Deep Brain Stimulation (Models, Theory, Techniques): Overview
Deep brain stimulation (DBS) of the subthalamic nucleus (STN) is a well-established treatment for medically refractory patients with advanced Parkinson’s disease (PD) (Benabid et al. 1991; Blond et al. 1992; Benabid et al. 2002; Deuschl et al. 2006) as well as for patients with early motor complications (Deuschl et al. 2006; Schuepbach et al. 2013). Several neurological diseases, such as Parkinson’s disease (PD) or essential tremor, are characterized by pathological synchronization (Nini et al. 1995; Brown et al. 2001). Parkinsonian resting tremor, for example, seems to origin from a pacemaker-like population of neurons of the basal ganglia firing in a synchronized and oscillatory manner (Hutchison et al. 1997; Hurtado et al. 1999; Magill et al 2001; Trottenberg et al. 2007). In contrast, under healthy conditions these neurons are active in an uncorrelated and desynchronized manner (Nini et al. 1995; Magill et al. 2001).
The standard DBS protocol employs permanent high-frequency (>100 Hz) pulse train stimulation (Benabid et al. 1991; Blond et al. 1992; Benabid et al. 2002). Symptom suppression by DBS is strongly dependent on stimulation frequency – with only high frequencies (>100 Hz) being effective and effects being rapidly reversible (Birdno and Grill 2008). High-frequency DBS was developed empirically, mainly based on clinical observations and experimental results (Volkmann et al. 2006), and the mechanism of high-frequency DBS is still a matter of debate (Benabid et al. 2005).
Experimental observations indicate that during high-frequency DBS, a regular bursting mode is induced (Beurrier et al. 2002), and after a reduction of stimulation artifacts, robust bursting activity in STN neurons was observed in slice experiments (Beurrier et al. 2001). In the same experiments, the offset of stimulation was followed by a blockade of activity, i.e., a depolarization blockade (Beurrier et al. 2001). These observations were made in anesthetized animals and are contradicted by measurements in awake behaving primates (Anderson et al. 2003; Hashimoto et al. 2003; Dorval et al. 2008) and rats (McConnell et al. 2012). Other groups argue that high-frequency DBS blocks neuronal activity in relevant target areas during stimulation and therefore mimics the effect of tissue lesioning (Benabid et al. 2002). In 2005, Benabid and coworkers summarized different hypothetical mechanisms: membrane inhibition, excitation of excitatory and inhibitory afferents, jamming, excitation of efferents, and plasticity (Benabid et al. 2005). Novel experimental techniques, such as optogenetics, enabled to further reveal the mechanism of DBS and, in particular, of the stimulation of afferent axons projecting to the target region (Gradinaru et al. 2009).
Spatially extended single- and multi-compartment neuron models were used to evaluate the contribution of these different mechanisms (Grill & McIntyre 2001; Terman et al. 2002; Rubin and Terman 2004). For example, Grill and McIntyre (2001) showed that depending on the stimulation amplitude and the shape of the stimulation pulses, cells were either activated directly or fibers mediating excitatory or strong inhibitory action were activated. The activation of a larger number of structures takes place on the single-neuron level with different and possibly conflicting impacts on single-neuron dynamics (Grill and McIntyre 2001). For example, in the same neuron, the cell body (soma) is inhibited as a result of activation of presynaptic axons and GABA release, while the efferent axon is activated by the stimulation pulses on an approximately one for one basis (McIntyre et al. 2004). The various reactions of neurons toward stimulation on the network level further add complexity that is important for the creation of a sound model: cells responding differently to external inputs, such as somatosensory stimulation or stimulation owing to active movements, are present in the target tissue together with so-called no-response cells (Lenz et al. 1994). Therefore, high-frequency stimulation has a complex impact on these structures (Benabid et al. 2002; Shen et al. 2003). However, surprisingly, even single STN model neurons – lacking synaptic dynamics, neural circuitry, and contributions of glial cells – subjected to high-frequency stimulation reproduce clinically observed response characteristics (Pyragas et al. 2013).
To study another aspect of DBS, several groups use physical models based on Maxwell’s equations to investigate the neuronal activation profile depending on electrode geometry and stimulation parameters (Butson and McIntyre 2005, 2006; Miocinovic et al. 2009; Yousif et al. 2008; Chaturvedi et al. 2010; Buhlmann et al. 2011).
While the standard DBS protocol was developed empirically (Volkmann et al. 2006), novel stimulation approaches are based on electrophysiological as well as computational concepts. Personalizing and optimizing high-frequency stimulation in real time by demand-controlled, adaptive DBS might constitute a superior high-frequency stimulation mode, as shown in an acute study in externalized patients (Little et al. 2013). In addition, modeling studies are used to further develop the stimulation algorithm, beyond standard high-frequency DBS, in order to finally establish superior stimulation mechanisms. For example, coordinated reset (CR), a patterned stimulation protocol specifically targeting the reduction of synchronized activity, was developed by means of mathematical models (Tass 2003) and essentially aims at an unlearning of both abnormal synaptic connectivity and synchrony (Tass and Majtanik 2006). CR was successfully tested in a preclinical study, where 5 days of low-dose CR stimulation induced long-lasting therapeutic effects for 30 days (Tass et al. 2012). Another example is a closed-loop approach, which was controlled by the extent of oscillatory beta-band activity. During stimulus delivery, this approach resulted in a better reduction of akinesia as well as pallidal firing rates as compared to classical DBS in parkinsonian nonhuman MPTP-treated primates (Rosin et al. 2011). Finally, modifications of the standard HF protocol might offer a novel approach to improve the efficacy of deep brain stimulation (Brocker et al. 2013; Hess et al. 2013).
The combination of all these modeling and experimental and technological approaches plays a vital role in shaping our understanding and helps to improve the promising therapeutic intervention DBS.