European Heart Journal Advance Access originally published online on December 1, 2006
European Heart Journal 2007 28(1):1-2; doi:10.1093/eurheartj/ehl397
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Will he ever be conscious again?
Department of Neurology, H-2-216, Academic Medical Center, University of Amsterdam, PO Box 226600, 1100 DD Amsterdam, The Netherlands
Corresponding author. Tel: +31 20 566 3842; fax: +31 20 566 9374. E-mail address: a.hijdra{at}amc.uva.nl
This editorial refers to ‘Prediction of neurological outcome after cardiopulmonary resuscitation by serial determination of serum neuron-specific enolase’
by J. Reisingeret al., on page 52
For relatives and doctors alike, crucial questions about the patient who remains comatose after having been resuscitated from cardiac arrest are ‘will he survive?’, ‘will he ever be conscious again?’, and ‘will he ever be independent again?’. Reisinger et al.1 provide data to predict outcome at an early stage that could be helpful to address the first two questions.
Attempts to predict outcome date from the first years after the introduction of cardiopulmonary resuscitation in 1960. Early retrospective case series have gradually made place for prospective cohort studies, but methodological problems remain. The two most important problems are the inevitability of treatment restrictions in selected patients and the definition and validity of outcome measurements. In clinical cohort studies, just as in regular clinical practice, non-treatment decisions are inevitable in some patients, given what is already known or presumed to be known about predictors of poor outcome and given the various patient preferences expressed by representatives. The tendency to restrict treatment selectively in patients with characteristics presumed to predict poor outcome may then result in the finding that such characteristics are indeed good predictors of poor outcome, the well known fallacy of a ‘self-fulfilling prophecy’. In studies designed to identify predictors of outcome, physicians who treat the patient or assess the outcome should, therefore, if possible, not be informed about the values of the variables studied. Guidelines for non-treatment decisions should be stated explicitly, and all non-treatment decisions should be documented to allow interpretation of the study results.
The second problem is that of the definition and validity of outcome measurements. Because the reliable prediction of poor outcome has proved to be more successful than that of good outcome, most studies use ‘poor outcome’ in the final analysis. Poor outcome is often defined as ‘death’ or ‘death or persisting coma’. Its counterpart ‘good outcome’ then necessarily includes survival with severe disability, which at least implies a round the clock need of supervision and nursing care. Many patients would consider such disability (caused by anterograde amnesia, personality changes with apathy and social blunting, and loss of motor control with spasticity or rigidity) as an outcome to be avoided. When disability is included in the outcome measurement, the timing of the outcome assessment is important. Severe disability after 1 month is not necessarily permanent, whereas after 6 months, no further improvement can be expected.2 Data from studies in which severe disability is not included in the poor outcome category are not helpful in answering the third question in the introduction.
A large number of prognostic studies in patients with post-resuscitation coma have been summarized in a number of structured reviews and meta-analyses, which have taken some or most of these considerations into account.3–5 These reviews have led to guidelines for the assessment of prognosis, most recently and comprehensively those of the American Academy of Neurology (Table 1).6 For many patients with an eventually poor outcome, this outcome can now be reliably predicted within the first days after resuscitation. The sensitivity of separate tests remains moderate, however, ranging from 
20 to 60%,6 but higher sensitivity may be reached with combination of tests.2
|
One of the tests is the determination of the serum level of neuron-specific enolase (NSE). NSE is a protein contained by neurons and neuroendocrine cells which is released in the blood after anoxic cell damage, the serum level presumably reflecting the amount of cell damage, comparable with the more familiar heart and liver enzymes. Reisinger et al.1 present a new cohort study in 177 patients to correlate serum levels of NSE with outcome. Serum NSE was determined on admission and after 1, 2, and 3 days. Serum levels generally peaked at the second and third day after admission. Poor outcome was defined as death or persistent coma within 6 months after the arrest and occurred in 59 patients, 55 of whom died in hospital, whereas four were in a vegetative state 6 months after the arrest. All patients with a peak level > 80 ng/mL had a poor outcome (specificity for predicting poor outcome 100%). Of all 59 patients with poor outcome, 37 had a peak level > 80 ng/mL (sensitivity 63%).
The treatment restriction guideline described by the authors is in agreement with clinical practice in most hospitals. The fallacy of a self-fulfilling prophecy cannot be entirely excluded. Instead of simply withholding the test results from the treating physicians, the investigators ‘eagerly avoided any influence of known NSE values on clinical treatment decisions’ (a policy that also throws some doubt on the alleged prospective nature of the study). Outcome assessment was well timed at 6 months after the resuscitation. Severe disability was not included in the poor outcome category and was not separately reported.
If the data from this study can be considered valid, how do they compare to those of the previous studies? Comparison of different series suggests a (biologically plausible) dose–response relation between serum NSE levels and outcome categories. In series using the combined outcome ‘death, vegetative state, or severe disability after at least 6 months’, cut-off values with 100% specificity (i.e. without false positive predictions) ranged from 14 to 33 ng/mL,2,7,8 whereas in series which did not include ‘severe disability’ in the poor outcome category, much higher values were found: 919 and 80 ng/mL.1 This emphasizes the importance of reporting data for more than one outcome measure, so that these data can be used to support relevant choices.
During the course of the study, induced hypothermia has gradually become the standard treatment of post-resuscitation coma in many hospitals. Can the information we now have about the predictive value of serum NSE levels be applied to patients treated with hypothermia? Reisinger et al. apparently did not find differences in NSE levels between treated and untreated patients (data not shown), but others did, albeit with conflicting results. In one study,8 serum NSE levels were found to be lower in patients treated with induced hypothermia (which again would be biologically plausible), with cut-off values for predicting death, vegetative state, or severe disability after 6 months of 13 ng/mL in the treated and 31 ng/mL in the untreated group. In another study,10 however, 13 patients were observed who underwent induced hypothermia and survived until discharge with no more than moderate disability, despite serum NSE levels ranging from 26 to 79 ng/mL (mean 44 ng/mL), i.e. well above levels which in previous studies were invariably associated with poor outcome. The investigators suggest that pulmonary complications may have contributed to the high serum levels, which were apparently not related to the amount of brain damage. This is clearly an issue that cannot be resolved with the available data and more prospective cohort studies in patients treated with induced hypothermia will be needed.
Reisinger et al. wisely conclude that ‘clinical decisions with potentially irreversible consequences should never rely on a single marker, but should only be made in the context of all available prognostic information’. As long as the predictive value of biochemical markers for poor outcome remains uncertain, the Practice Parameter of the American Academy of Neurology6 contains enough detailed information about clinical and neurophysiological variables that can be used to guide these decisions.
Conflict of interest: none declared.
Footnotes
The opinions expressed in this article are not necessarily those of the Editors of the European Heart Journal or of the European Society of Cardiology.
doi:10.1093/eurheartj/ehl316 
References
- Reisinger J, Höllinger K, Lang W, Steiner C, Winter T, Zeindlhofer E, Mori M, Schiller A, Lindorfer A, Wiesinger K, Siostrzonek P. (2007) Prediction of neurological outcome after cardiopulmonary resuscitation by serial determination of serum neuron-specific enolase. Eur Heart J 28:52–58 First published on October 23, 2006, doi:10.1093/eurheartj/ehl316.
[Abstract/Free Full Text] - Zandbergen EGJ, Hijdra A, Koelman JHTM, Hart AAM, Vos PE, Verbeek MM, De Haan RJ. (2006) Prediction of poor outcome within the first three days of postanoxic coma. Neurology 66:62–68.
[Abstract/Free Full Text] - Zandbergen EGJ, De Haan RJ, Stoutenbeek CP, Koelman JHTM, Hijdra A. (1998) Systematic review of early prediction of poor outcome in anoxic-ischaemic coma. Lancet 352:1808–1812.[CrossRef][ISI][Medline]
- Robinson LR, Micklesen PJ, Tirschwell DL, Lew HL. (2003) Predictive value of somatosensory evoked potentials for awakening from coma. Crit Care Med 31:960–967.[CrossRef][ISI][Medline]
- Booth CM, Boone RH, Tomlinson G, Detsky AS. (2004) Is this patient dead, vegetative, or severely neurologically impaired? JAMA 291:870–879.
[Abstract/Free Full Text] - Wijdicks EFM, Hijdra A, Young GB, Bassetti CL, Wiebe S. (2006) Practice parameter: prediction of outcome in comatose survivors after cardiopulmonary resuscitation (an evidence based review). Report of the Quality Standards Subcommittee of the American Academy of Neurology. Neurology 67:203–210.
[Abstract/Free Full Text] - Rosén H, Sunnerhagen KS, Herlitz J, Blomstrand C, Rosengren L. (2001) Serum levels of the brain-derived proteins S-100 and NSE predict long-term outcome after cardiac arrest. Resuscitation 49:183–191.[CrossRef][ISI][Medline]
- Tiainen M, Roine RO, Pettilä V, Takkunen OS. (2003) Serum neuron-specific enolase and S-100B protein in cardiac patients treated with hypothermia. Stroke 34:2881–2886.
[Abstract/Free Full Text] - Zingler VC, Krumm B, Bertsch T, Fassbender K, Pohlmann-Eden B. (2003) Early prediction of neurological outcome after cardiopulmonary resuscitation: a multimodal approach combining neurobiochemical and electrophysiological investigations may provide high prognostic certainty in patients after cardiac arrest. Eur Neurol 49:79–84.[CrossRef][ISI][Medline]
- Peters R, Oudemans-van Straaten HM, Van der Spoel JI, Bosman RJ, Wester JPJ, Zandstra DF. (2005) Prognosis of coma after cardiopulmonary resuscitation: reliability of serum neuron-specific enolase. Intensive Care Med 31:Suppl. 1, s120.
CiteULike 
Connotea 
Del.icio.us What's this?
Related articles in EHJ:
- Prediction of neurological outcome after cardiopulmonary resuscitation by serial determination of serum neuron-specific enolase
- Johann Reisinger, Kurt Höllinger, Wolfgang Lang, Christoph Steiner, Thomas Winter, Eduard Zeindlhofer, Michael Mori, Alexandra Schiller, Alexander Lindorfer, Kurt Wiesinger, and Peter Siostrzonek
EHJ 2007 28: 52-58.[Abstract] [Full Text]
This article has been cited by other articles:
|
|
 |
|
  G. Montalescot Editorial Eur. Heart J. Suppl., January 1, 2008; 10(suppl_A): A2 - A3. [Full Text] [PDF]
|
|
|
|
 |
|
 J. Reisinger and P. Siostrzonek Prediction of neurological outcome after cardiopulmonary resuscitation Eur. Heart J., October 1, 2007; 28(19): 2415 - 2415. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||