Expert Review of Neurotherapeutics
ISSN: 1473-7175 (Print) 1744-8360 (Online) Journal homepage: http://www.tandfonline.com/loi/iern20
Responsive neurostimulation in epilepsy Sofie Carrette, Paul Boon, Mathieu Sprengers, Robrecht Raedt & Kristl Vonck To cite this article: Sofie Carrette, Paul Boon, Mathieu Sprengers, Robrecht Raedt & Kristl Vonck (2015): Responsive neurostimulation in epilepsy, Expert Review of Neurotherapeutics, DOI: 10.1586/14737175.2015.1113875 To link to this article: http://dx.doi.org/10.1586/14737175.2015.1113875
Published online: 18 Nov 2015.
Submit your article to this journal
Article views: 10
View related articles
View Crossmark data
Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=iern20 Download by: [Monash University Library]
Date: 22 November 2015, At: 14:40
Review
Responsive neurostimulation in epilepsy
Downloaded by [Monash University Library] at 14:40 22 November 2015
Expert Rev. Neurother. Early online, 1–10 (2015)
Sofie Carrette *, Paul Boon, Mathieu Sprengers, Robrecht Raedt and Kristl Vonck Laboratory for Clinical and Experimental Neurophysiology, Neurobiology and Neuropsychology (LCEN3), Ghent University, Department of Neurology, Ghent University Hospital, Institute for Neuroscience, Ghent, Belgium *Author for correspondence: Tel.: +32 9 332 53 08 [email protected]
Various neurostimulation modalities have emerged in the field of epilepsy. Despite the fact that delivery of an electrical current to the hyperexcitable epileptic brain might, at first, seem contradictory, neurostimulation has become an established therapeutic option with a promising efficacy and adverse effects profile. In “responsive” neurostimulation the strategy is to interfere as early as possible with the accumulation of seizure activity to prematurely abort or even prevent an upcoming seizure. The design of technology required for responsive stimulation is more challenging compared with devices for open-loop neurostimulation. The achievement of therapeutic success is dependent on adequate sensing and stimulation algorithms and a fast coupling between both. The benefits of delivering current only at the time of an approaching seizure merit further investigation. Current experience with responsive neurostimulation in epilepsy is still limited, but seems promising. KEYWORDS: epilepsy ● neurostimulation ● closed-loop ● responsive ● deep brain stimulation (DBS) ● vagus nerve stimulation (VNS) ● responsive neurostimulation system (RNS) ● optogenetics ● drug delivery system
Neurostimulation for the treatment of epilepsy
A considerable number of patients with refractory epilepsy are not amenable to resective surgery, either due to multifocality of the epilepsy or due to the localization of the epileptic focus within functional cortex. For these patients, as well for those in whom resective surgery has failed, further options for seizure control are limited. Owing to the debilitating nature of recurrent seizures on the quality of life (QoL), new therapeutic strategies are warranted. To date, neurostimulation is more and more gaining its place in the therapeutic arsenal of the epileptologist. Already in the nineteenth century ‘counterirritation’ was suggested as a potential strategy to abate epileptic activity.[1–3] In the past decades an enormous amount of research has been devoted to the development and technological optimization of different neurostimulation techniques. Vagus nerve stimulation (VNS), deep brain stimulation of the anterior nucleus of the thalamus (ANT-DBS) and the responsive neurostimulation (RNS) system are the three neurostimulation techniques that have been granted Food and Drug Administration (FDA) approval for the treatment of refractory epilepsy. Several other modalities are being investigated in a preclinical and clinical setting.[4]
www.tandfonline.com
10.1586/14737175.2015.1113875
Stimulation of nerve tissue, albeit intra- or extra-cranially, can occur in a fixed, scheduled manner, also called open-loop stimulation, in which the electrical pulses are administered at certain pre-programmed given time points, either continuously or intermittently in cycles and independent of ongoing and variable neuronal activity. More novel is closed-loop or responsive neurostimulation, in which electrical pulses are delivered upon detection of seizure activity. Early detection of ictal electro-encephalographic (EEG) activity requires intracranial electrodes exactly at the site of seizure onset. However, seizure activity may also be recorded outside the brain, which allows a more accessible read-out, for instance by means of abnormal behavior, seizure-related muscle activity (EMG), or seizure-related cardiac changes (ECG). The latter is currently being investigated as a responsive vagus nerve stimulation approach (see Section 3.2.2). The rationale behind closed-loop stimulation is that the delivery of electrical stimuli upon detection of ictal activity is able to disrupt ongoing seizure activity. In an optimal scenario ‘sensing and pacing’ occurs before any clinical manifestation, so that clinically seizures are not only aborted, but even prevented. In this respect, the time window which can occur between intracranial encephalographic ictal onset and clinical
© 2015 Taylor & Francis
ISSN 1473-7175
1
Downloaded by [Monash University Library] at 14:40 22 November 2015
Review
Carrette et al.
seizure onset (e.g., temporal lobe seizures) may favor intracranial closed-loop stimulation systems compared with closed-loop treatments that measure ictal activity outside the brain. Closed-loop stimulation is technically more challenging, but carries several advantages over open-loop systems. Delivering electrical current only when ‘necessary’ lowers the total daily dose of delivered current, prolonging battery life and reducing adverse effects.[5] The absence of stimulation during normal brain activity prevents disruption of normal brain function in eloquent areas.[6] It has been postulated that closed-loop stimulation would offer higher efficacy due to the ‘on-demand’ stimulation strategy.[5] However, it might also be possible that for long-term neuromodulatory effects to develop, assuming that these occur, ‘a certain amount of stimulation’ is required. A consequence of the economic delivery of current in closed-loop systems would be that such effects occur at a slower pace compared with open-loop stimulation. It should be noted, however, that to this day these assumptions remain speculative. An additional challenge compared with open-loop stimulation is that, apart from an effective stimulation paradigm, the success of closed-loop stimulation is highly dependent on the quality of the seizure detection or prediction algorithm and its coupling to the stimulator. The algorithm should be highly sensitive, specific and fast. Technical optimization of the detection algorithm should strive for individualized ‘clever’ systems that are flexible and individually adaptable to the patient’s specific ictal pattern. Over the years of research and development, responsive neurostimulation has become a valid therapeutic option for the treatment of refractory epilepsy. Moreover, the development of reliable detection algorithms opens doors to the availability of seizure warning and alarm systems, addressing the most devastating characteristic of seizures, namely, their paroxysmal and unpredictable occurrence. This paper will review the different techniques that implement closed-loop neurostimulation. The RNS system (Neuropace, Inc.) has been studied extensively and has proven efficacy for the treatment of refractory epilepsy leading to its recent FDA approval (November 2013).[7] The AspireSR VNS system is a vagus nerve stimulator in which a cardiac-based seizure-detection (CBSD) algorithm was incorporated for additional on-demand stimulation. Very recently this device was also granted FDA approval (May 2015), but results from the efficacy study are pending. The limited preclinical experience with closed-loop optogenetic stimulation will be touched, as well as some potential future developments.
Responsive neurostimulation (RNS) system Background
The initial idea for a closed-loop DBS set-up was supported by experiments demonstrating that brief bursts of pulse stimulation had an abortive effect on electrically induced after discharges in patients undergoing invasive video-EEG monitoring.[8] Early proof-of-concept trials provided evidences for feasibility and safety of closed-loop stimulation of the epileptic focus and described a seizure reducing potential in focal epilepsy.[9–12] Further advancement in technology ultimately led to the design of the implantable RNS system (NeuroPace, Mountain View, CA, USA). 2
Technical aspects
The RNS system is a cranially implantable responsive neurostimulator which can be connected to one or two electrodes that are surgically placed in the brain according to the seizure focus (thus limited to patients with one or two seizure foci). The implanted electrodes can either be a 4-contact depth lead or a 4-contact cortical strip and serve a dual function: continuous monitoring of electrophysiological activity and delivery of electrical pulses.[7] As a result, the RNS system is a form of ictal onset zone stimulation, in contrast to ANT-DBS and VNS, which abort seizure activity via a ‘network approach’, by modulating certain ‘gating’ structures of brain activity, such as thalamic or brain stem nuclei respectively. Three detection tools are used for real-time detection of seizure activity.[13] The bandpass tool is used to detect spikes and rhythmic activity within specific frequency ranges by analyzing the amplitude and duration of recorded ‘half-waves’ which are segments of the EEG signal partitioned at local minima and maxima. For the line-length and the area tool, a short-term sliding window average (128 ms–4 s) is compared with a long-term sliding window average (4 s–16 min) to identify changes in signal amplitude and frequency (line-length algorithm) and overall signal energy (area feature). Detection occurs when the short-term measurements crosses an absolute or relative threshold derived from the long-term measurement. Using these detection tools, the algorithm is highly efficient (requiring low computational power) and able to detect electrographic events within a fraction of a second. The detection parameters are configurable and programmed by the physician to optimize sensitivity, specificity and latency of the detection.[13] The system automatically stored detected EEG epochs, allowing read-out during follow up. Note that the RNS system is not a detection system for clinical seizures. As soon as relevant EEG changes are identified stimulation is initiated, irrespective of whether or not these changes would have progressed to an electrographic (or clinical) seizure. Stimulation settings are adjustable by the physician. Pulses have a biphasic waveform and frequency can range from 1 to 333 Hz (standard 100–200 Hz), amplitude from 0.5 to 12 mA (standard 1.5–3 mA) and pulse width from 40 to 1000 µs (standard 160 µs). [13] Stimulation is mostly delivered in bursts of 100–200 ms between any combination of electrodes or between an electrode and the stimulator case. A retrospective analysis on stimulation data showed that closedloop stimulation occurred 600–2000 times a day in the majority of patients, resulting in a cumulative total of 80% of adult epilepsy patients.[32–35] Moreover, these cardiac changes may even precede electrical or clinical onset with several seconds,[36–38] resulting in a considerable amount of seizures that may be detected and prematurely aborted. The new type of VNS device, the AspireSR, offers closed-loop stimulation based on this concept. The device provides conventional open-loop stimulation, but has an incorporated CBSD algorithm that automates the delivery of additional stimulation upon seizure detection. The AspireSR was granted CE mark in Expert Rev. Neurother.
Responsive neurostimulation in epilepsy
Europe in February 2014 and FDA approval in May 2015. The results of the study assessing its add-on benefit in adult refractory epilepsy patients are pending.
Downloaded by [Monash University Library] at 14:40 22 November 2015
Technical aspects
The AspireSR pulse generator’s dimensions are 52 × 52 × 6.9 mm with a weight of ~25 g, which is 43% larger and 36% heavier than the Demipulse, which is the currently most widely used generator. These dimensions imply a larger skin incision. The actual surgical technique and the initiation and titration of the AspireSR VNS therapy does not differ much from the conventional dosing procedure, although an intra-operative heartbeat detection evaluation is recommended. The price of the new device is somewhat higher than for all other VNS models. The CBSD algorithm tracks relative heart rate as the ratio of two moving averages of instantaneous heart rate samples, a short window reflecting the current heart rate, divided by a long window. [39] The long window heart rate trend represents the background rate, which may change slowly over time based on the patient’s activity level. If the relative heart rate exceeds a preset threshold for a certain period of time, a tachycardia event is detected and stimulation is initiated, using conventional settings (30–60 s, 250–500 µs, 0–3.5 mA, 20–30 Hz). Because the threshold is based on the relative heart rate that takes into account the background rate, it automatically adjusts to the patient’s underlying activity as a source of non-ictal tachycardia, minimizing the false positive rate. The sensitivity of the detection algorithm is adjustable by the physician to six preset thresholds. The heartbeat (HB) sensitivity level 1–6 detects increases in heartbeat rate above baseline of 70–20% respectively (HB1 = 70%; HB2 = 60%; HB3 = 50%; HB4 = 40%; HB5 = 30%; HB6 = 20%). Clinical trials
The new AspireSR generator offers a unique technical improvement over the previous VNS devices. However, whether the CBSD feature will provide an additional benefit with regard to seizure control and QoL of refractory epilepsy patients is yet to be determined. The E36 trial is a prospective, unblinded multicenter study to evaluate the performance of the CBSD algorithm in 31 refractory epilepsy patients.[39] Patients were implanted with the AspireSR device and titrated according to standard VNS stimulation up to at least 0.75 mA. They were then admitted for an Epilepsy Monitoring Unit (EMU) stay up to 5 days, during which continuous EEG and ECG monitoring took place to identify seizures and collect heart rate data. During the EMU stay, the devices were randomized to one of three seizure detection algorithms and the normal mode setting (open-loop) was disabled to optimally evaluate seizure detection performance by means of sensitivity, potential false positive rate and latency (time between seizure detection and annotated electrical or clinical seizure onset). Results showed that the algorithm was able to detect >80% of seizures associated with tachycardia at a range of programmable settings, meeting up with the primary endpoint of the study. There is a low rate of potential false positives and the more sensitive the detection settings, the closer the detection and stimulation occurred relative to the seizure onset. In some cases stimulation www.tandfonline.com
Review
even preceded seizure onset, which is likely optimal for the purpose of acute stimulation. Seizure termination occurred in 40% of focal seizures (within ±2 min), which is consistent with the literature on magnet mode VNS therapy[24] and seizure duration was shorter compared with historical data. The impact of automated acute stimulation near seizure onset on seizure severity, intensity and post-ictal duration is being assessed in the E36 continuation study and the US IDE Study, E37. A recent paper by Hampel et al. investigated the effects of cardiacbased closed-loop VNS stimulation on seizure duration under controlled conditions in one patient.[40] A 29-year-old male patient with refractory epilepsy was implanted with the AspireSR device and underwent video-EEG monitoring (VEM) for 68 h with no changes in AEDs. On the first day following implantation the open-loop VNS duty cycle was inactive and the patient received closed-loop sham-stimulation (stimulus output 0.125 mA, pulse width of 500 µs, frequency 30 Hz during 60 s) as the control condition, while seizures were monitored (22 h VEM). After the 22 h, both open- and closed-loop stimulation were ramped up to 2 mA on days 2–4 without monitoring. Finally, on day 5 and 6 the patient underwent another 46 h of VEM, with closed-loop stimulation output at 2 mA (stimulus duration 60 s, pulse width 250 µs) and open-loop stimulation output at 0 mA (with a stimulus duration of 7 s, pulse width 500 µs and off-time of 180 min) as the test condition. The heartbeat sensitivity threshold of the detection algorithm was set at 50%. Seizure duration was determined by two independent epileptologists based on clinical signs. Twelve stereotypical seizures were recorded during VEM (six during sham- and six during the active stimulation). Meanwhile the CBSD algorithm detected a total of 160 events, of which 139 were classified as epileptic and 21 as nonepileptic. Eleven out of the 12 actual epileptic seizures were correctly identified as assessed by video-EEG-monitoring, resulting in a sensitivity of 92% and a specificity of 13.5%. Data analysis revealed a mean time delay between the clinical seizure onset and detection by the device of 7.4 ± 5 s. Closed-loop stimulation by the AspireSR significantly reduced the total seizure duration by 20% from 33.2 ± 4.8 to 26.5 ± 5.0 s (p = 0.039) and the remaining seizure duration after initiation of stimulation by 40% from 27.8 ± 4.3 to 16.2 ± 3.2 s (p = 0.001). No seizures were terminated immediately by acute stimulation. This case study is the first to illustrate the benefit of the incorporated CBSD-algorithm in the new VNS device with regard to seizure duration in controlled conditions. Larger studies, however, are required to confirm these findings. In conclusion, the AspireSR device offers closed-loop VNS therapy, providing additional acute seizure intervention on top of regular anti-epileptic VNS effect, potentially increasing overall therapeutic efficacy of conventional VNS therapy. Although the add-on benefit remains unclear, the consideration of the AspireSR in patients with documented ictal tachycardia is recommended to provide a substantial number of patients for later seizure outcome analysis. Closed-loop optogenetics Background
Although decades of research on electrical stimulation of nerve tissue has established its role as a treatment for refractory epilepsy, 5
Review
Carrette et al.
its inherent cell-type aspecificity is a potential limiting factor with regards to efficacy. Optogenetics is a new highly advanced technique that addresses this limitation and in contrast to electrical stimulation, is characterized by extreme cell-specificity, selectively activating or inhibiting a particular subset of neurons. Although still in its infancy, there are several groups exploring optogenetics for the treatment of refractory epilepsy.[41,42]
Downloaded by [Monash University Library] at 14:40 22 November 2015
Technical aspects
The cell-specificity in optogenetics is achieved by inducing expression of light sensitive ion channels (opsins) under control of promotors that are only expressed in specific cell types. Emission of particular wavelengths of light with an implanted optrode in the vicinity of the target neurons induces a conformational change of the expressed opsin, causing either depolarization of hyperpolarization of the neurons, enabling selective activation or inhibition of the neuronal circuit responsible for seizure initiation or propagation, allowing to terminate seizures. Because of control over the opsin-expressing cell population, the type of opsin expressed and the time of stimulation, optogenetics is a highly specific technique with exquisite spatiotemporal resolution. As for conventional DBS, the site of stimulation can either be at the ictal onset zone or in a remote gating structure of the epileptogenic network. Also, light may be emitted in an intermittent, scheduled open-loop or an on demand closed-loop set-up. Preclinical trials
In epilepsy research, optogenetics has only been explored in preclinical research with current efforts toward more sophisticated nonhuman primate models (for a detailed review, see Bentley et al.[41]). In an in vitro slice set-up Tonnesen et al. found that an optogenetic approach using halorhodopsin (light sensitive Cl– channel) was able to hyperpolarize hippocampal and cortical glutamatergic neurons, allowing inhibition of excessive hyperexcitability and epileptiform activity.[43] Four different in vivo rodent models of epilepsy have been used to assess the effect of optogenetics on epileptic activity and seizures. Wykes et al. assessed the rat tetanus toxin injection model for focal neocortical epilepsy, in which pyramidal cells in the epileptic focus were transfected with halorhodopsin.[44] Prior to illumination these neurons showed intrinsic hyperexcitability resulting in seizure generation. Optogenetic inhibition of these neurons allowed decrease of epileptiform activity and was sufficient to attenuate epileptic seizures. In a rat model for cortical stroke, Paz et al. demonstrated that the thalamus, a structure that is remote from, but connected to, the injured cortex, plays an important role in seizure maintenance.[45] Using optogenetics, the authors interfered with the thalamocortical circuit, inducing inhibition of the hyperexcitable thalamocortical neurons. A closed-loop stimulation set-up in response to epileptic activity measured in the cortex was able to immediately induce seizure arrest. Krook-Magnusen et al. performed several optogentic experiments in a mouse model of temporal lobe epilepsy (TLE).[46– 48] In an in vivo, real-time, closed-loop experiment, they demonstrated either optogenetic inhibition of hippocampal excitatory principal cells, or activation of a subpopulation of GABAergic cells 6
representing 10 years. Epilepsy & Behavior: E&B. 2011;20(3):478–483.
24.
Fisher RS, Eggleston KS, Wright CW. Vagus nerve stimulation magnet activation for seizures: a critical review. Acta Neurol Scand. 2015;131(1):1–8.
25.
Murphy JV, Torkelson R, Dowler I, et al. Vagal nerve stimulation in refractory epilepsy: the first 100 patients receiving vagal nerve stimulation at a pediatric epilepsy center. Arch Pediatr Adolesc Med. 2003;157(6):560–564.
9
Review
Downloaded by [Monash University Library] at 14:40 22 November 2015
26.
Carrette et al.
Boon P, Vonck K, Van Walleghem P, et al. Programmed and magnet-induced vagus nerve stimulation for refractory epilepsy. J Clin Neurophysiol: Official Publication Am Electroencephalographic Society. 2001;18(5):402–407.
27.
Morris GL 3rd. A retrospective analysis of the effects of magnet-activated stimulation in conjunction with vagus nerve stimulation therapy. Epilepsy & Behavior: E&B. 2003;4(6):740–745.
28.
Jeppesen J, Beniczky S, FuglsangFrederiksen A, et al. Detection of epileptic-seizures by means of power spectrum analysis of heart rate variability: a pilot study. Technol Health Care: Official J Eur Soc Eng Med. 2010;18(6):417–426.
29.
30.
31.
36.
Shoeb A, Pang T, Guttag J, et al. Noninvasive computerized system for automatically initiating vagus nerve stimulation following patient-specific detection of seizures or epileptiform discharges. Int J Neural Syst. 2009;19(3):157–172. Van Elmpt WJ, Nijsen TM, Griep PA, et al. A model of heart rate changes to detect seizures in severe epilepsy. Seizure: J Br Epilep Assoc. 2006;15(6):366–375. Eggleston KS, Olin BD, Fisher RS. Ictal tachycardia: the head-heart connection. Seizure. 2014;23(7):496–505.
33.
Sevcencu C, Struijk JJ. Autonomic alterations and cardiac changes in epilepsy. Epilepsia. 2010;51(5):725–737.
34.
Jansen K, Lagae L. Cardiac changes in epilepsy. Seizure-Eur J Epilep. 2010;19 (8):455–460.
35.
Lotufo PA, Valiengo L, Bensenor IM, et al. A systematic review and meta-analysis of heart rate variability in epilepsy
Jansen K, Varon C, Van Huffel S, et al. Peri-ictal ECG changes in childhood epilepsy: implications for detection systems. Epilep Behav. 2013;29(1):72–76.
37.
Leutmezer F, Schernthaner C, Lurger S, et al. Electrocardiographic changes at the onset of epileptic seizures. Epilepsia. 2003;44(3):348–354.
38.
Zijlmans M, Flanagan D, Gotman J. Heart rate changes and ECG abnormalities during epileptic seizures: prevalence and definition of an objective clinical sign. Epilepsia. 2002;43(8):847–854.
39.
Boon P, Van Rijckevorsel K, Elger CE, et al. Vagus nerve stimulation triggered by cardiac-based seizure detection, a prospective multicenter study. Epilep Curr. 2014;14(Suppl.):21.
Osorio I. Automated seizure detection using EKG. Int J Neural Syst. 2014;24 (2):1450001.
32.
10
and antiepileptic drugs. Epilepsia. 2012;53(2):272–282.
40.
Hampel KG, Vatter H, Elger CE, et al. Cardiac-based vagus nerve stimulation reduced seizure duration in a patient with refractory epilepsy. Seizure: The J Br Epilep Assoc. 2015;26:81–85.
41.
Bentley JN, Chestek C, Stacey WC, et al. Optogenetics in epilepsy. Neurosurg Focus. 2013;34(6):E4.
●●
The authors present an overview of the development of optogenetics and review recent studies investigating optogenetic modification of circuits involved in epileptic seizures.
42.
Ritter LM, Golshani P, Takahashi K, et al. WONOEP appraisal: optogenetic tools to suppress seizures and explore the mechanisms of epileptogenesis. Epilepsia. 2014;55 (11):1693–1702.
43.
Tonnesen J, Sorensen AT, Deisseroth K, et al. Optogenetic control of epileptiform activity. Proc Natl Acad Sci USA. 2009;106(29):12162–12167.
44.
Wykes RC, Heeroma JH, Mantoan L, et al. Optogenetic and potassium channel gene therapy in a rodent model of focal neocortical epilepsy. Sci Transl Med. 2012;4(161):161ra52.
45.
Paz JT, Davidson TJ, Frechette ES, et al. Closed-loop optogenetic control of thalamus as a tool for interrupting seizures after cortical injury. Nat Neurosci. 2013;16(1):64–70.
46.
Krook-Magnuson E, Armstrong C, Oijala M, et al. On-demand optogenetic control of spontaneous seizures in temporal lobe epilepsy. Nat Commun. 2013;4:1376.
47.
Krook-Magnuson E, Armstrong C, Bui A, et al. In vivo evaluation of the dentate gate theory in epilepsy. J Physiol. 2015;593(10):2379–2388.
48.
Krook-Magnuson E, Szabo GG, Armstrong C, et al. Cerebellar directed optogenetic intervention inhibits spontaneous hippocampal seizures in a mouse model of temporal lobe epilepsy. Eneuro. 2014;1:1.
49.
Sukhotinsky I, Chan AM, Ahmed OJ, et al. Optogenetic delay of status epilepticus onset in an in vivo rodent epilepsy model. Plos One. 2013;8(4): e62013.
50.
Salam MT, Mirzaei M, Ly MS, et al. An implantable closedloop asynchronous drug delivery system for the treatment of refractory epilepsy. IEEE Trans Neural Syst Rehabilitation Eng: a Publ IEEE Eng Med Biol Soc. 2012;20 (4):432–442.
51.
Mangubat EZ, Kellogg RG, Harris TJ Jr., et al. On-demand pulsatile intracerebral delivery of carisbamate with closed-loop direct neurostimulation therapy in an electrically induced self-sustained focalonset epilepsy rat model. J Neurosurg. 2015;122(6):1283–1292.
Expert Rev. Neurother.
ISSN: 1473-7175 (Print) 1744-8360 (Online) Journal homepage: http://www.tandfonline.com/loi/iern20
Responsive neurostimulation in epilepsy Sofie Carrette, Paul Boon, Mathieu Sprengers, Robrecht Raedt & Kristl Vonck To cite this article: Sofie Carrette, Paul Boon, Mathieu Sprengers, Robrecht Raedt & Kristl Vonck (2015): Responsive neurostimulation in epilepsy, Expert Review of Neurotherapeutics, DOI: 10.1586/14737175.2015.1113875 To link to this article: http://dx.doi.org/10.1586/14737175.2015.1113875
Published online: 18 Nov 2015.
Submit your article to this journal
Article views: 10
View related articles
View Crossmark data
Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=iern20 Download by: [Monash University Library]
Date: 22 November 2015, At: 14:40
Review
Responsive neurostimulation in epilepsy
Downloaded by [Monash University Library] at 14:40 22 November 2015
Expert Rev. Neurother. Early online, 1–10 (2015)
Sofie Carrette *, Paul Boon, Mathieu Sprengers, Robrecht Raedt and Kristl Vonck Laboratory for Clinical and Experimental Neurophysiology, Neurobiology and Neuropsychology (LCEN3), Ghent University, Department of Neurology, Ghent University Hospital, Institute for Neuroscience, Ghent, Belgium *Author for correspondence: Tel.: +32 9 332 53 08 [email protected]
Various neurostimulation modalities have emerged in the field of epilepsy. Despite the fact that delivery of an electrical current to the hyperexcitable epileptic brain might, at first, seem contradictory, neurostimulation has become an established therapeutic option with a promising efficacy and adverse effects profile. In “responsive” neurostimulation the strategy is to interfere as early as possible with the accumulation of seizure activity to prematurely abort or even prevent an upcoming seizure. The design of technology required for responsive stimulation is more challenging compared with devices for open-loop neurostimulation. The achievement of therapeutic success is dependent on adequate sensing and stimulation algorithms and a fast coupling between both. The benefits of delivering current only at the time of an approaching seizure merit further investigation. Current experience with responsive neurostimulation in epilepsy is still limited, but seems promising. KEYWORDS: epilepsy ● neurostimulation ● closed-loop ● responsive ● deep brain stimulation (DBS) ● vagus nerve stimulation (VNS) ● responsive neurostimulation system (RNS) ● optogenetics ● drug delivery system
Neurostimulation for the treatment of epilepsy
A considerable number of patients with refractory epilepsy are not amenable to resective surgery, either due to multifocality of the epilepsy or due to the localization of the epileptic focus within functional cortex. For these patients, as well for those in whom resective surgery has failed, further options for seizure control are limited. Owing to the debilitating nature of recurrent seizures on the quality of life (QoL), new therapeutic strategies are warranted. To date, neurostimulation is more and more gaining its place in the therapeutic arsenal of the epileptologist. Already in the nineteenth century ‘counterirritation’ was suggested as a potential strategy to abate epileptic activity.[1–3] In the past decades an enormous amount of research has been devoted to the development and technological optimization of different neurostimulation techniques. Vagus nerve stimulation (VNS), deep brain stimulation of the anterior nucleus of the thalamus (ANT-DBS) and the responsive neurostimulation (RNS) system are the three neurostimulation techniques that have been granted Food and Drug Administration (FDA) approval for the treatment of refractory epilepsy. Several other modalities are being investigated in a preclinical and clinical setting.[4]
www.tandfonline.com
10.1586/14737175.2015.1113875
Stimulation of nerve tissue, albeit intra- or extra-cranially, can occur in a fixed, scheduled manner, also called open-loop stimulation, in which the electrical pulses are administered at certain pre-programmed given time points, either continuously or intermittently in cycles and independent of ongoing and variable neuronal activity. More novel is closed-loop or responsive neurostimulation, in which electrical pulses are delivered upon detection of seizure activity. Early detection of ictal electro-encephalographic (EEG) activity requires intracranial electrodes exactly at the site of seizure onset. However, seizure activity may also be recorded outside the brain, which allows a more accessible read-out, for instance by means of abnormal behavior, seizure-related muscle activity (EMG), or seizure-related cardiac changes (ECG). The latter is currently being investigated as a responsive vagus nerve stimulation approach (see Section 3.2.2). The rationale behind closed-loop stimulation is that the delivery of electrical stimuli upon detection of ictal activity is able to disrupt ongoing seizure activity. In an optimal scenario ‘sensing and pacing’ occurs before any clinical manifestation, so that clinically seizures are not only aborted, but even prevented. In this respect, the time window which can occur between intracranial encephalographic ictal onset and clinical
© 2015 Taylor & Francis
ISSN 1473-7175
1
Downloaded by [Monash University Library] at 14:40 22 November 2015
Review
Carrette et al.
seizure onset (e.g., temporal lobe seizures) may favor intracranial closed-loop stimulation systems compared with closed-loop treatments that measure ictal activity outside the brain. Closed-loop stimulation is technically more challenging, but carries several advantages over open-loop systems. Delivering electrical current only when ‘necessary’ lowers the total daily dose of delivered current, prolonging battery life and reducing adverse effects.[5] The absence of stimulation during normal brain activity prevents disruption of normal brain function in eloquent areas.[6] It has been postulated that closed-loop stimulation would offer higher efficacy due to the ‘on-demand’ stimulation strategy.[5] However, it might also be possible that for long-term neuromodulatory effects to develop, assuming that these occur, ‘a certain amount of stimulation’ is required. A consequence of the economic delivery of current in closed-loop systems would be that such effects occur at a slower pace compared with open-loop stimulation. It should be noted, however, that to this day these assumptions remain speculative. An additional challenge compared with open-loop stimulation is that, apart from an effective stimulation paradigm, the success of closed-loop stimulation is highly dependent on the quality of the seizure detection or prediction algorithm and its coupling to the stimulator. The algorithm should be highly sensitive, specific and fast. Technical optimization of the detection algorithm should strive for individualized ‘clever’ systems that are flexible and individually adaptable to the patient’s specific ictal pattern. Over the years of research and development, responsive neurostimulation has become a valid therapeutic option for the treatment of refractory epilepsy. Moreover, the development of reliable detection algorithms opens doors to the availability of seizure warning and alarm systems, addressing the most devastating characteristic of seizures, namely, their paroxysmal and unpredictable occurrence. This paper will review the different techniques that implement closed-loop neurostimulation. The RNS system (Neuropace, Inc.) has been studied extensively and has proven efficacy for the treatment of refractory epilepsy leading to its recent FDA approval (November 2013).[7] The AspireSR VNS system is a vagus nerve stimulator in which a cardiac-based seizure-detection (CBSD) algorithm was incorporated for additional on-demand stimulation. Very recently this device was also granted FDA approval (May 2015), but results from the efficacy study are pending. The limited preclinical experience with closed-loop optogenetic stimulation will be touched, as well as some potential future developments.
Responsive neurostimulation (RNS) system Background
The initial idea for a closed-loop DBS set-up was supported by experiments demonstrating that brief bursts of pulse stimulation had an abortive effect on electrically induced after discharges in patients undergoing invasive video-EEG monitoring.[8] Early proof-of-concept trials provided evidences for feasibility and safety of closed-loop stimulation of the epileptic focus and described a seizure reducing potential in focal epilepsy.[9–12] Further advancement in technology ultimately led to the design of the implantable RNS system (NeuroPace, Mountain View, CA, USA). 2
Technical aspects
The RNS system is a cranially implantable responsive neurostimulator which can be connected to one or two electrodes that are surgically placed in the brain according to the seizure focus (thus limited to patients with one or two seizure foci). The implanted electrodes can either be a 4-contact depth lead or a 4-contact cortical strip and serve a dual function: continuous monitoring of electrophysiological activity and delivery of electrical pulses.[7] As a result, the RNS system is a form of ictal onset zone stimulation, in contrast to ANT-DBS and VNS, which abort seizure activity via a ‘network approach’, by modulating certain ‘gating’ structures of brain activity, such as thalamic or brain stem nuclei respectively. Three detection tools are used for real-time detection of seizure activity.[13] The bandpass tool is used to detect spikes and rhythmic activity within specific frequency ranges by analyzing the amplitude and duration of recorded ‘half-waves’ which are segments of the EEG signal partitioned at local minima and maxima. For the line-length and the area tool, a short-term sliding window average (128 ms–4 s) is compared with a long-term sliding window average (4 s–16 min) to identify changes in signal amplitude and frequency (line-length algorithm) and overall signal energy (area feature). Detection occurs when the short-term measurements crosses an absolute or relative threshold derived from the long-term measurement. Using these detection tools, the algorithm is highly efficient (requiring low computational power) and able to detect electrographic events within a fraction of a second. The detection parameters are configurable and programmed by the physician to optimize sensitivity, specificity and latency of the detection.[13] The system automatically stored detected EEG epochs, allowing read-out during follow up. Note that the RNS system is not a detection system for clinical seizures. As soon as relevant EEG changes are identified stimulation is initiated, irrespective of whether or not these changes would have progressed to an electrographic (or clinical) seizure. Stimulation settings are adjustable by the physician. Pulses have a biphasic waveform and frequency can range from 1 to 333 Hz (standard 100–200 Hz), amplitude from 0.5 to 12 mA (standard 1.5–3 mA) and pulse width from 40 to 1000 µs (standard 160 µs). [13] Stimulation is mostly delivered in bursts of 100–200 ms between any combination of electrodes or between an electrode and the stimulator case. A retrospective analysis on stimulation data showed that closedloop stimulation occurred 600–2000 times a day in the majority of patients, resulting in a cumulative total of 80% of adult epilepsy patients.[32–35] Moreover, these cardiac changes may even precede electrical or clinical onset with several seconds,[36–38] resulting in a considerable amount of seizures that may be detected and prematurely aborted. The new type of VNS device, the AspireSR, offers closed-loop stimulation based on this concept. The device provides conventional open-loop stimulation, but has an incorporated CBSD algorithm that automates the delivery of additional stimulation upon seizure detection. The AspireSR was granted CE mark in Expert Rev. Neurother.
Responsive neurostimulation in epilepsy
Europe in February 2014 and FDA approval in May 2015. The results of the study assessing its add-on benefit in adult refractory epilepsy patients are pending.
Downloaded by [Monash University Library] at 14:40 22 November 2015
Technical aspects
The AspireSR pulse generator’s dimensions are 52 × 52 × 6.9 mm with a weight of ~25 g, which is 43% larger and 36% heavier than the Demipulse, which is the currently most widely used generator. These dimensions imply a larger skin incision. The actual surgical technique and the initiation and titration of the AspireSR VNS therapy does not differ much from the conventional dosing procedure, although an intra-operative heartbeat detection evaluation is recommended. The price of the new device is somewhat higher than for all other VNS models. The CBSD algorithm tracks relative heart rate as the ratio of two moving averages of instantaneous heart rate samples, a short window reflecting the current heart rate, divided by a long window. [39] The long window heart rate trend represents the background rate, which may change slowly over time based on the patient’s activity level. If the relative heart rate exceeds a preset threshold for a certain period of time, a tachycardia event is detected and stimulation is initiated, using conventional settings (30–60 s, 250–500 µs, 0–3.5 mA, 20–30 Hz). Because the threshold is based on the relative heart rate that takes into account the background rate, it automatically adjusts to the patient’s underlying activity as a source of non-ictal tachycardia, minimizing the false positive rate. The sensitivity of the detection algorithm is adjustable by the physician to six preset thresholds. The heartbeat (HB) sensitivity level 1–6 detects increases in heartbeat rate above baseline of 70–20% respectively (HB1 = 70%; HB2 = 60%; HB3 = 50%; HB4 = 40%; HB5 = 30%; HB6 = 20%). Clinical trials
The new AspireSR generator offers a unique technical improvement over the previous VNS devices. However, whether the CBSD feature will provide an additional benefit with regard to seizure control and QoL of refractory epilepsy patients is yet to be determined. The E36 trial is a prospective, unblinded multicenter study to evaluate the performance of the CBSD algorithm in 31 refractory epilepsy patients.[39] Patients were implanted with the AspireSR device and titrated according to standard VNS stimulation up to at least 0.75 mA. They were then admitted for an Epilepsy Monitoring Unit (EMU) stay up to 5 days, during which continuous EEG and ECG monitoring took place to identify seizures and collect heart rate data. During the EMU stay, the devices were randomized to one of three seizure detection algorithms and the normal mode setting (open-loop) was disabled to optimally evaluate seizure detection performance by means of sensitivity, potential false positive rate and latency (time between seizure detection and annotated electrical or clinical seizure onset). Results showed that the algorithm was able to detect >80% of seizures associated with tachycardia at a range of programmable settings, meeting up with the primary endpoint of the study. There is a low rate of potential false positives and the more sensitive the detection settings, the closer the detection and stimulation occurred relative to the seizure onset. In some cases stimulation www.tandfonline.com
Review
even preceded seizure onset, which is likely optimal for the purpose of acute stimulation. Seizure termination occurred in 40% of focal seizures (within ±2 min), which is consistent with the literature on magnet mode VNS therapy[24] and seizure duration was shorter compared with historical data. The impact of automated acute stimulation near seizure onset on seizure severity, intensity and post-ictal duration is being assessed in the E36 continuation study and the US IDE Study, E37. A recent paper by Hampel et al. investigated the effects of cardiacbased closed-loop VNS stimulation on seizure duration under controlled conditions in one patient.[40] A 29-year-old male patient with refractory epilepsy was implanted with the AspireSR device and underwent video-EEG monitoring (VEM) for 68 h with no changes in AEDs. On the first day following implantation the open-loop VNS duty cycle was inactive and the patient received closed-loop sham-stimulation (stimulus output 0.125 mA, pulse width of 500 µs, frequency 30 Hz during 60 s) as the control condition, while seizures were monitored (22 h VEM). After the 22 h, both open- and closed-loop stimulation were ramped up to 2 mA on days 2–4 without monitoring. Finally, on day 5 and 6 the patient underwent another 46 h of VEM, with closed-loop stimulation output at 2 mA (stimulus duration 60 s, pulse width 250 µs) and open-loop stimulation output at 0 mA (with a stimulus duration of 7 s, pulse width 500 µs and off-time of 180 min) as the test condition. The heartbeat sensitivity threshold of the detection algorithm was set at 50%. Seizure duration was determined by two independent epileptologists based on clinical signs. Twelve stereotypical seizures were recorded during VEM (six during sham- and six during the active stimulation). Meanwhile the CBSD algorithm detected a total of 160 events, of which 139 were classified as epileptic and 21 as nonepileptic. Eleven out of the 12 actual epileptic seizures were correctly identified as assessed by video-EEG-monitoring, resulting in a sensitivity of 92% and a specificity of 13.5%. Data analysis revealed a mean time delay between the clinical seizure onset and detection by the device of 7.4 ± 5 s. Closed-loop stimulation by the AspireSR significantly reduced the total seizure duration by 20% from 33.2 ± 4.8 to 26.5 ± 5.0 s (p = 0.039) and the remaining seizure duration after initiation of stimulation by 40% from 27.8 ± 4.3 to 16.2 ± 3.2 s (p = 0.001). No seizures were terminated immediately by acute stimulation. This case study is the first to illustrate the benefit of the incorporated CBSD-algorithm in the new VNS device with regard to seizure duration in controlled conditions. Larger studies, however, are required to confirm these findings. In conclusion, the AspireSR device offers closed-loop VNS therapy, providing additional acute seizure intervention on top of regular anti-epileptic VNS effect, potentially increasing overall therapeutic efficacy of conventional VNS therapy. Although the add-on benefit remains unclear, the consideration of the AspireSR in patients with documented ictal tachycardia is recommended to provide a substantial number of patients for later seizure outcome analysis. Closed-loop optogenetics Background
Although decades of research on electrical stimulation of nerve tissue has established its role as a treatment for refractory epilepsy, 5
Review
Carrette et al.
its inherent cell-type aspecificity is a potential limiting factor with regards to efficacy. Optogenetics is a new highly advanced technique that addresses this limitation and in contrast to electrical stimulation, is characterized by extreme cell-specificity, selectively activating or inhibiting a particular subset of neurons. Although still in its infancy, there are several groups exploring optogenetics for the treatment of refractory epilepsy.[41,42]
Downloaded by [Monash University Library] at 14:40 22 November 2015
Technical aspects
The cell-specificity in optogenetics is achieved by inducing expression of light sensitive ion channels (opsins) under control of promotors that are only expressed in specific cell types. Emission of particular wavelengths of light with an implanted optrode in the vicinity of the target neurons induces a conformational change of the expressed opsin, causing either depolarization of hyperpolarization of the neurons, enabling selective activation or inhibition of the neuronal circuit responsible for seizure initiation or propagation, allowing to terminate seizures. Because of control over the opsin-expressing cell population, the type of opsin expressed and the time of stimulation, optogenetics is a highly specific technique with exquisite spatiotemporal resolution. As for conventional DBS, the site of stimulation can either be at the ictal onset zone or in a remote gating structure of the epileptogenic network. Also, light may be emitted in an intermittent, scheduled open-loop or an on demand closed-loop set-up. Preclinical trials
In epilepsy research, optogenetics has only been explored in preclinical research with current efforts toward more sophisticated nonhuman primate models (for a detailed review, see Bentley et al.[41]). In an in vitro slice set-up Tonnesen et al. found that an optogenetic approach using halorhodopsin (light sensitive Cl– channel) was able to hyperpolarize hippocampal and cortical glutamatergic neurons, allowing inhibition of excessive hyperexcitability and epileptiform activity.[43] Four different in vivo rodent models of epilepsy have been used to assess the effect of optogenetics on epileptic activity and seizures. Wykes et al. assessed the rat tetanus toxin injection model for focal neocortical epilepsy, in which pyramidal cells in the epileptic focus were transfected with halorhodopsin.[44] Prior to illumination these neurons showed intrinsic hyperexcitability resulting in seizure generation. Optogenetic inhibition of these neurons allowed decrease of epileptiform activity and was sufficient to attenuate epileptic seizures. In a rat model for cortical stroke, Paz et al. demonstrated that the thalamus, a structure that is remote from, but connected to, the injured cortex, plays an important role in seizure maintenance.[45] Using optogenetics, the authors interfered with the thalamocortical circuit, inducing inhibition of the hyperexcitable thalamocortical neurons. A closed-loop stimulation set-up in response to epileptic activity measured in the cortex was able to immediately induce seizure arrest. Krook-Magnusen et al. performed several optogentic experiments in a mouse model of temporal lobe epilepsy (TLE).[46– 48] In an in vivo, real-time, closed-loop experiment, they demonstrated either optogenetic inhibition of hippocampal excitatory principal cells, or activation of a subpopulation of GABAergic cells 6
representing 10 years. Epilepsy & Behavior: E&B. 2011;20(3):478–483.
24.
Fisher RS, Eggleston KS, Wright CW. Vagus nerve stimulation magnet activation for seizures: a critical review. Acta Neurol Scand. 2015;131(1):1–8.
25.
Murphy JV, Torkelson R, Dowler I, et al. Vagal nerve stimulation in refractory epilepsy: the first 100 patients receiving vagal nerve stimulation at a pediatric epilepsy center. Arch Pediatr Adolesc Med. 2003;157(6):560–564.
9
Review
Downloaded by [Monash University Library] at 14:40 22 November 2015
26.
Carrette et al.
Boon P, Vonck K, Van Walleghem P, et al. Programmed and magnet-induced vagus nerve stimulation for refractory epilepsy. J Clin Neurophysiol: Official Publication Am Electroencephalographic Society. 2001;18(5):402–407.
27.
Morris GL 3rd. A retrospective analysis of the effects of magnet-activated stimulation in conjunction with vagus nerve stimulation therapy. Epilepsy & Behavior: E&B. 2003;4(6):740–745.
28.
Jeppesen J, Beniczky S, FuglsangFrederiksen A, et al. Detection of epileptic-seizures by means of power spectrum analysis of heart rate variability: a pilot study. Technol Health Care: Official J Eur Soc Eng Med. 2010;18(6):417–426.
29.
30.
31.
36.
Shoeb A, Pang T, Guttag J, et al. Noninvasive computerized system for automatically initiating vagus nerve stimulation following patient-specific detection of seizures or epileptiform discharges. Int J Neural Syst. 2009;19(3):157–172. Van Elmpt WJ, Nijsen TM, Griep PA, et al. A model of heart rate changes to detect seizures in severe epilepsy. Seizure: J Br Epilep Assoc. 2006;15(6):366–375. Eggleston KS, Olin BD, Fisher RS. Ictal tachycardia: the head-heart connection. Seizure. 2014;23(7):496–505.
33.
Sevcencu C, Struijk JJ. Autonomic alterations and cardiac changes in epilepsy. Epilepsia. 2010;51(5):725–737.
34.
Jansen K, Lagae L. Cardiac changes in epilepsy. Seizure-Eur J Epilep. 2010;19 (8):455–460.
35.
Lotufo PA, Valiengo L, Bensenor IM, et al. A systematic review and meta-analysis of heart rate variability in epilepsy
Jansen K, Varon C, Van Huffel S, et al. Peri-ictal ECG changes in childhood epilepsy: implications for detection systems. Epilep Behav. 2013;29(1):72–76.
37.
Leutmezer F, Schernthaner C, Lurger S, et al. Electrocardiographic changes at the onset of epileptic seizures. Epilepsia. 2003;44(3):348–354.
38.
Zijlmans M, Flanagan D, Gotman J. Heart rate changes and ECG abnormalities during epileptic seizures: prevalence and definition of an objective clinical sign. Epilepsia. 2002;43(8):847–854.
39.
Boon P, Van Rijckevorsel K, Elger CE, et al. Vagus nerve stimulation triggered by cardiac-based seizure detection, a prospective multicenter study. Epilep Curr. 2014;14(Suppl.):21.
Osorio I. Automated seizure detection using EKG. Int J Neural Syst. 2014;24 (2):1450001.
32.
10
and antiepileptic drugs. Epilepsia. 2012;53(2):272–282.
40.
Hampel KG, Vatter H, Elger CE, et al. Cardiac-based vagus nerve stimulation reduced seizure duration in a patient with refractory epilepsy. Seizure: The J Br Epilep Assoc. 2015;26:81–85.
41.
Bentley JN, Chestek C, Stacey WC, et al. Optogenetics in epilepsy. Neurosurg Focus. 2013;34(6):E4.
●●
The authors present an overview of the development of optogenetics and review recent studies investigating optogenetic modification of circuits involved in epileptic seizures.
42.
Ritter LM, Golshani P, Takahashi K, et al. WONOEP appraisal: optogenetic tools to suppress seizures and explore the mechanisms of epileptogenesis. Epilepsia. 2014;55 (11):1693–1702.
43.
Tonnesen J, Sorensen AT, Deisseroth K, et al. Optogenetic control of epileptiform activity. Proc Natl Acad Sci USA. 2009;106(29):12162–12167.
44.
Wykes RC, Heeroma JH, Mantoan L, et al. Optogenetic and potassium channel gene therapy in a rodent model of focal neocortical epilepsy. Sci Transl Med. 2012;4(161):161ra52.
45.
Paz JT, Davidson TJ, Frechette ES, et al. Closed-loop optogenetic control of thalamus as a tool for interrupting seizures after cortical injury. Nat Neurosci. 2013;16(1):64–70.
46.
Krook-Magnuson E, Armstrong C, Oijala M, et al. On-demand optogenetic control of spontaneous seizures in temporal lobe epilepsy. Nat Commun. 2013;4:1376.
47.
Krook-Magnuson E, Armstrong C, Bui A, et al. In vivo evaluation of the dentate gate theory in epilepsy. J Physiol. 2015;593(10):2379–2388.
48.
Krook-Magnuson E, Szabo GG, Armstrong C, et al. Cerebellar directed optogenetic intervention inhibits spontaneous hippocampal seizures in a mouse model of temporal lobe epilepsy. Eneuro. 2014;1:1.
49.
Sukhotinsky I, Chan AM, Ahmed OJ, et al. Optogenetic delay of status epilepticus onset in an in vivo rodent epilepsy model. Plos One. 2013;8(4): e62013.
50.
Salam MT, Mirzaei M, Ly MS, et al. An implantable closedloop asynchronous drug delivery system for the treatment of refractory epilepsy. IEEE Trans Neural Syst Rehabilitation Eng: a Publ IEEE Eng Med Biol Soc. 2012;20 (4):432–442.
51.
Mangubat EZ, Kellogg RG, Harris TJ Jr., et al. On-demand pulsatile intracerebral delivery of carisbamate with closed-loop direct neurostimulation therapy in an electrically induced self-sustained focalonset epilepsy rat model. J Neurosurg. 2015;122(6):1283–1292.
Expert Rev. Neurother.
