This scientific commentary refers to ‘Stimulating at the right time: phase-specific deep brain stimulation’, by Cagnan et al. (doi:).
Immediate symptom control at the push of a button is the great dream of clinicians and patients alike. Anyone who has witnessed the striking moment when a tremor patient’s uncontrollable shaking breaks up and eventually disappears altogether with the onset of deep brain stimulation (DBS) will testify that there are few things more remarkable in neurology. Its reversible and adjustable character has contributed to an exponential growth in the use of DBS devices and an increased acceptance by clinicians. It is estimated that more than 100 000 patients suffering from Parkinson’s disease, pathological tremor and dystonia have been treated with DBS all over the world. Owing to this success, the key principle underlying DBS has not changed in the past 30 years. Contemporary DBS technology operates in an open-loop fashion in which electrical impulses are permanently delivered to subcortical target structures at high frequencies (typically > 100 Hz) via implanted depth electrodes. Currently, a DBS physician optimizes stimulation parameters in a time-consuming trial-and-error process until maximum therapy is achieved with minimal side effects. Consequently, there is a push to develop smarter DBS strategies and to go from open- to closed-loop control. In a closed-loop paradigm, a recorded biomarker signal is utilized to continuously adjust stimulation in a demand-controlled manner (adaptive DBS). The potential advantages of adaptive DBS technology have a great appeal to clinicians and patients alike: therapeutic efficacy may improve, side effects might be diminished and battery life would be prolonged to name the most important ones. Hitherto, however, only a few experimental studies have been performed on that subject. In this issue of Brain, Cagnan et al. pioneer a novel DBS approach that is based on real-time monitoring of neural-circuit pathophysiology in patients with pathological tremor. They show that effective tremor control can be achieved by utilizing the patient’s peripheral oscillations as a biomarker signal to trigger brief bursts of thalamic stimulation in a phase-specific manner (Cagnan et al., 2017).
Amplitude, frequency and phase are the fundamental properties of oscillatory neural signals. Dynamic phase coupling of rhythmic neural activities occurs at different spatial and temporal scales throughout the CNS and is considered a key principle of physiological information processing (Engel et al., 2013). In contrast, increased oscillatory activities in conjunction with excessive phase synchronization characterize network disorders with reduced information coding capacity. Disrupted motor output through abnormally synchronized network oscillations is an attractive hypothesis providing a link between the pathophysiology of tremor syndromes, Parkinson’s disease and dystonia (Schnitzler and Gross, 2005).
Pathological tremor occurs in all three conditions and can therefore be considered a prime example for an oscillopathy of the motor system. Pathological tremor is currently thought to represent a large-scale motor network disorder of the CNS involving massively synchronized neural oscillations along a highly connected set of regions such as brainstem, cerebellum, thalamus and cortex (Helmich et al., 2013). Oscillatory motor unit synchronization eventually results in large amplitude limb excursions, which disturb motor output and impose social stigma.
Initial, albeit indirect, evidence for the usefulness of a phase-based closed-loop DBS approach came from the landmark study of Rosin et al. (2011) in a non-human primate model of Parkinson’s disease. They showed that pallidal DBS, when triggered by and time-shifted to oscillatory cortical unit activity, was more effective than standard open-loop DBS. Proof-of-concept for phase-based stimulation in humans was then provided by the group of Peter Brown with a non-invasive stimulation approach. The authors were able to show that resting tremor in patients with Parkinson’s disease could be suppressed when transcranial alternating current stimulation was applied to motor cortex in a phase-specific manner (Brittain et al., 2013). Hitherto, the feasibility of phase-dependent stimulation using components of established implantable neurostimulation technology has not been demonstrated.
Simulation of phase-specific thalamic stimulation as in Cagnan et al. Intraoperative recordings from the motor thalamus (Vim) of a patient with essential tremor made during microelectrode-guided mapping of the surgical target. In this illustrative example, phase relationships at different levels of the oscillating tremor circuit are constant during tremor occurrence (left). Thus, phase of the peripheral tremor corresponds to a fixed central phase. Green: simulated phasic burst stimulation on ascending tremor-phase derived from accelerometer as in Cagnan et al. In this example, stimulation would hit the two simultaneously registered tremor cells after (neuron ‘A’) or during (neuron ‘B’) tremor-related burst discharges. In the absence of tremor (right), thalamic neurons are desynchronized and feedback-controlled stimulation would be switched off. Note that the superimposed green stimulation patterns are purely illustrative.
The present study of Cagnan et al. was designed to capitalize on and maximize these effects by precisely tuning thalamic stimulation to the optimal phase. Accordingly, the authors first determined which phase was most likely to deplete tremor in an individual patient. To this end, peripheral tremor phase was tracked in real time and then used as a trigger signal for the neurostimulator over a range of phases in a randomized fashion. Importantly, the authors boosted their stimulation effects by applying brief high-frequency bursts consisting of four to six pulses instead of delivering a single stimulation pulse per oscillation cycle (Fig. 1). In agreement with their previous observations, the phase in which thalamic DBS pulses hit the ongoing tremor oscillation was indeed critical to eliciting tremor suppression. The authors then went on to show that the tremor-suppressive effect of phasic burst stimulation, when time-locked to the optimal phase over sustained periods of time, was clinically significant in selected patients with essential tremor. Importantly, phase-specific tremor attenuation was twice as energy-efficient in these patients, when compared to conventional continuous high-frequency DBS.
Apart from its putative clinical significance, the study of Cagnan et al. is also a source of inspiration to those interested in tremor physiology and pathophysiology. In particular the demonstration of tremor-circuit responses to phase-selective perturbation—other than mere suppression—has great appeal. According to the results of Cagnan et al., phase can be used not only to attenuate, but also to enhance tremor. Augmentation of physiological neural oscillations through phase-specific DBS could thus become possible in the future. This could be used to enhance healthy brain function. Further, a phase-based DBS approach may help to increase specificity of stimulation and to ameliorate clinical conditions where phase synchronization in functional circuits is disrupted due to neurodegenerative processes. An obvious example is fornix DBS for Alzheimer’s disease (Lozano et al., 2016). Cagnan et al. also show that phase-selective stimulation could either accelerate or decelerate instantaneous tremor frequency. Only those patients whose tremor showed least variability in frequency exhibited this phenomenon and responded well to phase-specific stimulation. Resonant characteristics of the tremor circuit may thus determine whether a patient responds to a phase-selective intervention. Besides these effects on amplitude, frequency and regularity of tremor, perhaps the most important physiological implication of Cagnan et al. is that phase-specific DBS may be used as a tool to dissect individual tremor components. This function will provide important insights into the spatio-temporal dynamics of multiple juxtapositioned oscillating subcircuits, as in the sensorimotor thalamus of tremor patients. In conjunction with the newly developed field steering technique using directional electrodes, phase-specific DBS may not only account for temporal changes but also for the properties of the tracked oscillation in 3D space (i.e. adaptive field steering).
Finally, an interesting aspect of the study by Cagnan et al. is that phase-specific DBS worked well in selected patients with essential tremor, but was generally less effective in patients with dystonic tremor despite comparable phase tracking efficacy. Although the sample size was small, this observation deserves further inquiry and may eventually help us to understand the pathophysiological differences of isolated versus combined tremor syndromes.
It should be noted that physiological effects of thalamic low-frequency stimulation were already studied in the 1950s and 1960s during surgeries where thalamic lesions were placed to alleviate tremor. At that time, diagnostic electrical stimulation was applied on a routine basis during stereotactic interventions at a range of frequencies. It was noted that stimulation with high frequencies would mimic the therapeutic lesion effect (Ganglberger and Precht, 1964), which later became the basis for DBS. It was also discovered that thalamic stimulation at low frequencies could have manifold effects on frequency, amplitude and regularity of tremor. Already then, the phase at which thalamic stimulation hit the tremor cycle was discussed as an important parameter. Much like in the previous study of Cagnan et al. (2013), drift stimulation was applied as an experimental tool with individual stimulation pulses shifting through different phases of the tremor cycle (Ganglberger and Precht, 1964). However, most of these results were qualitative and inconclusive in nature, supporting the proposition that tremor may be easy to measure but difficult to understand, which is still valid today. Experimental utilization of the new phase-selective stimulation technique pioneered by Cagnan et al. will provide excellent opportunities to challenge this view.
Demand-controlled stimulation at the optimal phase may also be able to suppress pathological neural oscillations other than tremor. As an example, there is ample evidence that excessively synchronized beta-band (15–30 Hz) oscillations in basal ganglia-thalamo-cortical networks play a role in the pathophysiology of rigidity and akinesia in patients with Parkinson’s disease (Sharott et al., 2014). Previous work has shown that periods of high-frequency adaptive stimulation, triggered on elevated levels of beta power picked up from DBS electrodes, significantly improve akinetic-rigid motor disability (Little et al., 2013). An obvious question is whether DBS triggered on critical phases of the beta cycle may be equally or even more effective. The use of beta oscillations as a central biomarker for a phase-based approach, however, is challenging. An accurate estimation of instantaneous beta phase in real time may be difficult because waxing and waning of beta oscillations occurs on a much faster time scale compared to tremor and the tracked signal may be distorted by stimulation artefacts. To further complicate the matter, tremor-related and beta-band oscillatory activities are co-expressed at the level of the DBS target structure.
After three decades at a relative standstill, there are currently encouraging signs of innovation in the field of DBS technology. New pacemakers and electrode designs have been developed that enable less constrained programming of stimulation parameters and widen therapeutic windows. Translation of these advancements into clinical practice is of the utmost importance and will set the bar high for the development of closed-loop DBS technology. However, we believe that adaptive DBS has great potential to extend the therapeutic arsenal of DBS neurologists even further. Given the increasing complexity of devices, it will be essential to keep DBS optimization simple and sustainable. Autonomous DBS adjustment based on patient pathophysiology could provide a solution. Studies such as the one by Cagnan et al. are important first steps towards this vision, and will undoubtedly yield further important insights into motor system physiology and the pathophysiology of movement disorders.
Funding
The authors acknowledge support by the Deutsche Forschungsgemeinschaft (SFB 936, projects A3 and C8).
