High frequency spectral changes induced by single-pulse electric stimulation: Comparison between physiologic and pathologic networks

Highlights

• Physiologic and pathologic interactions in brain networks are tested with single-pulse electric stimulation (SPES).

• We found a global early excitation of fast rhythms and a delayed inhibition.

• Specific effects were detected, depending on the type of stimulated network.

Abstract

Objective

To investigate functional coupling between brain networks using spectral changes induced by single-pulse electric stimulation (SPES).

Method

We analyzed 20 patients with focal epilepsy, implanted with depth electrodes. SPES was applied to each pair of adjacent contacts, and responses were recorded from all other contacts. The mean response amplitude value was quantified in three time-periods after stimulation (10–60, 60–255, 255–500 ms) for three frequency-ranges (Gamma, Ripples, Fast-Ripples), and compared to baseline. A total of 30,755 responses were analyzed, taking into consideration three dichotomous pairs: stimulating in primary sensory areas (S1-V1) vs. outside them, to test the interaction in physiologic networks; stimulating in seizure onset zone (SOZ) vs. non-SOZ, to test pathologic interactions; recording in default mode network (DMN) vs. non-DMN.

Results

Overall, we observed an early excitation (10–60 ms) and a delayed inhibition (60–500 ms). More specifically, in the delayed period, stimulation in S1-V1 produced a higher gamma-inhibition in the DMN, while stimulation in the SOZ induced a higher inhibition in the epilepsy-related higher frequencies (Ripples and Fast-Ripples).

Conclusion

Physiologic and pathologic interactions can be assessed using spectral changes induced by SPES.

Significance

This is a promising method for connectivity studies in patients with drug-resistant focal epilepsy.

Introduction

Brain networks have drawn increased interest in the last years, further expanding on the localizationist-lesional model. This has been especially relevant for epilepsy, as brain networks best account for the pathophysiology of this condition (Spencer, 2002), with profound therapeutic implications (Jiruska et al., 2014).

The majority of the literature on brain networks derives from functional imaging studies and particularly resting state functional MRI (RS-fMRI). However this methodology suffers from limited temporal resolution. Furthermore, the hemodynamic response is only an indirect measure of neuronal activity (Yuan et al., 2016). On the other hand, surface electro-magnetic imaging has limited spatial resolution and a low sensitivity for the deep sources (Gallen et al., 1995, Ding and Yuan, 2011). Single pulse electrical stimulation (SPES) (<0.2 Hz) (Valentin et al., 2002) and cortico-cortical evoked potentials (CCEP) (0.2–1 Hz) (Matsumoto et al., 2004) applied to implanted electrodes in epilepsy surgery candidates, overcome these constraints, and have become the gold-standard for probing in vivo physiological (Guye et al., 2008) and pathological effective connectivity (Iwasaki et al., 2010, Alarcón and Valentín, 2012, Yaffe et al., 2015). Although they induce a modulation of the networks, the evoked responses have been demonstrated to replicate the spontaneous cortical response to epileptiform discharges, both at a macroscopic (Nayak et al., 2014) as well as a microscopic scale (Alarcón et al., 2012).

The study that introduced the SPES method (Valentin et al., 2002), described two types of responses. The first was ubiquitous in the early post-stimulation period, and thus considered as physiological cortical excitability. The second type appeared in the delay period, preferentially on contacts belonging to the seizure onset zone (SOZ) and thus thought as representing pathological, epilepsy-related, hyperexcitability. However, these are low frequency waves readily visible on the raw traces. Van ’t Klooster et al, (2011), demonstrated that higher frequencies, in particular ripples (R) and fast ripples (FR) are also modulated by SPES. These are especially relevant for epileptic pathophysiology as they represent biomarkers for the epileptogenic focus (van ’t Klooster et al., 2011) and they are more useful in understanding abnormal epilepsy-related neuronal bursting than single-unit recording on micro-electrodes. (Colder et al., 1996).

The default mode network (DMN) is the first resting-state network described in fMRI (Fox et al., 2005) and thought to be preferentially deactivated during loss of consciousness in epileptic seizures (Blumenfeld et al., 2009, Moeller et al., 2010). Being a “task-negative” network serving introspection and meta-consciousness, it is deactivated by salient sensory stimuli, which switch the cingular-opercular network via the primary sensory areas (Sridharan et al., 2008).

There are controversial results on how the epileptogenic networks engage these physiological networks. Although the majority of functional connectivity studies succeed in proving a significant distinction between the SOZ and the control brain areas (Yaffe et al., 2015), the direction varies widely. For example, in the archetypical focal epilepsy, hippocampal sclerosis – mesial temporal lobe epilepsy, analyzing the alterations between the lesioned hippocampus and the DMN, Zhang et al. (2010) found increased connectivity with the posterior cingulate and decreased connectivity with dorsal mesial prefrontal, inferior temporal cortex and medial temporal lobe. However, McCormick et al. (2013) found decreased connectivity between the posterior cingulate and the epileptogenic hippocampus, that was correlated with impaired presurgical memory, and predicted postsurgical cognitive decline.

In this context, the main objective of our study was to investigate cortico-cortical connectivity using spectral changes induced by direct electrical stimulation, which represents a controlled and task-independent method. Furthermore, we assessed whether SOZ engaged in these connections differently than the physiologic neuronal networks. We hypothesized that SPES would, in general, induce a significant spectral change compared to the baseline. Based on the fMRI literature we hypothesized that the DMN would be inhibited by the SPES applied to contacts in the sensory cortex more than by SPES applied to contacts outside these sensory areas. The third hypothesis was that SOZ had a specific effect on the induced spectral changes, especially in the frequency domains related to epileptogenic pathophysiology (R and FR).