Failure to breathe persists without air hunger or alarm following amygdala seizures

Abstract

Postictal apnea is thought to be a major cause of sudden unexpected death in epilepsy (SUDEP). However, the mechanisms underlying postictal apnea are unknown. To understand causes of postictal apnea, we used a multimodal approach to study brain mechanisms of breathing control in 20 patients (ranging from pediatric to adult) undergoing intracranial electroencephalography for intractable epilepsy. Our results indicate that amygdala seizures can cause postictal apnea. Moreover, we identified a distinct region within the amygdala where electrical stimulation was sufficient to reproduce prolonged breathing loss persisting well beyond the end of stimulation. The persistent apnea was resistant to rising CO2 levels, and air hunger failed to occur, suggesting impaired CO2 chemosensitivity. Using es-fMRI, a potentially novel approach combining electrical stimulation with functional MRI, we found that amygdala stimulation altered blood oxygen level-dependent (BOLD) activity in the pons/medulla and ventral insula. Together, these findings suggest that seizure activity in a focal subregion of the amygdala is sufficient to suppress breathing and air hunger for prolonged periods of time in the postictal period, likely via brainstem and insula sites involved in chemosensation and interoception. They further provide insights into SUDEP, may help identify those at greatest risk, and may lead to treatments to prevent SUDEP.

Keywords: Epilepsy; Neuroscience; Respiration; Seizures.

Figure 1. Stimulation-induced seizures in the amygdala evoked both ictal and persistent postictal apnea.

(A) Anatomical localization of right amygdala depth electrode contacts (black circles) in the coronal plane of P413. Numbers 1–8 specify electrode contacts from medial to lateral. Amygdala nuclei are represented as follows: La, lateral nucleus (royal blue); BL, basolateral nucleus (light blue); BM, basomedial nucleus (lavender); CEN, central nucleus (orange); CMN, cortical and medial nuclei (yellow); ATA, amygdala transition areas (light green); ASTA, amygdalostriatal transition area (forest green); AAA, anterior amygdala area (aqua); Hipp, hippocampus (brown). Short stimulation (gray shading) of contacts R2–R3 (red circles) in the right amygdala of P413 induced a focal seizure (blue shading). This resulted in postictal apneas (arrows) that became more profound 2.5 minutes after seizure termination and persisted for over 13 minutes beyond seizure end. iEEG signal is shown on top, and respiratory traces below (inspiration plotted up; conventions remain the same for B and C). (B) Anatomical localization of right and left amygdala depth electrode contacts in the coronal plane of P457. Stimulating right amygdala contacts R1–R2 (red circles) induced apnea during stimulation and during an induced unilateral right amygdala seizure. Postictal apneas persisted for over 90 seconds. Stimulating contacts L5–L6 in the contralateral left amygdala induced apnea during stimulation and induced unilateral focal seizures (bottom 3 stimulation trials). Normal baseline breathing resumed almost immediately following seizure termination. (C) Anatomical localization of left amygdala depth electrode contacts in the coronal plane of P466. Stimulating contacts L1–L2 in the left amygdala induced apnea during stimulation and during an induced unilateral left amygdala seizure. Postictal apneas persisted for over 60 seconds. (D) Summary of all 19 seizures elicited by stimulation in 7 patients. Duration of stimulation plus seizure (hatched gray bars), total apnea time (red bars), and total disrupted breathing time (black dot and line) are shown for each seizure elicited by stimulation.

Figure 2. Electrical stimulation of the amygdala evoked persistent post-stimulation apnea, an effect not seen with stimulation outside the amygdala.

(A) Anatomical localization of right amygdala depth electrode contacts (black circles) in the coronal plane of P352. Amygdala nuclei are defined as in Figure 1 (schematic remains the same for B and C). Stimulating contacts R2–R3 (red circles) in P352 resulted in apnea that was almost immediate in onset and lasted the duration of stimulation. After stimulation ended, apneas (black arrows) persisted in total for nearly 60 seconds. Repeated intervals of decreased oxygenation were observed with apneic periods. P352 was able to override stimulation-induced apnea through instructed voluntary breathing, but still exhibited post-stimulation apneas. iEEG signal is shown on top and respiratory trace from nasal pressure transducer shown below (inspiration plotted up; duration of stimulation depicted by shaded gray box; conventions remain the same for B and C). (B) Stimulating contacts L2–L3 in the left amygdala of P384 also resulted in post-stimulation apneas that lasted over 100 seconds in total and a total breathing disruption time of 5 minutes after stimulation ended. (C) Post-stimulation apneas were also observed with stimulation of contacts L1–L2 of the left amygdala in P466. (D) Summary of all stimulation trials (n = 51) for P352, P384, and P466, showing duration of stimulation (hatched gray bars), total apnea time (red bars), and total disrupted breathing time (black dot and line). Stimulation of amygdala sites in A–C led to persistent post-stimulation apneas with every trial at those sites, whereas amygdala stimulation outside these sites led to apnea of various degrees, lasting the duration of stimulation to no effect. Stimulation with the same amplitude (10–15 V) and frequency (50 Hz) outside the amygdala (white matter, WM; hippocampus, Hipp; and orbitofrontal sites) failed to induce any apnea.

Figure 3. Across-subject analysis localized post-stimulation apnea and postictal apnea to a specific site in the amygdala.

(A) Anterior-posterior, superior-inferior, and oblique views of all stimulated electrode pairs (n = 82 sites) in the temporal lobe and inferior frontal lobe across 18 subjects who had stimulation below seizure threshold (adult: triangles; pediatric: circles) plotted in a common coordinate system (MNI). Electrode contact pairs that produced apnea (red lines) were located in the medial amygdala. Electrode contact pairs that produced transient apnea (dark gray lines) were typically located just lateral or adjacent to this medial region. Electrode contact pairs that failed to induce apnea (light gray lines) were located in the lateral amygdala, areas just outside the amygdala, in hippocampus, Heschl’s gyrus, and orbitofrontal cortex. Electrode contacts may appear outside of the template brain due to anatomical variation across subjects relative to the MNI coordinate system. All electrode contacts were plotted in the right hemisphere for simplicity because no differences were observed between right and left amygdala stimulation. (B) Anterior-posterior, oblique, and superior-inferior views of all stimulated electrode pairs in the amygdala and hippocampus across the 5 subjects with persistent apnea, plotted in a common coordinate system (MNI). Electrode pairs that induced persistent post-stimulation and postictal apneas are denoted by magenta lines and clustered together mostly spanning the basolateral nucleus and including the cortical and medial nuclei and the medial aspect of the lateral nucleus. Electrode pairs that induced apnea are denoted by red lines, and transient apnea sites are denoted by dark gray lines; sites that did not induce apnea are depicted in light gray. See Supplemental Table 1 for a list of MNI coordinates and the respiratory effect for each contact pair. Nuclei are color-coded with the same conventions as in Figure 1.

Figure 4. Machine learning algorithm identified a site in the amygdala critical for persistent postictal apnea.

(A) Probability map of apnea (top) and persistent apnea (bottom) resulting from support vector machine classification of respiratory effects predicted from MNI coordinates of 87 stimulated electrode contact pairs across 20 subjects (Supplemental Table 1). (B) The persistent amygdala inhibition of respiration (pAIR) site (magenta) predicts persistent post-stimulation and postictal apneas based on the results of A, overlaid on amygdala (gray, FSL) and hippocampus (brown, FreeSurfer) (68, 69) in anterior-posterior, oblique, and lateral views. The pAIR site is located in a subregion of the AIR site (red). Probability map is plotted in the right hemisphere only for simplicity because no systematic differences were observed between right and left amygdala stimulation, and all left-sided contacts were projected to the contralateral hemisphere for the purpose of classification. For simplification, the results for transient apnea are not shown.

Figure 5. Amygdala stimulation evoked persistent post-stimulation hypoventilation despite hypercapnia.

(A) Electrode contacts superimposed upon P384’s temporal lobes (see Figure 1A). (B) While the patient was under anesthesia, intubated, and breathing independently, R2–R3 stimulation induced apnea during stimulation (as was observed at bedside). (C) etCO₂ increased after apnea followed by rapid increase in respiratory rate (RR), tidal volume (TV), and minute ventilation (VE) to normalize CO₂ levels. Dotted line indicates average pre-stimulation values. (D) L2–L3 stimulation induced long-lasting inhibition of independent breathing, causing post-stimulation apneas. Manual breaths and ventilator-dependent breathing were provided without difficulty (green shading) but did not initiate independent breathing. (E) Once independent breathing resumed, baseline RR resumed, but etCO₂ remained elevated and TV and VE decreased below baseline. Thus, P384 had persistent hypoventilation despite elevated etCO₂ for more than 10 minutes after stimulation. During this time, both etCO₂ and TV slowly returned toward baseline. (F) Comparison of respiratory measurements before and after apnea from R2–R3 (dark gray) and L2–L3 (magenta) stimulation. Site L2–L3 resulted in prolonged hypoventilation with elevated etCO₂ and lower TV and VE after independent breathing resumed compared with R2–R3. (G) Average ventilatory values before and 50 seconds after breathing resumed from stimulation of R2–R3 (m = 7) and L2–L3 (m = 3). etCO₂ was higher but TV was lower for L2–L3 50 seconds after independent breathing resumed, indicating persistent hypoventilation after L2–L3 stimulation. (H and I) Lateral amygdala (R4–R5), adjacent white matter (R5–R6), and hippocampus (H2–H3, H3–H4) stimulation failed to induce apnea or abnormal breathing (H) or changes in RR, TV, or VE (I). (J) Summary of all trials (m = 26) under anesthesia for P384. Only stimulation of L2–L3 led to persistent post-stimulation apneas. Stimulation in nearby white matter and hippocampus with the same parameters (10–15 V; 50 Hz) failed to induce apnea.

Figure 6. Persistent postictal apnea and post-stimulation apnea were not due to ongoing seizure activity or other correlated neural activity in amygdala.

(A) Time-frequency representation of the peri-ictal period for subjects P413, P457, and P466. iEEG during seizure trials indicates an increase in power (warmer colors) during the seizures but no consistent changes in the spectrotemporal response properties in the persistent apnea period. Respiratory trace shown above time-frequency plots identifies the apneas and disrupted breathing from stimulation-evoked seizure. (B) Stimulation without seizure also shows no consistent spectrotemporal changes associated with the post-stimulation apneas. (C) Extended view of 2 subjects who had persistent apneas longer than 5 minutes, P413 and P384. No changes were observed over the entire period of persistent postictal and post-stimulation apneas. (D) iEEG power changes before versus during postictal or post-stimulation apneas. Each dot represents a value from 1 patient; gray bars indicate means across patients. No significant power changes were seen in any of the canonical iEEG frequency bands (delta 1–4 Hz, theta 4–8 Hz, alpha 8–12 Hz, beta 13–30 Hz, low gamma 30–80 Hz, high gamma 80–150 Hz). (E) No significant correlation between the respiratory signal and the envelope of each canonical EEG band was observed. Each dot represents a value from 1 patient; gray bars indicate means across patients.

Figure 7. Amygdala is functionally connected to the pons, medulla, and insula.

(A) Schematic of electrical stimulation concurrent with functional MRI paradigm (es-fMRI; adapted from Rocchi, Oya, and colleagues, ref. 70). EPI, echo planar imaging; TR, repetition time. (B) Axial MRI of P352’s bilateral temporal lobes with zoomed view of right amygdala. Stimulated contacts R2–R3 (red circles) are located within the pAIR site. (C) Continuous stimulation of R2–R3 at bedside (light gray shading) induced persistant apnea. (D) During es-fMRI, the same site was stimulated with stimulation pulses (red lines) with some disruption to the subject’s normal breathing. (E) BOLD response associated with stimulation of site R2–R3 in P352. Stimulation of the R2–R3 site caused a significant decrease of BOLD activity within the medulla (t value = 3.89, P < 0.001; top panel) and superior part of the pons (t value = 3.85, P < 0.001; middle panel). Stimulation of the pAIR site significantly increased BOLD activation in the ventral part of the insula (t value = 3.74, P < 0.001; bottom panel). (F) Axial MRI of P352 with zoomed view of the right amygdala and anterior temporal cortex. Stimulated contacts R3–R4 are shown with red circles. (G) Stimulation of this site at bedside (light gray) was not associated with changes in breathing. (H) Stimulation during es-fMRI caused minimal or no changes in breathing. (I) Comparison of BOLD activity in each ROI by stimulation site. R2–R3 stimulation (pAIR site, magenta) significantly decreased BOLD activity in the medulla and pons while increasing BOLD activity in the ventral insula. In contrast, stimulation in the amygdala but outside the pAIR site and AIR site (dark gray) revealed no significant BOLD changes in the medulla or pons. Stimulation outside the amygdala in the contralateral left insula was used as a control site (white) and did not result in significant BOLD changes in the brainstem. *P < 0.05.