: Cardiac arrest; Prognosis; Lumbar puncture; Neurofilament protein (NFL)
1. Introduction In Sweden, between January 1990 and March 1995, a total of 10,966 persons had an out of hospital cardiac arrest where resuscitations efforts were attempted. Among these, 1692 patients were successfully resuscitated and brought to hospital. Only 544 were alive 1 month after admittance to hospital [1]. The initial high mortality in this patient group reaches a plateau after 2 weeks [2]. The global anoxia has neurological consequences to the majority of survivors. For proper prognostication, the magnitude of brain damage needs to be evaluated. The development of the clinical state is important and particularly persistent lack of motor response to painful stimuli is a reliable predictor of poor outcome [3]. Neuroimaging has reached increasing value and impaired separation between white and gray substance seems to forecast poor outcome [4]. The use of newer MRI techniques has given new opportunities to measure brain injury. Extensive pathology visible on diffusion weighted images (DWI) is indicative for a poor prognosis [5]. * Corresponding author. Tel.: +46-31-342-10-00; fax: +46-31-342-2467. E-mail address: [email protected] (H. Rose´n). 0022-510X/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.jns.2004.03.003
Biochemical methods to quantify brain damage have been sought. High levels of the brain specific proteins S-100 and NSE in serum have been shown to be related to clinical factors indicative of brain injury and can be used as predictors for poor short-term [6] and long-term outcome [7]. The neurofilament is a major structural element of neurons. The light subunit of the neurofilament protein (NFL) is the primary component of the neurofilament core. CSF-NFL have been analysed among patients with neurodegeneration and elevated levels were observed in patients with amyotrophic lateral sclerosis (ALS) [8], multiple system atrophy (MSA) [9] and normal pressure hydrocephalus (NPH) [10] but not after cardiac arrest. Several studies have analysed other brain damage markers, i.e. CK-BB and NSE in cerebrospinal fluid among cardiac arrest patients. Most studies have used early lumbar punctures and the results have been contradictory [11 – 13]. This work focuses on patients surviving the first two critical weeks after admission. The rationale of this study was to test a hypothesis that NFL is released into cerebrospinal fluid after cardiac arrest and can be measured at 2 to 3 weeks post arrest. Furthermore, we intended to verify that its determination would be correlated to the Glasgow Outcome Scale (GOS), activity of daily living (ADL) and mini mental state examination
20
H. Rose´n et al. / Journal of the Neurological Sciences 221 (2004) 19–24
(MMSE) as well as to compare the prognostic value of NFL with neurological signs of brain damage.
2. Methods 2.1. Study design The study was performed at Sahlgrens University Hospital in Go¨teborg, Sweden between November 1994 and April 1999. Patients with an out-of-hospital cardiac arrest brought to the emergency ward after successful resuscitation were enrolled in the study. Patients of >18 years with a return of spontaneous circulation for a duration of z 12 days were eligible. Exclusion criterias were acute neurological disease and trauma. After admission to hospital surviving patients were generally transferred to the intensive care unit and treated according to the general guidelines that applies for advanced cardiac life support and post resuscitation care. Patients whose clinical condition would not permit a lumbar puncture to be performed were excluded (e.g. raised intracranial pressure, brain death) as were patients with anticoagulant medication. The study was approved by the Medical Ethics Committee at the University of Go¨teborg. Informed consent was given by all patients or, in the cases of unconsciousness, by relatives. 2.2. Definitions The data of the arrest were collected from ambulance reports. A cardiac arrest was defined as the ceasing of cardiac mechanical activity and was verified by the absence of a palpable pulse, coma and apnoea. The aetiologies of the heart arrests were categorised as either cardiac or noncardiac. Bystander cardiopulmonary resuscitation (CPR) is an attempt to perform CPR by a person not included in an emergency rescue team. The anoxia time was defined as the interval between collapse and the return of spontaneous circulation (ROSC) [14]. Thus, it included both the no-flow (arrest time) and low-flow (CPR time) periods.
centrifuged and stored in aliquots at 80 jC until analysed. The NFL concentrations were measured using an enzyme-linked immunosorbent assay (ELISA) [8]. The detection limit was 125 ng/l. Cerebrospinal fluids from 22 age-matched volunteers were used as reference. These 22 persons were examined at our hospital, shown to be healthy and bore no sign of neurological and/or cardiologic disease. 2.4. Assessment of outcome The GOS was determined using the Pittsburgh cerebral performance modifications, at T2, T3, T4 and T5 [3,14]. The result was dichotomised as good (good recovery and moderate disability) or poor (severe disability, permanent vegetative state and death). We used the best-recorded result at any of the registrations above as outcome, as some patients awoke from the initial coma but died later for reasons other than anoxic brain damage, mostly cardiac. To determine the cognitive function of the patients we used the mini mental state examination (MMSE) at T3, T4 and T5 [17]. MMSE values were ranked between 0 and 30, with values < 28 regarded as low. The best value at any of the examinations was used as the MMSE-index in the outcome records. The overall functional performance of the patients was ranked by their dependence/independence in activity of daily living (ADL) according to Katz [18] (between 0 and 6). A value of < 4 was regarded as dependence. The best ranking at any of the examinations (T3, T4, T5) was used in the outcome records. 2.5. Statistical analysis Statistical analysis was performed using Stat View and SPSS. Nonparametric statistics was applied with Mann – Whitney U-test to compare groups. Correlations were tested using Spearman’s rank correlation test. A value of p < 0.05 was considered as significant. Cross-tables for diagnostic precision (sensitivity, specificity, positive and negative predictive values) were calculated at three levels of specificity, 100%, 90% and 80%.
2.3. Neurological investigations 3. Results The examinations of the patients were planned to occur at admission (T0), days 2 –4 (T1), days 12 – 14 (T2), day 45 (T3), 3 months (T4) and 1 year (T5). The coma level was assessed at T1 and T2 using the Swedish RLS 85 score [15]. The neurological examination was done by one of the authors (HR) at T1 and T2. The neurological status was indexed using the NIH stroke scale [16]. Computed Tomography (CT) scans and/or Magnetic Resonance Imaging (MRI) were done in the majority of the patients and in two cases at the follow up after 1 year. The intention was to lumbar puncture the patients 12 – 14 days post arrest. Routinely 12 ml of CSF was collected,
During the study period, 105 patients were successfully resuscitated and brought to our intensive care unit. Sixtytwo patients survived 12 days or more. Ten of these patients declined to participate in the study and eight were transferred to other hospitals. Lumbar puncture was contraindicated due to poor clinical status (n = 9) or warfarin treatment (n = 8) in 17 cases. Five patients were lost from this study due to practical reasons. Thus, in the present study 22 patients were lumbar punctured. The mean age of the study population (n = 22) was 59.6 F 3.4 years (S.E.M.) (range 23 to 81 years). Nineteen
H. Rose´n et al. / Journal of the Neurological Sciences 221 (2004) 19–24
21
signs of brain oedema, three white matter disease and one older brain infarctions. Twelve examinations were without evident pathology. The study population did not differ from the patients not lumbar punctured (n = 40) regarding anoxia time (11.9 min vs. 13.9), coma level (median coma level 3 vs. 2) or age (59.6 years vs. 66.5) on outcome ( p>0.3 in all cases). 3.1. Outcome variables
Fig. 1. Mean levels of neurofilament protein (NFL) in the cerebrospinal fluid after cardiac arrest according to outcome (S.E.M. within brackets). Outcome groups: GOS, good and poor; MMSE, normal (norm) and pathological result (path). ADL, independence (ind) and dependence (dep). ***: p V 0.001.
were men and three women. Eleven patients had a history of cardiovascular disease before the arrest and seven noncardiovascular. Of the latter patients two had several years earlier suffered from minor stroke, a third had previously been investigated due to vertigo (>1 year ago) and a fourth had multiple sclerosis, however with no bout during the last years. Only four patients were assumed to be in a healthy state pre-arrest. The arrest had a cardiac background in 21 of the patients. One case was caused by asphyxia. All but one case had been subjected to bystander CPR (cardiopulmonary resuscitation). Eighteen patients had ventricular fibrillation, one asystolia and one pulseless electrical activity (PEA) when seen by the emergency team. For two patients the initial rhythm could not be determined. The mean anoxia time was 11.9 min F 1.7 (S.E.M.) (range 1 to 25 min). The actual time points for the clinical examinations were T0; admission, T1; 3.8 days F 0.2 (S.E.M.), range 2 – 5, T2; 14.2 days F 0.7 (S.E.M.), range 10– 24, T3; 47.4 days F 2.4 (S.E.M.), range 32 – 57, T4; 99.5 days F 3.4 (S.E.M.), range 85 –113, T5; 402.5 days F 20.8 (S.E.M.), range 365– 502. The median of the RLS score at T1 was RLS 3. The median of the RLS score at T2 was RLS 2 (range 1 to 6). The median score of NIH scale at T1 was 13 and at T2 6. The mean CSF levels of NFL were 11,381 F 2393 (S.E.M.) ng/l (range 144 –37 231). Thirteen patients had elevated levels of serum CKMB (>15 Ag/l) and/or troponin T (>0.2 Ag/l) in blood as signs of acute myocardial infarctions. MRI and/or CT scans were done in 17 cases. One patient had
Two patients were dead (GOS 5), one in a permanent vegetative state (GOS 4), nine severely disabled (GOS 3), one moderately disabled (GOS 2) and nine had a good recovery (GOS 1). Thus, 12 patients were poor (GOS 3– 5) and 10 were good outcome (GOS 1 and 2). Thirteen patients had pathologic results according to the MMSE and eight were normal (>27). Of these five were scored 0 due to severe encephalopathy. One patient did not do the MMSE. Eleven patients were dependent and 11 independent according to ADL score. The sampling of the CSF was done at mean day 17.5 F 1.1 (S.E.M.) (range 12 –30). In a few cases a second (n = 4), a third and a fourth (n = 1) lumbar puncture was managed. The mean concentration of CSF-NFL was increased at the first lumbar puncture (11,381 ng/l F 2393 (S.E.M.)). The corresponding mean level of the agematched controls was 217 ng/l (S.E.M.) (range 125– 633). Levels of poor outcome, and dependent patients were higher as were those with a pathological MMSE ( p < 0.001) (Fig. 1). Correlations between the various NIH, coma level as well as NFL and the three outcome variables are given in Table 1. Very high correlations (r f 0.90) were observed between the scores of the NIH-scale as well as coma level at T2 and all outcome variables. The corresponding correlations for NFL were somewhat lower (r f 0.80). Outcome did not correlate with coma level at admission, age or CKMB-level (not shown). In Fig. 2, levels of NFL are plotted against the NIH score at T2. The predictive capacity of the various clinical variables as well as NFL with regard to the three outcome variables are given in Tables 2, 3 and 4. Sensitivity, positive and negative predictive values at various cut off levels rendering
Table 1 Correlations (Spearman) between outcome and NFL, coma level and NIH scoring
NFL RLS T2 NIH T2
GOS 1 – 5
MMSE
ADL
r = 0.79 p < 0.0003 0.90 0.0001 0.92 0.0001
r = 0.82 p < 0.0002 0.91 0.0001 0.87 0.0001
r = 0.79 p < 0.0003 0.86 0.0001 0.84 0.0002
NFL, CSF neurofilament protein concentration; RLS, level of coma according to the Swedish reaction level scale; NIH, score according to the NIH stroke scale; T2, see Methods.
H. Rose´n et al. / Journal of the Neurological Sciences 221 (2004) 19–24
22
Table 3 Cross-table ADL, predicting dependency at follow up at specificity levels 100%, 90% and 80%
Fig. 2. Correlations between levels of NFL protein (NFp) and scoring according to the NIH stroke scale at T2. Black dot indicates seven overlapping values.
an approximate specificity of 100%, 90% and 80%, respectively, were calculated.
4. Discussion If a patient survives the first 14 days after cardiac arrest intensive care generally can be discontinued. A decision regarding the future ward needs to be taken and rehabilitation can commence. Even though clinical signs are the main prognostic factors the rationale for future ward must be balanced between a high confidence that improvement can be achieved and that further treatment in many cases are unethical due to the extent of the brain damage. It is obvious
Cut off
NFL 18,668 (Ag/l)
Anoxia time 22 (min)
NIH T2 15
RLS T2 3
Specificity Sensitivity PPV NPV
100% 46% 100% 65%
100% 36% 100% 61%
100% 60% 100% 71%
100 46 100 65
Cut off
12,780 (Ag/l)
16 (min)
4
Specificity Sensitivity PPV NPV
91% 73% 89% 77%
91% 64% 88% 72%
91% 100% 92% 100%
* * * *
Cut off
9622 (Ag/l)
9 (min)
1
2
Specificity Sensitivity PPV NPV
82% 91% 83% 90%
82% 91% 83% 90%
82% 100% 85% 100%
82 100 85 100
PPV: Positive predictive value. NPV: Negative predictive value. *: Not possible to estimate the specificity at that respective level.
that adjunctive methods of measuring brain damage using biochemical means are needed. NFL is a biomarker previously used in a number of clinical situations. Chronic disorders with degeneration of white matter or myelinated tracts, such as vascular dementia, multiple sclerosis or amyotrophic lateral sclerosis were associated with moderately increased levels of NFL [8,19 – 21]. In acute conditions like neonatal asphyxia high concentrations of NFL were related with extensive ischemic brain damage and a poor prognosis [22]. Similarly, focal ischemia caused very high levels [8].
Table 2 Cross-table GOS, predicting poor outcome at specificity levels 100%, 90% and 80%
Table 4 Cross-table MMSE, predicting pathological result at MMSE at follow up at specificity levels 100%, 90% and 80%
Cut off
NFL 12,800 (Ag/l)
Anoxia time 22 (min)
NIH T2 4
RLS T23
Cut off
NFL 12,780 (Ag/l)
Anoxia time 9 (min)
NIH T2 4
RLS T2 2
Specificity Sensitivity PPV NPV
100% 75% 100% 77%
100% 33% 100% 55%
100% 100% 100% 100%
100% 42% 100% 59%
Specificity Sensitivity PPV NPV
100% 69% 100% 67%
100% 85% 100% 80%
100% 85% 100% 80%
100% 92% 100% 88%
Cut off
9622 (Ag/l)
16 (min)
1
2
Cut off
1326 (Ag/l)
8 (min)
1
Specificity Sensitivity PPV NPV
90% 92% 92% 90%
90% 58% 87% 64%
90% 100% 92% 100%
90% 100% 92% 100%
Specificity Sensitivity PPV NPV
88% 100% 93% 100%
88% 92% 93% 87%
88% 85% 92% 78%
* * * *
* * * *
* * * *
* * * *
* * * *
Cut off
5888 (Ag/l)
9 (min)
Specificity Sensitivity PPV NPV
80% 92% 85% 89%
80% 83% 83% 80%
PPV: Positive predictive value. NPV: Negative predictive value. *: Not possible to estimate.
Cut off
549 (Ag/l)
7 (min)
Specificity Sensitivity PPV NPV
75% 100% 87% 100%
75% 92% 86% 85%
PPV: Positive predictive value. NPV: Negative predictive value. *: Not possible to estimate.
H. Rose´n et al. / Journal of the Neurological Sciences 221 (2004) 19–24
One patient had the diagnosis of multiple sclerosis. However, it had been long since he had an active phase of the disease and according to Lycke et al. [21] it is only during the bout that MS patients have augmented levels of NFL. This patient combined a longstanding cardiac arrest with a following poor clinical status and deep coma. He carried extreme high levels of NFL (>30,000 ng/l) as signs of advanced brain damage. Three other patients had anamnesis of neurological disease (minor strokes and episodes of vertigo) but in all cases several years ago. NFL is abundant in myelinated axons and it has been suggested that degeneration of white matter causes the CSF increase [8]. The number of injured axons will rise with longer anoxia time and more extended brain damage. Since the neuronal compartment including axons accounts for a substantial part of the cerebral tissue volume, diffuse neuronal and axonal loss following the anoxic injury serves well to explain the extended increase of NFL in CSF after cardiac arrest. Focused neurological examinations and measurement of CSF NFL are different but complementary ways to evaluate consequences of brain damage. A very strong relationship between these two parameters was seen in the present study. However, it is obvious that a careful clinical examination is the most reliable prognostic instrument. The highest predictive values at 14 days were observed for the NIH classification. But, to correctly use the advantages of a clinical scale, the physician has to be neurologically trained. Although the predictive capacity of the NFL was somewhat lower the clinical benefit is at a comparable level. It is an important knowledge for the clinician that patients with the worst outcome according to GOS as well as functional outcome carried the highest NFL levels. Also, the results are not subjected to interrater variation and can be done in confused or sedated patients. Very high specificity and sensitivity for NFL predicting poor outcome according to the GOS at the cut off level about 10,000 Ag/l is shown in the present study. Thus, 9 out of 10 patients with levels exceeding this cut off have a poor outcome and 9 out of 10 patients with levels below have a good outcome according to GOS. Interestingly, already a modest increase of NFL was associated with worse performance at the MMSE but the cut off value for predicting dependence according to ADL was higher. This probably can be explained by different capacity of various brain regions to endure anoxia. In hippocampus neurons are selectively vulnerable even to relative mild anoxia and the patient is at risk of ensuing memory impairment. Dependency implies a more extensive destruction of neurons in larger regions of the brain as reflected by the high NFL values. However, a rather recent MRI study of Grubb and coworkers was at variance with this hypothesis. It was proposed that global cerebral atrophy rather than hippocampal lesions was correlated with memory impairment [23]. Relatively few other studies have been committed to investigate
23
radiological findings after cardiac arrest. Less than a decade earlier, Roine observed that patients surviving cardiac arrest often had deep cerebral infarctions in the white tissue. However, the findings did not correlate to the outcome but later studies have shown the benefit of modern MRI techniques [12,24]. Although our primary aim not was to perform a radiographic study we did CT scans or MRI among the majority of the patients. Their clinical values were minor. Most brain imagings were normal or with minor pathology. A single CT scan revealed signs of brain oedema in a deeply comatose patient. This is not the first study where brain specific proteins in the CSF are investigated after cardiac arrest, but no other group has previously analysed NFL. Already in the 1980s it was shown that CK-BB was augmented in CSF among victims of cardiac arrest [11]. The results were controversial but later it was shown that both S-100 and NSE (neurone specific enolase) are increased in CSF the first 48 h after circulatory arrest and levels correlate to the clinical course [13,25]. S-100 and NSE are easily soluble proteins and are also transiently found in serum during the first days after a cardiac arrest [7]. It has also been shown that these two brain damage markers, in contrast to NFL, rapidly disappear from the CSF after the acute stage of herpes encephalitis [26]. This dynamics was one of the main reasons why we chose the time of the lumbar tap to be at approximately 2 to 3 weeks post collapse. At this time CSF levels of NFL are approximately maximal, whereas NSE and S-100 have normalised. We did actually measure the CSF-levels of those markers as well as GFAp in this study. Although somewhat higher levels of GFAp could be observed in certain patients we found no correlation to our outcome variables. Regarding NSE and S-100 both were normal for all patients at this time point. Due to practical reasons the time points for the punctures varied between days 12 and 30, with the majority around day 18. However, it was possible for us to follow CSF NFL concentrations for a few months in some cases and the increased levels seemed to have reached a plateau between days 14 and 45. This indicates that the puncture can be performed at an optional time during the first 2 months. Contrary, a few occasional early samples measured at our laboratory have had substantially lower levels of CSF NFL but levels of the other markers were pathological. The dropout rate from the study was relatively high due to patients declining the puncture, poor clinical status or anticoagulant treatment. However, we managed to include 22 patients, which approximately represent every third patient eligible for the study surviving 12 days or more. The cases represent different times of anoxia as well as various stages of brain damage and no differences were observed between the study population and the dropout patients with respect to age, coma level or anoxia time. Thus, the dropout rate does not seem to affect the results or conclusions drawn from the study.
24
H. Rose´n et al. / Journal of the Neurological Sciences 221 (2004) 19–24
It is obvious that CSF NFL is a very accurate and sensitive biochemical marker of neuronal injury after a cardiac arrest. If levels are only moderately increased at 2 –3 weeks a hopeful prognosis with rehabilitation potential is possible. On the other hand, high levels presage a more pessimistic destiny with very low chances of improvement. Although lumbar puncture is fairly invasive it is a safe investigation when performed at correct indications. If the examination is preceded with a routine CT scan cases with increased intracranial pressure can be discovered and a routine blood sample reveals cases where bleeding might occur. Even so, in various clinical settings, it might not be feasible to lumbar puncture patients at the start of the rehabilitation period. However, the very precise results of our study are also interesting from a theoretical point of view. If levels of NFL can be measured in CSF, serum determination should be possible if more sensitive techniques were available. We have recently contributed to the development of very well characterised monoclonal antiNFL antibodies aiming at serum determination of NFL [27]. The present study urges this process.
[9]
[10]
[11]
[12]
[13]
[14]
[15]
Acknowledgements [16]
This study was supported by the John and Brit Wennerstro¨m Foundation, the Rune and Ulla Amlo¨v Foundation, the Hjalmar Svensson Foundation, the Laerdal foundation for Acute Medicine, the Heart– Lung Foundation, Go¨teborg foundation for Neurological research, the Edit Jacobsson Foundation and the Go¨teborg Medical Society. We also thank our statistician, Martin Gellerstedt for calculating the statistics and Shirley Fridlund for excellent technical assistance.
[17]
[18]
[19]
[20]
References [1] Holmberg M, Holmberg S, Herlitz J, Ga˚rdelo¨v B. Survival after cardiac arrest outside hospital in Sweden. Resuscitation 1998;36:29 – 36. [2] Snyder BD, Loewenson RB, Gumnit RJ, Hauser WA, Leppik IE, Ramirez-Lassepas M. Neurologic prognosis after cardiopulmonary arrest: II. Level of consciousness. Neurology 1980;30:52 – 8. [3] Edgren E, Hedstrand U, Kelsey S, Sutton-Tyrrell K, Safar P. BRCT I Study Group. Assessment of neurological prognosis in comatose survivors of cardiac arrest. Lancet 1994;343:1055 – 9. [4] Torbey MT, Selim M, Knorr J, Bigelow C, Recht L. Quantitative analysis of the loss of distinction between gray and white matter in comatose patients after cardiac arrest. Stroke 2000;31:2163 – 7. [5] Wijdicks EF, Campeau NG, Miller GM. MR imaging in comatose survivors of cardiac resuscitation. AJNR Am J Neuroradiol 2001;22: 1561 – 5. [6] Rose´n H, Rosengren L, Herlitz J, Blomstrand C. Increased serum levels of the S-100 protein are associated with hypoxic brain damage after cardiac arrest. Stroke 1998;29:473 – 7. [7] Rose´n H, Stibrant Sunnerhagen K, Herlitz J, Blomstrand C, Rosengren L. Serum levels of the brain-derived proteins S-100 and NSE predict long-term outcome after cardiac arrest. Resuscitation 2001;49:183 – 91. [8] Rosengren LE, Karlsson J-E, Karlsson J-O, Persson LI, Wikkelsø C.
[21]
[22]
[23]
[24]
[25]
[26]
[27]
Patients with amyotrophic lateral sclerosis and other neurodegenerative diseases have increased levels of neurofilament protein in CSF. J Neurochem 1996;67:2013 – 8. Holmberg B, Rosengren L, Karlsson J-E, Johnels B. Increased cerebrospinal fluid levels of neurofilament protein in progressive supranuclear palsy and multiple-system atrophy compared with Parkinson’s disease. Mov Disord 1998;13:70 – 7. Tullberg M, Rosengren L, Blomsterwall E, Karlsson J-E, Wikkelsø C. CSF neurofilament and glial fibrillary acidic protein in normal pressure hydrocephalus. Neurology 1998;50:1122 – 7. Kjekshus J, Vaagenes P, Hetland Ø. Assessment of cerebral injury with spinal fluid creatine kinase (CSF-CK) in patients after cardiac resuscitation. Scand J Clin Lab Invest 1980;40:437 – 44. Kjekshus J, Vaagenes P. Cerebrospinal fluid creatine kinase isoenzyme BB levels do not predict the clinical outcome in patients unconscious following cardiac resuscitation. Clin Cardiol 1988;11:286 – 364. Roine RO, Somer H, Kaste M, Viinikka L, Karonen S-L. Neurological outcome after out-of-hospital cardiac arrest. Prediction by cerebrospinal fluid enzyme analysis. Arch Neurol 1989;46:753 – 6. Cummins R, Chamberlain D, Abramson N, Allen M, Baskett P, Becker L, et al. Recommended guidelines for uniform reporting of data from out-of-hospital cardiac arrest: The Utstein Style. A statement for health professionals from a task force of the American Heart Association, the European Resuscitation Council, the Heart and Stroke Foundation of Canada, and the Australian Resuscitation Council. Circulation 1991;84:960 – 75. Starmark J-E, Sta˚lhammar D, Holmgren E. The reaction level scale (RLS 85): manual and guidelines. Acta Neurochir (Wien) 1988; 91:12 – 20. Brott T, Adams Jr HP, Olinger CP, Marler JR, Barsan WG, Biller J, Spilker J, et al. Measurements of acute cerebral infarction: a clinical examination scale. Stroke 1989;20:864 – 70. Folstein M, Folstein S, McHugh P. Mini mental state. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975;12:189 – 98. Katz S, Ford A, Moskowitz R, Jackson B, Jaffe M. Studies of illness in the aged. The index of ADL: a standardized measure of biological and psychosocial function. JAMA 1963;185:914 – 9. Rosengren LE, Karlsson J-E, Sjo¨gren M, Blennow K, Wallin A. Neurofilament protein levels in CSF are increased in dementia. Neurology 1999;52:1090 – 3. ˚ , Lind Sjo¨gren M, Blomberg M, Jonsson M, Wahlund L-O, Edman A K, et al. Neurofilament protein in cerebrospinal fluid: a marker of white matter changes. J Neurosci Res 2001;66:510 – 6. Lycke J, Karlsson J-E, Andersen O, Rosengren L. Neurofilament protein in cerebrospinal fluid: a potential marker of activity in multiple sclerosis. J Neurol Neurosurg Psychiatr 1998;64:402 – 4. Blennow M, Savman K, Ilves P, Thoresen M, Rosengren L. Brainspecific proteins in the cerebrospinal fluid of severely asphyxiated newborn infants. Acta Paediatr 2001;90:1171 – 5. Grubb NR, Fox KAA, Smith K, Best J, Blane A, Ebmeier KP, et al. Memory impairment in out-of-hospital cardiac arrest survivors is associated with global reduction in brain volume, not focal hippocampal injury. Stroke 2000;31:1509 – 14. Roine RO, Raininko R, Erkinjuntti T, Ylikoski A, Kaste M. Magnetic resonance imaging findings associated with cardiac arrest. Stroke 1993;24:1005 – 14. Martens P, Raabe A, Johnsson P. Serum S-100 and neuron-specific enolase for prediction of regaining consciousness after global cerebral ischemia. Stroke 1998;29:2363 – 6. Studahl M, Rosengren L, Gunther G, Hagberg L. Difference in pathogenesis between herpes simplex virus type 1 encephalitis and tickborne encephalitis demonstrated by means of cerebrospinal fluid markers of glial and neuronal destruction. J Neurol 2000;247:636 – 42. Norgren N, Karlsson J, Rosengren L, Stigbrand T. Monoclonal antibodies selective for low molecular weight neurofilaments. Hybrid Hybridomics 2002;21:53 – 9.