Abstracts

Rationale: Sudden unexplained death in epilepsy (SUDEP) is the leading cause of premature death in epilepsy. The symptoms in witnessed cases suggest a severe, seizure-related disturbance of the autonomic system. One possible explanation is an impairment of the central autonomic control due to seizure induced brainstem damage. Many epilepsy patients show a reduced heart rate variability (HRV) indicating a mild interictal autonomic dysfunction. The overall goal was to characterize the influence of structural and functional brainstem abnormalities on HRV in TLE. Methods: 18 TLE patients (mean age: 40.214.1) and 11 healthy controls (mean age: 30.77.8) underwent 5 min one channel ECG recordings in a relaxed state for HRV calculation immediately before structural (whole brain T1 and T2 images) and functional (3x 8 min task-free fMRI) 3T MR exams. Brainstem/cerebellum/diencephalon were extracted from the T1 and T2, warped onto a brainstem atlas and ICV and age-adjusted whole brainstem z-scores from the resulting Jacobian determinants calculated. The fMRI was processed in conn and BOLD activity from 30 cubic ROIs targeting brainstem and thalamic regions involved in autonomic control extracted. Graph analysis was used to characterize stationary and dynamic (sliding windows and cluster analysis of nodal strength and diversity) functional connectivity. Results: TLE had lower heart rate adjusted HRV z-scores (-0.1761.01) and more pronounced global brainstem atrophy (-0.390.92) compared to controls (0.3180.94 and -0.180.74) but the differences did not reach significance. Brainstem atrophy was negatively associated with HRV (p=0.02). In the stationary analysis TLE had increased global positive strength (4.561.00 vs 3.860.65, p=0.03) and increased nodal pos strength in 3 ROIs covering the high pons level reticular nuclei, the pulvinar and thalamic mediodorsal nucleus (FDR p < 0.05) compared to controls. Stepwise regression adjusting for atrophy identified the positive strength of ROIs covering the high pons level reticular nuclei and the mesiodorsal nucleus as being positively and that of a ROI covering the mid pons level reticular nucleus as being negatively associated with HRV. Dynamic resting state analysis identified two clusters, one large and one small, characterized by mildly (large cluster) and severely (small cluster) increased positive and decreased negative strength (hyperconnectivity states) that lasted significantly longer in TLE than in controls (large: 159.886.0 vs 87.279.4 windows/subject and small: 9.217.9 vs 31.649.00 windows/subject, p < 0.05). Stepwise regression analysis identified the same ROI strength/HRV association pattern in the large cluster as that found in the stationary analysis. Conclusions: Autonomic dysfunction manifesting itself as reduced HRV was associated with brainstem atrophy. Maintained HRV in the presence of brainstem atrophy was associated with prolonged phases increased functional connectivity in the mediodorsal thalamus, pulvinar and pons level reticular nuclei consistent with a functional protective compensatory mechanism. Funding: Epilepsy Foundation, UCSF REAC to SGM