Conflict of Interest Declaration
The ILCOR Continuous Evidence Evaluation process is guided by a rigorous ILCOR Conflict of Interest policy. The following Task Force members and other authors were recused from the discussion as they declared a conflict of interest: none
The following Task Force members and other authors declared an intellectual conflict of interest and this was acknowledged and managed by the Task Force Chairs and Conflict of Interest committees: Claudio Sandroni, Karen Hirsch, Jerry Nolan and Jasmeet Soar were coauthors of the systematic review used for adolopment. They did not participate in assessment of the systematic review for quality for adolopment. Markus Skrifvars and Jaana Humaloja were co-authors of some of the included studies. This intellectual COI was declared when discussing study inclusion, and bias assessments were performed by task force members not involved in the included studies.
CoSTR Citation
Insert citation for ILCOR.org posting of CoSTR
Skrifvars M, Humaloja J, Sandroni C, Hirsch K, on behalf of the ILCOR ALS Task Force, Use of brain injury biomarkers for the prediction of good outcome after cardiac arrest, Consensus on Science with Treatment Recommendations. Brussels, Belgium: International Liaison Committee on Resuscitation (ILCOR) Advanced Life Support Task Force, 2022 December 15. Available from: http://ilcor.org
Methodological Preamble and Link to Published Systematic Review
The continuous evidence evaluation process for the production of Consensus on Science with Treatment Recommendations (CoSTR) started with a systematic review of prognostication after cardiac arrest (PROSPERO: CRD 420 1914 1169). This review was conducted by a systematic review team with the involvement of clinical content experts from the ILCOR ALS Task Force and consisted of two parts. The first part was about prediction of poor neurological outcome and provided evidence for the 2021 ILCOR CoSTR. The second part was about prediction of good neurological outcome and it was completed after the publication of the 2021 ILCOR CoSTR. The two parts of this review have been published separately in 2020 and 2021 respectively (Sandroni C et al, DOIs 10.1007/s00134-020-06198-w and 10.1007/s00134-022-06618-z, respectively). As the systematic review of prognostication of favorable outcome was recent and met ILCOR criteria for being of sufficient quality, the TF deemed it appropriate to use the adolopment process for systematic reviews. Additionally, an updated search including the dates October 31, 2021- May 20, 2022 was conducted to capture any papers published since the search for the original systematic review. Task force members screened and selected all newly identified papers, extracted data and performed bias assessment using the QUIPS tool, which was also used in the original systematic review. The totality of this identified evidence was considered by the Advanced Life Support task force, and used to determine the certainty of evidence and formulate the Consensus on Science and Treatment Recommendations.
Systematic Review
Sandroni C, D'Arrigo S, Cacciola S, Hoedemaekers CWE, Westhall E, Kamps MJA, Taccone FS, Poole D, Meijer FJA, Antonelli M, Hirsch KG, Soar J, Nolan JP, Cronberg T. Prediction of good neurological outcome in comatose survivors of cardiac arrest. A systematic review. DOI: 10.1007/s00134-022-06618-z.)
PICOST
The PICOST (Population, Intervention, Comparator, Outcome, Study Designs and Timeframe)
Population: Population: Adults (≥16 y) who are comatose after resuscitation from cardiac arrest (either in-hospital or out-of-hospital), regardless of target temperature.
Intervention: A level below normal or a low level of one of the following brain injury biomarkers; neuron specific enolase (NSE), S-100B, neurofilament light (NfL), Tau, GFAP or UCH-1.
Comparators: None.
Outcomes: Prediction of good neurological outcome defined as Cerebral Performance Categories (CPC) 1-2 or modified Rankin Score (mRS) 1-3 at hospital discharge/1 month or later.
Study Designs: Prognostic accuracy studies where the 2 x 2 contingency table (i.e., the number of true/false negatives and positives for prediction of poor outcome) was reported, or where those variables could be calculated from reported data, are eligible for inclusion. Unpublished studies, reviews, case reports, case series, studies including less than 10 patients, letters, editorials, conference abstracts, and studies published in abstract form were excluded.
Timeframe: In 2015 and 2020, ILCOR evidence reviews identified four categories of predictors of poor neurological outcome after cardiac arrest, namely clinical examination, biomarkers, electrophysiology, and imaging. However, the prediction of good neurological outcome has never been systematically reviewed to date. We searched studies published from January 1, 2001, onwards. Our last search was on May 20th 2022.
PROSPERO Registration CRD42017080475
Consensus on Science
The original systematic review identified 37 studies on prediction of good neurological outcome, of which four investigated biomarkers. The updated review identified 6 studies, of which two investigated biomarkers.
The individual studies were assessed as being at low to high risk of bias, mainly due to the possible risk of a self-fulfilling prophecy due to withdrawal of life sustaining therapies in patients with a perceived poor prognosis. Because of a high degree of heterogeneity between studies we did not perform meta-analyses, but the certainty of the evidence was assessed using GRADE methodology even though results could not be pooled. Simplified GRADE Tables have been provided for the individual biomarkers. The overall certainty of the evidence was rated as very low, with common reasons for downgrading including risk of bias, inconsistency of cutoff biomarker values determined between studies, imprecision of sensitivity and specificity determinations, and indirectness.
Neuron specific enolase - NSE
NSE was investigated in four observational studies including a total of 2141 patients. [Zellner, 2013 1382; Moseby-Knappe, 2021 984; Streitberg 2017 1145, Wihersaari 2022 1]
In two studies [Moseby-Knappe, 2021 984;Zellner, 2013 1382] blood NSE values within the upper limit of the normal range (17–18 μg/L) at 24h predicted favorable neurological outcome at 6 months with specificities of 85% (95% CI 80-88%) and 89% (95% CI 77-96), respectively and sensitivities of 46% (95% CI 41-52%) and 26% (95% CI 15-40%), respectively. At 48h normal NSE values predicted favorable neurological outcome at 6 months with specificities of 84% (95% CI 79-88%) and 89% (95% CI 77-97%) (corresponding sensitivities 58% [95% CI 52-63%] and 41% [95% CI 25-58%]).
One study [Moseby-Knappe, 2021 984] reported that normal NSE blood levels ( =<17 μg/L) at 72h predicted good neurological outcome at 6 months after cardiac arrest with specificity of 80% (95% CI 75-85%) and sensitivity of 75% (95% CI 70-80%).
In one study [Streitberger, 2017 1145] normal blood NSE levels (=< 17 μg/L ) at 72h predicted good neurological outcome at ICU discharge determined as CPC scores 1–3 with specificity of 97% (95% CI 95-98%) and sensitivity of 33% (95% CI 29-37%).
In one study [Wihersaari, 2022 1] normal NSE values (=< 17 μg/L ) at 48 hours predicted favorable functional outcome at 12 months with a specificity of 54% (95% CI 44-64%) and sensitivity of 90% (95% CI 85-95%).
S-100B
Two studies (763 patients) were found investigating S-100B. [Zellner 2013 1382, Moseby-Knappe 2021 984]
In one study [Zellner 2013 1382] S-100B <0.61 ìg/L measured on ICU admission predicted good neurological outcome at 6 months with specificity of 89% (95% CI 78-96%) and sensitivity of 31% (95% CI 20-44%) .
In two studies [Moseby-Knappe 2021 984; Zellner 2013 1382] S-100B predicted good neurological outcome at 6 months with cut-offs of <0.105 [Moseby-Knappe 2021 984] and 0.12 ìg/L [Zellner 2013 1382] at 24 h after cardiac arrest with specificity of 74% (95% CI 69-79%) and 89% (95% CI 78-96%) respectively, and sensitivity of 69% (95% CI 64-74%) and 37% (95% CI 24-51%) respectively.
In one study [Moseby-Knappe 2021 984] S-100B < 0.105 ìg/L measured at 48 h after cardiac arrest predicted good neurological outcome at 6 months with specificity of 72% (95% CI 66-77%) and sensitivity of 73% (95% CI 68-78%).
In one study [Moseby-Knappe 2021 984] S-100B < 0.105 ìg/L measured at 72 h after cardiac arrest predicted good neurological outcome at 6 months with specificity of 63% (95% CI 57-69%) and sensitivity of 81% (95% CI 76-85%).
Glial fibrillary acid protein (GFAP)
GFAP was investigated in two observational studies (801 patients) [Moseby-Knappe 2021 984, Humaloja 2022 141].
GFAP was measured at different time points:
24h: One study [Moseby-Knappe 2021 984] used the normal value for GFAp (<22 pg/mL) at 24h to predict favorable neurological outcome at 6 months after cardiac arrest and reported specificity of 97% (95% CI 94-98%) and sensitivity of 41% (95% CI 35-46%).
48h: One study [Moseby-Knappe 2021 984] used the normal value for GFAP (<22 pg/mL) at 48h to predict favorable neurological outcome at 6 months after cardiac arrest and reported specificity of 97% (95% 95-99%) and sensitivity of 35% (95% CI 30-41%). One study [Humaloja 2022 141] determined cut-offs to predict favorable neurological outcome at 6 months with 100% and 95% specificities as 210 and 439 pg/mL, respectively and found sensitivities of 43% (95% CI 32-54%) and 75% (95% CI 65-85%) respectively.
72h: One study [Moseby-Knappe 2021 984] used the normal value for GFAp (<22 pg/mL) at 72h to predict favorable neurological outcome at 6 months after cardiac arrest and reported specificity of 95% (95% CI 92-97%) and sensitivity of 44% (95% CI 39-50%). One study [Humaloja 2022 141] determined cut-offs to predict favorable neurological outcome at 6 months with 100% and 95% specificities of 187 and 359 pg/mL, respectively and found sensitivities of 44% (95% CI 33-56%) and 73% (95% CI 62-83%), respectively.
Serum Tau Protein
Serum Tau was investigated in two observational studies (806 patients). [Moseby-Knappe 2021 984, Humaloja 2022 141]
TAU was measured at different time points:
24h: One study [Moseby-Knappe 2021 984] used the normal value for Tau-protein (<1.55 pg/mL) at 24h to predict favorable neurological outcome at 6 months after cardiac arrest and reported specificity of 94% (95% CI 90-96%) and sensitivity of 28% (95% CI 24-33%).
48h: One study [Moseby-Knappe 2021 984] used the normal value for Tau-protein (<1.55 pg/mL) at 48h to predict favorable neurological outcome at 6 months after cardiac arrest and reported specificity of 95% (95% CI 92-97%) and sensitivity of 41% (95% CI 36-47%). One study [Humaloja 2022 141] determined that a cut-off of 3.28 pg/mL predicted favorable neurological outcome at 6 months with 95% specificity (95% CI 89-100%) and a sensitivity of 53% (95% CI 42-65%).
72h: One study [Moseby-Knappe 2021 984] used the normal value for Tau-protein (<1.55 pg/mL) at 72h to predict favorable neurological outcome at 6 months after cardiac arrest and reported specificity of 93% (95% CI 89-96%) and sensitivity of 52% (95% CI 46-57%). One study [Humaloja 2022 141] determined that cut-offs of 2.10 and 3.37 pg/mL predicted favorable neurological outcome at 6 months with 100% and 95% specificities and found sensitivities of 21% (95% CI 12-31%) and 52% (95% CI 40-64%), respectively.
Serum Neurofilament Light Chain (NFL)
Neurofilament light chain was investigated in three observational studies (1047 patients). [Moseby-Knappe 2021 984, Wihersaari 2021 39, WIhersaari 2022 1].
NfL was measured at different time points:
24h: Two studies [Wihersaari 2021 39, Wihersaari 2022 1] determined that the NfL cut-offs to predict favorable neurological outcome at 6 months with 100% specificity were 12.5 and 30 pg/mL respectively with sensitivities of 12% (95% CI 6.3-18.3%) and 79% (95% CI 67.1–87.5%) respectively. One study [Wihersaari 2022] determined the cut-off to predict favorable neurological outcome at 12 months with 95% specificity (95% CI 91.8-99.4%) to be 21.5 pg/mL with a sensitivity of 37% (95% CI 28-46%). One study [Moseby-Knappe, 2021 984] used the normal value for NfL (55 pg/mL) as the cut-off, and found that it predicted favorable neurological outcome at 6 months with a specificity of 95% (95% CI 93-97%) and sensitivity of 65% (95% CI 60-70%).
48h: Serum NfL was investigated in three observational studies [Moseby-Knappe 2021 984, Wihersaari 2021 39, WIhersaari 2022 1] at 48h. Two studies [Wihersaari 2021 39, Wihersaari 2022 1] determined that cut-offs of 8 and 30 pg/mL predicted favorable neurological outcome at 6 or 12 months with 100% specificity (95% CI 92-100% and 95% CI 100-100% respectively) and with sensitivities of 6% (95% CI 1.3-10.6%) and 74% (95% 62.4-83.5%) respectively. One study [Wihersaari 2022 1] determined that a cut-off of 29 pg/mL predicted favorable outcome at 12 months with 95% (95% CI 90-99%) specificity and a sensitivity of 39% (95% CI 29-48%). One study [Moseby-Knappe, 2021 984] used the normal value for NfL (55 pg/mL) as the cut-off, and found that it predicted favorable outcome at 6 months with a specificity of 96% (95% CI 93-98) and sensitivity of 54% (95% CI 48-59%).
72h: Serum Nfl was investigated in two observational studies [Moseby-Knappe 2021 984, Wihersaari 2021 39] at 72h. One study [Wihersaari, 2021 39] determined that a cut-off of 27 pg/mL predicted favorable neurological outcome at 6 months with 100% specificity and with a corresponding sensitivity of 67% (95% CI 56-79%). One study [Moseby-Knappe, 2021 984] used the normal value for NfL (55 pg/mL) as the cut-off, and found that it predicted favorable neurological outcome at 6 months with a specificity of 97% (95% CI 90-100%) and sensitivity of 51% (95% CI 45-56%).
Ubiquitin carboxy-terminal hydrolase-L1 (UCH-L1)
Serum UCH-L1 was investigated in one observational study (693 patients)[Moseby-Knappe 2021 984]. They used the normal value of UHC-L1 (<327 pg/mL) at 24, 48 and 72 hours after cardiac arrest as cut-off and it predicted favorable neurological outcome at 6 months with specificities of 85% (95% 81-89%), 82% (95% CI 77-86%), and 70% (95% CI 65-76%), respectively with corresponding sensitivities of 64% (95% CI 59-69%), 74% (95% CI 69-78%), and 88% (95% 84-91%), respectively.
Treatment Recommendations
- We suggest using normal NSE (<17 μg/L) within 72 hours after ROSC, in combination with other tests, for predicting favorable neurological outcome in adults who are comatose after cardiac arrest (weak recommendation, very low-certainty evidence).
- We suggest against using serum levels of glial fibrillary acidic protein, serum tau protein, or neurofilament light chain in clinical practice for predicting favorable neurological outcome in adults who are comatose after cardiac arrest (weak recommendation, very low- certainty evidence).
Justification and Evidence to Decision Framework Highlights
The evidence suffers from problems with imprecision as the sensitivities and specificities vary a great deal between studies, and confidence intervals were often wide. In addition, there is the problem of indirectness as most studies only include patients from interventional randomized controlled trials with strict inclusion criteria. For example, there is little evidence in patients with a non-cardiac cause of the arrest.
The best evidence appears to be for NSE, given the number of patients included in trials and the similar cut-offs used to determine a normal value across studies. Importantly NSE is also already recommended as a part of multimodal prediction of poor outcome.
The evidence for the accuracy of the biomarkers S-100B, NfL, GFAP, Tau, and UCH-1 is mixed. NfL may be more accurate but there is limited data on feasibility of measuring these novel biomarkers in regular clinical practice as all analyses have included thawed samples measured at a later time point and in highly specialised laboratories. In addition, the cut-off levels for predicting a good functional outcome vary to a great degree making it difficult to provide a treatment recommendation suggesting use.
Importantly any possible withdrawal of life sustaining therapies in cardiac arrest patients should be undertaken using several prognostication modalities according to the in 2020 published COSTR on the prediction of poor outcome [Soar, 2020].
Knowledge Gaps
NSE
- The sensitivity and specificity of using a normal level of NSE are unclear and the studies conducted thus far provide highly variable results.
- There is limited data on the accuracy NSE in patients with in-hospital cardiac arrest and those with a non-cardiac cause of the arrest.
- There is limited data on the use of NSE in patients with variable degrees of haemolysis.
- There is limited data on the accuracy of NSE when used together with other means of predicting a good outcome such as clinical examination, brain imaging, EEG, SSEP and other biomarkers.
- There are no data on the cost-effectiveness of the use NSE in predicting outcome.
- There is limited data on whether the results of NSE measurements are consistent even if there is deviation from the recommended assessment time-point.
S-100B
- The sensitivity and specificity of using a normal level of S-100B are unclear and the studies conducted thus far provide highly variable results.
- There is limited data on the use of accuracy S-100B in patients with in-hospital cardiac arrest and those with a non-cardiac cause of the arrest
- There is limited data on the accuracy of S-100B when used together with other means of predicting a good outcome such as clinical examination, brain imaging, EEG, SSEP and other biomarkers
- There are no data on the cost-effectiveness on the use S-100B in predicting outcome
GFAP, NfL, TAU, UCH-1
- The used cut-offs for the studied biomarkers varied between studies making it difficult to assess whether the sensitivities and specificities are comparable.
- There is limited data on the use of these biomarkers in patients with in-hospital cardiac arrest and those with a non-cardiac cause of the arrest
- There is limited data on the accuracy of these biomarkers together with other means of predicting a good outcome such as clinical examination, brain imaging, EEG, SSEP and other biomarkers
- There are no data on the cost-effectiveness on the use of GFAP, NfL, TAU and UCH-1 in predicting outcome
Attachments:
References
Humaloja J, Lahde M, Ashton N J, et al. GFAp and tau protein as predictors of neurological outcome after out-of-hospital cardiac arrest: A post hoc analysis of the COMACARE trial, Resuscitation 2022, 170: 141–149
Moseby-Knappe M, Mattsson-Carlgren N, Stammet P, Backman S, Blennow K, Dankiewicz J, Friberg H, Hassager C, Horn J, Kjaergaard J, Lilja G, Rylander C, Ullén S, Undén J, Westhall E, Wise MP, Zetterberg H, Nielsen N, Cronberg T. Intensive Care Med. 2021 Serum markers of brain injury can predict good neurological outcome after out-of-hospital cardiac arrest. Intensive Care Med. 2021 Sep;47(9):984-994.
Streitberger KJ, Leithner C, Wattenberg M, Tonner PH, Hasslacher J, Joannidis M, Pellis T, Di Luca E, Födisch M, Krannich A, Ploner CJ, Storm C.Neuron-Specific Enolase Predicts Poor Outcome After Cardiac Arrest and Targeted Temperature Management: A Multicenter Study on 1,053 Patients.
Crit Care Med. 2017 Jul;45(7):1145-1151.
Wihersaari L, Reinikainen M, Furlan R, et al. Neurofilament light compared to neuron-specific enolase as a predictor of unfavorable outcome after out-of-hospital cardiac arrest, Resuscitation 2022 174:1–8
Wihersaari L, Ashton NJ, Reinikainen M, Jakkula P, Pettila V, Hastbacka J, Tiainen M, Loisa P, Friberg H, Cronberg T, Blennow K, Zetterberg H, Skrif- vars MB, Comacare Study G (2021) Neurofilament light as an outcome predictor after cardiac arrest: a post hoc analysis of the COMACARE trial. Intensive Care Med 2021, 47:39–48
Zellner T, Gärtner R, Schopohl J, Angstwurm M. NSE and S-100B are not sufficiently predictive of neurologic outcome after therapeutic hypothermia for cardiac arrest. Resuscitation. 2013 Oct;84(10):1382-6.