Somatosensory evoked potentials (SSEP) and Neuroprognostication after cardiac arrest

Abstract

Background:

Improved understanding of post-cardiac arrest syndrome and clinical practices such as targeted temperature management have led to improved mortality in this cohort. Attention has now been placed on development of tools to aid in predicting functional outcome in comatose cardiac arrest survivors. Current practice uses a multimodal approach including physical exam, neuroimaging, and electrophysiologic data, with a primary utility in predicting poor functional outcome. These modalities remain confounded by self-fulfilling prophecy and the withdrawal of life-sustaining therapies. To date, a reliable measure to predict good functional outcome has not been established or validated, but the use of quantitative SSEP shows potential for this use.

Methods:

MEDLINE and Embase search using words “Cardiac Arrest” and “SSEP”, “Somato sensory evoked potentials”, “QSSEP”, “quantitative SSEP”, “targeted temperature management in cardiac arrest” was conducted. Relevant recent studies on targeted temperature management in cardiac arrest, plus studies on SSEP in cardiac arrest in the setting of hypothermia and without hypothermia were included. In addition, animal studies evaluating the role of different components of SSEP in cardiac arrest were reviewed.

Results:

SSEP is a specific indicator of poor outcomes in post-CA patients but lacks sensitivity and has not clinically been established to foresee good outcomes. Novel methods of analyzing quantitative SSEP (qSSEP) signals have shown potential to predict good outcomes in animal and human studies. In addition, qSSEP has potential to track cerebral recovery and guide treatment strategy in post cardiac arrest patients.

Conclusions:

Lying beyond the current clinical practice of dichotomized absent/present N20 peaks, qSSEP has the potential to emerge as one of the earliest predictors of good outcome in comatose post-CA patients. Validation of qSSEP markers in prospective studies to predict good and poor outcomes in the cardiac arrest population in the setting of hypothermia could advance care in cardiac arrest. It has the prospect to guide allocation of health care resources and reduce self-fulfilling prophecy.

Introduction

Cardiac arrest (CA) remains a leading cause of morbidity and mortality in the United States and the world, with an incidence of 3,59,800 per year [1–3]. Of those, only 10.6% survive to the hospital [1], and most patients who obtain return of spontaneous circulation (ROSC) remain comatose on arrival to the hospital [4]. One-third of those comatose patients may have delayed awakening, greater than 72 hours after targeted temperature management, and 23% will remain comatose at one week [5, 6]. The most common cause of death among comatose cardiac arrest survivors remains withdrawal of life sustaining therapies (WLST) [7]. Clinician prognostication has been shown to greatly impact family decisions regarding WLST, and thus the need for a reliable prognostication approach remains paramount, both for providing families with clarity in making goals of care decisions and for proper allocation of critical care resources [8].

Initial research in this area had focused on reduction of brain injury secondary to cardiac arrest, in an attempt to improve functional outcomes in survivors [9]. These efforts led to the addition of therapeutic hypothermia in the international post-cardiac arrest guidelines since 2005 [10]. However, this practice has introduced new challenges for clinician prognostication in comatose post-cardiac arrest patients. The sedation and potential paralytics required during targeted temperature management may confound neurologic examination or neurophysiologic testing [11]. This has led to a proposal for a multimodal approach using clinical exam, electroencephalography (EEG), somato-sensory evoked potentials (SSEPs), serum biomarkers, and neuro-imaging modalities for neuro-prognostication in the post-cardiac arrest targeted temperature management population [12, 13].

Importantly, most of the tools in this model are used for prediction of poor neurological outcome after CA, which is in turn used in the decision to withdraw life-sustaining therapies. Thus, the data supporting each of these poor (negative) prognosticators is confounded by self-fulfilling prophecy [14]. The ability to predict good (or positive) neurological outcome remains a major limitation in the era of neuro-prognostication, thus limiting a clinician’s ability to properly identify those patients in whom life-sustaining therapies should be continued and critical care resources should be allocated. In this paper, we review the currently available prognostication tools and their validity in the era of targeted temperature management. We then investigate emerging preliminary studies for the potential use of somato-sensory evoked potentials to predict post-cardiac arrest patients who will have a positive neurological outcome.

Prognostication in targeted temperature management (TTM) Era

The widely accepted use of TTM has led to improved survival, which has increased the number of comatose post-cardiac arrest patients requiring intensive care [15, 16]. This has shed light on the need for adjunctive tools to assist intensive care providers in predicting prognosis of these patients and contributes to a growth in economic burden and demand for health care resources [17, 18].

There has been progress in the sophistication of several modalities utilized in prognosticating post-CA patients, but to date, there is no one tool or rating system to reliably dichotomize good and poor functional outcome. Hence, a multimodal approach which encompasses the neurological exam, neuroimaging, electrophysiology in the form of continuous encephalography (cEEG) and somatosensory evoked potentials (SSEP), and serum biomarkers has been adopted as the best strategic approach for prognostication and treatment in cardiac arrest [19, 20].

It is important to stress that many of these modalities are affected by the use of therapeutic hypothermia and sedation, which constitutes standard of care for these patients [21–27]. The accuracy of neurological examination is invariably compromised by the sedation and potential paralytics used during TTM. Further, metabolism of these drugs is changed by lowered body temperature, further altering time to wake [22]. Studies have shown that use of the Glasgow Coma Scale Motor Response (GCS-M) 72 hours after ROSC has a false positive rate (FPR) of 20% in prediction of poor outcome, defined as GCS-M≤2. Absent pupillary light reflexes at 72 hours from ROSC has an FPR of 2% for prediction of poor neurological outcome [9]. Conversely, the presence of bilateral pupillary reflexes lacked sensitivity in predicting favorable (positive) outcome, defined by Cerebral Performance Category (CPC) 1–2, Glasgow Outcome Score (GOS) of 4–5, or Modified Rankin Scale (mRS) of 0–3 [22].

Brain imaging has been used as a supplement to the neurological exam in patients who have remained persistently comatose [13]. A multicenter study found that ADC value of less than 650 × 10(−6) mm²/s in at least 10% of the brain volume was 90% predictive of poor prognosis, defined as failure to regain consciousness and failure to follow commands 14 days post-arrest [28]. Neuroimaging has the added benefit that it is not affected by TTM, however, absence of negative findings on MRI does not predict a positive prognosis.

Serum biomarkers for the prediction of outcome in comatose patients after CA have been proposed as they are easy to collect, inexpensive, and are not confounded by temperature management or sedation [29]. Neuron specific enolase (NSE) remains the best-studied and supported biomarker in the literature [30]. It was speculated that TTM may affect NSE release [31], however sub-analysis in the TTM trial confirmed clinically useful NSE cutoff values of >40–50ug/mL at 48–72 hours after ROSC with no significant temperature difference when comparing 33 vs 36 degrees Celsius [32]. However, despite these seemingly promising findings, multiple studies have reported patients with elevated NSE levels achieving positive neurological outcomes (CPC 1–3), leaving concern for false positive results [33].

Continuous EEG (cEEG) has been a widely accepted electrophysiological tool in prediction of outcome following cardiac arrest [22, 34]. As many as 30% of comatose post cardiac-arrest patients will have electrographic seizures during their intensive care unit (ICU) stay, but the presence of seizures does not reliably predict poor outcome [35]. Specific EEG patterns, termed “Malignant Rhythms”, have been proposed as prognostic signs for poor outcome (CPC 3–5) [26, 31, 36–38], though recent studies have shown good outcomes (CPC 1–2) even after malignant rhythm was identified [39, 40]. In contrast, early recovery of organized background EEG activity and reactivity to standardized external stimulus has been supported as a predictor of good neurological outcome and a high likelihood of awakening [13]. While EEG is one of the few tools that has shown promise as a positive predictive measure of outcome, its use is strongly confounded by the standard need for sedation in this patient cohort [41] as well as difficulties in technique standardization and resource availability [36].

SSEP and Neuroprognostication

While EEG has the ability to assess efferent functions of the brain, SSEP has the unique ability to assess afferent functionality of thalamocortical connections in comatose patients [42–46]. The short latency N20 peak (N10 in animals) is the negative cortical peak occurring approximately 20ms after stimulation of the median nerve [47]. Bilateral absence of this N20 peak has been regarded as a reliable tool to predict poor neurological outcome after cardiac arrest [24, 33, 48]. Recent studies have declared absent N20 peaks to be an early predictor of neurologic outcome even during hypothermia [9, 34].

In one study comparing 14 hypothermic to 27 normothermic patients who received SSEP days 1-2 after CA, bilaterally absent N20 was an invariable predictor of poor outcome defined as in-hospital mortality. In another study, comparing 30 hypothermic and 27 normothermic patients, 3 hypothermic and 8 normothermic patients had bilaterally absent N20, none of whom regained consciousness [49]. Further, a study of 46 hypothermic patients who received SSEP showed that 47.4% of the patients with poor outcome, defined as Glasgow Outcome Scale (GOS) 1–3 at 3 months, had bilaterally absent N20 responses [50]. In another study with 75 patients undergoing hypothermia, all patients had SSEPs recorded during hypothermia and again in 34 patients during normothermia after rewarming. During hypothermia, 13 patients had bilaterally absent N20, all of whom had poor outcome with GOS 1–3 at 30 days after discharge. This study also validates bilateral absence of N20 during hypothermia consistent with bilateral absence of N20 after rewarming [51]. Finally, in another CA study, 30 patients had SSEPs recorded approximately 72hrs after being rewarmed from hypothermia. Fourteen of these patients had bilaterally absent N20 and all died without regaining consciousness [30].

However, some studies reported rare false positive cases that have raised concern regarding the method’s reliability with TTM [52]. In a study comparing SSEP of 110 hypothermic patients at 33°C to 94 mildly hypothermic patients at 36°C, there was a false positive rate of 2.6% for bilaterally absent SSEP N20 [9]. Similarly, in a meta-analysis by Sandroni et al, there were no patients in the included studies that had neurological recovery when the N20 was bilaterally absent during TTM, but in 538 patients that had SSEP recordings following rewarming, there was 1 false positive [34]. Although experts have deemed these cases insufficient to counter the prognostic value of SSEP in post-CA prognostication [53], an ideal test of prognostication would have a false positive rate of 0%, particularly given the gravity of goals of care decisions and WLST in this patient population. Table 1 summarizes major studies investigating the use of SSEP in predicting poor functional outcome after cardiac arrest.

Table 1.

SSEP N20 Prognostic marker and clinical outcomes after cardiac arrest

Study type	Author	Poor outcome of total patients who had SSEPs	N20 Prognostic marker	Sensitivity	Specificity	Positive predictive value (PPV)	Outcome
Retrospective single-center	Bisschops [50]	67/103 (65 %)	Bilaterally absent cortical response at 72 h	N20 N/A	N/A	18/18 (100%)	GOS 3 months
Prospective multicenter	Bouwes [51]	51/77 (66 %)	Bilaterally absent cortical N20 response during hypothermia	N/A	N/A	13/13 (100%)	GOS 30 days after admission
Prospective multicenter	Bouwes [24]	207/391 (53 %)	Bilaterally absent cortical N20 response during hypothermia	28%	98%	40/43 (93.0%)	GOS 1 week, 1 month, 6 months
Prospective single-center	Cronberg [30]	53/111 (48 %)	Bilaterally absent cortical N20 response at 3 days after normothermia	N/A	N/A	N/A	CPC 6 months
Retrospective single-center	Leithner [25]	35/36 (97 %)	Bilaterally absent cortical N20 response at 3 days after cardiac arrest	N/A	N/A	35/36 (97.2%)	Coma recovery
Prospective single-center	Rossetti [101]	20/34 (59 %)	Bilaterally absent cortical N20 response at normothermia	N/A	N/A	100%	CPC 2 months
Prospective single-center	Rossetti [26]	66/111 (59 %)	Bilaterally absent cortical N20 response after passive rewarming	46%	100%	33/33 (100%)	CPC 3–6 months
Retrospective single-center	Maciel [33]	14/73 (19.2%)	Bilaterally absent cortical N20 response during hypothermia and after rewarming	38.9%	100%	14/14 (100%)	CPC 4–5 at discharge
Retrospective single-center	Moshayedi [47]	108/141 (76.5%)	Bilaterally absent cortical N20 response at normothermia	N/A	N/A	N/A	Survival to discharge
Prospective single-center	Oh [78]	141/192 (73.4%)	Absent N20, P25, or N20-P25 component after rewarming	N20: 30.5%, P25: 70.2%, N20-P25:71.6%	N20:100% P25: 100% N20-P25: 100%	N20: 43/43 (100%), P25: 99/99 (100%), N20-P25: 101/101 (100%)	CPC 6 months

SSEP (Somato-Sensory Evoked Potential), GOS (Glasgow Outcome Scale), CPC (Cerebral Performance Category)

Timing of SSEP Monitoring

There remain conflicting views among different groups regarding the ideal window of time in which SSEP measurements have the highest prognostic value. Some groups suggest that SSEPs should be measured 24 hours post-CA [54, 55]. In two large prospective studies performed in patients treated with TH, the bilateral absence of N20 at rewarming (48–72 hours after ROSC) was a reliable tool to predict poor outcome with significant accuracy [24, 26].

SSEPs and Good Outcome Prediction

While the bilateral absence of N20 peaks is a validated predictor of poor outcome, the presence of N20 peaks does not suggest good outcome [24]. Nearly half of the patients with present N20 peaks still go on to have poor neurologic outcomes of 6-month GOS 1–2 or 7-day CPC 3–5 [24, 34]. SSEP N20 is a binomial function, and therefore lacks specificity for more complex neural circuitry that plays a significant role in overall outcome. Also, withdrawal of life support has been a confounder in majority of the studies on prognostic outcome in this population [48, 56]. Despite this, the current clinical practice is defined utilizing parameters predicting poor neurological outcome [12]. This has led to the investigation of SSEP data in the prediction of good neurological outcome after cardiac arrest.

SSEP waveform: short-, mid-, or long- latency (Table 2)

Table 2:

Prognostic values of SSEP Components:

Components	Utility	Sensitivity	Specificity
N20 for poor outcome prediction	• Bilateral absence of the N20 wave of short-latency SSEPs predicts death or vegetative state (CPC 4–5) with 0 [0–5] % FPR as early as 24 h from ROSC [34]	(45–46%) [34]	~100% [34]
N20 for prediction of good outcome	• Positive predictive values of bilaterally present N20 range from 40% (29–50) [37] to 58% (49–68) [102]	N/A	N/A
N70 for poor outcome prediction	• Controversial clinical data in poor outcome prediction. • By using a cut-off of 130 ms, used to predict poor outcomes (CPC 3–5) [103] • 25% of failures to classify the N70-response, False positive rates (4–15%) [104]	94% [103]	97% [103]
N70 for good outcome prediction	• When N70 present, 28% had good outcome prediction (CPC1–3) at 1 mo [104]. • Cannot reliably predict good outcomes	N/A	N/A
ML-SSEP at 50 mA(milli-ampere)	• 6 mo CPC (1–3) good outcome prediction [63] • Consciousness recovery [63] • This study [63] needs further internal and external validation.	79.41% [63] 87.10% [63]	100% [63] 100% [63]
SL Amplitude	• Good outcomes (CPC 1–3) at discharge with amplitudes >0.62 mV[64] • Poor sensitivity • The amplitude threshold should be interpreted with caution considering the tolerable noise levels	65% [64]	N/A
Absent N20-P25 for poor outcome	• Absence of N20-P25 component improved the sensitivity for predicting a poor outcome compared to the absence of N20	71.6% [78]	100% [78]
Absent SSEP for poor outcome	• Absent SSEP, typically 48–72 h after resuscitation	28% [105]	100% [105]
SSEP Patterns A-E	• Poor outcome (CPC 4 and 5) and good outcome (CPC 1–3) were defined within four weeks of CPR. • A: Bilateral preserved short- and middle long latency cortical potentials • B: Unilateral preserved N20 peak and middle long latency cortical potentials • C: Bilateral preserved N20 peak, only unilateral detectable N35 peak • D: Bilateral/unilateral preserved N20 peak, absence of middle long latency cortical potentials • E: Bilateral absence of N20 peak	A-C: 92%, [96] D: 24.1% [96] E: 58%, [96]	A-C: 63%, [96] D: 93.3%, [96] E: 98.4%, [96]
PSA	• Only animal data [42] • Use to track early recovery and predict good outcomes	78% [42]	83–100% [42]

SSEP (Somato-Sensory Evoked Potential), CPC (Cerebral Performance Category), FPR (False Positive Rate), ROSC (Return of Spontaneous Circulation), ML-SSEP (Medium Latency-Somato-Sensory Evoked Potential), SL (Short Latency), PSA (Phase Space Area)

The SSEP waveform is conventionally viewed as comprised of two components - the short-latency (SL) and the long-latency (LL) complexes. Studies in the past have shown the SL-peak N20 as an indicator of thalamocortical integrity and the LL-peak N70 as an indicator of cortical function [57]. Clinical studies had suggested the LL-peak N70 to be a predictor of good outcomes. An animal study showed the importance of separation of long and short latencies during early recovery from brain ischemia during CA. With improved peak detection, the study showed LL-peak latencies to be predictive of good outcomes, assessed by Neurologic Deficit Score (NDS) ≥50, after asphyxia CA [58].

N60 or Mid Latency Somatosensory Evoked potential (MLSEP), a stable sequence of later-occurring potentials, is generated by complex cortico-cortical interactions and is modulated by the ascending reticular activating system. Clinical studies in stroke patients showed ML-peak is a robust marker of positive outcomes and survival with high specificity, positive predictive value, and likelihood ratios. They represent cortico-cortical synapses which are essential for arousal and recovery of coma [59–62]. A recent study showed ML-SSEP to be an earlier predictor of good outcomes, defined by coma recovery and 6-month CPC scores, compared to EEG [63].

Quantitative SSEPs

Multiple components of SSEP can be quantified, such as latency, amplitude, and shape. While various quantitative SSEP (qSSEP) algorithms have been developed, very few have been validated under temperature management conditions [64], especially during the early recovery period when other prognostic tools are not available.

Amplitude and Latencies

Perhaps the most obvious method of quantification of SSEP signals is the calculation of peak amplitude and latency. Short latency response (SLR) consists of N20/P25 and long latency response (LLR) consists of cortical potentials between 40–70ms. Multiple groups have studied the amplitude and latency of SSEP under the effect of temperature in both animals and humans [65–74]. There is prolongation of SSEP latencies in hypothermia (both SLR and LLR) and faster conduction during hyperthermia, suggesting that this test may be most valid once a patient has been re-warmed to normothermia. Studies have found that post-CA patients treated with hypothermia had significantly prolonged cortical N20 (N10 animal) peaks compared those treated with normothermia [75, 76]. Additionally, it was reported that normothermic patients with hypoxic-ischemic damage, defined by Cerebral Performance Category Scale (CPC) > 2 at 1 year, had significantly prolonged N20 latency and lower N20 amplitude compared to those without hypoxic-ischemic damage [75]. The effect of hypothermia on SSEP amplitudes remains unclear and unpredictable, with one study showing unanticipated increase in amplitude of SLR due to hypothermia [69]. This phenomenon could possibly be due to diminished activity of cortical inhibitory neurons, such as the thalamic reticular neurons, leading to a hyper-excitable state [77]. In one study, LLR disappeared earlier than SLR after the onset of CA insult and also had delayed recovery compared to SLR [58]. Thus, the study generated the hypothesis that the basic neural functions in the subcortical regions may be better preserved in HIE compared with higher-level cognitive function in the cortex. These studies have led to a better understanding of brain recovery post-CA and the effect of hypothermia on the neuronal structures.

Finally, a study of post-CA patients under TTM found a threshold amplitude of 0.62uV, at which lower amplitudes were associated with poor functional outcome, defined as CPC 4–5. [64] Contrarily, high amplitude SSEPs would argue against HIE. A study performed in South Korea, where WLST is outlawed, reported that an N20-P25 amplitude >2.31 had a sensitivity of 52.9% and specificity of 96.5% for predicting good functional outcome (CPC 1–2), though generalizability of this study remains limited [78].

Another recent study investigated the relationship between early and late High Frequency Oscillation (HFO) burst amplitude and clinical outcome [79]. HFO bursts constitute an early component (before N20) generated by thalamo-cortical action potentials and a late component (after N20) demonstrating highly organized spiking activity in the primary somatosensory cortex [80, 81]. Generation and modulation of HFO bursts is determined by the neuronal connections involved in arousal and consciousness [80–82]. The study revealed late HFO burst amplitudes above 70 nV to be a marker of the absence of severe hypoxic encephalopathy and hence predicting good outcomes [64, 79]. Absence of early or late HFO bursts did not indicate HIE or poor outcome [64, 79].

SSEP latencies and amplitudes hold potential promise in predicting positive outcomes in comatose post-CA patients, but these values require further validation and investigation.

Non- amplitude/latency quantitative SSEP - Phase Space Area

Phase space area (PSA) is a novel quantitative SSEP (qSSEP) technique that has been developed and tested in rats [42]. This qSSEP metric brings objectivity to SSEP interpretation and has been shown to predict good outcomes, assessed by NDS, in post-CA rats and track cerebral recovery and the effect of hypothermia [42, 76], while standard interpretation of absent N20 responses only predicts poor outcome.

The quantitative analysis uses the phase space curve (PSC), a plot of the first derivative against magnitude, to compute the phase space area (PSA), which is representative of signal power and quantified to obtain a single PSA value that represents the functionality of the somatosensory pathway [42]. The PSC differs from standard SSEP waveform analysis in that it incorporates multiple peak characteristics including peak amplitudes, slopes, and inter-peak latency [42]. In the initial study by Madhock et al, the animals with good (NDS ≥ 50) and bad (NDS <50) outcomes had significantly different PSA values in the first 3 hours of early recovery (85–190 min post-ROSC), with an outcome prediction accuracy of 80–93% (p<0.05) and sensitivity and specificity for prediction of good neurological outcome of 78% and 83–100% respectively [42]. Further, the early recovery PSA (within the first 4 hours after ROSC) remained consistent with 72 hour NDS scores [42]. To compare this method to standard SSEP analysis, N10-P15 peak-to-peak amplitude was shown to have similar trends for outcome prediction, but they were less separated between the outcome groups and had a higher variability. The use of PSA was tested using a graded hypothermia model, which showed that qSSEP-PSA had greater recovery in groups with deeper hypothermia and in subjects with good neurological outcomes (NDS ≥ 50 at 72 hours), corroborating the utility of PSA in tracking cerebral recovery and demonstrating the marker to have an association with good outcomes [83].

Quantitative SSEP-PSA provides the potential for easily-interpreted, continuous, quantitative criteria for determination of good prognosis after cardiac arrest, which encompasses multiple factors and conditions rather than the dichotomous N20 peak presence, and has been preliminarily validated in an animal hypothermia model. Still, this must be interpreted with caution, as the existing studies are preliminary and have not yet been tested in humans [83].

Table 2 summarizes the reported prognostic value of variable SSEP waveform components.

Role of SSEP in Tracking Recovery

SSEP, in contrast to EEG, also provides detailed information on degree of functional damage to specific central nervous system pathways: somatosensory, motor, auditory, visual. SSEPs can be used to assess the integrity of the somatosensory pathways, repair of normal thalamocortical coupling, and onset of arousal [42]. One animal study showed prominent and quantifiable differences existed in the SSEP based on the severity of neurologic injury, and SSEP evolved in a predictable manner during the recovery phase [43]. The study showed a significant difference between rats with good (NDS ≥50) and bad (NDS < 50) outcomes, in terms of peak to peak amplitudes and mean N10 during the early recovery phase [43]. In addition, animal studies demonstrated that various SSEP components such as N10 (N20 in humans) Peak Variance and HFO are able to illustrate pathology of the sensory pathway due to CA [44–46]. Figure 1 shows representative figures from Jia lab in a rat after 8min asphyxial cardiac arrest with comparison between traditional SSEP waveforms and qSSEP during different time periods.

Figure 1 shows representative figures from Jia lab in one rat after 8min asphyxial cardiac arrest with comparison between traditional SSEP waveforms and qSSEP during different time periods in A (baseline), B (30min after ROSC), C (2h after ROSC), and D (3h after ROSC).

Role of SSEP in Other Pathologies

SSEPs have been used in neuromonitoring outside of cardiac arrest prognostication. One such utility is the monitoring of central conduction time (CCT) and N20 amplitudes during cerebral aneurysm clipping to predict impending strokes and neurologic deficits intra-operatively [84]. There have also been multiple studies addressing SSEP as a prognostic marker of recovery post-stroke [85, 86], as well as a tool for neuromonitoring during endarterectomy [87] and spine procedures [88]. SSEPs and motor evoked potentials may provide prognostic information on functional recovery after spinal cord injury [89, 90]. SSEP was also used as a comparative tool, in combination with MRI, in patients with cervical spondylotic myelopathy (CSM). Patients with CSM who had abnormal SSEPs showed a decrease in myelin water fraction (MWF) (p < 0.05). MWF was also correlated with SSEP latencies [91].

In traumatic brain injury (TBI), SSEPs may help detect patients with good or poor functional outcome, assessed by GOS and Barthel Index [92, 93]. In one study, SSEP on day three after TBI was shown to predict cognitive functions and neuropsychologic measures at one year [94]. In addition, SSEP was used as a monitoring tool in a neurological decompression sickness model of rats. Results indicated that abnormal SSEPs were seen in rats with spinal cord ischemia resulting from decompression sickness. SSEPs were improved when rats were pretreated with helium [95]. Thus, SSEP has come to the forefront as an early prognostic and intra-operative neuro-monitoring modality for multiple neuropathologies.

Limitations of SSEP and potential for improvement

SSEP has been accepted into standard practice for use in predicting poor neurological outcome in comatose patients post-cardiac arrest and post-TTM, but assessment for absent bilateral N20 peaks remains, to some degree, subjective, even among highly trained experts, with only moderate inter-observer consistency [96, 97]. Hence, an objective and reliable measurement of qSSEP (Quantitative SSEP) could achieve these requirements, if verified to predict outcome under temperature management. Moreover, the utility of SSEP is limited by availability of skilled personnel with requisite expertise and equipment resources, as methodological differences can greatly affect results. Given this potential for variability, difficulty with generalizability of studies in this field must be considered. A simple automated algorithm might be helpful in addressing this problem. The other future directions in this field lie in predicting good functional outcomes, understanding neuronal connections, and establishing qSSEP beyond a dichotomous interpretation. Despite the merits of SSEP, there have also been reported cases of recovery of absent bilateral N20 responses and positive functional outcome despite absent N20 responses [24, 25, 52, 98]. In a large prospective study, SSEPs of 3 patients with good outcome were initially classified as bilaterally absent. Retrospective assessment indicated that noise levels had impeded reliable interpretation [24]. It is important to acknowledge that in the intensive care setting, interference from muscle artifacts, muscle relaxants, electrical beds, infusion pumps, ventilators and low signal to noise ratio make detection of the N20 SSEP wave difficult and potentially cause falsely pessimistic predictions. As referenced previously, withdrawal of life sustaining therapies has been a confounder in testing poor (negative) prognostic abilities of several tools, including SSEPs, in this population [48, 56].

Conclusion

Establishing early prognosis of functional outcome in comatose post-cardiac arrest patients remains an enticing area of research. Current practice surrounds a model of predicting poor functional outcomes by utilizing a multimodal approach involving clinical exam, neurological imaging, cEEG, and bilateral N20 responses on SSEP. This prognostic information is being used by families to make end-of-life decisions for their loved ones, thus stressing the importance of sensitivity and specificity of these tests and the gravity of false positive findings.

SSEP remains a promising modality of neurophysiological testing, which remains unique in its evaluation of afferent thalamic and cortical connectivity. It has been shown to maintain validity in the era of targeted temperature management and is not confounded by sedation or other medications [9, 99]. However, the current dichotomy of N20 response in the prediction of poor functional outcome continues to have limitations.

The ability to predict positive outcome in coma following cardiac arrest may allow better appropriation of critical care resources and negates the possibility of self-fulfilling prophecy, yet a tool for this capability does not yet exist. Review of the current literature suggests that newer methods of analyzing SSEP may open this door [42, 64, 100]. The qSSEP model may provide a quantitative assessment of injured neuronal circuitry and real-time tracking of cerebral recovery [42, 43]. Further study of qSSEP in the setting of targeted temperature management and in human subjects holds potential for improved quality of neuro-prognostication in coma after cardiac arrest and should be an area of future study.