Acquired brain injuries: neurophysiology in early prognosis and rehabilitation pathway

2.1 Somatosensory evoked potentials

SEPs are sensitive to diffuse grey-matter lesions, such as those observed in anoxic brain injury, as well as to a mix of focal and diffuse lesions (cerebral cortex, hemispheric white matter, brainstem) detected in head trauma. Furthermore, at least in the carotid territory, SEPs are sensitive to ischaemic damage [27]. As a result, SEPs are a good indicator of overall cerebral function [28], allowing for functional assessment of neural pathways in addition to neuroimaging. SEPs are important mainly in the prognosis of a poor long-term neurological outcome (Table 1). SEP waveforms can be grouped into several patterns, based on their presence and amplitude in both hemispheres of the main and early cortical complex (N20/P25), regardless of the aetiology of coma onset. SEPs are classified as normal (N) if the N20/P25 amplitude is normal; pathological (P) if the central conduction time (CCT) is abnormally prolonged, and/or the N20/P25 amplitude is less than 1.2 mV or the left-right amplitude asymmetry is greater than 50%; and absent (A) if cortical responses are absent with preserved cervical N13 [7]. Examples of different SEP patterns are reported in Fig. 1.

Fig. 1.Examples of different SEP patterns obtained after median nerve stimulation. (A) Normal on both hemispheres. (B) Normal on left hemisphere and absent on the right hemisphere C. Absent on the right hemisphere and normal on the left hemisphere. (C) Pathological for amplitude (<1.2 µV) on both hemispheres. (D) Absent on the right hemisphere and pathological for amplitude (<1.2 µV) on the left hemisphere. (E) Absent on both hemispheres. In left columns are showed SEP components after right median nerve stimulation (cortical responses of the left hemisphere). In right columns are showed SEP components after left median nerve stimulation (cortical responses of right hemisphere).

According to different aetiologies of brain damage, the same SEP pattern can vary in prognostic meaning. In ICH, a bilaterally absent SEP pattern (AA), in which the cortical responses of both hemispheres are absent, usually predicts death [29]. In TBI, the prediction for the SEP pattern AA may be that of death, a UWS or severe disability [30]. In particular, the prognostic power of the SEP pattern AA reaches 100% specificity only if severe neurological disability (GOS 3 [according to the Glasgow Outcome Scale]) is included in the poor outcome group [30]. Concerning this particular aetiology of brain damage, caution is due in the prognosis of patients with a bilaterally pathological SEP pattern (PP) because of the possibility of diffuse axonal damage. Thus, in TBI, a good recovery is possible despite severe alteration of SEPs.

Concerning HIE, the SEP pattern AA predicts either death or non-recovery of a consciousness state (death or UWS corresponding to CPC 5 or CPC 4 according to the Cerebral Performance Categories) [1, 2, 3, 8, 9, 10, 26, 31]. In post-anoxic patients, the bilateral absence of cortical SEP is a robust poor prognostic indicator of consciousness recovery independent of the recording time. Recently, Carrai et al. [5] showed, even if in a small group of patients, that this SEP pattern also maintained its predictive value for poor neurological prognosis a few hours after cardiac arrest (CA).

Concerning this particular aetiology of brain damage, recent studies have also shown a similarly poor prognosis associated with the SEP pattern absent in one hemisphere and pathologic in the other hemisphere (AP) [2, 8, 9, 26]. Finally, in the last five years, researchers investigated the possible role of cortical SEP amplitude as a poor prognostic indicator of HIE. The authors reported that a cortical SEP present but of bilaterally low amplitude (PP) was also associated with a poor neurological outcome (CPC 5 or CPC 4) [4, 31, 32, 33, 34]. What differed across these studies was the amplitude cut-off on the better hemisphere, which varied from 0.6 to 0.3 μV [4, 32, 33, 34, 35].

Concerning SAH, the literature data are scant. SEPs seem to be more accurate than both the World Federation of Neurological Societies (WFNS) grades and the modified Fisher scale in the prediction of poor long-term neurological outcomes prior to surgical or interventional treatment, and thus it may be applied as an effective aid in preoperative assessment [36]. Therefore, according to evidence in the literature, the prognostic power of SEPs in predicting a poor neurological outcome, both in terms of consciousness and disability, strictly depends on coma aetiology.

According to the aetiology of coma, some SEP patterns, particularly the bilaterally normal (NN) or the normal-pathologic pattern (NP), can also be useful in predicting a good neurological outcome in the acute stages in a limited number of cases. In more detail, in TBI patients, a NN or NP SEP pattern may predict a good neurological outcome (GOS 4–5) [37]. Concerning HIE instead, a recent study suggested that the bilateral presence of cortical SEPs with an N20-P25 amplitude greater than 3 μV is often associated with a recovery of the consciousness state [35].

Altogether, recent data showed that SEPs could be used not only to predict the long-term neurological prognosis but also to predict an early neurological deterioration evolving toward BD. With the exclusion of HIE, the SEP patterns AA and AP predict evolution toward BD within the first 72 hours of coma with good sensitivity (75%) and specificity (84.9%) [7]. Concerning HIE patients, however, SEP pattern findings different from AA and AP always excluded BD [24].

It must be pointed out that all data reported above were collected only in adult patients. To date, fewer data are available for comatose children; however, a systematic literature review showed that, even in these subjects, SEPs should be integrated into the process of neurological outcome prediction [38].

2.2 Electroencephalography

EEG has a weaker neurological prognostic value compared with SEPs in all the aetiologies of coma, excluding HIE patients, in which, instead, it may have both a good and poor prognostic meaning (Table 1). A possible explanation is that all coma patients, except for those with CA, are treated with high doses of propofol or midazolam at an early stage after ABIs to control intracranial pressure, creating unreliable EEG findings. In HIE, the EEG instead shows a high prognostic value for both poor and good neurological outcomes, because lower dosages of the interfering anaesthetic drugs are usually used. In this condition, in fact, the EEG usually remains continuous [8, 10].

Concerning the neurological prognosis of HIE, EEG, in contrast to SEPs, is a time-dependent indicator [3]. In fact, when specific EEG patterns are found at a particular time after coma onset, it is possible to predict a neurological outcome with 100% specificity (Fig. 2A). The continuous and nearly-continuous EEG patterns (Fig. 2B–C) [39, 40] found within the first 12 hours after CA are always associated with a good outcome (recovery of consciousness), whereas an isoelectric pattern (amplitude <2 µV) after 12 hours, a burst-suppression pattern [39] after 24 hours and a suppression pattern [39] after 48 hours from CA are always associated with a poor outcome (CPC 4–5) [8, 10].

Fig. 2. Examples of EEG patterns according to Hirsch’s: malignant patterns. (A) Isoelectric. (B) Burst-suppression. (C) Suppression. Sweep time: 1 sec/div (A,C) or 2 sec/div (B); sensibility: 7 µV/mm.

Concerning TBI, evidence in the literature shows that EEG has a predictive value for a good outcome when, approximately one week after coma onset, the reactivity of the background activity is detected (Fig. 3A), whereas its absence usually marks a poor prognosis, above all when it is associated with severe SEP patterns (AP and PP) [30]. In conclusion, to date, when EEG is used at an early stage of coma, except in HIE patients, it is generally employed mainly for its diagnostic value for the detection of a non-convulsive epileptic status, a common consequence of ABIs. Furthermore, EEG has also shown its value as a guide to the treatment of refractory epileptic states or dosing of anaesthetic drugs [41].

Fig. 3.Examples of EEG patterns according to Hirsch’s classification: benign patterns: (A) reactive background; (B) continuous; (C) nearly-continuous. Sweep time: 1 sec/div; sensibility: 7 µV/mm.

2.3 Protracted phase

Establishing the neurological prognosis in post-acute stages, when patients are already admitted to the IRUs, is also important mainly for individualised multidisciplinary rehabilitation planning, given the realistic expectations of the patient’s recovery, for assisting physicians in communicating with the patient’s family members and caregivers, as well as for research trials focusing on assessment of the intervention effect. However, in contrast to the literature about the acute stages, there are no robust data about the long-term neurological prognosis in patients already admitted to the IRUs. In fact, to date, only a few studies [12, 42, 43, 44, 45, 46] have analysed post-acute clinical or instrumental predictors for late neurological prognosis, and they have shown some limitations and conflicting results.

One of the main reasons for the lack of post-acute robust neurological predictors has been the evaluation of only one neurological item at a time in most recent studies. In fact, if on one hand, some authors have focused on the evaluation of clinical or instrumental predictors of the recovery of consciousness [12, 42, 43], other authors have limited their assessment to clinical or instrumental parameters for motor recovery after CNS and/or neuromuscular involvement [13, 21, 47, 48]. Moreover, concerning the prognosis of the recovery of consciousness, another limitation is represented by the evaluation of only one clinical or instrumental parameter at a time in most recent studies.

In fact, only Scarpino et al. [43] accounted for the association of post-acute clinical and instrumental parameters for neurological outcome prediction, as suggested by Kotchoubey and Pavlov [49]. Until then, regarding clinical variables, only the total score on the Coma Recovery Scale–Revised (CRS–R) at patient admission to the IRUs and the improvement in the score during the first 4 weeks after admission had been investigated as post-acute neurological predictors [12]. Concerning instrumental parameters, EEG is the most investigated test for both diagnostic and prognostic purposes.

In contrast to neuroimaging (brain CT and MRI), as already underlined, EEG is a simple, risk-free, and inexpensive test that can be performed at the patient’s bedside. For these reasons, EEG findings represent the neurological predictor most investigated, also in the post-acute phases in patients with severe ABIs [1, 44]. However, while the 2012 American Clinical Neurophysiology Society (ACNS) terminology for EEG in the critical care setting [39] was accepted by researchers as standards for identifying specific EEG patterns as indicative of poor and good prognosis in the acute stages; data for the post-acute stages are still conflicting. This knowledge gap exists mainly because researchers have not yet agreed on specific terminology and classification of post-acute EEG findings in patients with DOC and because there is still no consensus on which EEG features are the most useful for prognostic purposes.

Another limitation is that most of the previous studies focused on one EEG descriptor at a time as the neurological indicator. Background reactivity was the most studied EEG descriptor [42, 43, 46, 50], and, despite being tested in a variety of ways, it was strongly associated with a better long-term neurological outcome when observed in a typical 30-min EEG recording. Another EEG descriptor that has been studied extensively is represented by the detectable transient patterns of stage II sleep (Fig. 4), which are used as a diagnostic method for assessing the extent of DOC as well as a late neurological predictor [43, 50, 51]. In most of the studies, this EEG feature was investigated through prolonged recordings. However, when this descriptor is examined through a standard 30-min EEG recording, its presence is also related to an improvement in the neurological outcome, albeit with reduced sensitivity, because of the lower probability of occurrence in a short EEG recording.

Fig. 4.Example of spindle activity. Sweep time: 1 sec/div; sensibility: 7 µV/mm.

Epileptic discharges have also been investigated as a neurological predictor, albeit with conflicting results. Some authors have shown that, when present, epileptic discharges hamper the recovery of consciousness [43, 52], whereas Bagnato et al. [53] reported that the occurrence of structural epilepsy does not affect the recovery of consciousness.

Some classifications for the post-acute EEG of patients with DOC have been proposed in order to standardize EEG interpretation and therefore to identify particular EEG descriptors associated with a poor or good neurological prognostic meaning. In more detail, Bagnato et al. [42] randomly assigned a specific score to each pattern of specific EEG descriptors considered as strong neurological predictors (reactivity, voltage and frequency), resulting in a total score ranging from 3 to 7. Higher scores were associated with a higher likelihood of raising one’s level of consciousness. Scarpino et al. [43] instead used the 2012 ACNS EEG terminology [38], which previously had only been used in the acute setting, to interpret the post-acute EEG. The authors observed that the presence of higher frequencies (alpha waves), detectable transient patterns of stage II sleep, reactivity and variability in background EEG activity (a descriptor never considered previously) were associated with an improvement in the consciousness level. In contrast, the presence of epileptic discharges and lower frequencies (delta waves) and the absence of reactivity and variability in the background activity were associated with a poor prognostic meaning [43]. Estraneo et al. [45] also proposed a post-acute EEG classification even though the authors showed that its reliability was greater when used as a diagnostic tool to better assess a patient’s consciousness level, rather than when used as a prognostic tool. This classification was based mainly on the assessment of the EEG reactivity associated with the evaluation of specific parameters of the background activity, such as frequency, voltage or the presence of an anterior–posterior gradient.

Recent investigations [44] have attempted to compare these previously proposed EEG classifications in order to identify which of them had the best predictive power for long-term neurological outcomes and thus the potential to be used in routine clinical practice among IRUs. Scarpino et al. [44] showed that the classification based on ACNS EEG terminology [39] had better predictive power than those of Bagnato et al. [42] and Estraneo et al. [45]. However, these data were retrospective and monocentric [44]. Thus, further prospective, multicentre studies are needed to obtain more solid and reliable data.

Concerning neuromuscular disorders, some neurophysiological parameters, such as nerve conduction (electroneurography-ENG) and electromyography (EMG), for the evaluation of spontaneous activity, have been used for CIPNM detection in critically ill patients [13, 21, 47]. The authors showed that the presence of CIPNM may prolong recovery time in IRUs, worsen the final outcome and increase the cost of hospitalisation [54, 55].

Intiso et al. [21] observed that despite a full motor recovery (40%–50% of CIPNM survivors [56, 57]), most of the patients reverted to a lower health status, including a problematic return to active daily living, resocialisation and participation. Patients perceived reduced endurance and poor physical stress tolerance as particularly debilitating.

Concerning CNS involvement, motor evoked potentials (MEPs), in association with clinical evaluation performed with the Medical Research Council scale (MRC), have been evaluated as predictors of motor function recovery. In a previous study [48], the authors showed that MEPs could be a supportive tool to increase the prognostic accuracy of upper limb motor and functional outcomes in hemiparetic patients, especially in those with severe initial paresis (MRC <2) and/or with motor evoked potentials absent in the post-stroke acute phase.

Most of these previous studies [13, 21, 42, 43, 44, 48, 54] evaluating instrumental parameters for prognostic purposes, focused only on motor recovery or consciousness recovery. Despite this limitation, the evaluation of an instrumental parameter for prognostic purposes is an important strength of these studies because it is well known that, in patients with DOC, clinical evaluation alone may not always be reliable, given the presence of several confounding factors. For example, in some cases, patients might show a lower CRS-R score because of motor impairment; on the other hand, clinical evaluation of motor disorders might also be questionable and inconclusive, especially in non-cooperative or severely cognitively impaired patients.

However, in contrast to the acute phase, in which in recent years, authors have proposed a multimodal approach for patients with severe ABIs, there is still no evidence about a multimodal clinical and instrumental evaluation of critically ill patients in the sub-acute stages, keeping account of both potential consciousness and motor recovery. This is an important topic because most rehabilitation treatments focusing on motor and functional recovery require the patient’s cooperation. For this reason, in recent years, many studies have pointed out the need to implement strategies for treating neuromuscular and CNS disorders in non-cooperative patients as well. On the other hand, knowing the expected level of motor and functional recovery after CNS damage or neuromuscular complications may help in early decision-making on medical and rehabilitation treatment, especially in patients who do not show a severe DOC and who can immediately cooperate during the rehabilitation treatment.