Mapping musical automatism: Further insights from epileptic high-frequency oscillation analysis

Abstract

As ictal semiology is increasingly understood to arise from epileptogenic networks, high-frequency oscillation propagation patterns are helping elucidate networks relevant for surgical planning. Musical automatisms, a well-documented but very rare phenomenon of epilepsy, have yet to be examined as a manifestation of high-frequency propagation in the public literature. In our current study, we report a rare case intractable epilepsy with ictal humming whose epileptogenic zone was associated with the non-dominant left anterior medial temporal region. Mapping our case's ictal semiology and high-frequency propagation pattern both facilitated treatment and further supports prior observations that the rare phenomena of musical automatisms localize to a non-dominant frontal-temporal network rather than a specific cortical territory.

1 ∣. INTRODUCTION

As ictal semiology is increasingly understood to arise from epileptogenic networks, high-frequency oscillation (HFO) propagation patterns are helping elucidate networks relevant for surgical planning. High-frequency oscillation (HFO, >80 Hz) network mapping is a particularly important biomarker for differentiating epileptogenic from symptomatogenic zones given its ability to improve the resolution and localization of relevant anatomy for surgical planning.^1,2

The relation between music and the brain has been extensively researched over the past century. Despite these works, the neural networking that supports musical experience and behavior remains poorly understood. Musical vocalization is a complex phenomenon requiring widespread coordination of perioral, oral and respiratory muscles whose presence in multiple species belies an intricate network embedded in our brain's primitive circuitry.³ Musical automatisms, a well-documented but very rare phenomenon of epilepsy, have yet to be examined as a manifestation of HFO propagation in the public literature.

In our current study, we report a rare case intractable epilepsy with ictal humming whose epileptogenic zone was associated with the non-dominant left anterior medial temporal region. The patient's symptomatogenic zone for ictal humming included neural networks in both hemispheres, whose propagation mapped using HFO analysis has been consistent with previous research regarding both anatomy and networks associated with musical behavior as a whole.

2 ∣. CASE REPORT

We report a 62-year-old left-handed female with intractable localization-related complex partial seizures who presented to our institution as a candidate for surgical management of epilepsy.

The patient reported a personal history of tonic-clonic seizures as an infant that self-resolved by approximately 3 years of age. The patient was clinically seizure-free until the age of 46 years when she became aware of clinical seizure activity. The patient reported that a stereotyped aura preceded her typical episodes described as an epigastric sensation of warmth alongside a ringing sound in both ears. The patient would then become unresponsive following which would begin to demonstrate manual automatisms of mouth, hands, and feet described as picking at her clothing, walking around, and non-specific abnormal writhing. Initially, when these seizures first manifested, she would involuntarily repeat the phrase "cash over short" (an accounting term that signals a discrepancy between a company's reported and audited figures); over time, this behavior changed to a rhythmic, melodic humming. Within the year previous to evaluation, the family had described episodes with singing, but could not identify the song. Of note, though the patient had no formal musical education, she was an avid singer and took much joy in the act. During the past 16 years, the patient had attempted multiple antiepileptic drug regimes, yet her seizures had remained pharmacoresistant. The patient's neurological examination and medical history were otherwise unremarkable, and family history was significant for epilepsy in two first first-degree cousins with onset in their 20-30 seconds.

A 3T magnetic resonance imaging (MRI) brain with and without contrast did show mild asymmetry between the mesial temporal lobes. The right mesial temporal lobe and right mesial parieto-occipital area demonstrate increased FLAIR signal intensity and T2 with effacement of the gray and white matter junction.

Non-invasive continuous video-EEG monitoring was performed using a 10-20 international system with subtemporal electrodes. Interictal EEG demonstrated intermittent focal spike-wave complexes at T3/T5 with no evidence of contralateral epileptiform activity. Ictal scalp EEG demonstrated brief rhythmic focal 5 Hz discharges at T3/T5 regions with rapid propagation to the ipsilateral frontal lobe and slowing of the contralateral frontal hemisphere (Figure 1A). Ictal event seizure semiology was similar in both captured events. First episode seizure occurred during sleep. The episode began with a drawing in of legs with non-purposeful movement followed by fumbling of arms and hands, followed by humming, musical in nature, with rhythmic movements of both hands and both feet. This episode lasted 15-20 seconds, and afterward, the patient was confused for 5 minutes. The second captured episode was while awake. This episode began abruptly with light monotone humming for 20 to 30 seconds, followed by lip-smacking and small movements of her hands and feet for roughly the same duration. The patient was slightly disoriented afterward.

FIGURE 1.

Non-invasive EEG with standardized international 10-20 lead placement. A, Prolonged scalp EEG demonstrated focal and evolving rhythmic 5 Hz discharge in the T3/T5 region, followed by propagation to the ipsilateral frontal lobe with slowing of the contralateral frontotemporal region. B, Source analysis using CURRY 8 Neuro Imaging software (top left: representation of single epileptic spike-wave complex, bottom left: 2D voltage distribution map, right: 3D cortical surface representing a current density distributed dipole) localized spike-wave discharges to the left mesial anterior lobes. C, Source analysis demonstrated that within milliseconds of onset, the ictal wave propagated along the ventral mesial cortex. D, By 24 ms, the seizure has propagated along supramarginal gyrus, Heschl's gyrus and eventually to the ipsilateral frontal lobe

In both situations, the patient could not identify the hummed song. No convulsions or secondary generalizations were observed. Source analysis was provided by CURRY 8 Neuroimaging software platform, which combines multiple image datasets (EEG, ECoG, MRI, fMRI, CT) to allow for electrode position obtained from the CT to be superimposed on the MRI for validation of location accuracy. Source analysis localized ictal spike-wave discharges to the left mesial anterior lobes (See Figure 1B). Furthermore, source analysis demonstrated that within milliseconds of onset, the ictal wave propagated along the ventral mesial cortex (See Figure 1C), then by 24 ms had propagated along supramarginal gyrus, Heschl's gyrus, and eventually to the ipsilateral frontal lobe (Figure 1D).

Following localization, the patient was planned and prepped for surgical intervention. WADA testing was consistent with right-sided memory dominance and right-sided language dominance. The patient was admitted to the operating room for left craniotomy with grid and strip placement. (See Figure 2). The procedure was without complication. Intracranial video-EEG monitoring again demonstrated ictal spike and wave discharges in the left inferior anterior temporal lobe followed by HFO in the same leads with evolving polyspike and wave discharges. Ictal semiology was again associated with rhythmic humming and oral automatisms that persisted for several seconds after the ictal rhythm propagated to the ipsilateral frontal hemisphere (See Figure 2A). Intracranial ictal high-frequency oscillation (HFOs, >80 Hz) ripple analysis (80-600 Hz) revealed two zones of ictal onset localizing to the left mesial anterior temporal lobe and lateral inferior temporal cortex (See Figure 2B). Frequency domain analysis revealed two areas of predominant >200 Hz activity during the epileptic seizure onset localizing to electrodes 39 and 50 (See Figure 2C). Following ictal onset, there was polyspike evolution and propagation that resulted in HFO ripple activity localized to temporal-parietal region and continued interictally.

FIGURE 2.

Invasive intracranial EEG monitoring with 60 strip/grid electrodes. Intracranial grids included placement of one platinum electrode grid (4 × 8) was placed anterolaterally over the left temporal lobe, two 1 × 6 platinum electrode strips placed anterolaterally and mesially, and four 1/4 platinum electrode strips inferiorly along the temporal pole in a posterolateral fashion. A, Clinical humming seizure correlated with spike and wave discharges in contacts 39-43 (left anterior medial temporal lobe) followed by evolving polyspike and wave discharges in theta range with subsequent discharges in 39, 40. B, High-frequency oscillation (HFOs, >80 Hz) ripple analysis (80-600 Hz) revealed two zones of ictal onset localizing to the left mesial anterior temporal lobe and lateral inferior temporal cortex. C, Better visualized with frequency domain analysis showing >200 Hz activity localizing to electrodes 39 and 50. D, Following ictal onset and propagation, HFO ripple activity localized at to temporal-parietal region (electrode 52) and continued interictally

The patient subsequently underwent a left anterior temporal lobectomy with hippocampectomy and has been seizure-free four years since surgery.

3 ∣. DISCUSSION

The relationship between music and epilepsy is complex and poorly understood. Clinical studies suggest that the processing of music within the human brain involves numerous cortical networks of which have been hypothesized to be the same underlying networks involved in the generation and propagation of seizures that manifest with musical semiology. The pathological activation and disinhibition of these networks during a seizure is now recognized to be in part responsible for the musical phenomena associated with epilepsy. Musicogenic epilepsy (such as seizures triggered by music) may be due to hyperexcitable cortical areas sensitized to specific musical triggers, while musical ictal semiology (such as auditory hallucinations, musicophilia, ictal singing, or humming) may be due to pathological hyperactivation of normal circuits.³

Musical automatisms as ictal phenomena are rare. There have been ten cases of ictal singing reported in the literature⁴-¹¹ and a minimum of seven cases of ictal humming^8,12,13 to the best of our knowledge. Singing automatisms tend to have epileptogenic foci more often within the frontal lobe, specifically within the right/non-dominant prefrontal cortex.^8-9,14 Concerning the symptomatogenic zone of ictal singing, there have been controversies regarding its anatomical localization given the diverse cortical regions and networks activated. Some authors have speculated that ictal singing, therefore, occurs either from a release phenomenon or by activation of learned motor patterns.⁵

Humming automatism, as is in our case, appears to occur more often with lesion/foci in the temporal region.^8,15 Previous research regarding these “symptomatogenic zones” of ictal humming is rare. Bartolomei et al have suggested that ictal humming occurs only under a particular definable set of condition and ictal activity in anteromesial limbic regions followed by discharge activity in the superior temporal gyrus (STG), and the frequency of this discharge activity must be within a specific range (~6 Hz in the anterior STG and 15 Hz in the posterior STG) and that the humming starts at 0-4 seconds after the onset of discharge in STG.¹² These findings are consistent with our case's propagation pattern whereby it began with a focal rhythmic 5 Hz discharge at T3/T5 region spreading within milliseconds to the ventral medial cortex followed by rapid (approx. 24 ms) propagation to the STG via the inferior longitudinal fasciculus and ipsilateral frontal via the uncinate fasciculus lobe, whereby humming started after that.

Our current patient's seizure-free status following a left anterior temporal lobectomy with hippocampectomy demonstrates that the epileptogenic focus was indeed localized in the left anterior temporal lobe. An intracranial study of our patient revealed that ictal humming started after the ictal rhythm propagated from the non-dominant anterior temporal region to the ipsilateral frontal lobe and superior temporal gyrus (STG). To the best of our knowledge, our patient is the first surgically proven case of ictal humming due to left non-dominant mesial temporal lobe epilepsy.

These findings suggest that even though the epileptogenic zone was located in the left medial temporal area, the symptomatogenic zone for ictal humming included the frontal and temporal-parietal areas, and it is activated in a spatial-temporal specific manner. These findings were consistent with the previous hypothesis that ictal musical semiology is, in part, manifested via the step-wise recruitment of music-related neural networks in different regions rather than the activation of a specific cortical region.¹² The mapping of our case's ictal semiology and HFO propagation pattern both facilitated treatment and further supports prior observations that the rare phenomena of musical automatisms localize to a non-dominant frontal-temporal network rather than a specific cortical territory.