Adaptive Deep Brain Stimulation for Movement Disorders

Currently, we typically stimulate patients with Parkinson’s disease (PD) with 24/7 stimulation parameters, and what we know is that the symptoms of PD patients fluctuate over the day. So you often see that the tremor sometimes is less severe and sometimes patients feel “off”, meaning that their motor symptoms are more severe. We know that there are neurophysiological physiomarkers signals, especially neural oscillations in the beta range, from 13 to 30 Hertz approximately. And we see that there’s a significant correlation between the amplitude of these oscillations in the local field potentials, that’s the potential surrounding the DBS electrodes, and the severity of contralateral symptoms. So we think that by recording those local field potentials and quantifying them, we can better adjust the stimulation based on the needs of patients. We know that PD patients still use their medication after the DBS surgery, typically around 50%, but we have also shown that the beta oscillations decrease when patients take their medication. So we think we can make the medication and the stimulation work in synergy when we adapt the amount of DBS to the actual states of the patients. That’s one advantage because we know that we can stimulate too little, but we can also stimulate too much – and if you stimulate too much, you can get dyskinesias or other stimulation-induced side effects.

There’s one aspect that controls the stimulation when patients are treated, based on the best stimulation parameters. Another approach is that we can use neurophysiological signals but also more advanced neuro and anatomical features to optimise the way of stimulation. So I think currently it’s a “trial and error” and we use a combination of different contacts different stimulation parameters. We can soon rely more on tensor imaging to see the connectome of the target structure and more selectively influence fibre bundles that project to critical structures. On the other hand, by more rigorously measuring the neural activity surrounding all the electrodes, we can also find neurophysiological sweet spots, which will help find the optimal stimulation targets.

I think there are still plenty of disadvantages, which is one of the reasons why not every patient is getting adaptive DBS yet. In our study, we used a very basic control algorithm that is a patient’s specific power spectral density in a single frequency range, and for this study, we used beta oscillations. One of the important findings of this study, but also in earlier work on correlations between symptoms and the neurophysiological signals, is that we do see that bradykinesia and rigidity can improve with applying adaptive DBS. That’s not the case for tremor. If you have tremor-dominant PD patients, it seems that applying adaptive DBS based on beta oscillations is not the best feedback algorithm. Having said that, we’re still beginning to explore the neurophysiological signatures, so we might be able to find a more tremor-specific physiomarker or combination between neurophysiological signals or other signals. So that’s a clear limitation at this moment.

There’s a need for more clever algorithms; on the other hand, we can’t even manage to suppress bradykinesia and rigidity in a setting outside the hospital. So we could proceed in small steps in non-tremor-dominant patients and look on whether we can make sure that aDBS is of any help for them. One tricky thing about the accelerometer is that it detects tremor, but it has a certain delay. So when you see that the tremor emerges in, for example, essential tremor, you can be too late. In PD it’s often resting tremor, so that’s less of a problem. However, we can think of combining signals to find the optimal stimulation setting.

This is the highlight of the interview. If you like to explore more, please visit our YouTube Channel.