Authors :
Presenting Author: Sina Sadeghzadeh, AB – Stanford University School of Medicine
David Purger, MD PhD – Stanford University; Casey Halpern, MD – University of Pennsylvania; Cornelia Drees, MD – Mayo Clinic Arizona; Mike Sather, MD – Penn State Health; Shirley McCartney, PhD – Oregon Health & Sciences University; Ahmed Raslan, MD – Oregon Health & Science University; Christopher Skidmore, MD – Thomas Jefferson University; Chengyuan Wu, MD MSBmE – Thomas Jefferson University; Ioannis Karakis, MD, PhD, MSc – Emory University School of Medicine; Barbara Jobst, MD – Dartmouth Hitchcock Medical Center; Ji Yeoun Yoo, MD, FAES, FACNS – Icahn School of Medicine at Mount Sinai; Saadi Ghatan, MD – Icahn School of Medicine at Mount Sinai; Jonathon Parker, MD PhD – Mayo Clinic Arizona; Vivek Buch, MD – Stanford University
Rationale:
The effectiveness of the NeuroPace Responsive Neurostimulation System (RNS) has been established through pivotal clinical trials and long-term studies. There has been a growing trend towards the use of stereoelectroencephalography (sEEG)-guided depth electrode implantation in place of cortical strip electrodes. However, the impact of sEEG guidance and the efficacy of neocortical depth leads for RNS therapy has not yet been explored. Here, we retrospectively analyze the relationship between seizure reduction and distance between the sEEG epileptic focus (sEEG-EF) and implanted RNS depth electrode.
Methods:
Patients enrolled at participating institutions with at least one depth electrode placed in a neocortical structure (non-hippocampal) were included. Pre-operative T1-weighted magnetic resonance imaging and post-operative computed tomography images (post-sEEG and post-RNS placement) were merged to ascertain the precise location of lead contacts. The Euclidean distance between the sEEG-EF and the nearest RNS depth electrode contacts was calculated and correlated with the percent reduction in post-operative seizure frequency. Linear regression models and Pearson correlations were employed to analyze outcomes.
Results:
Data from 46 patients across eight institutions was analyzed. RNS leads were implanted across multiple neocortical locations including frontal, insular, occipital, parietal, and temporal lobes, as well as non-neocortical hippocampal and thalamic structures (Table 1). Patients were stratified into three groups based on seizure decline percent at one year (1-Y) (strong-responders: >50%; weak-responders: >0% and ≤50%; and seizure worsening: ≤0%). The distribution of Euclidean distances did not vary significantly across the three groups (p=0.16). Interestingly, a smaller Euclidean distance predicted enhanced seizure reduction among strong-responders and non-responders but not among weak-responders (Figure 1). In long term follow up (LTFU; defined as two to three years post-operatively), seizure decline for weak-responders showed significant improvement (1-Y: M=41.48, SD=11.79; LTFU: M=61.66, SD=26.40; t(11)=2.09, p=0.06), with larger distances demonstrating more substantial improvement (β=4.57, p=0.03). However, long term outcomes for strong-responders trended towards a decline in efficacy (1-Y: M=91.13, SD=13.35; LTFU: M=86.56, SD=15.63; t(41)=-1.051, p=0.29), irrespective of distance (β=0.85, p=0.10). Long term response among the seizure worsening group was more variable, but general improvement was also seen with two patients demonstrating a positive response in LTFU. These results were not driven by the presence of hippocampal leads.
Conclusions:
In this first of its kind multi-center analysis, neocortical depths for RNS therapy are shown to be effective, and sEEG-guidance plays an important role in dictating initial outcomes for a subset of patients. However, long-term enhancements in clinical response among initially weak-responders and the seizure worsening group are not dependent on sEEG-guided lead proximity and may potentially be related to the critical benefit of network remodeling over time. Further investigation into these non-distance related patient factors is warranted.
Funding: NA