FV 1. Can EEG predict present cortical somato-sensory evoked potentials after cardiac arrest?

Abstract

Background . Despite high specificity for poor outcome after cardiac arrest (CA), somatosensory evoked potentials (N20 responses of SSEPs) have low predictive value for good outcome and might be redundant with EEG. Purpose . To determine if EEG, and whether specific features or rather global standardized EEG assessments, can reliably predict cortical SSEPs occurrence, in the early evaluation of post-CA coma. Methods . In a prospective adult CA registry, EEGs recorded within 72hours were scored according to the ACNS nomenclature and classified into pre-defined categories ( Westhall E. et al, Neurology 2016 ) “benign”, “malignant”, “highly malignant”. Correlations between EEG patterns of the best recording and SSEPs (bilaterally absent versus present) and between EEGs/SSEPs and outcome (good: CPC 1-2) at three months were assessed. Results . Among 708 CA episodes, 532 (75.1%) had present N20. Best EEG (obtained after a median of 24, range 2-72 hours) resulted “benign” in 366 (51.7%), “malignant” in 220 (31.1%), “highly malignant” in 122 (17.2%); 441 (62.3%) showed continuous background, 485 (68.5%) preserved reactivity and 140 (19.8%) epileptiform features. While all EEG patterns differed significantly (p<0.001) between patients with and without N20, only “benign EEG” was almost universally associated with present N20. Only two subjects displayed “benign EEG” and absent N20, one with a medullary lesion. “Benign EEG” was associated with N20 occurence with 98.7% (95%CI:95.9-99.9) specificity and 99.5% (95%CI:97.9-99.9) positive predictive value (PPV). Good outcome was reached in 283/533 (53.1%) of patients with N20 responses and in 1/176 (0.6%) of those without: PPV 54% (95%CI 52.8-56.9). The combination of “benign” EEG and present N20 showed similar PPV for good outcome as “benign” EEG alone: 69.0% (95% CI: 65.2-72.4) vs 68.6% (95% CI: 64.9-72.0). Conclusion . Global EEG (“benign”) assessment, rather than single features, can reliably predict SSEP responses. Since SSEP adjunction does not increase EEG prognostic performance towards good outcome, SSEP recordings could be omitted in the multimodal post-CA prognostication in patients with “benign EEG” and reserved for those with “malignant” and “highly malignant” EEGs. Reference Westhall E, Rossetti AO, van Rootselaar AF, Wesenberg Kjaer T, Horn J, Ullen S, et al. Standardized EEG interpretation accurately predicts prognosis after cardiac arrest. Neurology. 2016;86(16):1482-90.