Neurophysiological and neuroradiological test for early poor outcome (Cerebral Performance Categories 3–5) prediction after cardiac arrest: Prospective multicentre prognostication data

The data presented here are related to our research article entitled “Neurophysiology and neuroimaging accurately predict poor neurological outcome within 24 hours after cardiac arrest: a prospective multicentre prognostication study (ProNeCA)” [1]. We report a secondary analysis on the ability of somatosensory evoked potentials (SEPs), brain computed tomography (CT) and electroencephalography (EEG) to predict poor neurological outcome at 6 months in 346 patients who were comatose after cardiac arrest. Differently from the related research article, here we included cerebral performance category (CPC) 3 among poor outcomes, so that the outcomes are dichotomised as CPC 1–2 (absent to mild neurological disability: good outcome) vs. CPC 3–5 (severe neurological disability, persistent vegetative state, or death: poor outcome). The accuracy of the index tests was recalculated accordingly. A bilaterally absent/absent-pathological amplitude (AA/AP) N20 SEPs wave, a Grey Matter/White Matter (GM/WM) ratio <1.21 on brain CT and an isoelectric or burst suppression EEG predicted poor outcome with 49.6%, 42.2% and 29.8% sensitivity, respectively, and 100% specificity. The distribution of positive results of the three predictors did not overlap completely in the population of patients with poor outcome, so that when combining them the overall sensitivity raised to 61.2%.

We report a secondary analysis on the ability of somatosensory evoked potentials (SEPs), brain computed tomography (CT) and electroencephalography (EEG) to predict poor neurological outcome at 6 months in 346 patients who were comatose after cardiac arrest. Differently from the related research article, here we included cerebral performance category (CPC) 3 among poor outcomes, so that the outcomes are dichotomised as CPC 1e2 (absent to mild neurological disability: good outcome) vs. CPC 3e5 (severe neurological disability, persistent vegetative state, or death: poor outcome). The accuracy of the index tests was recalculated accordingly. A bilaterally absent/absent-pathological amplitude (AA/AP) N20 SEPs wave, a Grey Matter/White Matter (GM/WM) ratio <1.21 on brain CT and an isoelectric or burst suppression EEG predicted poor outcome with 49.6%, 42.2% and 29.8% sensitivity, respectively, and 100% specificity. The distribution of positive results of the three predictors did not overlap completely in the population of patients with poor outcome, so that when combining them the overall sensitivity raised to 61.2%.
© Value of the Data A description of which SEP, EEG and brain CT features are relevant for neurological prognostication in CA comatose surviving patients and a description of the statistical analysis. A statistical analysis including patients with severe disability (CPC 3) in the poor outcome group (a neurological outcome aggregation similar to most of previous published data). This aggregation allows a generalizability of the data. These data has been derived from a population of patients where withdrawal of life sustaining treatment (WLST) based on prognostication of poor neurological outcome was not performed, except when brain death occurred. In most of previous prognostication data censoring from WLST, which is often based on the same predictors under investigation, has an important confounding effect. Table 1 shows the demographic characteristics of the 346 enrolled patients. Based on ROC curve analysis (Fig. 1aec), the optimal cut-off for SEPs, EEG, and brain CT that maximised sensitivity for poor outcome prediction while maintaining 100% specificity was identified, and index test data were dichotomised accordingly. SEP patterns were dichotomised as grade 2 (AA-AP) vs. grade 1 (NN-NP-PP). EEG patterns were dichotomised as malignant (isoelectric, burst-suppression) vs. non-malignant (continuous, nearly continuous, discontinuous, epileptic discharges, low-voltage, and suppression), whereas for brain CT a threshold of <1.21 of GM/WM ratio was identified (Fig. 2). Grade 2 SEP predicted poor outcome with 49.6% sensitivity and 100% (CI 43.3e55.8) specificity (AUC ¼ 0.86; CI 0.82e0.89). A GM/WM ratio <1.21 predicted poor outcome with 42.2% sensitivity and 100% (CI 36.1e48.5) specificity (AUC ¼ 0.78; CI 0.73e0.83). The cut-off ensuring 100% specificity for EEG was identified as malignant pattern (isoelectric, burst suppression) vs. non-malignant patterns (continuous, nearly continuous, discontinuous, epileptic discharges, low-voltage, suppression). Malignant patterns predicted poor outcome with 29.8% (CI 24.3e35.8) sensitivity and 100% specificity (AUC ¼ 0.90; CI 0.86e0.93). Data are reported in Table 2. As a further analysis we combined the data of the three tests and calculated the cumulative proportion of patients who were correctly identified as having poor outcome (CPC 3e5) when at least one among the three investigated indices reached the threshold for 100% specificity. According to our data, 166 patients had at least one poor prognostic criterion.  The distribution of positive data of different tests overlapped only partially in patients with poor outcome. Only 21 of these patients were detected by all three tests. In 41/128 patients with positive SEP data, in 28/116 patents with positive brain CT data, and in 1/36 patients with positive EEG data true positives were identified by only one test (Venn diagram in Fig. 3). Consequently, the cumulative sensitivity increased by adding the individual test sensitivities. When two tests were considered, if at least one of the patterns predicting poor outcome was present, the sensitivity increased from 49.6% (obtained with the best single performing test, SEPs), to 60.8% (obtained by the combination of SEPs and brain CT). When all three tests were considered, the sensitivity for poor prognosis increased to 61.2%, while maintaining 100% specificity.

Patient management
In all comatose patients included in the analysis, brain CT, EEG and SEPs were performed within 24 hours after CA. The data of these instrumental tests did not affect ongoing patient management. Patients were sedated using either propofol (range 1e2mg/Kg/h) or midazolam (range 0.03e0.1mg/Kg/h) as recommended. The choice of the targeted temperature management i.e., 34 C vs. 36 C was at the discretion of the participating centre [2].

Index tests
SEP analysis was based on the evaluation of the presence, amplitude [3]or absence of the cortical responses (N20/P25 complex) on both hemispheres [4]. We identified six SEP patterns: NN, NP, PP, AN, AP and AA, in which N stands for normal (N20/P25 amplitude is normal), P stands for pathological (N20/P25 amplitude is < 1.2mV or the difference between the two sides is greater than 50%) and A stands for absent if no reproducible cortical components could be identified in the presence of a lemniscal potential [5,6]. SEPs were elicited via stimulation of the median nerve at the wrist to an intensity of 4e5mA greater than that needed to evoke a muscular response. In the case of the use of neuromuscular blocking, Erb potential amplitude was used to estimate the intensity of the stimulation. Pulse duration was 0.2 m s and stimulus rate was 3 Hz. A portable digital 4-channel EPA apparatus was used. Recording stainless steel needle electrodes were placed at Erb's point (referred to contralateral Erb's point), spinous process CV7 (referred to the anterior neck) and C3 and C4 (referred to Fz and ipsilateral mastoid). At least two repetitions (averages of 300 responses) were needed to assess the reproducibility of waveforms. The analysis time was 100 m s and bandwidth was 5Hze3 kHz. EEGs were classified according to the ACNS terminology for EEGs in critical care [7]. The "continuity" and the "voltage" of the background activity were the main parameters taken in to account for EEG classification. Thus, the main patterns identified were: continuous; nearly continuous; discontinuous; burstsuppression; suppression; epileptiform discharges, low voltage (voltage <20mV) and isoelectric. Isoelectric (voltage <2mV) recordings were identified, although the original classification did not distinguish them from suppressed activity (voltage <10mV) [8]. Brain CT prognostic power is based on the GM/WM ratio density. In particular in our analysis we performed density measurements limited to the basal ganglia level, according to previously reported method [9e11], as GM/WM ratio ¼(Caudate Nucleus þ Putamen)/(Corpus Callosum þ Posterior limb of the Internal Capsule). For further details regarding patient management, SEP and EEG recording and Brain CT acquisition refer to the related research article [1].

Ethical approval
The protocol was approved by the Regional Ethics Committee of Tuscany (Ref OSS.15.009). Written informed consent was obtained from the patient's authorized representative prior to the subject enrolment.

Statistical analysis
Continuous variables were reported as median and inter-quartile range (IQR), whereas categorical variables were reported as numbers and percentages. Normality of baseline distribution was tested using the Shapiro-Wilk test. The Pearson's chi-square and the ManneWhitney U tests were used for comparing categorical and continuous variables, respectively. For these data, sensitivity and specificity of SEPs, EEG, and brain CT were calculated. In addition, receiver operating characteristics (ROC) curve and the corresponding area under the curve (AUC) are reported. When predicting poor outcome, maximising specificity is essential in order to avoid a falsely pessimistic prediction leading to treatment limitations in patients with a chance of neurological recovery [12,13]. Thus, data of the investigated predictors were dichotomised based on the value or category that ensured 100% specificity, as identified on the ROC curve. Both the individual and the combined prognostic accuracy of index tests have been investigated. We performed a tree-based analysis to identify the best combination of different predictors in order to maximise sensitivity for outcome prediction, and calculated the sensitivity of the possible combinations of two criteria indicating poor outcome with 100% specificity. Neurological status was determined using CPC at two follow-up points: at hospital discharge, looking at the chart review, and, for patients surviving at hospital discharge, at least 6 months after CA, by telephone interview [14,15]. A p-value<0.05 was considered statistically significant. Statistical analysis was performed using Wizard 1.9 version (Evan Miller, USA) and IBM-SPSS Statistics for Windows 25.0 version (IBM Corp., Armonk, NY, USA).