Abstracts

Rationale: Deep brain stimulation (DBS) of the thalamic centromedian nucleus (CM) is an emerging therapy for patients with drug-resistant generalized epilepsies, including Lennox-Gastaut syndrome (LGS). Previous trials of DBS have relied on stereotactic atlases to target the CM, potentially failing to accommodate inter-patient variability in thalamic anatomy. We aimed to improve neurosurgical targeting by (i) developing a structural MRI approach for patient-specific CM visualisation, (ii) identifying the CM’s neurophysiological signatures, and (iii) mapping the CM’s connectivity with functional MRI (fMRI). Methods: Nineteen patients with LGS (13 females and six males; mean age=28 years) underwent pre-surgical 3T MRI using Magnetisation-Prepared 2 Rapid Gradient Echoes (MP2RAGE) and resting-state fMRI sequences; 14 patients proceeded to CM-DBS device implantation (bilateral Medtronic 3389 leads; four electrode contacts per lead) and intraoperative microelectrode recordings from the thalamus (Medtronic LeadPoint system). Visualization of the CM was achieved by processing MP2RAGE scans using a Sobel image filter. DBS electrode positions were reconstructed from post-surgical CT scans co-registered to the pre-surgical MRI. Linear mixed-effects analysis compared three microelectrode recording features (spike firing rate, spike burst index, and background noise-level) between three thalamic nuclei sampled along the surgical trajectory: ventrolateral, CM, and parafasicular nuclei. Whole-brain resting-state fMRI connectivity was explored using CM-DBS electrode contact positions as regions-of-interest; areas with which the CM showed significant connectivity (p<0.05, family-wise error corrected) were summarised by calculating spatial overlap with the ‘BrainMap’ meta-analytic database of 20 intrinsic connectivity networks (J Cogn Neurosci, 2011;23(12):4022-4037). Results: The CM was visualizable on pre-surgical Sobel-filtered MP2RAGE scans (Figure 1A). Intraoperative microelectrode recordings revealed that, at a group level, reductions in spike firing rate and background noise-level, and an increase in spike burst index, distinguished the CM from the ventrolateral thalamic nucleus (Figure 1B-D); however, results in individual patients were more variable. For the CM-DBS electrode contacts chosen for stimulation, fMRI connectivity was maximal with intrinsic networks that normally support emotional/interoceptive, motor/visuospatial, cerebellar, and auditory/motor speech processing, consistent with known anatomical pathways. In contrast, connectivity was less apparent with default-mode, fronto-parietal, and visual networks (Figure 2). Conclusions: Accurate targeting of the CM is achievable using pre-surgical MRI. A three stage approach allowed reliable identification of the CM: (i) ~10mm lateral and 2mm anterior to the posterior commissure; (ii) approximate location from a thalamic atlas; and (iii) patient-specific identification of CM from Sobel-filtered MP2RAGE MRI. Microelectrode recordings may provide additional important information for intraoperative CM localization; however, further characterization of individual patient variability in neurophysiological features is warranted. Therapeutic effects of CM-DBS may be mediated via modulation of distributed subcortical and cortical networks with which CM stimulation sites show strong functional connectivity.  Funding: National Health and Medical Reach Council of Australia (NHMRC, project grant number #1108881).Lennox-Gastaut syndrome Foundation (LGS Foundation) Post-doctoral Research Fellowship.