Hello. My name is Dr. Afijer Ohayranor. I'm going to be talking about grander causality analysis and how it can be used to identify DBS targets in patients with secondary dystonia. So, patients with secondary dystonia are challenging because the targets for DBS targets can vary with each patient. So, to get around this problem in CHLA, patients undergo a chronic, they're implanted with temporary depth electrodes and undergo chronic recordings. So, the children are monitored for about a week or so with continuous recordings, and occasionally they get tested at those electrodes using, record from two different types of electrodes, macro and micro electrodes, and record from 10 depth electrodes implanted in multiple brain areas. This is just a little scan showing the types of electrodes we've used. Once you do this, you can generate a large volume of information. We collect EMG, EEG, local field potential, individual neuron activity, multi-unit activity. And just shown on the left is a small snippet of this data. On top, you can see individual spikes. On the bottom, you can see how they're associated with EMG activity. But the question that I was interested in is to try to understand how this data can be put together to understand control circuitry and control points that modulate the basal ganglia and thalamic circuitry. So, to help address this, I looked at one patient, so this is a very preliminary analysis, who has a history of acquired dystonia, secondary to autoimmune striatal degeneration, and he had left-side dystonia. So, the patient underwent depth electrode placements in the right VPL, VIM, VOA, STN, and GPI, and he was asked to complete a finger-nose-finger task. And we recorded LFP data from macro electrodes, and these were conditioned on which side of the body the patient was moving at the time. And just so you know, all the slides going forward, left movement is shown in blue, right movement is shown in red, and no movement is shown in black. So, one way to sort of try and get at the interactions and try to understand interactions between brain areas is to use coherence. Coherence is a normalized measure of the correlation between two signals, and you can do this by basically calculating the coherence between all possible pairs. And so, once we did this, we found that for only a subset of these nuclei, there are actually movement-dependent changes in coherence. So, top left, you can see a decrease in coherence when the patient moves to the left, really only for VIM to VPL. And then, really, that change was most prominent in the gamma band, so above 35 hertz. And then STN to VIM, VIM to VOA, and GPI to VOA, you can see increases in coherence when the patient moves to the left side, but not to the right side, and not with no movement. So, coherence is nice because it tells you which nuclei are interacting, but it doesn't give you a good sense of which nuclei is maybe driving another one. So, to try to address this, one way of doing this is using something called Granger causality. So, Granger causality is basically a way of identifying or coming up with some way of understanding the direction of information flow, and it tests, really, whether or not a prediction of any particular time series, in this case, Y, is improved by including another time series, X. And if so, then you say X Granger causes Y. And this has been used to identify information flow in different areas of the cortex, but as far as we know, nobody's really looked at this to understand information flow within the basal ganglions and subcortical circuitry. So, we did this, played this game. I basically looked at the Granger causality between all possible peers, and again, in all possible directions. And what we found was, really, in this one particular patient, VPL drove VIM, GPI, and STN in a direction-specific manner. So, the most prominent sort of peak you can see on the chart on the right, and you'll see a broad-based peak that basically includes beta and gamma bands associated with when the patient moves to his left side. Again, that's blue. And then you can see smaller increases in Granger causality when the patient moves his left side for VPL to GPI and VPL to STN. So, this is basically a way of saying that VPL is, in this particular patient, driving VIM, GPI, and STN. So, just some points of discussion. So, VPL was identified actually through a separate process as a putative target for DBS in this patient, and he underwent implantation with leads in VPL, VOA, and GPI. One year later, he was doing very well. He was weaned off his dystonia medications and had no further progression of disease, but the VPL lead had to be turned off secondary to paresthesias. The point of this is basically that this is one way that chronic recordings can be used to tailor DBS treatments to identify specific targets in different patients. And then just as a bigger sort of point about trying to understand dystonia, VPL is a sensory nuclei which incorporates sending information from the cerebellum, and there's one theory about dystonia which basically says that it's generated by an inability to parse out sensory input. And so our finding that VPL might be, at least in this one patient, associated or drive dystonia supports that. Any questions? And these are my acknowledgments. Thank you. Any questions? This isn't specifically about your results, but as a matter of course for that clinical protocol, don't you have to stimulate for quite a while to see benefits for dystonia? How is that sorted out in such a short period of time with so many targets? Only really for GPI you have to stimulate for a while. So really the way targets were tested was just whether or not the patients had side effects. So it was basically sort of a guessing game. But then ultimately these patients may still get multiple targets. Correct. Yes. Exactly. Multiple targets. And then leads are turned on and off clinically a couple months later or so to see how they do. How do you ever get an insurance company to pay for that? Is this a research protocol? Yes. It's pretty crazy because basically, at least in California, they'll pay for this. And with these guys, for most of these patients now, we're doing bilateral implantations. John, actually, I'm the surgeon. So I can answer some of those questions later. So the actual procedure is actually done under an IRB. And it's actually surprisingly we do get insurance to authorize it. If you actually look at the hospitalization, the entire test hospitalization, it's lower than the cost of a single generator. So a lot of the insurance companies sort of view it as a way to potentially get out of paying for the generator if it doesn't work. And so we've actually with California, too, we have the benefit of having California Children's Services, which is very receptive to this particular procedure, surprisingly. So then you just have to demonstrate that it's not likely to be futile. Right. Yeah. And actually, you know, in answer to your earlier question, the thoracic stimulation does show very early effect. We basically implant GPI in all the patients because we can't disprove that as a target in such a short-term testing. But if patients respond to additional targets or if based on electrophysiology, we believe that additional target may be involved in there. Because usually what we do is we test the patients, then race over the following two weeks to try to analyze all the data. And then based on data analysis and the response to stimulation, we'll select a potential secondary target. So a lot of these kids end up getting four leads. We do actually occasionally have kids who just seem to not show a response and actually will exclude them from placement of the permanent simulators. Thank you.

grander causality analysis

deep brain stimulation targets

secondary dystonia

chronic recordings

coherence and Granger causality