Therapeutic brain hypothermia, its mechanisms of action, and its prospects as a treatment for epilepsy


  • Gholam K. Motamedi,

    Corresponding author
    1. Department of Neurology, Georgetown University Hospital, Washington, District of Columbia, U.S.A
    • Address correspondence to Gholam K. Motamedi, Department of Neurology, PHC 7, Georgetown University Hospital, 3800 Reservoir Rd., NW, Washington, DC 20007, U.S.A. E-mail:

    Search for more papers by this author
  • Ronald P. Lesser,

    1. Department ofNeurology, Johns Hopkins University School of Medicine, Baltimore, Maryland, U.S.A.
    2. Department ofNeurosurgery, Johns Hopkins University School of Medicine, Baltimore, Maryland, U.S.A.
    Search for more papers by this author
  • Stefano Vicini

    1. Department of Pharmacology & Physiology, Georgetown University School of Medicine, Washington, District of Columbia, U.S.A
    Search for more papers by this author


Cooling the core body temperature to 32–35°C, is almost standard practice for conditions such as cardiac arrest in adults, and perinatal hypoxic ischemic encephalopathy in neonates. Limited clinical data, and more extensive animal experiments, indicate that hypothermia could help control seizures, and could be applied directly to the brain using implantable devices. These data have fostered further research to evaluate whether cooling would be a viable means to treat refractory epilepsy. Although the effect of temperature on cellular physiology has long been recognized, with possibly dual effects on pyramidal cells and interneurons, the exact mechanisms underlying its beneficial effects, in particular in epilepsy, are yet to be discovered. This article reviews currently available clinical and laboratory data with a focus on cellular mechanisms of action and prospects of hypothermia as a treatment for intractable seizures.

Clinical evidence and data from animal models have shown the neuroprotective effects of hypothermia and have suggested that it might be an effective treatment for drug-resistant epilepsy. It now seems possible to implant microscale thermoelectric devices to apply direct cortical hypothermia (Rothman, 2009; Fujii et al., 2012).

Cooling the core body temperature (Tc), usually down to 32–35°C, is becoming a standard practice for adults resuscitated after cardiac arrest and for infants treated for perinatal hypoxic ischemic encephalopathy (HIE) (Polderman & Herold, 2009). Hypothermia may have beneficial effects on pathophysiologic cascades initiated by brain injuries, cascades that can result in edema, seizures, and apoptosis (Schmitt et al., 2006; Rossetti & Lowenstein, 2011).

Clinical Experience with Therapeutic Hypothermia

Historical background

Fay reviewed prior experiences with surface application of therapeutic hypothermia, from as early as 1829 (Fay, 1959). He introduced cold saline or boric acid in metal capsules surgically positioned adjacent to the lesions, and reported gratifying results with respect to swelling, pain, healing, and even infection control in patients with cerebritis or brain abscess. The lowest focal temperature achieved was 22°C. Patients were stable and tissue viable after 3 months of exposure to hypothermia. Tissue biopsy showed no inflammation, cellular infiltration, or cerebritis around the metal capsule. Generalized hypothermia with Tc of 24°C was also tolerated for many hours without major side effects. Ommaya and Baldwin reported a patient in status epilepticus (SE) who responded to systemic (body surface) cooling applied to control fever (Ommaya & Baldwin, 1963). At rectal temperatures slightly below normal, the seizure would slow down. Direct subdural and ventricular irrigation with 10–12°C saline resulted in successful seizure control, while avoiding systemic side effects of hypothermia such as ventricular fibrillation. However, they reported a 30 mm Hg drop in systolic and diastolic blood pressures, areflexia, mild respiratory slowing, and later pneumonia and death. They treated two other patients with direct cold (2–4°C) irrigation of exposed cortex in the operating room with no surgical resection. The patients became unresponsive to stimuli with each course of cooling; the unresponsiveness reversed within 10 min of rewarming while seizures remained under control. The authors reported that these patients had better seizure control subsequently but did not provide details.

Disorders other than epilepsy

In recent years, data have accumulated indicating a therapeutic role for hypothermia in several clinical conditions. Systemic hypothermia (32–34°C Tc) after cardiac arrest may reduce mortality and increase the rate of favorable neurologic outcome after 6 months (Hypothermia after Cardiac Arrest Study Group, 2002). It is gaining popularity as a therapy, typically started at 33 ± 0.5°C Tc within 6 h after delivery in patients with perinatal HIE (Shankaran et al., 2012). Cellular changes following hypoxic ischemic brain injury may trigger a pathologic cascade that can result in delayed neuronal apoptosis. Hypothermia started within 6 h of injury and continued for 2–3 days, may protect against this, possibly by inhibiting the inflammatory cascade (Drury et al., 2010; Zhao et al., 2012). Cooling the brain by a few degrees can be protective in rat models of stroke (Busto et al., 1987) and can lower metabolic activity in patients with ischemic stroke, but the optimal time window, best delivery system, and exact clinical indications are not known (Groysman et al., 2011). There have been promising results when using hypothermia to treat traumatic brain injury (TBI) and spinal cord injury (Sydenham et al., 2009; Dietrich et al., 2011). Possible mechanisms underlying the beneficial effects of hypothermia in TBI include reductions in several cell injury cascades, maintenance of increased neurogenesis, and neural repair occurring as natural protective responses to TBI (Bregy et al., 2012). However, mechanism(s) of action of neuroprotection in these neurologic conditions may differ from those in epilepsy.

Therapeutic Hypothermia in Epilepsy

Clinical data

Unlike in postanoxic encephalopathy, there is a paucity of information on efficacy and safety of therapeutic hypothermia in epilepsy. Pouring cold saline on exposed spike generating cortex can cause complete, but transient, abortion of spike discharges (Karkar et al., 2002). Data from animal models of SE indicate efficacy and neuroprotective effects of hypothermia. However, there is no standard delivery method, and timing of initiation, duration, and depth of hypothermia are unknown. The degree of cooling has been classified based on the achieved Tc as mild (34–35.9°C), moderate (32–33.9°C), moderate-deep (30–31.9°C), and deep (<30°C), but a Tc of <32°C is generally avoided because of potential side effects (Polderman & Herold, 2009). Available data show possible efficacy for different methods of cooling, cooling durations of 20–96 h, and temperature ranges of 30–36°C Tc, all used along with antiepileptic drugs (AEDs). The effect can last for the duration of cooling, but seizures tend to recur upon rewarming (Rossetti, 2011; Shorvon, 2011).

In one of the oldest trials, Surek and Travnicek applied a combination of whole body cooling and local brain hypothermia via burr holes in 25 patients with refractory epilepsy. Using a cooling chamber and subarachnoid extravascular irrigation with 0°C saline under general anesthesia, they achieved Tc of 32–29°C and brain surface temperature of 5–6°C. In 1 year follow-up, these patients reportedly had no morbidity and few seizures, four of them seizure-free, while on AEDs (Sourek & Travnicek, 1970). In a recent study in 15 patients with refractory epilepsy, external head cooling through a specially designed helmet worn for 60 min per session resulted in 12.2°C scalp, 1.7°C tympanic, and 0.12°C rectal mean temperature drops, respectively. Average weekly seizure frequency decreased from 2.7 to 1.7 (Bagic et al., 2008). In a case-control study of 15 neonates with perinatal stroke and encephalopathy, five neonates treated with systemic hypothermia did not develop seizures, whereas seven untreated cases had seizures and worse cognitive outcome (Harbert et al., 2011).

Status epilepticus (SE)

Patients with SE who remain unresponsive to standard protocols, typically including a benzodiazepine as well as other AEDs as needed, are considered to have refractory SE. Furthermore, seizures that continue or recur ≥24 h after onset or upon tapering of anesthetic agents, have been categorized as super refractory SE. Status epilepticus is refractory in 23–43% of patients and commonly carries a high risk of major morbidity or mortality (Novy et al., 2010; Holtkamp, 2011; Rossetti & Lowenstein, 2011; Shorvon & Ferlisi, 2011).

In patients with refractory SE, in particular those with generalized convulsive seizures, continuous intravenous anesthetics are used despite potential side effects including respiratory depression, hypotension, and immunosuppression. When AEDs and anesthetics fail, nonpharmacologic modalities may be added (Rossetti & Lowenstein, 2011). Vastola et al. (1969), reported successful treatment of five out of six patients with refractory SE using hypothermia through external cooling to 31–36°C. Three children with generalized convulsive SE who had failed conventional treatment with AEDs were successfully treated with moderate systemic hypothermia (30–31°C) and thiopental coma with burst-suppression or isoelectric tracing on electroencephalography (EEG) (Orlowski et al., 1984). In an infant with hemimegalencephaly and refractory SE, high doses of intravenous AEDs converted generalized status to focal status in the right occipital region (Elting et al., 2010). Upon (body surface) cooling to about 36°C, the seizure activity suddenly terminated. With continuing mild hypothermia there was a significant reduction in seizure frequency, with decreased dosage of AEDs.

Using an endovascular cooling system, Corry et al. (2008) reported complete abortion of ictal activity at 31–35°C Tc in four patients with SE of different etiologies (Fig. 1). These patients had failed benzodiazepines and/or barbiturates. After rewarming, all patients had a marked reduction in seizure frequency, with two becoming seizure-free. Shivering and coagulopathy with venous thromboembolism were reported side effects. Moreover, a patient with refractory SE secondary to a stroke and treated with thiopental and mild systemic hypothermia developed acute bowel ischemia with cecum necrosis requiring resection surgery (Cereda et al., 2009). Therapeutic systemic hypothermia (33–35°C Tc) was effective in controlling seizures, with full recovery, in a dog with refractory epilepsy secondary to TBI (Hayes, 2009).

Figure 1.

Scalp EEG in a patient with status epilepticus before (left panel) and after (right panel) cooling through an endovascular cooling system that circulated cooled saline through balloons placed into the inferior vena cava; the rhythmic epileptiform discharges are aborted after cooling to 32°C core temperature (Tc). From (Corry et al., 2008).

Continuous EEG monitoring should be performed throughout the cooling process to minimize morbidity. There are reports of subclinical SE with high mortality in comatose patients with cardiac arrest treated with systemic hypothermia along with neuromuscular blockade (Legriel et al., 2009). Subclinical electrographic seizures are also common in neonates with HIE, especially in moderate or severe cases, with incidence rates reported at 22–64%. Hypothermia may not always be effective. In a group of such term neonates systemically cooled to 34–35°C while monitored with video-EEG, 65% continued to have seizures, 47% had nonconvulsive electrographic seizures, and 23% continued to have SE. The majority of the seizures were seen within the first 48 h of the recording (Wusthoff et al., 2011). However, a retrospective review reported beneficial effects among 107 neonates with HIE monitored using continuous EEG. The effects of systemic cooling in neonates with electrographic seizures were compared between 16 cooled and 15 uncooled patients. The seizure duration was shorter in the former group (60 (39–224) vs. 203 (141–406) min), although this effect was significant only in moderate HIE (Low et al., 2012). These results and the high incidence of seizures despite mild hypothermia in patients with HIE highlight the need for optimal paradigms, as described above. It is also possible that the response to hypothermia might depend to a great extent on the pathophysiology involved.

Hypothermia in Animal Models of Epilepsy

The temperature dependence of biological processes is well recognized as reflected in the reaction rate constants in Arrhenius and Nernst equations. Delay at neuromuscular junction at lower temperatures has been attributed to desynchronized release of acetylcholine from presynaptic terminals (Katz & Miledi, 1965). There can be an augmenting effect of temperature increase on ion currents. For example the Na2+ current in the node of Ranvier has a Q10 (change in conductance rate per 10°C increase in the temperature) of 2.34 (Collins & Rojas, 1982).

In vitro epilepsy models

Experiments using slice models of epilepsy have established the anticonvulsive effects of cooling more definitively than any prior clinical efforts. Thompson et al. (1985) reported that cooling guinea pig hippocampal slice from 37°C to 33–27°C increased resting input resistance in CA1 pyramidal neurons, increased amplitude and duration of afterhyperpolarization, and decreased spike amplitude. The latter changes were attributed to an increase in Ca2+-activated K+ current resulting in enhanced spike adaptation, hence reduced network excitability during hypothermia. Shen & Schwartzkroin (1988) showed that hypothermia at <30°C reversibly decreased the amplitude of population spikes, increased spike duration, and induced a reversible depolarizing shift in membrane potential along with increased input resistance.

Applying focal hypothermia through a thermoelectric (Peltier) device placed directly over rat hippocampal-entorhinal slices successfully aborted epileptiform discharges induced by 4-aminopyridine (4-AP) (Hill et al., 2000). This effect was not seen when the activated Peltier device was not in direct contact with the slice, indicating temperature change, rather than electrical field, as the mechanism. A computer-generated automatic seizure detection algorithm could activate the device and suppress the epileptiform discharges in <4 s suggesting the potential for developing an implantable Peltier device for use in human epilepsy. The device could rapidly drop surface temperature of exposed cortex of a newborn pig from 36 to 26°C, measured 1.7 mm deep under the device (Fig. 2-1). Typically epileptiform discharges would reappear upon rewarming the slice to the baseline temperature of 33°C.

Figure 2.

(1) Suppression of ictal-like discharges in hippocampal dentate granule cells treated with 4-aminopyridine (4-AP) (50 pm). Top panel (A), onset of a 35-s normothermic seizure; middle panel (B), cooling terminates seizure discharges in the same slice within 7 s; and lower panel (C), ictal-like discharges resume after rewarming but are shorter in duration than during the pre-cooling period; V, voltage; T, temperature. From (Hill et al., 2000). (2) Rapid cooling results in temperature drop from 28 to 27°C and almost immediately abort spontaneous epileptiform discharges recorded in CA3, 4-AP 50 μm. Upon rewarming (1°C/s) the slice back to baseline temperature the spontaneous activity reappears (b1 and b2). The time course of the temperature line is shown above each trace. (3) Slow cooling (0.2°C/s) gradually drops the temperature from 32 to 23°C resulting in progressive decrease in amplitude of spontaneous discharges and finally complete cessation followed by recovery upon rewarming (0.2°C/s) to baseline. Top and bottom panels are continuous. The arrow indicates a brief increase in burst frequency early on during hypothermia that often appears, typically at around 27°C. Temperature recordings are shown above the field recording traces. From Motamedi et al. (2006).

In other experiments cooling the recording chamber consistently aborted field excitatory synaptic potentials as well as spontaneous epileptiform activity in rat brain slices exposed to 4-AP, 4-AP plus bicuculline, and Mg2+-free artificial cerebrospinal fluid (aCSF) at 28–34°C (Motamedi et al., 2006). The study used both rapid and slow cooling paradigms (perfusion with cold aCSF at rates of 2–5 and 0.1–1°C/s, respectively). Both completely, but reversibly, aborted epileptiform discharges in all three seizure models and in different recording sites. However, rapid cooling could effectively terminate the discharges with a slight temperature drop, as little as 1–2°C (Fig. 2), whereas with slower cooling the discharges would gradually decrease in amplitude, requiring a much longer time before complete cessation. Slow cooling required absolute temperature drops to 21–23°C for a 90% reduction in discharge frequency and 14–15°C for complete termination (Fig. 2). Disruption of synchrony seemed a possible mechanism of action. In these models, hypothermia seemed quite safe as the slices could tolerate as long as 2 h of cooling at temperatures as low as 8°C.

Single- and dual-patch clamp recordings from hippocampus and cortex in the 4-AP model of epilepsy in mice have revealed depolarization block in current clamp recordings in CA3, and hyperpolarization block in CA1/CA2, with a significant synchrony between CA1 and CA3 pyramidal neurons, but interestingly not between the pyramidal neurons and interneurons (Motamedi et al., 2012). Hypothermia may disrupt synchrony between CA1 and CA3 neurons, whereas in the interneurons it could block only rhythmic discharges without any effect on action potentials (Fig. 3). In recordings with perforated multielectrode array (pMEA), hypothermia disrupts synchronized discharges without affecting multiunit activity (Motamedi et al., 2012).

Figure 3.

Differential effect of cooling on the firing rates of pyramidal neurons and interneurons. Hypothermia disrupts the synchrony and terminates rhythmic action potentials in pyramidal neurons but not in the interneurons. Temperature recording is shown below the cell recording traces; 4-aminopyridine (4-AP) 50 μm, Green GAD gfp mice at postnatal day 15; hippocampal slice; CA3; dual current clamp recordings. From (Motamedi et al., 2012).

In vivo epilepsy models

Moseley et al. (1972) compared unit discharges, recorded via intracellular and extracellular microelectrodes, in cat and monkey sensory motor cortices before and after focal cooling with a Peltier device. Rapid hypothermia to 19–21°C eliminated unit activity and decreased amplitude of action potentials. Reynolds et al. (1975) used intracellular and extracellular recording from neurons and glial cells to study hypothermia in monkeys with seizures induced by alumina hydroxide. After cooling the cortex from 37°C to 32–34°C by applying cold Ringer's solution to a metal block, extracellular discharges were suppressed and remained so for the duration of cooling. Upon rewarming, the spikes were polyphasic and prolonged, with this interpreted as a change in seizure focus characteristics. Intracellular recording showed membrane depolarization in 10 of 12 glial cells in the visual cortex; in one case this was followed by hyperpolarization. Depolarization shift occurred in four of six cortical neurons with coincidental spikes on extracellular recording, whereas two showed hyperpolarization. These changes were thought secondary to alterations in either passive exchange or active pumping of Na2+ and K+ ions across the cell membrane.

In another study, 15 cats received penicillin injection to the right hippocampus and visual cortices to induce focal seizures (Voiculescu & Voinescu, 1992). External systemic hypothermia decreased the amplitude and frequency of the epileptiform discharges. This effect was more pronounced in occipital seizures, whereas hippocampal seizures seemed more refractory. Similar results were obtained in another study that compared various degrees of hypothermia and hyperthermia in rats and also found indications for both anticonvulsant and neuroprotective effects of hypothermia (Liu et al., 1993). For seizures induced by intraperitoneal injection of kainic acid, hypothermia at 28°C Tc reduced ictal discharges by 50%. Further cooling down to 23°C Tc resulted in almost complete cessation of the epileptiform discharges. There was no hippocampal cell loss in animals treated with hypothermia, whereas all of the animals kept at normal temperature had significant cell loss. On the other hand, at 42°C hyperthermia, both seizures and hippocampal cell loss were more pronounced, resulting in severe tonic seizures and death in all animals within 2 h. Some animals were injected with kainate at either 41–42°C Tc or normal temperature. The former group showed severe hippocampal cell loss and continuous ictal EEG activity, whereas the latter group had neither. Hypothermia is equally effective in models of focal seizures. After prior cooling to 30°C Tc, intraamygdala kainate injection in rats resulted in longer latency, shorter duration, and less frequent limbic seizures in both amygdala and hippocampus (Maeda et al., 1999). These animals also showed decreased glucose utilization both locally and diffusely.

With data indicating the effectiveness in slice models of hypothermia induced by focal cooling with a Peltier device, the same method was tested in vivo in halothane-anesthetized rats. Recurrent and prolonged seizures were induced by subpial injection of 4-AP into the surface of motor cortex. As seen in slice models, placing the thermoelectric device in direct contact with cortex dropped the temperature to 20–25°C with a significant reduction in seizure duration. This effect seemed to persist for a while after rewarming the tissue. Vital signs remained stable during the cooling and there were no acute or delayed neuronal injuries noted in histology (Yang & Rothman, 2001). Using this method, hypothermia penetrates about 4 mm below the surface, indicating a limited area of involvement (Rothman, 2009).

Schmitt et al. (2006) showed anticonvulsant effects of hypothermia in 10–12 week old rats with self-sustaining SE induced by electrical stimulation through electrodes implanted in the perforant path. The animals were treated with external systemic cooling for 3 h (minimum 29°C), with diazepam, with both, or were untreated controls. During 5 h monitoring, the severity of motor seizures and frequency and amplitude of spontaneous seizure discharges were compared among the groups. Animals treated only with diazepam showed just a slight decrease in the amplitude of the discharges with no reduction in seizure activity, although in later stages of SE motor seizures reduced markedly. The cooled animals showed significant clinical seizure reduction but no changes in the electroencephalography (EEG) discharges recorded from the dentate gyrus through intracerebral electrodes. Those treated with both cooling and low-dose diazepam showed a significant drop in frequency and amplitude of EEG discharges (Fig. 4-1). These effects were reversible: the seizure discharges reappeared upon rewarming the cooled animals. This experiment suggested that hypothermia might be a safe adjunct to pharmacotherapy in patients with refractory SE. The same group has compared the anticonvulsant effect of moderate and deep hypothermia (30 and 20°C) in the electrical stimulation induced model of self-sustaining status epilepticus (SSSE) (Kowski et al., 2012). With application of systemic cooling using ice packs only, animals cooled to 20°C showed complete suppression of SSSE. This effect was sustainable after rewarming with no mortality or discernible morbidity (Fig. 4-2).

Figure 4.

(1) Epileptic discharges recorded from the dentate gyrus in rats with self-sustaining status epilepticus (SSSE) induced by electrical stimulation. Hypothermia through external body surface cooling down to 29°C significantly reduced motor seizure activity but not the EEG discharges; however, when combined with low-dose diazepam it dramatically reduced both amplitude and frequency of epileptiform discharges. Ictal discharges continue to decrease as temperature is lowered. At about 29°C, the discharges are almost abolished. Following rewarming the epileptiform discharges reappeared to pretreatment level. From (Schmitt et al., 2006). (2) Epileptiform discharges recorded from the right dentate gyrus in an untreated (left panel) and treated rat (right panel). The animal was cooled to 20°C. The tracing on top shows ≥1 Hz SSSE following electrical stimulation and cooling; the second row, 60 min after cooling shows decreased frequency; at 20°C, the third row, the SSSE is terminated. There are no significant changes in the control. After rewarming to 37°C (bottom row), the spontaneous discharges and SSSE did not reappear after hypothermia, whereas the SSSE continued in the control. From (Kowski et al., 2012). (3) Video-electrocorticography (ECoG) recording from a rat model of traumatic brain injury with chronic spontaneous recurrent focal seizures both in mock-cooled and cooled animals. Prophylactic focal cooling by 2°C, started 3 days post injury and continued for 5.5 weeks, aborted seizures for the 10 weeks following the last cooling session. (Top, A) Exploratory behavior is shown in frames captured prior to the seizure onset in a control. Ictal discharges are first detected by the perilesional electrode e4. The following 12 consecutive frames are captured during ictal discharges with normal movement resumed after the seizure. (Bottom, B) A rare brief seizures recorded in a treated animal. The images show grooming behavior, that is, changes in the position of head and paws before seizure onset. Upon seizure onset there is ictal freezing indicated by the animal's static posture associated with ictal discharges only in the perilesional electrode (e4–e5). The Ictal onset is indicated by the gray arrows; insets show expanded ictal discharges. From (D'Ambrosio et al., 2012).

A device made of a thermoelectric chip based on a silver plate has been used in rats with focal seizures induced by cortical application of penicillin G or cobalt powder (Fujii et al., 2012; Kida et al., 2012). Cooling cortex focally to 10–20°C resulted in significant seizure control, but functional sensorimotor deterioration at 15°C suggests this temperature as the safety cutoff point. Different frequency ranges were sensitive to particular temperatures, that is, 25, 20, and 15°C attenuating beta, alpha to beta, and delta to beta frequency discharges, respectively.

The therapeutic effects of hypothermia might not be limited to seizure control. It not only reduced seizures, but also decreased brain edema, and improved cognitive function in rats with spontaneous kainate-induced recurrent seizures (Wang et al., 2011). Forced swimming in cold temperature is known to have anticonvulsant and neuroprotective effects in animals. In a rat model of lithium-pilocarpine seizures, duration and low temperature have been shown to be the defining factors (Fournier et al., 2008). Rats swimming for at least 5 min in temperatures of ≤20°C showed the anticonvulsant effect with no tolerance even after repeated swimming. These animals had significantly less hippocampal neuronal degeneration.

Pilocarpine-induced SE in rat pups has been used as a model of pediatric SE. Animals kept mildly hypothermic before pilocarpine injection show longer seizure latency, longer SE latency, lower spike frequency, smaller spike area, and less hippocampal apoptosis than controls (Yu et al., 2011). However, timing of intervention was critical in preventing neural damage and seizures in a group of near-term fetal sheep. Moderate to severe brain cooling after 8.5 h of cerebral ischemia induced by bilateral carotid occlusion and after onset of postischemic seizures, did not confer neuroprotection (Gunn et al., 1999).

Neuroprotective effect of cerebral hypothermia during seizures have been associated with decrease in both nitric oxide production, and hippocampal cell loss as shown in kainate induced seizures in immature rabbits (Takei et al., 2004). These findings were confirmed in immature rabbits cooled to 33°C Tc and then given kainic acid followed either by cooling or rewarming to 37°C Tc. Rapid rewarming resulted in increased nitric oxide production in and around hippocampus, and decreased cortical cerebral blood flow and mean arterial blood pressure during seizures (Takei et al., 2005).

In a model of TBI in adult rats (parasagittal fluid-percussion brain injury), exposing animals to 4 h of moderate systemic hypothermia resulted in significantly fewer seizures after pentylenetetrazole injection (Atkins et al., 2010). Hypothermia also significantly suppressed mossy fiber sprouting; however, it did not have any effect on chronic cell loss in the dentate hilar region. Most recently, using the same model of posttraumatic epilepsy, D'Ambrosio et al. (2012) showed that focal cortical cooling by as little as 0.5–2°C inhibited seizure onset, and cooling by 2°C for 5.5 weeks, starting 3 days following the head trauma, abolished seizures, an effect lasting for 10 weeks after the last cooling session (Fig. 4-3). These studies suggest a potential for hypothermia to not only act as an anticonvulsant but also to block epileptogenesis.

Molecular and Cellular Mechanisms of Action of Therapeutic Hypothermia in Epilepsy

Considering the immediate effect of cooling on epileptiform discharges in animal models, its mechanism of action in epilepsy seems to be different from those speculated in other neurologic conditions. Initially, the effects of hypothermia were attributed to alterations either in the Na2+-K+ pump or in passive Na2+ exchange mechanism across the cell membrane. This was based on the more prolonged rising phase of depolarization compared to the falling phase, suggesting that passive Na2+ exchange was more affected than outward flux of K+. However, because both neurons and glia showed membrane depolarization, the Na2+-K+ pump exchange was proposed to be the main target (Moseley et al., 1972; Reynolds et al., 1975). Volgushev et al. (2000), studied alteration of basic membrane properties in rat visual cortex cooled from 35 to 7°C and found increased input resistance with decreased K+ conductance but no changes in Na2+ current or its activation threshold. Masino & Dunwiddie (1999) have suggested a temperature dependent role for adenosine in regulating glutamatergic synaptic transmission. Boucher et al. (2010) have established the temperature dependency of synaptic diffusion including that of glutamate and glutamate receptor kinetics. Temperature changes significantly affect glutamate binding affinity and may also alter speed of diffusion. However, in vitro studies are typically conducted at room temperature (27–34°C), which may not be reflective of normal human cellular physiology.

Hypothermia can affect both evoked synaptic activity and synchronized spontaneous epileptiform activity. Considering the key role of ion channels in synaptic transmission and epileptiform discharges, it is possible that hypothermia could affect ion channel gating properties. This is consistent with the finding of a Q10 value of 1.8 for the effect of cooling on the rate of epileptiform discharges, which is similar to that of voltage-gated sodium channels. However, temperature dependence of local field potentials (LFP) in field recording cannot be directly related to the Q10 of ion channel gating. This change in channel gating could be attributed to phase transitions of membrane phospholipids, which can indirectly affect channel gating, possibly through affecting both inactivation and activation gates (Rosen, 1996, 2001; Aihara et al., 2001; Motamedi et al., 2006).

Yang et al. (2005) used 2-Photon microscopy in submerged hippocampal slices from rats placed over a Peltier device to measure vesicular release at 33 and 20°C. They facilitated presynaptic loading of the fluorescent vesicular label FM1-43 by the stimulation of Schaffer collaterals. Hypothermia resulted in significant reduction in presynaptic vesicular release as averaged over 100-square-pixel regions (Fig. 5). These findings were consistent with studies on synaptic delay at the neuromuscular junction (Katz & Miledi, 1965). A morphologic study of rat slices at different temperatures revealed reversible changes in dendritic spines. At the baseline temperature of 32°C and also after cooling down to 10°C for 2 h, there was no loss of dendritic spines. After cooling the slice to 5°C there was a significant loss of dendritic spines with the appearance of beading, but this was completely reversed after 15 min of rewarming. There was no increase in the number of spines with either cooling or rewarming (Yang et al., 2006; Rothman, 2009). These changes support the involvement of synaptic mechanisms during hypothermia.

Figure 5.

The effect of hypothermia on neurotransmitter exocytosis. (A) Single 512 × 512 pixel image plane showing FM1-43 fluorescence imaged over a 15-min period at 33°C in the absence of electrical stimulation. The final image in each series is the background (maximal destaining) obtained after an additional 10 Hz stimulation for 2 min. (B) Another image plane viewed over 15 min of 1 Hz electrical stimulation at 33°C. At 15 min almost all FM1-43 fluorescence has been eliminated, that is, reduction in presynaptic vesicular release, and the image is similar to background. (C) Stimulation at 1 Hz causes very little FM1-43 release at 20°C. Most of the dye finally destained after 10 Hz stimulation for 2 min, indicating that it was in a functional compartment. Scale, 5 μm. From (Yang et al., 2005).

Intracellular current clamp in hippocampal pyramidal neurons during hypothermia has shown blockade of action potentials in hippocampal neurons, either through a tetrodotoxin (TTX)–sensitive depolarization or hyperpolarization, both associated with action potential firing block and increased membrane input resistance (Payton et al., 1969; Thompson et al., 1985; Volgushev et al., 2000; Motamedi et al., 2012). These findings suggest closure of a tonically active inward current that would normally be activated by warmer temperatures possibly involving transient receptor potential (TRP) channels such as TRPV4, which is expressed in hippocampal neurons and is activated by temperatures >30°C (Shibasaki et al., 2007). However, a selective TRPV4 agonist and nonspecific agonists and antagonists of several other TRP channels failed to induce currents during hypothermia, making involvement of TRP channels less likely (Motamedi et al., 2012).

The sudden disruption of synchronized epileptiform discharges by cooling in slice models raises the possibility of interference with network synchrony. This preferential effect of hypothermia on synchrony between CA1 and CA3 pyramidal neurons but not on action potential firing in interneurons (Fig. 3) could be due to different synaptic afferents to pyramidal cells and interneurons, different intrinsic action potential firing mechanisms in these neurons, or different thresholds in voltage-gated channels (Motamedi et al., 2012). Hippocampal interneurons and pyramidal neurons express two different subunits of voltage-gated Na2+ channels, that is, NaV1.1 and NaV1.6, respectively. Mutations in the latter have been linked to generalized epilepsy with febrile seizures plus (Catterall et al., 2010). Gamma oscillations (30–120 Hz) are believed to reflect neuronal network synchrony through γ-aminobutyric acid (GABA)Aergic depolarizations as are seen in preictal transition in slice models of epilepsy (Kohling et al., 2000), and brief high-frequency GABAergic discharges can result in seizures due to synchronization of neural networks (Jiruska et al., 2010). Cooling of rat brain slices (from 34 to 21°C) has been shown to reversibly terminate gamma oscillations and block transition to ictal discharges without affecting normal synaptic transmission (Javedan et al., 2002). Therefore, a decrease in fast interneuron synchronous discharges during hypothermia may disrupt network synchrony thereby terminating synchronous ictal discharges. It is also possible that persistence of action potential firing in interneurons during hypothermia results in seizure suppression through tonic GABA release, an effect similar to the benzodiazepine effect due to prolongation of the decay of phasic inhibitory synaptic current. An opposite effect has been shown with hyperthermia (Qu & Leung, 2009). It is possible that subtypes of interneurons have different temperature sensitivities, with specific subtypes modulating GABAergic mechanisms underlying the effects of hypothermia. The differential effect between the pyramidal cells and interneurons may also relate to the expression of distinct K+ channels between these two types of neurons (Coetzee et al., 1999; Volgushev et al., 2000).

Safety Concerns

Based on the available clinical studies, hypothermia seems to be reasonably safe, particularly at mild to moderate levels. Coagulopathy, venous thromboembolism, or ventricular fibrillation have been reported at temperatures <30°C (Benson et al., 1959; Corry et al., 2008). Lomber et al. (1999), implanted multiple cooling devices (cryoloop) in a single cat or monkey for focal cooling of the cortex or midbrain in order to inactivate and study different neural circuits. Intermittent cooling, to as low as 0°C, for about 3 years proved to be safe with no structural or functional damage. In vivo experiments in rats and cats using Peltier device, including daily 2-h long sessions for 10 months, resulted in minimal gliosis near the area of cortical contact with the device. However, these changes also were seen in sham-operated rat cortex and may be a physiologic response to a foreign body and not necessarily to the cooling (Yang et al., 2006; Rothman, 2009). Histopathologic examination of these animals using Nissl and TUNEL staining has not revealed any abnormalities after 2-h long exposure to as low as 5°C. There was no neuronal loss or activation of apoptotic pathways, and dendritic beading and spine loss were reversible (Yang & Rothman, 2001; Yang et al., 2006). In slice model tissues remain viable even after prolonged and extreme hypothermia (Motamedi et al., 2006). However, in addition to the concerns for long-term tissue damage, there is a need to define the safety limit of hypothermia before it might cause even temporary dysfunction, for example, language or memory dysfunction with temporal lobe or a loss of motor function with frontal lobe hypothermia.

Brain tissue has low heat conductivity, in part due to its high fat content, and single- or two element probes may not be able to rapidly cool the target area without causing tissue damage. A multiprobe device might solve this problem. Using modeling and computer simulations, Osorio et al. (2009) reported probe density (25 vs. 7 elements) in a cooling device to be likely a more significant factor than other parameters but this requires further investigation. The main endpoints of this study were tissue safety and simplicity of engineering. However, the large number and high density of cooling elements as well as their size may raise safety and practicality concerns.

Conclusions and Future Directions

Hypothermia can terminate epileptiform discharges both in vitro and in vivo. It is becoming a standard therapy in some clinical conditions and is used in others with encouraging results. Further data including clinical trials are necessary to assess the efficacy of hypothermia as a viable chronic treatment option for refractory epilepsy and other neurologic disorders. Due to study designs, the available data primarily indicate an anticonvulsant effect of hypothermia, and there are only limited data suggesting an antiepileptogenic effect (Atkins et al., 2010; D'Ambrosio et al., 2012). If further studies confirm findings of efficacy and limited morbidity, hypothermia could become a useful anticonvulsant and antiepileptogenic agent.

Clinical use of hypothermia most likely would require an implanted cooling device. Currently, there are sparse but promising data on implantable devices; however, so far their use has been limited to experiments in animals (Yang et al., 2002; Fujii et al., 2012). An effective method of power storage and delivery might require a combination of small batteries within the implanted device, and transcranial recharging or power induction to facilitate ongoing treatment delivery. A device would have to be sufficiently powered to cool the target tissue quickly and efficiently with a safe and efficient method for heat transfer and dissipation. For example, with a Peltier device, cooling of the side that is in contact with brain tissue results in heat production on the opposite side, and the heat needs to be conducted away from the device, for example, through metal wires or fluid circulation. Either of these may impose new technical challenges. It is not presently clear if such a cooling device would be more or less power intensive than the currently available implantable brain stimulators. However, given that those devices have been only modestly effective in controlling refractory epilepsy in human trials (Kossoff et al., 2004; Fisher et al., 2010; Morrell, 2011), greater power consumption for a cooling device might be acceptable, if the device proved to better control seizures. Changes in seizure frequency and intensity after applying a brief period of hypothermia, and the effects of various rates of cooling and rewarming, also would need further in vivo studies. Implantable stimulators currently use one of two treatment paradigms. With one paradigm, therapy occurs intermittently at set intervals and for set durations; with the other, it occurs only when the device detects possible epileptogenic activity. We think it likely that cooling devices would be of the latter type; but the former might also be effective. If detections occurred much more frequently than clinical seizures, this could result in multiple unnecessary cooling episodes, with unknown tissue effects.

As with stimulation, it is not necessarily clear where the target tissue is. If stimulation is not completely effective because it occurs in the wrong place, the same would probably occur with cooling. Furthermore, it is not clear what volume of tissue constitutes a “critical mass” to be cooled and over what “time window” cooling needs to be performed, in order to prevent a seizure from spreading and becoming symptomatic. Although the safety of prolonged cooling is unknown, brief responsive and rapid cooling may be as effective and safer for a prospective therapeutic device. The fastest rapidity of cooling (2–5°C/s), as reported in slice models, was achieved through rapid switch between perfusion solutions in experimental chamber, a method that may not be feasible in humans through implanted devices. However, if technically possible, a smaller magnitude drop in brain temperature might be effective enough, which would be of importance given potential technical challenges in removing the heat generated by an intracranial cooling device. Therefore, there is a need to further study the speed and duration of cooling in clinical setting to produce beneficial effects, and also to further study how to overcome any technologic obstacles to achieving this goal with an implanted device.

A clinically useful device would have to be small enough to be implantable, but large enough to cover the target area. There would have to be a reliable method for target determination and device placement, especially with respect to less accessible areas of the brain, and an optimized method for controlling the depth, radius, and duration of cooling. In one in vivo primate experiment the effect of cooling was restricted to 1.5–2.5 mm distance from the cooling device, even at 1°C (Lomber et al., 1999). As with intracerebral stimulators, the idea of direct treatment to the epileptogenic region presumes that this region can actually be found, so that the right treatment can be delivered to the right place. Furthermore, current experimental data may apply best to seizures with focal or regional onset; the efficacy of hypothermia in human primary generalized epilepsy, where there is no particular “focus” to target, is unknown. It is also unclear if the effects of cooling delivered to human brain would be limited to the area actually cooled or if cooling would have remote effects. Finally, it is possible that a combination of methods would work best, for example, cooling plus stimulation or cooling plus medication.

These considerations emphasize the importance of studying the exact cellular and molecular mechanism(s) underlying the effects of hypothermia. For example, if hypothermia affects a particular subtype of interneurons, a medication or genetic manipulation targeting this subtype might be as or more effective in blocking or reversing the processes that lead to clinical seizures and chronic epilepsy.


Dr. Lesser is co-inventor for three patents related to cooling (USPTO # 6,248,126; 6,882,881; 7,228,171). Dr. Motamedi is co-inventor for one of these patents (USPTO # 6,882,881. Dr. Vicini has no disclosures to report. 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.