- Top of page
Summary: Purpose: To assess the safety, tolerability and efficacy of high-frequency periodic thalamic stimulation in inoperable mesial temporal epilepsy and the usefulness of intracranially evoked responses for assessment of anatomical uniformity of lead placement.
Methods: Four subjects were implanted with leads aimed at the anterior thalamic nuclei. Six weeks later, Soletra IPGs were activated using parameters similar to the closed-loop trial's (mean: 175 Hz; 4.1 V; 90 μs; 1 min ON, 5 min OFF). Efficacy was assessed by comparing percentage change in seizure frequency over a 6-month baseline versus a 36-month treatment period, using a within-subjects repeated measures design. Tolerability and safety were similarly monitored. Evoked responses elicited by thalamic stimulation were recorded from depth electrodes in the amygdalo-hippocampal regions and compared intra and interindividually.
Results: All subjects completed the study, tolerated stimulation and had no serious adverse effects. Mean reduction in seizure frequency was 75.6% (t =−8.24; p ≤ 0.01) (range: 92% to 53%). Quality of life improved in all. Verification of electrode placement with a software function indicated that stimulated structures were presumably, Anterior thalami, Latero-polaris, Reticulatus Polaris, Ventro-oralis Internus, and Campus Forelii. Evoked responses from stimulated sites were heterogeneous, intra and interindividually, also suggesting a lack of uniformity in lead placement.
Conclusions: High-frequency, periodic, round-the-clock thalamic stimulation seems safe, well tolerated and efficacious for inoperable mesial temporal epilepsy. Identification of clinically useful parameters may be facilitated by brief closed-loop trials. Selective stimulation of a single structure may not be feasible at certain intensities, nor required for efficacy. Evoked responses may be useful for verification of uniformity of target acquisition.
There is renewed interest in the development of alternative therapies to drugs or ablative surgery, for subjects with pharmaco-resistant epilepsies. Delivery of electrical currents directly to structures such as the anterior thalamic nuclei (ATN) (Hoadie et al., 2002; Kerrigan et al., 2004; Andrade et al., 2006) are among the more recent efforts aimed at addressing this important medical need. This study assesses under rigorous conditions the effects of electrical stimulation of the ATN on intractable epileptics by (1) enrolling subjects, whose epileptogenic zones are directly connected to these nuclei and (2) identifying for each subject, using a detection algorithm (Osorio et al., 2002) safe and tolerable electrical stimulation parameters that are also likely to reduce seizure frequency, intensity, or duration (Osorio et al., 2005). Specifically, this study investigates the safety, tolerability, and efficacy of bilateral, high frequency (>100 Hz), periodic, round-the-clock stimulation of the ATN, on subjects with inoperable pharmacoresistant seizures of mesial temporal origin. It also investigates the potential utility of evoked responses elicited from the stimulation targets, in assisting in the determination of uniformity and precision of lead placement intra and interindividually—a potentially valuable tool, particularly for targets such as the ATN, lacking specific functional markers.
- Top of page
This study approved by the Human Subjects Committee and the FDA (IDE # G990238), was conducted at the University of Kansas Medical Center between March 2000 and July 2004 on four subjects with inoperable pharmaco-resistant seizures who met all the inclusion and none of the exclusion criteria (see Appendix 1). These subjects were enrolled in the order of admission for invasive surgical evaluation.
The choice of the ATN as stimulation targets was based on the following considerations: (1) experimental evidence of their role as seizure gating structures (Mirsky et al., 1997) even though the model used in those investigations may not apply to human mesial temporal epilepsy (Loscher et al., 1991); (2) direct and extensive connections with mesial temporal structures (Swanson, 1978; Segal, 1979; Brodal, 1981; Jenkins et al., 2002); and (3) their larger distance from specific sensory or motor thalamic nuclei, allows delivery of higher-intensity currents, than to targets such as the centromedian or subthalamic, while also keeping subjects blinded as to times of stimula- tion.
The experimental protocol consisted of: (1) a baseline (6 months) during which monthly seizure type and frequency were recorded and medications remained unchanged; (2) inpatient closed-loop stimulation (several days) to identify safe, tolerable and efficacious stimulation parameters (Osorio et al., 2005); (3) postsurgical recovery (6 weeks) without electrical stimulation; and (4) outpatient periodic, round-the-clock electrical stimulation (36 months).
Following completion of the surgical evaluation, for which multiple bilateral depth electrodes had been placed within or near the amygdalae, pes hippocampi and body of hippocampi, through a lateral approach, subjects who were deemed inoperable (bitemporal independent seizures) and had consented in writing to participate in this study were taken back to surgery for stereotactic implantation of leads (3387 DBS Leads; Medtronic Inc, Minneapolis, MN, U.S.A.) into each ATN (see Appendix 2). MRIs were obtained in each subject after electrode implantation to visually verify correctness of placement and select the contacts, through which electrical currents would be passed. Given the size of the ATN in relation to the number, length and intercontact distances, only two out of the four available contacts/electrode were selected for delivery of current; these two contacts were visually chosen by the neurosurgeon as the most likely to be within the ATN.
To investigate the existence of modulatory influences of the structures where the stimulating contacts resided upon the epileptogenic zones (mesial temporal regions), biphasic square pulses (0.7 Hz; 0.1 ms/phase; 5.1 mA/phase), were delivered through the two contacts presumably located in the ATN; recording of intraindividual reproducible and stable responses with well-defined waveforms from the epileptogenic zones (EZs), would be interpreted as evidence of modulation of thalamus upon these zones. Additionally, the morphology, latency and polarity of the responses would provide information about the uniformity of electrode placement intra- and interindividually since they depend on the nature, location and orientation of the current sources, as well as on volume conduction characteristics which are determined by the electrical and geometric properties of the tissue (Nunez, 1990).
The subjects were allowed to recover for 24 h before the inpatient quantitative search for safe, tolerable and potentially efficacious parameters was initiated. This was carried out with a custom bedside system (Flint Hills Scientific, Lawrence, KS, U.S.A.) (Peters et al., 2001) consisting of two PC's connected to a constant current Grass S-12 stimulator (Astro-Med, Inc., West Warwick, RI, U.S.A.) for delivery of charge-balanced square waves. Thalamic stimulation was triggered by automated seizure detections (closed-loop phase) using a validated algorithm (Osorio et al., 2002). High frequency electrical stimulation, defined as 100Hz minimum, was selected because: (1) it may induce synaptic plasticity in the form of short-term depression, long-term depression, or both: (2) it upregulates glutamic acid decarboxylase and downregulates calcium-and calmodulin-dependent protein kinase II, the net effect of which is to enhance inhibition at or near the stimulated site, and (3) it increases the seizure threshold in the rat pentylenetetrazol model when delivered to the ATN (Mirski et al., 1997; Osorio et al., 2005). Once tolerable, safe and efficacious stimulation parameters had been identified, the inpatient closed-loop phase was terminated and the thalamic electrode leads were internalized under general anesthesia and connected to two Soletra IPGs (Medtronic Inc. Minneapolis, MN, U.S.A.) placed in the right and left infraclavicular regions; the electrodes that had been used for electrocorticogram (ECoG) and event response acquisition were removed at this time.
No earlier than 6 weeks after the subjects had been discharged from the hospital, the Soletra IPGs were programmed (Table 1) and turned on. Intensity was identical to that used in the closed-loop trial; pulse width and frequency had to be lowered due to the limitations of the Soletra IPGs vis-à-vis the custom system used in the closed-loop trial. Also, since these devices have no means for automated seizure detection, the stimulation cycle was periodic and around the clock: 1 min ON (twice the stimulation duration of the inpatient phase) and 5 min OFF for each generator. To encompass as much of the stimulation target as possible, the two contacts selected were made the cathode (−) and the device case in the chest, the anode (+) (“double monopolar” configuration). Use of rescue drugs or emergence of new seizure types or serious and/or intolerable adverse events during the 36-month trial, would result in termination of enrollment.
Table 1. Stimulation parameters, active contacts, and polarity
| ||Patient A||Patient B||Patient C||Patient D|
|Pulse width (μs)||60; Δ to 90 on 7-20-01||90||90||90|
|Rate (Hz)||145||145; Δ to 170 on 01-08-03||145||145; Δ to 170 on 2-20-02|
|On time||1 min||1 min||1 min||1 min|
|Off time||5 min||5 min||5 min||5 min|
|Impedance||477–891 Ω||516–659 Ω||588–802 Ω||516–2,000 Ω|
Data collection and analysis
Subjects were required to keep a daily seizure (by seizure type) and adverse event diary for the duration of the study (42 months). Neurological status, adverse events and drug serum concentrations were monitored every month during the baseline (6 months), for the first 18 months of the outpatient phase and twice per year thereafter. A quality-of-life evaluation (QOLIE-31 Version 1.0) and formal neuropsychological testing were performed during baseline and repeated approximately 6 and 18 months later.
Lead/contact placement was quantitatively assessed several months after implantation with a surgical navigation system (Stryker Leibinger Inc., Kalamazoo, MI, U.S.A.). Each subject's MRIs in compatible electronic format were imported into this system that was outfitted with the Schaltenbrand–Wahren atlas that had been used to guide the stereotaxic electrode placement. Each subject's contacts, through which currents were being passed (“active”) and not passed (“inactive”) into brain tissue, were visually identified on the digitized MRs and their centers electronically registered so that the nucleus within which each contact was placed was provided automatically by a software routine available in this navigation system (see Appendix 3).
Efficacy of periodic stimulation was assessed by comparing seizure frequency obtained from daily seizure diaries over the 6-month baseline versus the 36-month outpatient phase, using a within-subjects repeated measures design where each patient served as her/his own control. Baseline seizure frequencies were established by averaging the total number of seizures patients had reported during each 4-week period prior to implantation of the DBS system. Effects of stimulation on percentage changes in seizure frequency were evaluated using 3-month interval counts (range: 89–92 days) and compared to baseline (i.e., tested to see if different from zero). Each measured interval represented a major time-point in the study and was used as a repeated measure in the analyses. Statistical significance (p < 0.05) was calculated for each 12-week block, independent of any other block. Subjects were deemed “responders” if seizure frequency decreased by a minimum of 50% during the outpatient phase compared to baseline; and “nonresponders” if not.
- Top of page
Ostensibly, long-term, periodic, round-the-clock, bilateral, asynchronous high-frequency (mean: 157 Hz) stimulation of certain thalami nuclei, is safe, well tolerated and efficacious in subjects with inoperable bitemporal lobe epilepsy. The latency between device activation and the appearance of a beneficial effect (Fig. 5) was short in three of the four subjects and it has neither increased nor decreased (tachyphylaxis) as a function of time, in any of them. The worthwhile reduction in seizure frequency (mean: −75.6%) in all subjects, however, cannot be unequivocally attributed to electrical stimulation, since the experimental design did not include a crossover phase. The beneficial response to closed-loop stimulation, ascribed to the passage of currents into the thalamus (Osorio et al., 2005), the lack of effect following electrode implantation and before device activation and the increased probability of seizures in Subject A, each time the device output was accidentally turned off, provide evidence in support of a beneficial role of electrical stimulation. A lesion effect (microthalamotomy) that was invoked in recent publications (Hoadie et al., 2002; Andrade et al., 2006) as the cause for improvement, was indirectly investigated in this study over the immediate postoperative period (6 weeks) and could not be detected.
The effects of closed-loop (local or remote) electrical stimulation on epileptogenic tissue are both closely temporally correlated with delivery of currents (immediate effect) and persist for some time (unknown as of now) beyond cessation of stimulation (“carry-over” effect) (Osorio et al., 2005). Given that in this study stimulation was delivered over a small fraction of each day's duration (240/1440 min or 17%), it is reasonable to posit that if electrical stimulation, contributed to the significant reduction in seizure frequency, the “carry-over” effect played a prominent role. The clinical implications of this phenomenon may be useful: depending on the duration and degree of “carry-over” effect, stimulation parameters, most notably the interstimulation interval, may be adjusted to attain a “steady state” of protection that minimizes seizure occurrence probability.
The concordance in therapeutic response between the closed-loop and the outpatient trial, if replicated in larger studies, would have important practical implications. Brief closed-loop trials with portable devices could be used to identify efficacious parameters, thus increasing efficacy and decreasing the search time period for said parameters. Also, treatment accessibility to patients would be increased, as this approach could obviate the need for devices with seizure detection capabilities, which would translate into smaller size, lower cost and longer battery life, thus reducing the number of surgical interventions required for its replacement.
The finding that only 3 out of 16 active contacts in 2 out of 4 subjects may have been within the ATN or its projections, yet all subjects benefited from electrical stimulation, raises the possibility that delivery of currents to other nuclei such as the N. Reticulatus polaris, the N. Lateropolaris (Ventral anterior in Walker's nomenclature) and the Fields of Forel may have antiseizure effects. Electrical stimulation of the reticular nucleus decreases the intensity and duration of kindled limbic motor seizures in rats (Nanobashvili et al., 2003) and electrolytic lesions of the N. Lateropolaris (or Ventral Anterior) had a similar effect on focal neocortical (Kusske et al., 1972) or generalized (Feeny and Gullotta, 1972) seizures in cats and rhesus monkeys. Similarly, destruction of the fields of Forel improved seizure control in humans with pharmaco-resistant epilepsy (Dennosuke and Mukawa, 1987).
The rostral pole of the reticular nucleus has topographically mapped connections to limbic cortex and to ATN in the rat (Lozsadi, 1994; Lozsadi, 1995; Guillery 1998) and via these structures to mesial temporal regions (Swanson, 1978). The ventral anterior nucleus may, through its connections to midline thalamus (Kusske et al., 1972; Miller et al., 1993) influence hippocampus, amygdala, and entorhinal cortex (Bertram et al., 2001; Zhang and Bertram, 2002). However, identification of which of these nuclei was responsible for the beneficial effect observed in this pilot study is a difficult task for two reasons: (i) The N. Reticulatus polaris and the N. Lateropolaris (ventral anterior) abut the ATN (Schaltenbrandt and Wahren, 1977) and (ii) Electrical currents likely spread outside the nuclei where the leads resided. A quantitative study of the volume of axonal tissue directly activated by electrical “bipolar” stimulation of the subthalamic nucleus with parameters (3 V; 0.1 ms; 150 Hz) similar to those used in this study showed axon activation as far as 4 mm from the electrode contact (McIntyre et al., 2004). While in this study, no attempt was made to model the spread of current outside the intended anatomical target, the likelihood of this occurring is likely greater than in that recently estimated (McIntyre et al., 2004), given the “double monopolar” configuration [brain contacts (−) to IPG case (+)] and the higher stimulation intensity (mean: 4.1 V). These factors and the proximity of the ATN, Lateropolaris and Reticular nuclei to each other support the conjecture that the “loci” of this trial's therapeutic effect may have been more likely at the regional (ATN, Lateropolaris, Reticular, and possibly other nuclei) than at the single thalamic nucleus level. Selective nuclear stimulation at certain intensities and/or electrode configurations/geometry may be neither feasible, nor required for clinical efficacy. If confirmed in larger trials, this observation will markedly simplify stereotaxic and surgical electrode implantation procedures for the treatment of intractable epilepsies. Furthermore, interpretation of efficacy and adverse effects and investigation of mechanisms of action based on a “regional” or multinetwork” concept may be more fruitful than those that focus on a single structure or nucleus.
A relevant question this trial raises is: Were the electrodes/contacts in two recently published studies (Hoadie et al., 2002; Kerrigan et al., 2004) in the ATN? Elicitation of the “recruiting response” (Dempsey and Morrison, 1941) which is construed by one group of investigators (Kerrigan et al., 2004) as evidence that the leads were in the ATN and by the other (Hoadie et al., 2002) as a predictor of efficacy is of little localizing value since this response can be elicited from several thalamic nuclei (Dempsey and Morrison, 1941). The finding that the morphology of cortical responses to anterior thalamic stimulation in four subjects (Zumsteg et al., 2006) that participated in one of these studies (Hoadie et al., 2002) was very heterogeneous, suggests lack of uniformity of thalamic lead placement in them. Also, given the intensities used for stimulation in said studies (Hoadie et al., 2002; Kerrigan et al., 2004; Andrade et al., 2006) current spread outside the intended target was also highly likely (McIntyre et al., 2004).
Although evoked responses techniques, as used in this study, do not provide direct information about lead localization, they may be used to investigate if the EZ is modulated by the stimulation targets and also as tools to assess intra as well as interindividual uniformity and precision of lead placement. The intra and interindividual differences in evoked responses in these subjects accurately predicted the probable differences in lead location. The basis for this claim is that the morphology and polarity of evoked responses at any location in the brain, depend on the nature, location and orientation of the current sources, as well as on volume conduction characteristics which are determined by the electrical and geometric properties of the stimulated tissue (Nunez, 1990). That is, responses of different morphology and polarity, recorded from the same site, are generated by different structures or sources.
The elicitation of mesial temporal responses by contralateral thalamic stimulation (Fig. 3) is of functional and clinical significance. While actual contact electrode locations are unknown and no reference to known neuroanatomical connections if existent is thus possible, the presence of mesial temporal responses to contralateral thalamic stimulation suggests that unilateral stimulation may have bilateral effects.
To conclude, high-frequency stimulation of certain thalamic nuclei appears as a safe, well-tolerated and efficacious therapeutic modality for inoperable mesial temporal epilepsy. The utility of regional thalamic, as opposed to single nucleus, stimulation for control of extratemporal seizures deserves investigation.