Combination of Single- and Paired-Pulse Somatosensory Evoked Potentials in Ischemic Monitoring: Preliminary Investigation in Carotid Endarterectomy

Introduction Severe ischemia induces cerebral excitability imbalance before completion of infarct. To investigate the clinical availability of this imbalance with ischemic monitoring, paired-pulse somatosensory evoked potentials (SEPs) were performed in conjunction with conventional SEPs during carotid endarterectomy. Methods For carotid endarterectomy patients with hemodynamic deficits of the middle cerebral artery area (n = 34), the excitability imbalances (Q) were measured by paired-pulse SEPs, wherein the second response (A2) was divided by the first (A1; Q = A2/A1). Regional cerebral saturation (rSO2) was also measured. Occlusion was performed twice using shunting. Results Each carotid occlusion induced a significant decrease in mean A1 and rSO2, and an increase in mean Q values (p < 0.001), which returned to the baseline level after occlusion. While neuronal imbalances were mostly transient, persistently increased Q values were observed in four cases (11.8%), all indicating postoperative abnormalities in diffusion-weighted magnetic resonance imaging (100%). Meanwhile, A1 detected the postoperative abnormality in only one case (25%). Preoperative Q values at the time of surgery were significantly higher in symptomatic patients having the upper limb deficits than those without (p < 0.01), indicating persistent or permanent imbalances. Conclusion Paired-pulse SEPs reliably identified transient, persistent or permanent neuronal imbalances, depending on the ischemic severity. These preliminary results indicated that paired-pulse SEPs, in combination with conventional SEPs (A1), may offer better ischemic monitoring.


Introduction
Somatosensory evoked potentials (SEPs) can identify cerebral ischemic changes by detecting electric failure in the penumbra, which lies within a narrow hemodynamic window (cerebral blood flow of 12-14 ml/100 g/min). While SEPs are widely used for ischemic monitoring, the unsolved issues of low sensitivity (58%) and false negatives (0 to 3.5%) suggest the need for developing another approach [1].
Growing evidence has indicated that ischemia before completion of infarct, which includes transient ischemic attacks [2,3], can induce changes in cerebral excitability through the imbalances of excitatory and inhibitory neuronal activities [4,5]. Further, recent findings on the rescued penumbra or chronic hypoperfusion have demonstrated a salvageable tissue damage, called selective neuronal loss (SNL) [6,7]. SNL is characterized by a reduction in benzodiazepine receptors in normally appearing cortex [6,8], which has dynamic properties for its rapid induction (<6 hours from stroke onset in humans [9], and in the order of minutes in animals) [10,11], and reversibility [10][11][12][13]. Although these dynamic properties suggest the potential availability with ischemic monitoring, no previous studies, to our knowledge, have addressed the issue. Cerebral excitability imbalances can be electrophysiologically evaluated by paired-pulse protocols, wherein evoked responses to two successive stimuli, typically within 100 ms in human SEPs, are evaluated [14]. Assuming that paired-pulse SEPs, in conjunction with conventional single-pulse SEPs, could contribute to the better diagnostic accuracy, this study addressed this issue in patients undergoing carotid endarterectomy (CEA).

Patients
In this observational study, we analyzed the data of patients with ipsilateral or bilateral carotid artery stenosis who underwent CEA with the following three inclusion criteria: (i) preoperative condition of stage 1 or 2 ischemia upon single photon emission computed tomography (SPECT) in the middle-cerebral artery (MCA) area of the affected hemisphere, and the availability of (ii) intraoperative shunting and (iii) conventional and paired-pulse SEPs (described below). The research was approved by the institutional review board. Informed consent to participate in the study was obtained in written form from each patient.

CEA and the electrophysiological evaluation
Preoperative 123 I-IMP dual-table autoradiography SPECT and magnetic resonance imaging (MRI, 1.5 Tesla) with plaque imaging and carotid duplex were routinely performed. Head computed tomography (CT) was performed before and on the next day after surgery, and diffusion-weighted MRI (DWI) was performed within one week after surgery. Cerebral angiography was performed on high-risk patients.
Occlusion was induced twice during the shunting maneuver (LeMaitre Vascular GmbH, Germany). Anesthesia was maintained by sevoflurane with normocarbia. The monitoring was performed by trained technicians blinded to the objective of the study.
Paired-pulse SEPs were performed by applying the median nerve stimuli at the wrist contralateral to the affected side, where the peak-to-peak amplitude (measured by N 20 /P 25 complex) of the second stimuli (A 2 ) was divided by that of the first (A 1 ; paired-pulse index Q = A 2 /A 1 , see C 1 in Figure 1). The inter-stimulus interval was 30 ms, wherein paired-pulse suppression (A 1 > A 2 ) was prominent [14].    Figure 3). The bars indicate peak-to-peak amplitudes. Each number in parenthesis shows amplitudes (uV) of A 1 and A 2 respectively. Ischemic warning criteria were set as a >50% decrease in the amplitude of A 1 [15].
The regional saturation level of oxygen (rSO 2 ) of the middle cerebral artery territory was monitored using 2channel near-infrared spectroscopy (NIRS; INVOS 5100C, Medtronic Inc., Minneapolis, MN, USA), following the focal shaving of the hair over the bilateral fronto-parietal areas. A warning sign was set at a >15% decrease (usually set as >20%) [16].

Data analysis
Off-line analyses of A 1 , Q, and rSO 2 values were performed for pre-occlusion (C 1 ), first occlusion (O 1 ), after occlusion with use of shunting (C 2 ), second occlusion (O 2 ), and after occlusion (C 3 ). The data were analyzed by Wilcoxon ranked tests using Bonferroni correction for the multiple datasets, or Mann-Whitney tests for the paired datasets, using the R 3.5.2 software program (https://www.r-project.org/). A p < 0.05 was considered significant. Data were presented as the mean ± standard error (SE) unless otherwise noted.  Figure 3A, 3B]).

Discussion
Compared to the investigation of immediate neuronal imbalances after the completion of infarct [17,18], investigation done before completion of infarct is substantially limited [2,3]. To our knowledge, this is the first clinical report investigating the real-time changes in neuronal imbalances seen during ischemic monitoring. By complementing false negatives in conventional SEPs [1], paired-pulse SEPs elucidated qualitative criteria for transient, persistent or permanent imbalances, according to the severity of ischemic load. In particular, the association of persistent imbalances with DWI abnormalities implied the clinical availability of ischemic monitoring.
Areas in the penumbra that do not progress to infarction develop SNL, which is rapidly induced (<6 hours from the stroke onset in humans [9][10][11]), reversible [13], and characterized by a reduction in benzodiazepine receptors in normally-appearing cortex [6,8]. Animal studies have further shown (i) the induction of SNL in the order of minutes (as early as 5 minutes) through major artery occlusion [5,19] and (ii) the reversibility of SNL if ischemia is limited to 20 to 60 minutes [10][11][12]. Given the congruency between SNL and neuronal imbalances in the present study, paired-pulse SEPs may be able to evaluate real-time changes in SNL.
Although the physiological mechanisms underlying paired-pulse SEPs are yet to be elucidated [20], the potential involvement of inhibitory processing has been reported in both basic [21] and clinical studies [20]. A 1 , or the first N 20 -P 25 component, reflects the overall intracortical activation of the somatosensory cortex.
In addition, N 20 is associated with Brodmann area 3b and the P 25 component is associated with Brodmann areas 2 and 1, as well as Brodmann area 4 in part [22]. When sensory stimuli were applied at intervals less than 100 ms in humans, paired-pulse suppression (i.e., A 1 > A 2 ) was observed [14]. This suppression is associated with γ aminobutyric acid A (GABA A ) receptor mediated mechanism [19] through thalamocortical processing [23]. GABA A receptors form GABA A /benzodiazepine receptor complex. Thus, given the fact that SNL is associated with the reduction in benzodiazepine receptors, it is physiologically reasonable to assume that changes in GABA A -mediated mechanisms can affect the dynamics of Q values. From this view, one simple but plausible explanation for such dynamics would be that, while A 1 is relatively resistant to prepenumbral ischemia, GABAergic changes are more susceptible to ischemic changes, which result in neuronal imbalances.
Postoperative DWI abnormalities in the present study were seen in 11.8% (4/34) of the patients, which were roughly in agreement with a meta-analysis study done in CEA patients that demonstrated prevalence of postoperative abnormalities in 10.6% [24]. Whereas this fact potentially indicates a high clinical demand for intraoperative evaluation of DWI abnormalities, the conventional SEPs in the present study indicated low sensitivity. For reference, when DWI abnormalities are targeted, the sensitivity and specificity of conventional SEPs were 25% and 82.4%, respectively. Paired-pulse SEPs reliably detected DWI abnormalities in all cases. There is a possibility, however, that persistent imbalances detected in this study were induced by SNL since SNL is not always associated with DWI abnormalities [25,26]. This issue must be further investigated using SPECT or positron emission tomography (PET).
Persistent imbalances, if intraoperatively identified, will be managed in the same way as conventional SEPs through intentional hypertension, minimization of time for declamping, and administration of edaravone [27]. Identification of such imbalances will aid in timely and proper postoperative management not only by performing DWI MRI but through prescribing medications, which turned out to be effective for treating impaired cognitive function and inducing functional recovery of the rescued penumbra [7,8]. Further, paired-pulse SEPs may aid in rapid and proper decision-making for thrombolysis or endovascular therapies.
The limitations of the present study are as follows: (i) the study design, being observational, has inherent limitations; (ii) the small sample size and the limited cases of DWI abnormalities seen in the single institution analyzed in this study may have affected the generalization of the findings; (iii) the physiological mechanisms of Q values and the involvement of SNL were left untouched, with the latter ideally investigated using iomazenil SPECT; and (iv) the paired-pulse SEPs in the study merely presented qualitative criteria of transient, or persistent or permanent imbalances, with the possible existence of ischemic thresholds, as used in Figure 2, still being unknown.
There are still many issues to be addressed, and further research involving large cohorts is needed. Issues of clinical importance will be whether paired-pulse SEPs could be applied to other cerebrovascular surgeries such as aneurysmal clipping, superficial temporal artery to middle cerebral artery (STA-MCA) bypass, or tumor surgeries involving MCA areas. Further, given the increasing opportunities for the perioperative assessment of ischemia due to the development of thrombolysis or endovascular therapies such as thrombectomy, clinical availability of paired-pulse SEPs, if proved useful, will provide rapid and proper decision making with aggressive perioperative management. Finally, it should be clarified whether pairedpulse SEPs can provide a novel diagnostic tool for cognitive impairment in patients with chronic cerebral hypoperfusion.

Conclusions
Paired-pulse SEPs, in combination with conventional SEPs, may offer better ischemic monitoring. They reliably identified transient, persistent or permanent neuronal imbalances, depending on the ischemic severity, which also suggested that ischemia-induced excitability imbalance may be much more pervasive than previously believed. Taken together, the results are suggestive of potential diagnostic targets for cerebral ischemia.

Additional Information Disclosures
Human subjects: Consent was obtained by all participants in this study. Animal subjects: All authors have confirmed that this study did not involve animal subjects or tissue. Conflicts of interest: In compliance with the ICMJE uniform disclosure form, all authors declare the following: Payment/services info: All authors have declared that no financial support was received from any organization for the submitted work. Financial relationships: All authors have declared that they have no financial relationships at present or within the previous three years with any organizations that might have an interest in the submitted work.
Other relationships: All authors have declared that there are no other relationships or activities that could appear to have influenced the submitted work.