ERNA: A newer neurophysiological biomarker to automate DBS programming

She's a neurologist at the Florey Department of Neuroscience at the University of Melbourne in Australia and a PhD candidate. Welcome, Dr. Nagao. 

Dr. Kanae J. Nagao: Thank you for having me. 

Dr. Mitra Afshari: We're very happy to have you and today, we wanna discuss, a recent article that you published called "Automating Subthalamic Deep Brain Stimulation Programming with Evoked Resonant Neural Activity, ERNA in Parkinson's Disease."

And this was published just this past April, in the "Movement Disorder" journal, the senior PI on the work being Wesley [00:01:00] Thevathasan your PhD advisor. And so Dr. Nagao, could you just start off by giving just a little bit of background about yourself and the, work you've done more recently?

Dr. Kanae J. Nagao: So I'm a neurologist currently working in Melbourne in Australia, and I'm currently undertaking a PhD. And my work is mainly focused on exploring how we can translate this potential new biomarker, ERNA, or evoked resonant neural activity, and hopefully try and improve DBS programming and potentially as lead placement.

Dr. Mitra Afshari: That's wonderful. So Dr. Nagao, there's a lot of talk in the neuromodulation world about beta oscillations. That's the 13 to 30 hertz oscillations. They also call it beta band local field potentials, beta power. And so beta power is a neurophysiological marker of bradykinesia and underlying pathology in Parkinson's disease.

[00:02:00] And actually, beta power is the basis of our current commercially available closed-loop DBS system, where we can both sense in real time when there's dysfunction and then stimulate in real time only when that dysfunction occurs. But in addition to that real-time closed-loop stimulation, what these neurophysiological markers that you're studying can also help us with is streamlining the DBS process like you were alluding to really help programmers determine where might be best to stimulate patients, since there really are many options along the lead. At least in the United States, we're just getting more comfortable using leads with up to 16 contacts that also have directionality. And as you mentioned in the paper, traditional clinician-based programming is largely heuristic. So wouldn't it be nice to have something a little bit more objective?

My first question to you, Dr. Nagao, before we really dive deeper [00:03:00] into the manuscript that you've published is, can you tell us just a little bit more about ERNA and what advantages that it might provide as compared to beta power as a neurophysiological marker for neuromodulation in PD?

Dr. Kanae J. Nagao: Sure. So ERNA is a large amplitude neural signal that is evoked by applying stimulation in certain areas of the brain, such as the STN and the GPi. And it is a oscillatory waveform, so it looks sinusoidal, and if you apply a burst and stop stimulating, it decays over time. And there are several groups looking at the neurobiological mechanisms of ERNA.

But the understanding is that perhaps it reflects engagement of the networks that may be involved in Parkinson's disease and other conditions like dystonia. So other groups using computational modeling and rat models have proposed that perhaps it reflects reciprocal activity between the STN and [00:04:00] GPe.

And other groups have found that perhaps it's more activation of the STN by the cortex through the hyperdirect pathway. And there has been a lot of promising work prior to this that potentially shows that ERNA could be a great biomarker for programming. So ERNA amplitude appears to be largest in the dorsal STN.

And when using non-directional leads, there have been studies that have shown a correlation between where ERNA amplitude is largest and it correlates well with where the clinician has selected as the optimal programming contact. Additionally, the contact where ERNA amplitude was largest also was the same as the contact that had the highest clinical benefit for a patient clinically. But these were not prospective studies per se. But we do know that ERNA localizes to the dorsolateral and has a positive correlation with beta power. And I guess where it [00:05:00] may have an advantage over local field potentials is how large the amplitude So it's several hundred microvolts, and that means the signal-to-noise ratio is a lot better.

And even though it is attenuated by, for example, general anesthesia, because of its large amplitude, it's a little bit more robust to its effects, so we can still record it quite easily. And what we found that the spatial distribution of ERNA across a lead, so their relative amplitudes at each electrode remain preserved under general anesthesia.

So that's really helpful. And I guess the positive is that it can be rapidly recorded intraoperatively, which is very important when we want to try and minimize the impact on the surgical workflow.

Dr. Mitra Afshari: Absolutely. So what I'm hearing as a summary is that ERNA is evoked by stimulation, and ERNA may be especially relevant in asleep DBS surgeries where general anesthesia is being used, and, these are the types of surgeries that [00:06:00] patients prefer, right? Logically. and I, think you alluded to this, in your explanation, was that in 2022, your group published work, where you recorded both beta oscillations and ERNA intraoperatively, and you looked into which of the three, beta versus ERNA versus anatomy, would best predict the contacts ultimately chosen for chronic stimulation in 47 patients with STN DBS with PD.

And as you mentioned, this specific study was a retrospective study. And of those three, you found that ERNA came in at 80%, anatomy came in at 67%, and beta came in at 50%. So it seems like, this earlier finding is what kind of fueled your next steps and the study that we're discussing today.

Dr. Kanae J. Nagao: Yes, that's correct. And though that study was in a non-directional lead cohort, and now as [00:07:00] we have these amazing directional leads that have multiple independent current control. So we have this ability to fractionate current with a lot more spatial resolution. And I think this is where ERNA has a role because of its large amplitude, we can potentially provide that spatial resolution of how to fractionate current. That would be a lot easier than with local field potentials where they're very small. And so I guess the point is we're not trying to say ERNA is better or is going to replace local field potentials. We're trying to look at adding a tool for DBS programming that could be potentially useful in a different aspect especially trying to improve the efficiency of that initial DBS programming period, which is really quite arduous for patients.

Dr. Mitra Afshari: Absolutely. That really helps put things in perspective. So in this study that we're discussing today, Dr. Nagao, the objective was to first see if you could use intraoperative ERNA to generate a programming [00:08:00] configuration for STN DBS, and then second, to see how, that configuration compares to traditional clinical programming based on-- that experts do in the clinic and image-guided programming.

And so I'm just gonna summarize what I understand as the methods of your study, and, obviously correct me if I'm wrong or if there's a detail that I missed. So the study was performed in 12 STN DBS PD patients. These are patients recruited across three centers in Melbourne.

All the patients were implanted by the same neurosurgeon. The surgeries were performed under a standard workflow of stereotactic frame-based and MER-guided surgery. Some surgeries were performed under general anesthesia, some were performed awake. The eight-contact Boston Scientific Cartesia leads were used where the two middle tiers are segmented leads and the system operates on multiple [00:09:00] independent current control, that's MICC, which allows for the fractionation that you were talking about.

The dorsolateral STN was targeted, and when you systematically analyzed whether the leads were well-placed by essentially looking how far they were from the individual sweet spots, they were well-placed within one point two millimeters from the estimated sweet spots. The image-guided programming was done via Guide XT, which utilizes BrainLab, and you looked at programming outcomes from three different programming paradigms from one side, the most affected hemibody, about four to six months out from DBS implantation.

Is that all correct?

Dr. Kanae J. Nagao: That's all correct, yes. And I think the key is that these patients were, as you said, well-placed. So not just anatomically, but when we looked at their four-month postoperative outcome, their off-medication on stimulation scores [00:10:00] almost matched their preoperative on-medication score. So they had done well.

So there is that to take into consideration

Dr. Mitra Afshari: That's a good point. So I know what the results are with respect to the motor outcomes comparing the three different groups. But I'd love to hear, and I think our audience would love to hear the results of the study in your own words as the lead researcher

Dr. Kanae J. Nagao: Sure. So what we did was these patients had been programmed as per standard of care by the clinician. So we had their optimized clinician program to compare to, and as you said, had the GuideXT generated or GuideXT-guided program to compare to. But these people underwent an acute assessment four to six months post-op.

So on one day, they had these three programs compared, so it was quite an arduous four to five-hour session. They had a stimulation washout. Their programs were then applied in counterbalanced [00:11:00] order with a washout in between, so to minimize any confounding from order effects. And what we were interested in was comparing their on DBS MDS-UPDRS III hemi body scores for each program compared to their OFF stimulation scores on that day, and then looked at that percent improvement as well. And very pleasingly, we did find that first of all, all three programs performed better on stim compared to OFF stim, which is always reassuring. But that the ERNA program on stimulation scores was similar to clinician and imaging, and that percent improvement which was about seventy-six percent, was comparable to the clinician program, which was sixty-nine percent, and imaging, which was eighty-two percent.

And we were also wanting to account for potentially how different their preoperative levodopa response was because that's important. And so we calculated a motor response ratio, which basically compares that percent MDS-UPDRS [00:12:00] III improvement to their preoperative levodopa response. Meaning that if they had a ratio of one or above, they basically matched that levodopa response or exceeded it, so meaning a good outcome. And again, we found that ERNA performed well, so their motor response ratio was one, and they didn't differ from i-imaging guide, which was one point one, and clinician, which was point nine. Importantly, we wanted to make sure that it was a tolerable program, and their side effect thresholds were, again, very similar. And their therapeutic plateaus were also similar. Now, that therapeutic plateau is a bit different to what some people would understand as a therapeutic threshold. So a therapeutic threshold is often the minimum current where people start getting a significant clinical benefit. But this therapeutic threshold is more the current where the patient had their nadir or their best MDS-UPDRS III score and were stable at that score.

So The numbers do look a little bit higher than what you'd expect for a typical [00:13:00] therapeutic threshold. But the point is that these were very similar.

Dr. Mitra Afshari: And so in summary, you found that ERNA is a good predictive paradigm for programming.

Dr. Kanae J. Nagao: Yes. We also wanted to then compare, is there a difference systematically of where we're programming using ERNA compared to the clinician and the imaging? And so we've represented each of these programs as like a weighted midpoint of where the current was fractionated.

And interestingly, we found that the ERNA program seemed to stimulate about one millimeter more ventrally to the imaging-guided and about 0.8 millimeter more ventrally to the clinician-guided program.

Dr. Mitra Afshari: I did read that and I was very curious about that. So the ERNA on average was about one millimeter ventral and about 0.3 millimeters posterior. Is that correct? The clinical programming paradigm.

Dr. Kanae J. Nagao: Yeah, sorry. I misspoke [00:14:00] before. It was one millimeter more than the clinical.

Dr. Mitra Afshari: No worries. And and you found that as a commonality. That's really interesting. So there might be something to this. And what do you think about that? What is your hypothesis there? 

Dr. Kanae J. Nagao: I think it's a really interesting finding in that, it raises the possibility that perhaps ERNA is localizing to a distinct location to what we traditionally think as the clinical hotspot. And, is that because of it proximity to certain trajectories or certain fibers? It's unclear.

We didn't do any DTI mapping to explore that further, but that would be an interesting avenue to explore. Certainly, it has been raised by other groups. So for example, one defined an ERNA hotspot, and that was a little bit more ventral and closer to that border between the association and the motor subdomains of the STN.

So it does raise an interesting point. But I guess we are limited in that we can't completely discount it's a way of [00:15:00] recording ERNA that we've found this potential ventral bias. So we need to explore whether there is confounds from our recording protocol that potentially raise this.

Dr. Mitra Afshari: Understood. It's really fascinating what you mentioned about maybe potentially incorporating DTI. I could definitely see that, as a next step. In the paper, Dr. Nagao, you have a figure of, an example current fractionation using ERNA for one of your participants. And what I see in this figure is that there's a significant amount of fractionation.

So just to summarize, there's 22% on one of the contacts, one of the segmented contacts. A neighboring contact has 8%, and then right above it, there's 52% on a contact and the neighboring segmented contact has 18%. We don't have a table of all of the final stimulation amplitudes for the 12 patients in each of the three programming paradigms, but you do [00:16:00] mention that directional stimulation was only chosen for one of 12 patients when, they performed clinical programming, which means that almost all of the clinical programming settings for the patients provided omnidirectional stimulation versus for the ERNA-programmed settings nine of those patients were directional, and for the image-guided programming settings, nine of those patients were directional.

So this was actually really interesting to me. So my question would be, if there's no difference in the motoric outcome of stimulation between these three programming paradigms, is it possible that, stimulating at such a high resolution with complex fractionation is not necessarily important, and what is important is that you're just stimulating in a nice dorsolateral STN [00:17:00] and perhaps that these results could be a reflection of very well-implanted leads?

What do you think about that?

Dr. Kanae J. Nagao: I think that's a good point and it does go back to the fact that with our population, we did have well-placed leads. And while that's great for our population, it's also a limitation of how generalizable these results may be. And to your point about the directionality and how significant and what is that resolution of directionality that's important, I don't think that is known yet.

Certainly with-- even with the imaging-guided new software like Illumina that provide crazy fractionation and combinations, how much of a clinical difference does that make? We're still not sure. Of interest though, when we did look at our results, the imaging-guided imagingguided software did perform better than the clinician programming despite them being in a relatively similar location of where they were stimulating.[00:18:00] 

So perhaps that raises the possibility that because the main difference between these were the directionality. It raises the possibility that it is still significant even within a well-placed population. But certainly, we have a very small population size. We have well-placed leads.

We do need to expand our research into larger populations where there's more of an offset to the ideal stimulation location and see whether the directionality confers a degree of benefit that we're not seeing here. But I guess the other limitation is that we did one hemibody and assessed the improvement in their worst affected hemibody.

So there are limitations with that as well.

Dr. Mitra Afshari: And so there's a lot of future studies that can be done here, right? So expanding your cohort, doing both sides, looking at chronic stimulation, right? And whether ERNA changes over time potentially. So I could see all of those being [00:19:00] done. Another important finding that you talked about earlier was, the percentage improvement in motor UPDRS didn't differ between patients who were programmed using ERNA configurations, whether the ERNA was captured in the awake state versus whether it was captured under general anesthesia.

 And so in other words, the ERNA predictions were accurate whether captured awake or asleep. And what are the implications of that? How does that perhaps differ with beta power, if you want to reiterate kind of the points you made earlier in our discussion?

Dr. Kanae J. Nagao: I think this is one of the key advantages of ERNA, is that how robust it is under general anesthesia. Different centers utilize different surgical methods, but we want this to be available to a wider population. We want good outcomes to be possible for the entire DBS population. We don't want it to be restricted to patients that are awake.

So the fact that being under general anesthesia doesn't influence [00:20:00] this is a really great potential advantage of ERNA.

Dr. Mitra Afshari: Absolutely. I think that's a really good point, is that the diversity of the way people are implanted these days is huge. And so this could be, a potential of using ERNA across multiple centers. And then could you explain to us, Dr. Nagao, and this will be my final question. I know I've posed some pretty tough questions to you. What are the potential practical implications of the fact that the ERNA recording in the OR required only 45 seconds per lead? I think you mentioned earlier that is also a nice practical advantage of ERNA as well.

Dr. Kanae J. Nagao: Yeah. So it just speaks to how ERNA could be quite seamlessly integrated into the surgical workflow. Just to clarify, it is recorded from the DBS lead itself, so it's not that we need to add another electrode to record it. So it has a minimal impact on surgical risk and surgical time, and [00:21:00] it can be then interpreted after the surgery.

And as we said, it's quite objective with just looking at the amplitude of the signal. So I think this is another significant advantage because not only have to know how good ERNA is as a, as a biomarker, but it has to be practical to be clinically useful.

Dr. Mitra Afshari: Wonderful. So I wanna thank you, Dr. Nagao. This was really a rich discussion. These were pretty complex topics. It's a bit hard to convey over a podcast, but you did such a great job breaking everything down, and I wanna thank you for the work that you're doing and your group is doing to ultimately improve DBS for all of our patients, and really to, streamline these processes for our providers as well. So I'm really looking forward to all of your future work, with your amazing group. So thank you for taking the time to be with us on this podcast today, and goodbye.

Dr. Kanae J. Nagao: Thank you. [00:22:00] [00:23:00]