Deep Brain Stimulation for Obsessive-Compulsive Disorder: evolution of tractography-based targeting
Presented During:
Monday, June 24, 2024: 5:45 PM - 7:00 PM
COEX
Room:
Grand Ballroom 104-105
Poster No:
20
Submission Type:
Abstract Submission
Authors:
Alba Segura-Amil1, Ki Sueng Choi2, Sonya Olson2, Jip de Bruin2, T. A. Khoa Nguyen1, Andrew Smith3, Brian Kopell2, Helen Mayberg2, Martijn Figee2
Institutions:
1University Hospital Bern, Bern, Swaziland, 2Icahn School of Medicine at Mount Sinai, New York, NY, 3Icahn School of Medicine At Mount Sinai, New York, NY
First Author:
Co-Author(s):
Introduction:
Deep brain stimulation (DBS) in the anterior limb of the internal capsule (ALIC) for obsessive-compulsive disorder (OCD) can result in large improvements in symptoms and quality of life. However, without a clear stimulation target within the anatomically variable ALIC region, clinical benefits require long trial-and-error periods of parameter optimization.
We report an evolution toward precision ALIC DBS targeting for OCD based on patient-specific tractography. Our aim was to develop a targeting method that would result in a more uniform target location and symptom response pattern. We generated a responder common map of ALIC connectivity and used it to target a new cohort of patients prospectively. To validate our targeting method, we generated a tractography-based stimulation model to predict clinical outcomes.
We report an evolution toward precision ALIC DBS targeting for OCD based on patient-specific tractography. Our aim was to develop a targeting method that would result in a more uniform target location and symptom response pattern. We generated a responder common map of ALIC connectivity and used it to target a new cohort of patients prospectively. To validate our targeting method, we generated a tractography-based stimulation model to predict clinical outcomes.
Methods:
We included 13 OCD patients implanted with bilateral DBS leads in the ALIC. In the preliminary cohort (n=3), tractography-based DBS targeting was informed by the reconstruction of the lateral ALIC connected to the brainstem and the medial ALIC connected to the dorsomedial thalamus. In cohort A (n=7), we refined our tractography-based target by including additional cortical ALIC projections [1].
Data from patients in the preliminary cohort and cohort A was used to generate a responder common map as in [2]. For each patient, the stimulation volumes were generated with the therapeutic stimulation settings. Probabilistic tractography was performed with FSL bedpostx from the stimulation volumes (seed regions) to the rest of the brain. 5000 streamlines were generated per seed voxel, and a cerebrospinal fluid avoidance mask was used. To obtain the responder common map, individual white matter tracts were warped into MNI space, thresholded at 1% to remove false positives, binarized, and averaged. This map was used to prospectively target patients in cohort B (n=7).
A tractography-based stimulation model was generated with data from cohorts A and B (n=10). For each patient, we generated a whole-brain tractogram of 10 million streamlines in MRtrix3 as in [3]. We segregated the ALIC streamlines into different pathways according to their connectivity to the PFC [1] and the medial and lateral midbrain regions highlighted by the common map. Then, we estimated the activation of these ALIC pathways for each stimulation setting and related it to symptom improvement. We fitted a linear regression model to predict Y-BOCS improvement based on the individual pathway activation.
Data from patients in the preliminary cohort and cohort A was used to generate a responder common map as in [2]. For each patient, the stimulation volumes were generated with the therapeutic stimulation settings. Probabilistic tractography was performed with FSL bedpostx from the stimulation volumes (seed regions) to the rest of the brain. 5000 streamlines were generated per seed voxel, and a cerebrospinal fluid avoidance mask was used. To obtain the responder common map, individual white matter tracts were warped into MNI space, thresholded at 1% to remove false positives, binarized, and averaged. This map was used to prospectively target patients in cohort B (n=7).
A tractography-based stimulation model was generated with data from cohorts A and B (n=10). For each patient, we generated a whole-brain tractogram of 10 million streamlines in MRtrix3 as in [3]. We segregated the ALIC streamlines into different pathways according to their connectivity to the PFC [1] and the medial and lateral midbrain regions highlighted by the common map. Then, we estimated the activation of these ALIC pathways for each stimulation setting and related it to symptom improvement. We fitted a linear regression model to predict Y-BOCS improvement based on the individual pathway activation.
·Methodology for the tractography analysis. A: Input data used in B and C. B: Methodology to generate the common map of responder patients. C: Methodology to generate the tractography-based stimulation
Results:
Tractography-based ALIC DBS improved OCD symptoms in cohorts A and B by 41.2% over 6 months with an overall 80% response rate. OCD improvement was faster and larger in cohort B using the common map target versus cohort A, targeted without these data (Month 1, median Y-BOCS improvement 7.7% in cohort A and 19.4% in cohort B; Month 6, 38.9% in cohort A and 45.6% in cohort B). Targeting based on the common map also required fewer trial-and-error parameter adjustments.
At the cortical level, the responder common map highlighted connectivity to the vmPFC and vlPFC. At the midbrain level, we observed two tracts: 1) lateral to the red nucleus and medial to the subthalamic nucleus, 2) medial to the red nucleus.
A linear model with L2 regularization was implemented and evaluated with 10-fold cross-validation. The model predicted Y-BOCS improvement (R2 = 0.53, mean squared error (MSE)=167). The empirical Y-BOCS scores and model-predicted Y-BOCS scores were also significantly correlated (Pearson's r=0.77, p=8.4e-32).
At the cortical level, the responder common map highlighted connectivity to the vmPFC and vlPFC. At the midbrain level, we observed two tracts: 1) lateral to the red nucleus and medial to the subthalamic nucleus, 2) medial to the red nucleus.
A linear model with L2 regularization was implemented and evaluated with 10-fold cross-validation. The model predicted Y-BOCS improvement (R2 = 0.53, mean squared error (MSE)=167). The empirical Y-BOCS scores and model-predicted Y-BOCS scores were also significantly correlated (Pearson's r=0.77, p=8.4e-32).
·Stimulation model prediction of Y-BOCS reduction (%). A. Predicted Y-BOCS reduction versus empirical values. r, Pearson’s correlation coefficient; p: p-value. B. Predictive weights for each feature in
Conclusions:
This study demonstrates that patient-specific tractography can help precisely select targets for ALIC DBS. This approach leads to a consistent and selective reduction of OCD symptoms with minimal parameter adjustments. Our therapeutic map of ALIC connections represents progress in overcoming the limitations of ALIC DBS, potentially contributing to a more widespread clinical use of DBS for OCD.
Brain Stimulation:
Deep Brain Stimulation 1
Disorders of the Nervous System:
Psychiatric (eg. Depression, Anxiety, Schizophrenia) 2
Modeling and Analysis Methods:
Classification and Predictive Modeling
Diffusion MRI Modeling and Analysis
Neuroanatomy, Physiology, Metabolism and Neurotransmission:
White Matter Anatomy, Fiber Pathways and Connectivity
Keywords:
Obessive Compulsive Disorder
Other - DBS, ALIC, targeting, tractography
1|2Indicates the priority used for review
Provide references using author date format
[2] P. Riva-Posse et al., “Defining critical white matter pathways mediating successful subcallosal cingulate deep brain stimulation for treatment-resistant depression,” Biol. Psychiatry, vol. 76, no. 12, pp. 963–969, Dec. 2014.
[3] A. Segura-Amil et al., “Programming of subthalamic nucleus deep brain stimulation with hyperdirect pathway and corticospinal tract-guided parameter suggestions,” Hum. Brain Mapp., Aug. 2023.