ALS 713: Prognostication in Normothermia

Transcription

1 ALS 713: Prognostication in Normothermia TFQO: Clifton Callaway (COI #214) EVREVs: Claudio Sandroni (COI #134); Tobias Cronberg (COI #35) Taskforce: ALS

2 COI Disclosure (specific to this systematic review) Commercial/industry Claudio Sandroni: none Tobias Cronberg: none Potential intellectual conflicts Claudio Sandroni: first author of two systematic reviews on prognostication after cardiac arrest and of the ERC-ESICM Advisory Statement on prognostication in comatose survivors of cardiac arrest Tobias Cronberg: author of the ERC-ESICM Advisory Statement on prognostication in comatose survivors of cardiac arrest

3 2010 CoSTR ALS-PA-041A (Clinical Examination) ALS-PA-051A (Electrophysiology) ALS-PA-052A-B (Biomarkers) ALS-PA-059 (Imaging)

4 2010 Clinical Examination The following parameters predicted poor outcome (CPC 3 or 4, or death) with a false-positive rate (FPR) of 0%: absent vestibulo-ocular reflexes at 24 h [(95%CI 0%-14%)] (LOE P1) absence of pupillary light and corneal reflex at 72 h (95% CI 0% to 9%) (LOE P1) GCS <5 at 48 h (95% CI 0% - 13%) (LOE P1) and on day 3 (95% CI 0% - 6%) (LOE P2) a clinical examination score<15 on day 4 [(95%CI 0% -18%) (LOE P1) However, in 1 study an absent motor response (GCS motor =1) at 72 h after cardiac arrest predicted poor outcome with a FPR of 5% [(95%CI 2% to 9%] (LOE P1) The presence of myoclonus status in adults was strongly associated with poor outcome (LOE P1; LOE P3; LOE P4), but rare cases of good neurological recovery have been described and accurate diagnosis was problematic.

5 2010 Clinical Examination Treatment Recommendations There are no clinical neurologic signs that reliably predict poor outcome <24 hours after cardiac arrest. In adult patients who are comatose after cardiac arrest, have not been treated with hypothermia and have no confounding factors (e.g., hypotension, sedatives or neuromuscular blockers), the absence of both pupillary light and corneal reflex at 72 hours reliably predicts poor outcome. Absence of vestibulo-ocular reflexes at 24 hours and a GCS motor score of 2 or less at 72 hours are less reliable. Other clinical signs, including myoclonus, are not recommended for predicting poor outcome

6 2010 Clinical Examination Knowledge Gaps The reevaluation of prognostic indicators during therapeutic hypothermia and in the presence of other confounders needs to be completed to guide current post cardiac arrest care.

7 2010 Electrophysiology SEP measured between 4 h and 2 weeks after cardiac arrest were associated with poor outcome in 14 studies (LOE P1, P2, P3) The absence of cortical N20 response to median nerve stimulation at 24 to 72 hours after cardiac arrest predicted poor outcome (CPC 3 or 4, or death) with a FPR of 0.7% (95% CI ) (LOE P1) Electroencephalography predicted poor outcome in comatose survivors of cardiac arrest within 1 week after cardiac arrest in 12 studies (LOE P1, P3, P4, P5). In a meta-analysis, EEG showing generalized suppression to less than 20 µv, burst-suppression pattern associated with generalized epileptic activity, or diffuse periodic complexes on a flat background 12 to 72 hours after sustained ROSC predicted a poor outcome (FPR of 3%, 95% CI 0.9% to 11%).

8 2010 Electrophysiology Treatment Recommendations No electrophysiological study reliably predicts outcome of comatose patient after cardiac arrest in the first 24 hours. After 24 hours, bilateral absence of the N20 cortical response to median nerve stimulation predicts poor outcome in comatose cardiac arrest survivors not treated with therapeutic hypothermia. In the absence of confounding circumstances, such as sedatives, hypotension, hypothermia, or hypoxemia, it is reasonable to use unprocessed EEG interpretation (specifically identifying generalized suppression to less than 20 µv, burst suppression pattern with generalized epileptic activity, or diffuse periodic complexes on a flat background) observed between 24 and 72 hours after sustained ROSC to assist the prediction of a poor outcome in comatose survivors of cardiac arrest not treated with hypothermia.

9 2010 Electrophysiology Knowledge Gaps More data are needed about the performance and timing of somatosensory evoked potentials and electroencephalography criteria for aiding prognostication in patients treated with induced hypothermia.

10 2010 Biomarkers Serum neuronal-specific enolase (NSE) elevations are associated with poor outcome for comatose patients after cardiac arrest (LOE P1; LOE P2; LOE P3). Although specific cutoff values with a FPR of 0% have been reported, clinical application is limited due to variability in the 0% FPR cutoff values reported among various studies. Serum S100 elevations are associated with poor outcome for comatose patients after cardiac arrest (LOE P1; LOE P2; LOE P3) Worse outcomes for comatose survivors of cardiac arrest are also associated with increased levels of cerebrospinal fluid (CSF)-CK (LOE P2), CKBB (LOE P1; LOE P2; LOE P3). However, 1 study found no relationship between cerebrospinal fluid-ckbb and prognosis (LOE P2). Outcomes are also associated with increased cerebrospinal fluid levels of other markers including NSE (LOE P1; LOE P2; S100 (LOE P2); LDH, GOT (LOE P2); neurofilament (LOE P3); and acid phosphatase and lactate (LOE P2).

11 2010 Biomarkers Treatment Recommendations Evidence does not support the use of serum or cerebrospinal fluid biomarkers alone as predictors of poor outcomes in comatose patients after cardiac arrest with or without treatment with therapeutic hypothermia. Limitations included small numbers of patients and/or inconsistency in cutoff values for predicting poor outcome.

12 2010 Biomarkers Knowledge Gaps Future studies should identify and resolve the heterogeneity of cutoff values used to predict poor outcome with a FPR of zero. Studies also must account for confounders that may alter levels or predictive performance of various markers (eg, hypothermia, underlying disease, pregnancy, intra-aortic balloon pump, brain instrumentation, hemodialyisis, or other organ failure). Studies examining whether biomarkers can be used to monitor ongoing injury and response to therapy may be useful.

13 2010 Imaging MRI There are no LOE P1-P2-studies that support the use of MRI. In 32 studies (P3-P5) the timing of MRI ranged from 1 day to 10 months after ROSC. Overall these studies were limited by small sample sizes, variable time of imaging (many very late in the course of the event), lack of comparison with a standardized method of prognostication, often nonmodern MRI techniques, and early withdrawal of care.

14 2010 Imaging CT There are no LOE P1-P2-studies that support the use of CT. In 22 studies (P3-P5) the timing of MRI ranged from 1 h to 20 days after ROSC. Overall these studies were limited by small sample sizes, variable time of imaging (many very late in the course of the event), lack of comparison with a standardized method of prognostication, and early withdrawal of care.

15 2010 Imaging Treatment Recommendations There is insufficient evidence to recommend for or against the routine use of neuroimaging to predict outcome of adult cardiac arrest survivors. Knowledge Gaps Adequately powered prospective studies are required to evaluate the accuracy of CT, MRI, or both in prognosticating outcome of comatose cardiac arrest survivors. Outcome prediction should include a comparison with more conventional methods. All studies should allow for sufficient time to realize patient recovery, avoiding the bias of self-fulfilling prophecy and premature withdrawal of care.

16 C2015 PICO Population: Adults with ROSC who are not treated with targeted temperature management (TTM) Intervention: any clinical variable when normal (e.g. 1. Clinical Exam, 2. EEG, 3. SSEP, 4. Imaging, 5. Other) Comparison: any clinical variable when abnormal Outcomes Survival with Favorable neurological or functional outcome at discharge, 30 days, 60 days, 180 days AND/OR 1 year, Survival only at discharge, 30 days, 60 days, 180 days AND/OR 1 year

17 Inclusion/Exclusion & Articles Found Inclusion: adult ( 16 years) pts, comatose (unconscious, unresponsive, or GCS 8) Exclusion: Studies including pts. in hypoxic coma from causes other than CA (e.g., respiratory arrest, CO poisoning) except when a subpopulation of cardiac arrest patients could be evaluated separately. Included: 46 articles

18 2015 Proposed Treatment Recommendations Clinical Examination - 1 We recommend using the absence of pupillary reflex to light (or the combined absence of both pupillary and corneal reflexes) at 72h from ROSC to predict poor outcome. We suggest against using an absent or extensor motor response to pain (M 2) alone to predict poor outcome, given its high false positive rate. However, due to its high sensitivity, this sign may be used to identify the population with poor neurological status needing prognostication or to predict poor outcome in combination with other more robust predictors.

19 2015 Proposed Treatment Recommendations Clinical Examination - 2 We suggest using the presence of a status myoclonus within 72 h from ROSC in combination with other predictors for prognosticating a poor neurological outcome. We suggest prolonging the observation of clinical signs when interference from residual sedation or paralysis is suspected, so that the possibility of obtaining false positive results is minimized.

20 2015 Proposed Treatment Recommendations Electrophysiology We recommend using bilateral absence of N20 SSEP wave at 24h after ROSC to predict poor outcome. We suggest using the presence of burst-suppression on EEG at 72h from ROSC in combination with other predictors for prognosticating a poor neurological outcome. We suggest against using EEG grades due to the inconsistencies in their definitions. We also do not suggest using low-voltage EEG given the potential interferences of technical factors on EEG amplitude.

21 2015 Proposed Treatment Recommendations Biomarkers We suggest using high serum values of NSE at h from ROSC in combination with other predictors for prognosticating a poor neurological outcome. However, no threshold enabling prediction with zero FPR can be recommended. We suggest using utmost care and preferably sampling at multiple time-points when assessing NSE, to avoid false positive results due to hemolysis.

22 2015 Proposed Treatment Recommendations Imaging We suggest using the presence of a marked reduction of the GM/WM ratio on brain CT within 48 h after ROSC or the presence of extensive reduction in diffusion on brain MRI at 2-6 days after ROSC in combination with other predictors for prognosticating a poor neurological outcome. We suggest using brain imaging studies for prognostication only in centers where specific experience is available.

23 Risk of Bias in studies Author, year Index Blinding (treating team) Blinding (index or outcome evaluators) Excludes sedation or paralysis Best CPC Excludes previous neurol. diseases Interobserv. agreement assessed Length of follow-up Index test used for WLST Overall Bassetti 1996 SSEP No no N/A no yes no 1 year No Bassetti, 1996 EEG, BR No no no no yes no 1 year No Bauer, 2003 SEP N70 No no N/A no no no 1 year No Berkhoff, 2000 EEG, SSEP No no yes no no no 1 year Not reported Bertini, 1989 PLR No no no no yes no discharge Not reported very Brunko, 1987 SSEP No no N/A no no no 30 days Not reported very Chockroverty, 1975 EEG No no no no no no 30 days Not reported very Choi, 2008 CT No no N/A no no yes discharge Not reported very Choi, 2010 MRI No no N/A no no no 90 days No Earnest 1979 PLR, MR No no no no yes no discharge Not reported very Edgren 1987 PLR, CR, MR No no yes no yes no 180 days Yes Edgren, 1987 EEG No yes yes no yes no 180 days No Serious Els, 2004 MRI No yes N/A yes no no 180 days Not reported very Fischer, 2006 GCS No no yes yes no no 1 year Not reported very Gendo 2001 SEP N70 No no N/A no no no 180 days Not reported very Grindal 1977 EEG No no no no no no 60 days Not reported very

24 Risk of Bias in studies Author, year Hachimi-Idrissi, 2002 Index Blinding (treating team) Blinding (index or outcome evaluators) Excludes sedation or paralysis Best CPC Excludes previous neurol. diseases Interobserv. agreement assessed Length of follow-up Index test used for WLST S-100B Yes yes N/A yes no N/A discharge No No Kaplan, 1999 EEG No no no no no No discharge Not reported very Krumholz, 1988 Myoclonus No no no no yes No 180 days Not reported Madl 2000 SEP N70 No no N/A no no No 180 days Not reported very Mlynash, 2010 MRI No yes N/A yes no Yes 180 days Not reported very Morimoto, 1993 CT no no N/A no no No discharge Not reported very Mussack, 2002 S-100B no no N/A no no N/A 1 year Not reported very Nakabayashi 2001 SEPs no no N/A no no No 1 year Not reported Pfeifer 2005 GCS no no yes no yes No 30 days Not reported Reisinger, 2007 NSE No no N/A no no N/A 180 days No Serious Rosen, 1998 S-100B no no N/A no no N/A discharge Not reported very Rosen, 2001 NSE, S-100B no no N/A no no N/A 1 year Not reported very Rothstein, 1991 EEG no no no no no No 60 days Not reported very Rothstein, 2000 SSEP no no N/A no no No 30 days No Serious Samaniego, 2011 NSE no no N/A no no N/A 90 days Not reported very Scollo-Lavizzari 1987 Overall EEG no no no no no No discharge Not reported very Steffen, 2010 NSE no no N/A no no N/A discharge no Stelzl, 1995 SSEP, NSE no no N/A no no No discharge Not reported very

25 Risk of Bias in studies Author, year Index Blinding (treating team) Blinding (index or outcome evaluators) Excludes sedation or paralysis Best CPC Excludes previous neurol. diseases Interobserv. agreement assessed Length of follow-up Index test used for WLST Overall Thomke, 2005 Myoclonus no no yes no no No 1 year no Tiainen, 2003 NSE, S-100B no no N/A no no N/A 180 days Not reported very Tianen, 2005 SSEP No no N/A no no No 180 days no Topcuoglu 2009 GCS no yes no no no No 90 days no Topcuoglu, 2009 MRI yes yes N/A no no No 90 days no No Torbey, 2004 CT no no N/A no no No discharge Not reported very Vignaendra, 1974 EEG no no no no no No discharge Not reported very Wijdicks, 1994 Myoclonus no no no yes yes No 180 days no Serious Wijdicks, 2001 MRI no yes N/A no no No discharge Not reported very Wijman, 2009 MRI limited yes N/A yes no No 180 days No Serious Wu, 2009 MRI No no N/A no no No 180 days Not reported very Young 2005 SSEP No no N/A no yes No 90 days Not reported very Young 2005 PLR, CR, OVR No no yes no yes No 90 days Not reported very Zanatta, 2012 SSEPs No no N/A no no No 90 days Not reported very Zandbergen, 2006a PLR, CR, MR, myoclonus, EEG no no yes no yes No 1 year Yes (PLR, MR) very Zandbergen 2006a NSE, S-100B Yes no N/A no yes No 1 year No No Zandbergen 2006a SSEPs Limited no N/A no no No 1 year Yes Zandbergen 2006b SEP N70 Limited no N/A no no No 1 year No Serious

26 Risk of Bias in studies Author, year Index Blinding (treating team) Blinding (index or outcome evaluators) Excludes sedation or paralysis Best CPC Excludes previous neurol. diseases Interobserv. agreement assessed Length of follow-up Index test used for WLST Overall Zhang, 2011 EEG no yes yes no no No 180 days Not reported very Zhang, 2011 SSEP no yes N/A no no No 180 days Not reported very Zingler, 2003 SSEP no no N/A no no No 90 days Not reported very Zingler, 2003 NSE, S-100B no no N/A no no No 90 days Not reported very

27 Clinical Examination - 1 For the critical outcome of survival with unfavorable neurological status or death at 180 days, we identified 4 studies on brainstem reflexes, motor response, or myoclonus (650 patients, low or very low QOE) For the critical outcome of survival with unfavorable neurological status or death at one year, we identified 2 studies on brainstem reflexes, motor response, GCS, or myoclonus (172 patients, very low QOE)

28 Clinical Examination - 2 For the critical outcome of survival with unfavorable neurological status or death at discharge, we identified two studies on pupillary reflex and motor response or oculocephalic reflex (151 patients, very low QOE) For the critical outcome of survival with unfavorable neurological status or death at 30 days, we identified one study on GCS (97 patients, very low QOE). For the critical outcome of survival with unfavorable neurological status or death at 90 days, we identified two studies on corneal reflex, pupillary reflex, motor response, oculovestibular reflex, GCS, or myoclonus (97 patients, very low QOE).

29 Evidence profile table Predictor (studies) Timing Risk of bias Indirectness Inconsistency Imprecision QOE TP FP FN TN Sensitivity FPR LR+ Corneal reflex (3) At 24h No Serious low [32-42] 17 [9-27] 2 [1-4] Corneal reflex (2) At 48h No Serious low [25-35] 7 [2-20] 3 [1-8] Pupillary reflex (3) At 24h No No very low [17-25] 9 [4-18] 3 [1-5] Pupillary reflex (3) At 48h No No very low [15-23] 4 [0-12] 4 [1-14] Pupillary reflex (2) At 72h No Serious Serious low [15-23] 0 [0-8] 10 [1-71]

30 Evidence profile table Predictor (studies) Timing Risk of bias Indirectness Inconsistency Imprecision QOE TP FP FN TN Sensitivity FPR LR+ Absent CR and PLR At 24h No - low [10-18] 5 [1-17] 3 [1-11] Absent CR and PLR At 48h No - low [9-17] 3 [0-17] 4 [1-26] Absent CR and PLR At 72h No - low [10-18] 0 [0-15] 6 [0,4-87] MR 2 (2) At 24h No Serious Serious low [71-80] 27 [12-48] 3 [1-21] MR 2 (2) At 72h No Serious very low [33-44] 15 [5-31] 3 [1-7]

31 2015 Proposed Treatment Recommendations Clinical Examination - 1 We recommend using the absence of pupillary reflex to light (or the combined absence of both pupillary and corneal reflexes) at 72h from ROSC to predict poor outcome (strong recommendation, QOE very low). We suggest against using an absent or extensor motor response to pain (M 2) alone to predict poor outcome, given its high false positive rate (weak recommendation, QOE very low). However, due to its high sensitivity, this sign may be used to identify the population with poor neurological status needing prognostication or to predict poor outcome in combination with other more robust predictors.

32 Evidence profile table Predictor (studies) Timing Risk of bias Indirectness Inconsistency Imprecision QOE TP FP FN TN Sensitivity FPR LR+ Myoclonus Admission Serious No - low [35-57] 0 [0-14] 19 [1-302] Myoclonus At 24h No - Serious low [75-100] 0 [0-5] 10 [7-1646] Status myoclonus At 36-48h No Serious Serious low [5-11] 0 [0-5] 4 [0-72] Status myoclonus At 72h No low [1-4] 0 [0-14] 1 [0,1-14]

33 2015 Proposed Treatment Recommendations Clinical Examination - 2 We suggest using the presence of a status myoclonus within 72 h from ROSC in combination with other predictors for prognosticating a poor neurological outcome (weak recommendation, QOE very low). We suggest prolonging the observation of clinical signs when interference from residual sedation or paralysis is suspected, so that the possibility of obtaining false positive results is minimized.

34 Knowledge Gaps Prospective studies are needed to investigate: the pharmacokinetics of sedative and NMB drugs the reproducibility of clinical signs used to predict outcome. There is no universally accepted definition of status myoclonus Recently proposed definition = generalised myoclonus persisting for 30 minutes.

35 Electrophysiology - 1 For the critical outcome of survival with unfavorable neurological status or death at discharge, we identified 2 studies on short-latency somatosensory evoked potentials (SSEP) (63 patients, very low QOE) and 3 studies on EEG (46 patients, very low QOE). For the critical outcome of survival with unfavorable neurological status or death at 30 days, we identified 2 studies on SSEP (80 patients, very low QOE). For the critical outcome of survival with unfavorable neurological status or death at 60 days we identified 2 studies on EEG (54 patients, very low QOE).

36 Electrophysiology - 2 For the critical outcome of survival with unfavorable neurological status or death at 90 days, we identified 2 studies on SSEP or EEG (102 patients, very low QOE). For the critical outcome of survival with unfavorable neurological status or death at 180 days, we identified 6 studies on SSEP, SEP or EEG (733 patients, very low QOE). For the critical outcome of survival with unfavorable neurological status or death at 1 year, we identified 6 studies on SSEP or EEG (829 patients, low or very low QOE).

37 Evidence profile table Predictor (studies) Timing Risk of bias Indirectness Inconsistency Imprecision QOE TP FP FN TN Sensitivity FPR LR+ Absent N20 SSEP (4) At 24h very no no very low [38-48] 0 [0-3] 24 [6-93] Absent N20 SSEP (4) At 48h very no very low [40-52] 0 [0-5] 11 [3-43] Absent N20 SSEP (2) At 72h no very low [40-52] 0 [0-9] 18 [3-122] Absent N20 SSEP (3) At 12-72h very no very low [39-51] 1 [0-5] 17 [4-65]

38 Evidence profile table Predictor (studies) Timing Risk of bias Indirectness Inconsistency Imprecision QOE TP FP FN TN Sensitivity FPR LR+ EEG Grade h Serious Serious EEG Grade h Serious Serious very low [13-65] 0 [0-22] 10 [1-156] very low [11-69] 0 [0-24] 9 [1-150] EEG Grade 4-5 (3) Serious no no Serious low [33-54] 0 [0-11] 7 [1-33] Burst suppression 48h no very low [28-56] 5 [0-26] 8 [1-55] Burst suppression 72 h no very low [6-12] 0 [0-11] 5 [0,3-81] EEG 20 µv 48 h no very low [7-28] 0 [0-15] 6 [0,4-104] EEG 21 µv 72 h Sery no very low [25-37] 0 [0-11] 17 [1-272]

39 2015 Proposed Treatment Recommendations Electrophysiology We recommend using bilateral absence of N20 SSEP wave at 24h after ROSC to predict poor outcome (strong recommendation, QOE very low). We suggest using the presence of burst-suppression on EEG at 72h from ROSC in combination with other predictors for prognosticating a poor neurological outcome (weak recommendation, QOE very low). We suggest against using EEG grades due to the inconsistencies in their definitions (weak recommendation, QOE very low). We also do not suggest using low-voltage EEG, given the potential interferences of technical factors on EEG amplitude (weak recommendation, QOE very low).

40 Knowledge Gaps In most prognostication studies on TTM-treated patients, SSEPs have been used as a criterion for WLST Blinded studies are needed to assess the relevance of self-fulfilling prophecy for SSEPs. Definitions of EEG-based predictors are inconsistent among prognostication studies. Future studies should comply with recently recommended definitions (e.g. Hirsch and coll., 2013).

41 Biomarkers For the critical outcome of survival with unfavorable neurological status or death at discharge, we identified 2 studies on S-100B (99 patients, low or very low QOE) and 1 study on NSE (73 patients, very low QOE). For the critical outcome of survival with unfavorable neurological status or death at 90 days, we identified 1 study on NSE (32 patients, very low QOE) and 1 study on S-100B (27 patients, very low QOE). For the critical outcome of survival with unfavorable neurological status or death at 180 days, we identified 3 studies on NSE or S- 100B (618 patients; moderate, low or very low QOE). For the critical outcome of survival with unfavorable neurological status or death at one year, we identified 2 studies on NSE or S- 100B (86 patients, very low QOE)

42 Evidence profile table Predictor, timing (studies) Threshold (mcg L -1 ) Risk of bias Indirectness Inconsistency Imprecision QOE TP FP FN TN Sensitivity FPR LR+ NSE 24 h (3) 13,3 no - very low [33-82] 0 [0-17] 20 [1-313] 33 no no - Serious moderate [42-55] 0 [0-8] 36 [2-563] 47.6 no - very low [28-77] 0 [0-26] 12 [1-180] 8,8 very no very very low [48-93] 0 [0-17] 25 [2-389] NSE 48 h (4) very very no no very very very low [22-54] 0 [0-12] 18 [1-286] very low [64-99] 0 [0-26] 19 [1-286] 80 no no [50-77] 0 [0-3] 89 [6-1447]

43 Evidence profile table Predictor, timing, (studies) Threshold (mcg L -1 ) Risk of bias Indirectness Inconsistency Imprecision QOE TP FP FN TN Sensitivity FPR LR+ 15 No - low [55-84] 0 [0-12] 34 [2-532] NSE 72h (5) 22.4 Serious No - 33 No Serious low [78-92] 10 [2-27] 9 [3-25] low [-48-63] 3 [0-14] 9 [0-292] 90.9 no low [50-93] 0 [0-26] 17 [1-251]

44 2015 Proposed Treatment Recommendations Biomarkers We suggest using high serum values of NSE at h from ROSC in combination with other predictors for prognosticating a poor neurological outcome. However, no threshold enabling prediction with zero FPR can be recommended (weak recommendation, QOE very low). We suggest using utmost care and preferably sampling at multiple time-points when assessing NSE, to avoid false positive results due to hemolysis.

45 Knowledge Gaps There is a need for standardisation of the measuring techniques for NSE and S-100 in cardiac arrest patients. Little information is available on the kinetics of the blood concentrations of biomarkers in the first few days after cardiac arrest.

46 Imaging For the critical outcome of survival with unfavorable neurological status or death at discharge, we identified 3 studies on CT (113 patients, very low QOE) and 2 studies on MRI (40 patients, very low QOE). For the critical outcome of survival with unfavorable neurological status or death at 90 days, we identified 2 studies on MRI (61 patients, low or very low QOE). For the critical outcome of survival with unfavorable neurological status or death at 180 days, we identified 3 studies on MRI (34 patients, very low QOE).

47 Evidence profile table CT Index Timing from ROSC Risk of bias Indirectness Inconsistency Imprecision QOE TP FP FN TN Sensitivity FPR LR+ GWR<1.22 BG <24 h No low [38-84] 0 [0-45] 13 [1-190] GWR<1.18 BG <48 h No low [30-70] 17 [0-64] 3 [0,5-19] Brain swelling 72 h very no very very low [37-67] 0 [0-45] 6 [0,4-90]

48 Evidence profile table MRI Index Timing from ROSC Risk of bias Indirectness Inconsistency Imprecision QOE TP FP FN TN Sensitivity FPR LR+ Extensive DWI changes 4-32 h Serious low [65-100] 0 [0-45] 11 [1-161] Diffuse DWI FLAIR cortical changes DWI changes in BG 80h (IQR ) No No low [28-99] 0 [0-35] 12 [1-183] low [55-100] 43 [10-82] 2 [1-5]

49 Evidence profile table MRI Index Timing from ROSC Risk of bias Indirectness Inconsistency Imprecision QOE TP FP FN TN Sensitivity FPR LR+ Whole-brain ADC<665 x10-6 mm 2 /sec 2 (0-10) days very very very low [28-53] 0 [0-21] 11 [1-175] ADC < mm 2 /sec in >10% of brain volume h Serious No low [47-100] 0 [0-78] 5 [0,4-64] Occipital cortex ADC < 740 x 10-6 mm 2 /sec low [66-100] 0 [0-31] 16 [1-241] Putamen ADC < 590 x 10-6 mm 2 /sec <120 h low [66-100] 0 [0-31] 16 [1-241] Thalamus ADC < 647 x 10-6 mm 2 /sec low [57-98] 0 [0-31] 15 [1-224]

50 Evidence profile table MRI Index Timing from ROSC Risk of bias Indirectness Inconsistency Imprecision QOE TP FP FN TN Sensitivity FPR LR+ Extensive global FLAIR changes 7d very no very very low [61-100] 0 [0-78] 6 [0,4-71] Extensive cortical DWI and FLAIR changes 7d no no very low [66-100] 0 [0-78] 5 [0,4-68]

51 2015 Proposed Treatment Recommendations Imaging We suggest using the presence of a marked reduction of the GM/WM ratio on brain CT within 48 h after ROSC or the presence of extensive reduction in diffusion on brain MRI at 2-6 days after ROSC in combination with other predictors for prognosticating a poor neurological outcome (weak recommendation, QOE very low). We suggest using brain imaging studies for prognostication only in centers where specific experience is available.

52 Knowledge Gaps Prospective studies in unselected patient populations and including inter-rater agreement are needed for evaluating the prognostic accuracy of imaging studies in comatose patients resuscitated from cardiac arrest.

53 Next Steps Consideration of interim statement Person responsible Due date