Next, we have Dr. Nanda, and he will be speaking to us on Structural Connectivity Associated with Effective Capsulotomy Lesion Volumes for Refractory OCD. Dr. Nanda. Hi, my name is Pranav Nanda. I'm an incoming intern at MGH. And for the past few years, I've been working with Dr. Sheth in his Cognitive and Functional Neurophysiology Lab at Columbia. Today, I'll be talking about some of our work, particularly regarding the connectivity of lesions created by stereotactic radiosurgical capsulotomy for OCD. So, as is well known, OCD is a prevalent and debilitating disorder, but behavioral and medical therapies are effective in 80 to 90 percent of cases. For severe and refractory cases, there's the option of neurosurgical interventions, including stereotactic radiosurgical or SRS capsulotomy. An example lesion of SRS capsulotomy can be seen on the right-hand side. And it's been found to be effective in about 55 to 70 percent of cases, but its mechanism of action and predictors of its efficacy remain fairly uncertain. One hypothesis involves the corticostreatothalamocortical, or CSTC, model of OCD. And this model suggests that capsulotomy operates by modulating abnormally hyperactive CSTC loops, as illustrated at the bottom of the slide. And our goal was to try to improve our understanding of this procedure by identifying the structural and functional connectomes of effective lesions. So, in this effort, we masked lesions on postoperative T1 MRIs from 28 patients who had underwent SRS capsulotomy at Brown, Columbia, and Sao Paulo. And you can see a sample mask on the right-hand side. Clinical data for these patients were measured as a percent reduction in the Yale-Brown obsessive-compulsive scale, or the YBOX, with a 35 percent reduction being treated as a response. Because this cohort was accrued over a long period of time, the patients did not have consistent connectomic imaging, and so we used healthy control data sets as proxies. For the structural connectomes, we used 40 randomly selected patients from the Human Connectome Project, or the HCP. And for functional connectomes, we used data from the 1,000 Functional Connectomes Project. So, to create structural lesion connectomes, we nonlinearly transformed each of the lesions into the spaces of the 40 HCP patients. And we then performed probabilistic tractography within these 40 patients, or rather, within these 40 subjects, seeding from the transformed lesion, and then thresholding the tracts at two standard deviations above the mean. Finally, we transformed the resulting lesion connectomes in the 40 HCP subjects into standard space, and there we averaged the connectomes in order to generate a single structural connectome for each lesion. And a similar process was performed using 1,000 Functional Connectomes Project in order to create functional connectome maps for each lesion. Positive and negative functional connectivity maps were generated, but only the positive maps were found to have significant results, and so discussion will be limited to them. So, having created these structural and these functional connectomes for each lesion, we then looked to see if they had an association with clinical response and clinical outcome. Specifically, we used the randomized platform in FSL to perform voxel-by-voxel generalized linear models over the unions of connectomes. And within these GLMs, we associated structural and functional lesion connectomes with corresponding reductions in Y-box scores, co-varying for post-operative time. So, here we can see the descriptive heat maps of the responder and non-responder structural and functional connectomes. So, as can be seen, the overall patterns of connectivity are quite similar, suggesting that differences might be more quantitative than qualitative across these groups. Also, these similarities across response groups suggests the possibility that information is lost by binarizing patients into two discrete categories. However, when we perform the statistical analysis correlating lesion connectomes with clinical response, we find significant regions of significant association. And at bottom, we can see the heat maps of the areas of significance. Clinically significant structural connectomes included the orbitofrontal cortex and thalamus, and conversely, clinically significant functional lesion connectomes included anterior cingulate, caudate, and putamen. And although the different modalities reveal different areas, it's notable that these regions are all key players in the CSTC loops, which, as we described earlier, have been implicated in the neuropathophysiology of OCD. So, these results indicate that the therapeutic mechanism of SRS capsulotomy may indeed involve the modulation of CSTC loops, with the structural findings suggesting that the procedure targets tracts running between thalamus and OFC, and the functional findings suggesting that lesions influence CSTC circuits more broadly. And it appears that different imaging modalities do capture different aspects of the effects of capsulotomy that information may be gained by performing both methods of analysis and by comparing and corroborating the results. These findings may inform and improve the preoperative targeting of capsulotomies, as they may provide methods whereby to use connectomic imaging to predict the optimal patient specific location for clinically successful ablation. And as we move forward, it'll be useful to analyze connectomic imaging of patients themselves, both to more directly assess the structural and functional connectivity of lesions, but also to correlate lesions' features with patient outcomes and functional changes, as might be measured by postoperative EEG. It'll also be useful to develop increasing precision for capsulotomy lesioning in order to effectively leverage improved methods of target localization. The eventual goal will be to prospectively use structural and functional imaging in order to identify ideal patient specific capsulotomy targets that might optimize modulation of salient circuitry. So thanks to our collaborators for their help with this work, and thanks particularly to Dr. Banks and Pathak, whose help was instrumental for these analyses, and thanks to Dr. Sheff, whose guidance has been really instrumental over the years, for which I'm really grateful. Thank you all for listening. Any questions? Sorry to interrupt. Congratulations on your graduation. Thank you, sir. This lesion is made in variable ways and variable targets, and in variable sizes that arise in patients. Did you find any variation in this work related to the actual targets that came in your research? Yes, sir, absolutely. We actually ended up finding... Can you repeat the question? Absolutely. The gist of the question is, was there variability that we found, both in the creation of the targets and the creation of the lesions, and in that relationship, perhaps, to the effectiveness of the lesions? And we find that the lesions are actually quite idiosyncratic between patients, both in terms of their location and in terms of the timing of their development, with lesions taking six-plus months to develop. And, in fact, the locations are quite different as well, and the sizes are quite different. In our analysis, we found that the size of the lesions were not, in fact, correlated with change in YBOX, but there seems to be some association with location. Particularly, lesions which affect tracts running between the thalamus and orbitofrontal cortex appear to be more effective. So there is a region in the sort of ventral anterior limb of the internal capsule that appears to be optimal. Okay. Thanks. Wonderful. Thank you.

structural connectivity

capsulotomy lesion volumes

refractory obsessive-compulsive disorder

neurosurgical interventions

stereotactic radiosurgical capsulotomy