Deep Brain Stimulation for Epilepsy

Introduction

Epilepsy is a chronic seizure condition, classically characterized as “a continuing tendency to seizure relapse” ( ). It was estimated to affect 50 million people worldwide in 2012. As a chronic condition, epilepsy confers significant mortality, including suicide and SUDEP (sudden unexpected death in epilepsy patients) ( ), and morbidity, including cognitive, psychological, and social impairment, risk of injury, and socioeconomic consequences ( ). While antiepileptic drugs have been transformative for a majority of sufferers, around 20% of patients continue to have poor seizure control despite optimal medical therapy ( ). Some of these patients with medically refractory epilepsy are candidates for resective surgery and can achieve up to 66% long-term seizure freedom ( ). However, many are ineligible due to comorbidities, or the number and location of epileptic foci. Further, the risk of permanent memory impairment is significant with resection of the dominant mesial temporal structures ( ).

It is clear from decades of research that seizures do not simply originate from one location (e.g., temporal lobe) and spread haphazardly throughout the cortex, but rather propagate and are sustained via distinct cortical–subcortical networks ( ). Since the 1980s attempts have been made to modulate these networks to disrupt or prevent seizures in medically intractable epilepsy using deep brain stimulation (DBS), with some success ( ). Although the “mechanism of action” of DBS is still hotly debated ( ), it is clear that DBS disrupts neuronal networks (whether by activation, inhibition, or lesion), and that this is likely the source of therapeutic benefit ( ).

Numerous DBS targets have been proposed and studied in the treatment of refractory epilepsy, including the anterior nucleus of the thalamus (ANT), the hippocampus, the centromedian nucleus of the thalamus, the subthalamic nucleus, the caudate nucleus, the hypothalamus, the globus pallidus, the nucleus accumbens, the mamillothalamic tract, the zona incerta, the ventralis oralis posterior nucleus of the thalamus, and the cerebellum ( ). Currently the targets with the most promising clinical evidence are the ANT and hippocampus. Both targets are part of a wider limbic network with extensive reciprocal cortical connections. Studies of both have focused on focal seizures with or without secondary generalization, and DBS may be most effective in cases where seizure origination has propagated along limbic pathways. The success of the first randomized controlled trial of DBS of the ANT, the SANTE trial, the recent publication of its long-term follow-up data ( ), the favorable safety profile of DBS when compared to surgical resection ( ), the ability to adjust stimulation parameters, and its reversibility have all moved DBS therapy from the experimental category to a promising treatment option for many patients.

In this chapter we briefly review relevant seizure circuitry, and then focus on evidence for effective DBS of the ANT and the hippocampus for medically refractory epilepsy. We also discuss the problem of anatomical localization both preoperatively and postoperatively in light of several recent studies highlighting the importance of precise localization in the ANT for favorable outcomes. We finally review the safety evidence for DBS.

Limbic Network as Conduit of Seizure Propagation

Limbic circuitry, originally proposed by Papez in as the circuit of emotional processing, may be important to seizure propagation. Papez’s goal was to provide an anatomic basis of the theory that emotion resulted from the interaction between the cortex and these subcortical structures (a network now thought to be involved primarily in memory encoding). Much of the cerebral cortex projects, via the cingulum, to the parahippocampal gyrus, which projects to the entorhinal cortex. The entorhinal cortex in turn projects into the hippocampus via the perforant pathway. The hippocampus projects outward via the fimbria and fornix to the mammillary bodies, which in turn project to both the hypothalamus and the ANT (via the mamillothalamic tract). From here the ANT completes the cortico–limbic loop by projecting to the cingulate cortex. Thus Papez described a network that funnels cortical fibers into the hippocampus on one end and subsequently routes pathway-connected fibers in the thalamus back to the cortex on the other.

Much animal and human evidence exists that the hippocampus and subcortical structures, as well as their interaction with the cortex, are key to seizure propagation ( ). As one example, Gale describes a consistent network that is activated during limbic motor seizures in rodents (thought to be the corollary of partial complex seizures in humans) that includes (but is not limited to) the entorhinal cortex, hippocampus, amygdala, and thalamus. Importantly, these regions were consistently activated regardless of which area of the limbic region initiated seizures ( ). If seizures propagate along predictable pathways, then interruption of those pathways via DBS could potentially prevent the generation and propagation of seizures.

There is also evidence that seizures themselves may induce cortical–subcortical network changes, possibly making epilepsy more difficult to treat, and may have implications for patient selection and timing of intervention. Prolonged seizures in rats lead to progressive recruitment of subcortical structures ( ). In humans not only does the interictal extent of cortical metabolic derangement expand as seizure frequency increases in children with intractable epilepsy ( ), but interictal thalamic and hippocampus metabolic derangements worsen with duration of epilepsy, and are worse in patients with secondarily generalized seizures ( ). These results may simply reflect coactivation of nodes in a known cortical–subcortical network, but alternatively may hint at secondary epileptogenesis, a phenomenon described in animal models whereby seizure activity in one site eventually leads to independent seizure activity in a second remote site ( ). Clinical outcomes data on long-term prognosis for epilepsy could be consistent with this idea: the authors of a recent comprehensive review conclude that the longer an epilepsy is active the poorer is the long-term outcome, and further, the largest predictor of poor long-term outcome is pretreatment seizure frequency ( ). However, the extent to which secondary epileptogenesis exists in humans is unclear, and the further question of whether it is reversible is unknown. Nonetheless, it is interesting to note that several studies of DBS for epilepsy, including the recent SANTE trial, report a seizure reduction rate of at least 50% in some patients only after 3–6 years of stimulation ( ). While these results could be explained by evolving stimulation parameters or antiepileptic drug regimens, it is intriguing to speculate that DBS may exert a chronic plastic effect on epileptogenic circuitry, as is seen in DBS therapy for other conditions such as dystonia or obsessive–compulsive disorder ( ).

Deep Brain Stimulation of the Anterior Nucleus of the Thalamus