Address correspondence to Masami Fujii, Department of Neurosurgery, Yamaguchi University School of Medicine, 1-1-1 Minamikogushi, Ube, Yamaguchi 755-8505, Japan. E-mail: firstname.lastname@example.org
Purpose: Focal brain cooling is effective for suppression of epileptic seizures, but it is unclear if seizures can be suppressed without a substantial influence on normal neurologic function. To address the issue, a thermoelectrically driven cooling system was developed and applied in free-moving rat models of focal seizure and epilepsy.
Methods: Focal seizures limited to the unilateral forelimb were induced by local application of a penicillin G solution or cobalt powder to the unilateral sensorimotor cortex. A proportional integration and differentiation (PID)–controlled, thermoelectrically driven cooling device (weight of 11 g) and bipolar electrodes were chronically implanted on the eloquent area (on the epileptic focus) and the effects of cooling (20, 15, and 10°C) on electrocorticography, seizure frequency, and neurologic changes were investigated.
Key Findings: Cooling was associated with a distinct reduction of the epileptic discharges. In both models, cooling of epileptic foci significantly improved both seizure frequency and neurologic functions from 20°C down to 15°C. Cooling to 10°C also suppressed seizures, but with no further improvement in neurologic function. Subsequent investigation of sensorimotor function revealed significant deterioration in foot-fault tests and the receptive field size at 15°C.
Significance: Despite the beneficial effects in ictal rats, sensorimotor functions deteriorated at 15°C, thereby suggesting a lower limit for the therapeutic temperature. These results provide important evidence of a therapeutic effect of temperatures from 20 to 15°C using an implantable, hypothermal device for focal epilepsy.
Despite the long history of investigation, the clinical feasibility of focal brain cooling remains unclear. One of the crucial but unresolved issues is to clarify whether “therapeutic temperatures” really exist, since focal brain cooling can suppress epileptic seizures, but is also associated with suppression of synaptic transmission. Indeed, a series of in vivo experiments have shown cooling-induced deterioration of various neurologic functions, including visual function in cats (Lomber et al., 1996; Lomber & Payne, 2004), auditory function in cats (Malhotra et al., 2004), and motor function in monkeys (Sasaki & Gemba, 1984; Brinkman et al., 1985). The cooling temperatures of the cortical surface in these experiments were not explicitly described, but neurologic deterioration was presumably induced by excessive suppression of synaptic transmission. Therefore, to address the issue of clinical feasibility, it is necessary to clarify whether seizure suppression can be achieved with a minimal influence on neurologic function.
This issue was preliminarily investigated by Karkar et al. (2002) wherein bath application of 4°C saline on the cortex in a patient with epilepsy did not influence the amplitudes of motor-evoked potentials. Another study of cooling showed that network synchronization in hippocampal slices was terminated without blocking normal synaptic transmission (Javedan et al., 2002). Although these results are promising, they do not provide direct evidence. Therefore, we addressed the effect of cooling on seizure and neurologic functions in awake, free-moving rats using an implantable cooling system, with a focus on the eloquent cortex.
A cooling device originally developed in our laboratory (Imoto et al., 2006; Oku et al., 2009; Fujioka et al., 2010) was used in the study. This device includes a cooling component and a heat-processing component. The cooling component (about 11 g in weight) consists of a proportional integration and differentiation (PID)–controlled thermoelectric chip (6.0 × 6.0 mm; maximum current (I max) 1.8A, maxmum voltage (V max) 2.5V, maximum power (Q max) 2.4W; Ferrotec Corp., Tokyo, Japan). The cooling side of the thermoelectric chip is attached to a pure silver plate (thickness of 1 mm) for direct cooling of the cortex. To avoid contact injuries with the brain, a fine thermocouple (Physitemp 23T, Clifton, NJ, U.S.A.) is embedded in the silver plate (Fig. 1A). The thermoelectric chip is controlled by a PID controller (Yamatake Corp., Tokyo, Japan). Each PID value was selected by automatic tuning of the controller to minimize overshooting or undershooting of a target temperature. The temperatures of the brain surface were cooled to 20, 15, and 10°C. Heat from the thermoelectric chip was transferred via a copper-made heat sink (6 × 6 mm with a thickness of 4 mm; see Imoto et al., 2006; Tanaka et al., 2008; Fujioka et al., 2010). The heat sink, with two water channels inside, was connected to the heat processing component via medical catheters (TYGON, R-3603, Saint-Gobain Performance Plastics, Akron, OH, U.S.A.) filled with Ringer’s lactate. The heat-processing component includes a helium-gas cooler (TwinBird Corp., Tsubame, Japan) and a direct current (DC)-driven pump (flow rate of 200 ml/min), which circulates Ringer’s solution at a controlled temperature of 20°C. Cooling was started manually in the current study.
Focal seizure and epilepsy models
Animal experiments were performed using protocols approved by the Yamaguchi University School of Medicine Institutional Animal Care Committee. Male Wistar rats (Chiyoda Kaihatsu Co. Ltd., Tokyo, Japan) (450 ± 50 g) housed in a temperature-controlled room (23.0 ± 2.0°C) were used in the study (n = 29 in total). Following induction of anesthesia by 4% sevoflurane, atropine (0.01 mg/kg) was injected subcutaneously and a mixture of ketamine (40 mg/kg, i.m.) and xylazine (4 mg/kg, i.m.) was administered for maintenance of anesthesia. The rectal temperature was maintained at 37 ± 0.2°C using a heating pad. The skull of the rat was fixed using a stereotactic apparatus (Narishige, Tokyo, Japan) and the skin on the skull was cut following injection of lidocaine (2%). A craniotomy was made with a dental drill over the ipsilateral sensorimotor (SI-MI) area (1.5–7.5 mm lateral, 3.0 mm anterior, and 4.0 mm posterior to the bregma). The cooling device was implanted and fixed in place with medical resin (Unifast II; GC Corp., Tokyo, Japan). The cooling component (6.0 × 6.0 mm) cools the entire somatotopic representation center, except for the tongue and lips, in rats (Fig. 1B,C) (Hall & Lindholm, 1974).
We used a focal seizure model (n = 12) and a focal epilepsy model (n = 6) for induction of focal seizures limited to the unilateral forepaw area. Rats with seizures outside the forepaw area were excluded from the study. Focal seizures were induced by intracortical infusion of a 4% NaCl solution of penicillin G (PG) using a syringe pump (0.3 μl/min at a concentration of 200 IU/μl, up to 1,200 IU) until continuous seizures were stably but minimally induced. Intracortical infusion was performed using a fine needle (28 gauge) with a tip length of 0.8 mm, which was attached to the center of the cooling component. The needle was stereotactically implanted on the eloquent area (i.e., the forepaw area of the sensorimotor [SI-MI] cortex at a depth of 0.8, 1.0 mm anterior, and 3.6 mm lateral to the bregma) using medical resin. Once seizures were induced, experiments were performed within 30 min. The frequency of seizures was stable over this time window. To investigate the effect of cooling when the seizure focus extends out of the forelimb area, >1,200 IU (up to 2,200 IU) of PG solution was also applied (n = 3).
Focal epilepsy was induced by direct application of cobalt powder on the same area of the cortex (Dow et al., 1962). Following a small craniotomy made with a dental drill, cobalt powder (8 mg; Sigma-Aldrich Co. LLC., Tokyo, Japan) was applied on the dura over the eloquent cortex. A sterilized cotton sheet was placed and the skin was sutured. Following a recovery period of 3 days, the rat was reanesthetized and the cooling device was implanted using dental resin at the center of the forepaw area, which became an epileptic focus. Cooling experiments were performed at 9 ± 2 days after implantation.
The effect of cooling on the frequency of seizures before and during cooling was evaluated by the number of involuntary lifts of the forepaw from the floor in 3 min for both the focal seizure and epilepsy models. All tests were recorded by video camera (60 frames/s) and forepaw lifting was evaluated blindly by at least two of four researchers. Comprehensive neurologic functions before and during cooling were assessed on a 21-point neurologic scale, which was originally developed for assessment of a cerebral ischemia model in rodents (Hunter et al., 2000). This scale comprises a battery of 10 items: assessment of paw placement, righting reflex, ability to grip a horizontal bar, time on an inclined platform, rotation, visual forepaw reaching, circling, contralateral reflex, motility, and general condition. This kind of scale is commonly used in behavior assessments of rats (McGill et al., 2005). The assessment was performed three times within 30 min in the current study.
Sensorimotor functions of the limbs were investigated in foot-fault tests and according to the receptive field size in the forepaw area. Foot-fault tests were evaluated using the following formula: (foot faults per limb/steps per limb) × 100 (Soblosky et al., 1996). The rat was placed gently on an elevated grid and the number of slips into the grid (i.e., foot faults) in 25 paired steps was calculated. The trial was performed three times and mean scores were calculated. Sensory function was evaluated by measuring the receptive field (RF) size of the forepaw area contralateral to the cooling cortex (layer iv; depth 450–800 μm) under the ketamine anesthesia described above (n = 5) (Fujioka et al., 2004). Tactile stimuli were applied on the forepaw areas of the skin (40 points) with von Frey hair-type probes (calculated force of 0.6 g) before and during cooling. The number of reaction fields was counted and defined as the RF size. All behavioral experiments were performed at cooling temperatures of 20, 15, and 10°C. All tests were also recorded by video camera (60 frames/s) and were evaluated blindly by at least two of four researchers.
An electrocorticography (ECoG, 1 Ch) over the epileptic focus in all rats was differentially recorded using a pair of needle-type electrodes (impedance 500 kOhm at 500 Hz) attached to the bottom of the cooling device (Fig. 1C). Data were amplified and recorded in Powerlab (ADInstruments, Colorado Springs, CO, U.S.A.) with a sampling rate of 2 kHz (low-cut filter 5 Hz, high-cut filter 100 Hz).
Before implantation of the device, the skin of the right chest was incised in the supine position and a telemetry system (PhysioTel, DSI, St. Paul, MN, U.S.A.) was implanted in normal rats (n = 4) and in normal sham rats with a cooling device implanted in the brain (n = 4). Two-lead ECG was sampled at 2 kHz with a duration of 1 min and recorded in a PC via a Powerlab instrument (ADInstruments).
Following the experiments, the rats were sacrificed and hematoxylin and eosin (H&E) staining was performed (5-μm sections).
Statistical analyses were performed by paired Student t-tests, Dunnett post hoc tests, or Steel-Dwass tests using the R software package (see the homepage; http://www.R-project.org.). p < 0.05 was considered to be statistically significant. Analysis of variance (ANOVA) was performed to evaluate the significance of differences between the means of all groups. A Dunnett test or Steel–Dwass post hoc test for multiple comparisons was used to compare groups with parametric or nonparametric data and unequal sample size or sample variance. Data are shown as the mean ± standard deviation (SD) in Student t-tests and Dunnett tests, and as the mean ± standard error of the mean (SEM) in the Steel–Dwass test.
Temperature gradient of the cooling area
The temperature gradient under the cooling device was evaluated thermographically on an agar surface warmed to 37°C. This surface was cooled to 20°C with the cooling device (6 × 6 mm). The cooling effect was limited to the contact area and did not reach the perimeter (Fig. 2).
Effects of cooling on focal seizures
The implanted device (Fig. 1A–C) did not influence ordinary behaviors in sham rats, such as eating, moving, grooming, or sleeping. ECG did not show any cooling-associated changes in rate rhythms (194 ± 6.32 in normal rats vs. 198.5 ± 14.73 in normal sham rats, p = 0.65 by Student t-test) and did not induce arrhythmia before, during, or after cooling. Although cooling to a target temperature was achieved without overshooting or undershooting, such precise temperature control was generally obtained at the cost of time. The times to reach the target temperatures of 20, 15, and 10°C were 9 ± 0.2, 12 ± 0.4, and 20 ± 0.4 s, respectively (mean ± SD, each n = 4).
Focal seizure model
Intracortical application of a PG solution reliably induced focal seizures limited to the unilateral forelimb area. Cooling of the seizure focus immediately and significantly reduced the frequency of seizures per minute at surface brain temperatures of 20°C (48.7%, p < 0.0001), 15°C (11.8%, p < 0.0001), and 10°C (0%, p < 0.0001), in comparison to the noncooling ictal group (100%, n = 6, Fig. 3A). The reduction of the seizure frequency was coincident with the suppression of epileptic discharges (EDs) in the ECoG (Fig. 4A,D). The effect of cooling-induced seizure suppression was continuous and was not diminished as long as cooling was performed. There was no apparent difference in the extent of suppression between rapid and slow cooling to a target temperature.
The significant reduction of the seizure frequency was also associated with improvement of neurologic scores within the range of 20–15°C. Induction of seizures caused a significant deterioration of neurologic scores (15.6 ± 0.43, p = 0.023). These scores were improved to 18.33 ± 0.21 at 20°C (p = 0.027) and 19.5 ± 0.22 at 15°C (p = 0.029), compared to those of the noncooling group (Fig. 3B). These effects disappeared soon after cessation of cooling. Cooling to 10°C also achieved a seizure-free condition, but neurologic scores remained low (15.93 ± 0.21, p = 0.979). Additional injection of a PG solution (>1,200 IU) induced seizures outside the forelimb area, which made it impossible to inhibit seizures in the forelimb area, as well as in areas outside the forelimb (Fig. 4B,E).
Implantation of the device for 1 month with 1 h cooling per day did not result in detrimental changes in H&E staining (n = 5), except for partial fibrosis of the subarachnoid region under the device (Fig. 5).
Focal epilepsy model
Cobalt-induced epileptic seizures limited to the unilateral forelimb area were sufficiently severe and continuous to be suggestive of a state of epilepsia partialis continua. The number of seizures per minute in the cobalt model was double that in the PG model. Seizure frequency was reduced in association with improvement of neurologic scores in a cooling range of 20–15°C and was coincident with suppression of EDs during cooling (Fig. 4C,F). The frequency of seizures per minute was reduced by 54.4% at 20°C (p < 0.0001), 3.9% at 15°C (p < 0.0001), and 0% at 10°C (p < 0.0001) (Fig. 6A). The significantly lower neurologic scores under noncooling, ictal conditions (14.5 ± 0.34, p = 0.023) were improved by 16.88 ± 0.28 at 20°C (p = 0.027) and 18.38 ± 0.2 at 15°C (p = 0.029, Fig. 6B), in comparison with the noncooling group (n = 6). As in the PG model, cooling to 10°C did not improve the neurologic scores (14.88 ± 0.41, p = 0.979, Fig. 6B). The therapeutic effect was not diminished as long as cooling was performed.
Histology in cobalt-treated rats shows bowl-shaped necrotic changes (Dow et al., 1962), which were limited to the shallow cortex in our study. There were no other particular cooling-associated changes.
Effects of cooling on neurologic functions
Because neurologic improvement was limited in the cooling ranges of 20–15°C, we hypothesized that cooling below 15°C induced excessive blockage of synaptic transmission. Therefore, we investigated the effects of cooling on normal neurologic functions in sham rats. Apparent neurologic deficits in ordinary behaviors (walking, eating, grooming, and so on) were not observed by cooling to 15°C. Neurologic functions (n = 6) were robust with cooling to 20°C (20.9 ± 0.06, 99.8%, vs. sham group, p = 0.92), but a trend for deterioration began at 15°C (20.6 ± 0.20, 98.1%, p = 0.089), and these changes reached statistical significance at 10°C (18.2 ± 0.21, 86.7%, p = 0.012), in comparison to the noncooling group (20.95 ± 0.05, 100%) (Fig. 7A).
A subsequent investigation of sensorimotor functions invariably revealed significant deterioration at 15°C. In foot-fault tests (n = 6), cooling to 20°C did not induce substantial changes (% error of 1.1 ± 1.31%, p = 0.957), but the findings reached statistical significance at 15°C (3.34 ± 0.73%, p = 0.018) and 10°C (18 ± 3.45%, p < 0.0001, Fig. 7B). In anesthetized sham rats (n = 5), RFs of the forepaw under the cooling area began to diminish at 20°C (84.1%, p = 0.099) and reached statistical significance at 15°C (30.6%, p < 0.0001) and 10°C (1.2%, p < 0.0001), in comparison to the noncooling group (100%) (Fig. 7C).
This study provided important evidence for a therapeutic effect of low temperature on focal seizure and epilepsy in an animal model. The results build on findings in previous studies (Yang & Rothman, 2001; Rothman et al., 2005; Yang et al., 2006; Rothman, 2009). Temperatures from 20°C down to 15°C significantly suppressed seizures and were associated with improvement of neurologic function. The effect was powerful, instantaneous, and continuous, which suggests advantages over other existing epileptic therapies.
Limitations and feasibility of focal brain cooling for epilepsy treatment
An indication of focal brain cooling for focal epilepsy requires accurate identification of epileptic foci of a size that is well within the cooling area. These conditions were produced in two animal models. The PG model of focal seizure allows adjustment of the size and extent of seizures (Elger & Speckmann, 1983). Therapeutic effects were obtained when the focal seizure was within the cooling area. However, application of PG >1,200 IU caused the focal seizures to no longer be limited to the forelimb, but to extend over the hind limb and body. In such cases, the suppressive effects of focal cooling were limited, even in the epileptic focus (Fig. 4B,E).
In our study, the epileptic focus in the cobalt model was clearly identifiable (Chang et al., 2004). The epileptic seizures in this model were more severe than those in the PG model, but seizure control was as prominent as that in the PG model. Therapeutic temperatures were also identified in the cobalt model, suggesting the feasibility of focal brain cooling as therapy for focal epilepsy.
Factors influencing the therapeutic cooling temperature
The results of the study show that therapeutic temperatures are not uniquely defined, but are changed by factors such as seizure severity and the size of the focus. Other factors that can influence the therapeutic temperatures include antiepileptic drugs (AEDs) and neuroplasticity. We did not use AEDs in the study, but the assumed synergistic effects of AEDs during cooling (Sourek & Travnicek, 1970) may increase the upper limit of the therapeutic temperature. Another important aspect of focal brain cooling is the involvement of functional compensation, presumably due to behavioral adaptation or neuronal plasticity. This property has been reported in a series of studies in normal monkeys, wherein cooling-induced functional deficits began to be ameliorated over a time course of months (Sasaki & Gemba, 1984; Brinkman et al., 1985). Identification and evaluation of these factors are important issues that remain to be addressed.
Determination of the therapeutic cooling temperature
Our data showed that cooling to 15°C reliably suppressed focal seizures and improved neurologic function (21-point scores), but a detailed investigation of sensorimotor functions (foot-fault tests and receptive field size) in normal sham rats revealed significant deterioration. Clinicians who place an emphasis on seizure suppression may prefer lower therapeutic temperatures at the cost of functional deterioration, whereas those who wish to avoid neurologic dysfunction may prefer higher therapeutic temperatures. Therefore, determining the therapeutic temperature in patients with epilepsy will depend not only on objective criteria but also on subjective criteria that maximize the quality of life.
Mechanism of seizure suppression by focal brain cooling
Focal brain cooling is generally considered to induce reduction of transmitter release (Eilers & Bickler, 1996), kinetic alteration of voltage-gated ion channels (Traynelis & Dingledine, 1988; Hill et al., 2000; Volgushev et al., 2000), and network desynchronization (Javedan et al., 2002). Although the precise antiepileptic mechanisms remain to be determined, it is generally recognized that suppression of synaptic transmission is involved in reduction of seizures. An in vitro study showed that synaptic transmission begins to decrease below 20°C (Volgushev et al., 2000). In a case in which the temperature is <20°C at 1 mm under the cortical surface, but >20°C at a depth of 2 mm, it is reasonable to assume that synaptic transmissions and EDs in the shallow cortex (layer II/III) are selectively suppressed because of the spread through neurons in the shallow layer with horizontal connections to the ipsilateral or contralateral cortex (Nolte, 2009). Selective suppression of synaptic transmission due to a cooling-induced thermogradient in the cortex may contribute to the vulnerability of somatosensory processing, as indicated by the reduction of RFs during cooling. Because the neurons that form a pyramidal tract (layer V) lie deep in the sensorimotor cortex, selective transmission failure may have occurred during surface cooling.
Pathologic changes due to cooling were not observed in the PG and cobalt models. Although partial fibrosis under the cooling device did occur, this was probably not caused by cooling, given the histologic tolerance to focal brain cooling even down to 5°C (Yang et al., 2006; Oku et al., 2009). Rather, it is likely that this histologic change was caused by direct contact with the cooling part of the device (i.e., pure silver) because inflammation of the contact area cannot be avoided under free-moving conditions.
Clinical advantages and requirements of the cooling device
Temperature control is of crucial importance in therapeutic applications, given that the range of therapeutic temperatures is narrow and that a small deviation from this range may lead to neurologic dysfunction. Furthermore, varying brain temperatures in the ictal stage may further complicate temperature control. In this regard, thermoelectronic devices have an advantage over traditional circulatory-cooling devices, since the thermoelectronic devices are small but have sufficient cooling power and precise temperature control. An alternative approach using systemic hypothermia has been used in refractory status epilepticus (Corry et al., 2008), but clinical use of this method is limited by adverse effects and limitations on the cooling temperature (31–35°C) and period.
Clinical demand for an implantable cooling device will not be limited to the epileptic field. Other potential applications include treatment for cerebrovascular diseases, including poststroke rehabilitation (Clark & Colbourne, 2007), neurotrauma (Clark & Colbourne, 2007), and pain (Fujioka et al., 2010), all of which will depend on thermal modulation of neuronal excitability.
Application of the cooling device for treatment of epilepsy
Focal brain cooling may be applied therapeutically for patients who have an epileptic focus on the eloquent cortex (i.e., motor or language area) or those who cannot be treated with AEDs. Cooling may also be used as a diagnostic tool in intracranial ECoG monitoring of patients with potential neurosurgical indications, but in whom the focus cannot be clearly defined. In such cases, the final surgical indication would be decided by preliminary application of cooling to the focus. There are several physiologic and technical issues to be solved before the device can be applied in intractable epilepsy. However, this study is an important step toward medical use of an implantable hypothermal device for treatment of focal epilepsy and other neurologic disorders.
This work was supported by a Grant-in-Aid for Specially Promoted Research (No.20001008) granted by MEXT of Japan.
None of the authors has any conflict of interest to disclose. We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.