Comparison between AC and DC stimulation for the suppression of epileptiform activity - mechanism and clinical relevance
Low-amplitude DC electric fields are effective in suppressing epileptiform activity in several in vitro models. The suppression thresholds are far below the values required to produce excitation (Warren & Durand, 1998), i.e. trigger neuronal firing. The inhibition mechanism is well established as membrane hyperpolarization (Durand & Bikson, 2001). There are several drawbacks associated with DC stimulation: (1) even at low current, erosion and electrochemical damage occur during continuous DC stimuli due to unbalanced charge (Brummer & Turner, 1977); (2) termination of DC stimulation generates immediate ‘excitation rebound’ of epileptiform activity, (Ghai et al. 2000) and (3) the suppression efficiency is highly dependent on the relative orientation between the electric field and the somatic-dendrite axis (Fig. 7; Jefferys, 1981; Chan et al. 1988; Gluckman et al. 1996; Durand & Bikson 2001) Thus, in vivo, the intended neuronal target must be anatomically uniform for DC stimulation to be effective. Moreover, inhibition (by hyperpolarization) could be changed into excitation (by depolarization) if the implanted electrode was slightly displaced.
AC stimulation is also highly effective at suppressing neuronal activity in various models of epilepsy in vitro and has several advantages over DC stimulation including: (1) termination of AC stimulation is often associated with a prolonged disruption of epileptiform activity (compare Fig. 3A with 3B and Fig. 4A with 4B); and (2) AC stimulation is not orientation-dependent (Fig. 7).
DBS, which uses exclusively AC stimulation, has been used to suppress seizure activity in humans (Cooper et al. 1973; Faber & Vladyka, 1983; Velasco et al. 2000). Although AC stimulation protocols require higher currents than DC stimulation (see above), DBS has been proven to be safe and well tolerated in patients with epilepsy (Fisher et al. 1992). Paradoxically, the clinical effect of high frequency stimulation of a targeted structure mimics the effect of lesioning the structure, which suggests that high frequency stimulation is inhibiting neuronal firing (Benazzouz et al. 1995).
It is well established that high frequency (AC) stimulation of excitable tissue results in extracellular potassium accumulation (Gardner-Medwin, 1983a,b; Gardner-Medwin & Nicholson, 1983; Bawin et al. 1986; Poolos et al. 1987; Kaila et al. 1997). This increase can, in turn, depolarize neurons sufficiently to tonically inactivate Na+ channels such that action potentials cannot be initiated (Traub et al. 1991; Hille 1992). While small increases in potassium can promote epileptiform activity, large increases could thus suppress spontaneous bursting (Rutecki et al. 1985). Consistent with our previous report using uniform electric fields (Bikson et al. 2001) the present study shows that local suppression of epileptiform activity in vitro with monopolar AC stimulation is always associated with a large increase in extracellular potassium and ‘depolarization block’ of neurons. Moreover, post-stimulus suppression is linked to an undershoot of potassium below baseline levels.
Are potassium transients sufficient to induce depolarization block? Under normal conditions a tonic depolarization of ∼20 mV will suppress regenerative action potentials in most classes of neurons (Hille, 1992). Potassium-selective microelectrodes measure an average potassium concentration at the interface of the bath and slice surface, and so underestimate the potassium concentration at the surface of the cells during endogenous potassium release. Current clamp (Takahashi & Tsuruhara, 1987; Traynelis & Dingledine, 1988) and voltage clamp (Chvatal et al. 1999) recordings from glial cells, as well as modelling studies (Hounsgaard & Nicholson, 1983), approximate this underestimate at 25–75 %. Thus, in this study, the 2 mm increase measured by potassium-selective electrodes during suprathreshold stimulation could reflect a ∼6 mm rise at the surface of cells which would result in depolarization block. Moreover, pro-convulsant conditions can significantly reduce the level of depolarization required for action potential block (Hahn & Durand, 2001). Potassium levels in the brain during normal function, though not during seizures (Pumain et al. 1985; Jefferys, 1995), are lower than those used in the present report. Induction of depolarization block at ‘normal’ potassium levels would presumably require high stimulation intensities. Lastly, because in vitro the bath fluid can serve as a potassium sink, larger potassium rises might be expected in vivo (Krnjevic et al. 1980; Pumain et al. 1985; Jefferys, 1995).
Alternatively, other factors could contribute to depolarization block of neurons in response to high frequency stimulation. Stimulus-induced reductions in extracellular Ca2+ (Krnjevic et al. 1980) would depolarize neurons and shift the activation characteristics of ion channels to facilitate depolarization block at lower potentials. However, Ca2+ reductions are not sufficient in themselves to induce depolarization block, since epileptiform activity persists under reduced Ca2+ conditions. Stimulus-induced increases in intracellular K+ would depolarize neurons while large increases in intracellular Cl− could result in ‘depolarizing GABA’ (Bragin et al. 1999). The latter would promote further extracellular potassium release, though could not play a role during suppression of low-Ca2+ or picrotoxin activity. Further studies are required to determine the contribution of each of these factors in inducing depolarization block. There is no evidence suggesting these other factors are sufficient in themselves to induce depolarization block, however, they would all be expected to: (1) promote extracellular potassium accumulation and (2) reduce the level of extracellular potassium required to induce depolarization block.
Although extracellular potassium accumulation and intracellular recordings during DBS are currently not available, clinical high frequency stimulation would be expected to induce large extracellular potassium transients (Krnjevic et al. 1980) which, in turn, would have a dramatic effect on neuronal function. Our results suggest that DBS may suppress neuronal firing (mimicking the effect of a lesion), and hence the symptoms of epilepsy and Parkinson's disease, by a depolarization block mechanism mediated, at least in part, by a potassium rise.
Comparison between uniform and local stimulation for the block of epileptiform activity - clinical relevance
Stimulation using both large parallel (uniform field) electrodes and monopolar electrodes is effective in suppressing epileptiform activity. Uniform electric fields affect the entire tissue between the electrodes and, in vitro, do not have to be in direct contact with the targeted tissue, thereby minimizing the electrochemical damage (Durand & Bikson, 2001). While this method is convenient in a brain slice preparation, clinical implementation might be difficult because of the anatomical constraints of placing two large electrodes across a targeted structure. In addition, in vivo, the field electrodes would inevitably be in contact with nervous tissue. In contrast, monopolar electrodes are more selective (Fig. 5) and could be placed directly within a seizure focus.
Low-duty cycle stimulation and clinical relevance
In clinical practice, low-duty cycle stimulation protocols are used mainly to minimize tissue damage and improve battery efficiency (Agnew & McCreery, 1990b; Benabid et al. 2000). In the present study, low-duty cycle (1.68–50 %) AC fields were capable of completely suppressing epileptiform activity in all slices in which 100 % duty cycle was also effective. Moreover, lowering the duty cycle from 100 % to 50 % resulted in only a small increase in suppression threshold (far less than double) and a suppression protocol used clinically with 1.68 % duty cycle resulted in only ∼5-fold increase in threshold. Thus, the results of this study suggest that a lower-duty cycle stimulation protocol could effectively suppress activity with less charge delivered.
In conclusion, our results show that localized AC stimulation can suppress epileptiform activity in a localized region in vitro. The effective duty cycle and frequency are comparable with those used in clinical DBS (Loddenkemper et al. 2001). Compared with DC stimulation, AC stimulation was not orientation-dependent, was associated with extracellular potassium accumulation, and was more spatially selective. DC suppression is generated as a result of membrane hyperpolarization while AC suppression results from depolarization block. The results are consistent with the hypothesis that neuronal suppression by DBS is due to a depolarization block of neurons (partially) as a result of raised extracellular potassium concentration.