Ictal onset on intracranial EEG: Do we know it when we see it? State of the evidence

Transcription

1 FULL-LENGTH ORIGINAL RESEARCH Ictal onset on intracranial EEG: Do we know it when we see it? State of the evidence *Shaily Singh, *Sherry Sandy, and * Samuel Wiebe SUMMARY Dr. Shaily Singh is a clinical fellow in epilepsy and electrophysiology at the University of Calgary, Canada Objective: A major limitation of intracranial electroencephalography (ieeg) is recording from a confined region. This may falsely localize seizure onset if the distinction between ictal onset zone, proximity, and spread is unclear, or if the ictal rhythm is not clearly identified. Delineation of the ictal onset zone is crucial for surgical success. We appraised the evidence to determine whether specific ieeg ictal patterns are associated with the ictal onset zone. Methods: We searched Embase for articles in English until September 30, 2014, with MeSH keywords related to intracranially implanted electrodes and seizures. Two authors independently screened abstracts, reviewed full text articles, and abstracted data. The association between seizure outcome and type of ictal onset pattern (IOP), and its extent, location, and spread were explored visually or by univariate analysis when sufficient data were provided. Methodologic quality of each study was assessed. Results: We reviewed 1,987 abstracts from which 21 articles were analyzed. Fifteen IOPs were reported. Low frequency high amplitude repetitive spiking (LFRS) was the most frequently reported IOP by studies that dealt with mesial temporal lobe epilepsy (mtle) and investigated with depth electrodes. In neocortical epilepsy, low voltage fast activity (LVFA) was the most commonly described IOP. Delta activity was an infrequently reported IOP and was described mostly as a spread pattern. Significance: LFRS is associated with good surgical outcome in mtle and has a strong relation with mesial temporal pathology and its severity. LVFA is associated with neocortical temporal epilepsy and focal LVFA is associated with better surgical outcome. Electrodecrement may be associated with regional or widespread onsets. Rhythmic delta is a propagation rhythm rather than an IOP. Focal IOPs and slower propagation times are associated with better outcomes. The quality of the studies is suboptimal and there are methodological problems. Interobserver agreement is poorly documented. KEY WORDS: Seizure, Epilepsy, Electrode, Implanted. Delineation of the ictal onset zone is crucial for surgical success. Intracranial electroencephalography (ieeg) is an important tool for this purpose, especially in nonlesional, multifocal, network epilepsies, and in epilepsies involving eloquent areas. 1,2 The ieeg registers electrical activity Accepted July 19, 2015; Early View publication August 21, Departments of *Clinical Neurosciences and Community Health Sciences, University of Calgary, Calgary, Alberta, Canada; and Hotchkiss Brain Institute and O Brien Institute for Public Health, University of Calgary, Calgary, Alberta, Canada. Address correspondence to Samuel Wiebe, Foothills Medical Centre, St NW, Calgary, AB, Canada T2N 2T9. swiebe@ ucalgary.ca Wiley Periodicals, Inc International League Against Epilepsy from a very confined region, and this may lead to false localization of the ictal onset zone. This is likely if there is no clear distinction among ictal onset patterns (IOPs) corresponding to the ictal onset zone proper or areas of ictal spread. High frequency oscillations (HFOs) are increasingly reported as useful markers of the ictal onset zone. However, the process is labor intensive and not standardized, and it is not yet broadly used in clinical practice. Determining what constitutes an IOP on ieeg may affect clinical practice and surgical decisions, and agreement would help clinicians and researchers communicate meaningfully. Several authors 3 6 have attempted to establish a relationship between various 1629

2 1630 S. Singh et al. Key Points Low voltage fast activity is a neocortical ictal rhythm and is associated with better surgical outcomes Low frequency repetitive spikes is a mesial temporal ictal rhythm, associated with a better outcome in this topography and is associated with MTS Sinusoidal theta and delta activity are propagation rhythms IOPs, the epileptogenic zone, and the underlying pathology. However, it is evident that there is considerable variability to date as to what constitutes an IOP. It is common to assume that the ictal onset zone corresponds to the site where the first focal electrical changes are detected, often irrespective of their morphology, if these changes precede any clinical ictal manifestations. 7 It therefore becomes necessary to differentiate IOPs from those denoting ictal spread or late ictal rhythms that denote ictal evolution rather than onset. This can be challenging due to sampling limitations, as the pattern of ictal activity can differ considerably, even between two adjacent ieeg contacts. 8 Even with extensive ieeg coverage, the ictal onset zone may still be elusive, and clinical experience coupled with local clinical culture at individual centers often determines the interpretation of ieeg. The lack of a standardized definition of what constitutes an IOP and the subjectivity inherent in its identification can lead to substantial variation in clinical practice and arguably to differences in clinical decisions and surgical results. Moreover, the methodologic quality of studies describing IOPs can also play a crucial role in the credibility and applicability of the evidence. We systematically appraised the evidence to determine how IOPs are determined, whether specific IOPs on ieeg are associated with the ictal onset zone, and also whether these IOPs have consistent clinical associations. Methods Literature search strategy and selection criteria We systematically searched the MEDLINE database using a comprehensive strategy that incorporated Medical Subject Headings and variations of text words for intracranially implanted electrodes and seizures, up to September 2014 (see Appendix S1 for literature search strategy). We also reviewed relevant studies identified in the reference sections of included articles. We included studies if they contained original research in English that included more than five patients irrespective of age, describing IOPs identified by visual inspection, or by spectral analysis if it was correlated to specific IOPs. We did not analyze HFOs as markers of seizure onset. We excluded studies that reported only quantitative ieeg data without reference to specific IOPs, review articles with no original data (although their references were checked for additional original studies), and studies describing only interictal data. Study selection and data collection Two authors independently performed the literature search, screened all study titles and abstracts, selected and reviewed studies in full text, and extracted data. Disagreements were solved through consensus and discussion with a senior author. From each eligible study we extracted publication year, country of origin, study design (prospective vs. retrospective), and number of participants. Clinical data abstracted included sex, mean or median age at onset of epilepsy and at ieeg evaluation, duration of epilepsy, lobe(s) of seizure onset, presence of a lesion on structural imaging or histopathology, type of ieeg electrodes and coverage, type of analysis (visual or quantitative plus visual), criteria used for determining an IOP, unit of analysis (patients or seizures), number of seizures assessed overall and per patient, type of seizures analyzed, how multiple IOPs in the same patient were analyzed, the extent, topography and spread of IOPs, and surgical outcome using Engel s classification. 9 Definitions To allow for meaningful comparisons of IOPs, which were described using a large variety of terms in the literature, we grouped them under 15 categories (Table 1). For example, we categorized low voltage frequencies >15 Hz as low voltage fast activity (LVFA), and low frequency high amplitude epileptiform discharges as low frequency Table 1. Frequency with which IOPs are reported in the literature Ictal pattern a Studies, N (%) Electrodecremental (ED) 5 Low voltage fast activity (LVF) b 19 ED plus LVFA 3 <2 Hz rhythmic waves 1 <2 Hz spikes Hz rhythmic waves Hz spikes Hz rhythmic waves Hz spikes Hz rhythmic waves Hz spikes 4 >13 Hz spikes 2 Repetitive spikes (no frequency specified) 6 Delta brush 1 Burst suppression 1 a Not mutually exclusive. b See text for description.

3 1631 Intracranial Ictal Onset Patterns repetitive spikes (LFRS). Several studies reported the frequency of IOPs and the morphology of IOPs separately. We combined the two whenever possible. Rhythmic activity denotes sinusoidal morphology, and epileptiform activity denotes sharply contoured morphology such as spikes and sharp waves. Extent of IOP from here on refers to the number of electrodes displaying the initial IOP. We categorized IOPs as focal if 3 adjacent electrodes were involved, and as regional if >3 electrodes were involved. Location of IOP refers to areas from which the ictal rhythm emanates, for example, mesial temporal. IOP propagation time refers to the number of seconds required for an IOP to spread from one region to contiguous or noncontiguous regions. Good outcome refers to Engel class I and poor outcome to Engel class II IV. Analysis Descriptive statistics for normally and nonnormally distributed data were used to tabulate variables of interest. Univariate analyses of association used parametric and nonparametric methods as appropriate. When possible, data were statistically pooled using a random effects model, and assessing heterogeneity using the Q statistic. Potential sources of heterogeneity were explored when sufficient information was provided. The association between seizure outcome and type of IOP, and its extent, location, and spread was explored visually or statistically with univariate analysis when sufficient data were provided. We assessed the methodologic quality of each study using five pragmatic criteria, as follows: (1) EEG interpretation by >1 reader independently, (2) blinding of EEG readers to diagnosis and outcome, (3) interpretation requiring agreement between EEG readers, (4) inclusion of consecutive patients, and (5) accounting for all patients in the analyses. Because there is no gold standard to determine what constitutes an IOP, we used a set of four external criteria as proxies to determine the strength with which an observed ictal pattern could be defined as an IOP. We postulated that these external criteria could be graded from strongest to weakest as follows: (1) good surgical outcome, (2) association with histopathology or magnetic resonance imaging (MRI) lesion, (3) classical semiology corresponding to topography, and (4) number of studies describing a particular ictal pattern as an IOP (as a proxy measure for consensus). Results The literature search yielded 1987 citations, of which 159 studies were potentially eligible. After full text review, 17 studies fulfilled eligibility criteria and were selected for analysis. References from these articles and hand searching of other sources yielded four additional articles. Thus, 21 articles published between 1992 and 2014 and comprising 974 participants were included in the analysis (3 8, 10 24) (Fig. 1). Study characteristics are presented in Table 2A,B. All studies were retrospective and all but one focused on adults. All the studies were from single centers representing five countries (U.S.A., United Kingdom, Canada, Czechoslovakia, and Korea). The median number of patients per study was 33 (range 9 177), and that of seizures was 152 (range ). Figure 1. Preferred Reporting Items for Systematic Reviews and Meta- Analyses (PRISMA) chart of literature search results, and excluded and included studies. Epilepsia ILAE

4 1632 S. Singh et al. Table 2. Characteristics of included studies Author Year N Total Age (mean) Age of onset (mean) Duration of epilepsy (mean) Lesional or nonlesional Topography Electrode type Seizures assessed Seizures per pt (mean) (A) Lee Both Mixed Mixed 333 Park Nonlesional Temporal Depths 153 Wennberg Both Mesial temporal Depths Toczek Both Frontal Mixed Wetjen Nonlesional Neocortical extratemporal Subdural Dolezalova Both Temporal Mixed Spanedda Both Mesial bitemporal Depths Faught Both Mesial temporal Subdural 140 Spencer Both Mixed Mixed 166 Perucca Lesional Mixed Depths Velasco Both Mesial temporal Depths Holtkamp Both Frontal Mixed Schuh Both Mesial temporal Mixed Jimenez-Jimenez Both Temporal Mixed 373 Kutsy Both Neocortical mixed Subdural 5.5 Turkdogan Both Neocortical mixed Mixed Schiller Both Mixed Mixed 121 Jung Both Neocortical temporal Subdural 152 Weinand Nonlesional Temporal Subdural Kim Both Neocortical mixed Subdural Alarcon Nonlesional Mixed Mixed 78 Author Pts operated (%) Events excluded EEG seizures excluded Criteria for IOP Analysis of multiple IOPs Quality score a (B) Lee 53 (100) Auras, specific patterns, unclear onset Yes 1st sustained EEG change Predominant pattern 2 Park 33 (100) Auras, unclear onset, seizures contralateral to surgery Yes 1st sustained EEG change, typical semiology Not mentioned 5 Wennberg 11 (60) Multiple patterns, clinical onset before EEG onset No Not mentioned Excluded 1 Toczek 9 (100) Not mentioned No 1st sustained EEG change Not mentioned 1 Wetjen 31 (60) Not mentioned NA Earliest EEG change Not mentioned 3 Dolezalova 51 (100) Specific patterns, seizures <5 s Yes 1st sustained EEG change Predominant pattern 3 Spanedda 15 (70) Unclear onset, multiple patterns, seizures <5 s No 1st sustained EEG change Excluded 4 Faught 28 (100) Auras NA Earliest EEG change Predominant pattern 0 Spencer 26 (100) Specific patterns Yes Earliest EEG change Not mentioned 0 Perucca 28 (80) Clinical onset before EEG onset No Earliest EEG change Predominant pattern 5 Velasco Multiple and specific patterns Yes Not mentioned Excluded 3 Holtkamp 25 (100) Not mentioned NA Earliest EEG change Not mentioned 1 Schuh 40 (100) Not mentioned No 1st sustained EEG change Not mentioned 3 Jimenez-Jimenez 69 (100) Seizures in 1st 24 h No 1st sustained EEG change Predominant pattern 4 Kutsy 26 (100) Not mentioned NA 1st sustained EEG change Not mentioned 1 Turkdogan 25 (100) Not mentioned No Earliest EEG change Not mentioned 3 Continued

5 1633 Intracranial Ictal Onset Patterns Table 2. Continued. EEG seizures excluded Criteria for IOP Analysis of multiple IOPs Quality score a Author Pts operated (%) Events excluded Schiller 57 (70) Not mentioned No Earliest EEG change Not mentioned 4 Jung 31 (100) Not mentioned NA 1st sustained EEG change Not mentioned 2 Weinand 89 (100) Not mentioned NA 1st sustained EEG change Not mentioned 2 Kim 175 (100) Not mentioned NA 1st sustained EEG change Not mentioned 2 Alarcon 13 (90) Not mentioned NA Earliest EEG change Not mentioned 0 NA, not available or not mentioned. a Score range = 1 5 (higher = better); score = 0 if individual items not described by study. Ictal onset patterns and associations The most common IOP reported was LVFA (n = 19, 90%) and the least frequent were burst suppression (n = 1) and delta brush (n = 1) (Table 1). The median proportion of patients demonstrating the various IOPs has been detailed in Fig. 2. IOP and topography LVFA was most frequently reported in neocortical epilepsy, both temporal and extratemporal (n = 9). 4,5,8,10 15 One study demonstrated that LVFA in the beta frequency range was associated with neocortical temporal epilepsy, whereas LVFA within the gamma frequency range was associated with extratemporal epilepsy. 5 Low frequency repetitive spikes 2 Hz (LFRS) was more frequently reported as an IOP in mesial temporal lobe epilepsy (mtle; n = 5). 4,6,16 18 Spanneda et al. 17 found that TLE patients with hippocampal and amygdalar atrophy exhibited LFRS, whereas those with isolated amygdalar atrophy and normal temporal lobe showed LVFA as the predominant IOP. All studies on mtle that described LFRS were investigated using depth electrodes. One study that did not demonstrate this pattern in mtle used epidural electrodes. 19 The studies did not identify any differences in IOPs based on the type of electrode used. An electrodecremental pattern (ED) was reported as an IOP in five studies. 7,9,17,19,20 In three of these studies 9,17,20 it occurred infrequently (4 19%). ED was seen more frequently in the frontal lobes (46%). 20 Two other studies reported ED in a higher proportion of patients (71 80%), 7,19 and both studies described this as part of the initial sequence of ictal changes and not as the first EEG change. Figure 2. Frequency with which different IOPs are reported in the literature. LVFA, low voltage fast activity; LFRS, Low frequency repetitive spikes; Rep Spikes, spikes of unspecified frequency. See text for descriptions. Epilepsia ILAE

6 1634 S. Singh et al. Less frequently reported IOPs ED and simultaneous LVFA were reported in two studies. 20,21 The occurrence of this pattern was variable, occurring more commonly in neocortical epilepsy (23 30%). 20,21 Rhythmic delta sinusoidal waves was reported in six studies, and its occurrence ranged from 3 15% of patients. 5,15,19 22 Epileptiform activity in the delta range was reported by three studies and ranged from 4 15%. 4,18,20 Rhythmic activity in the alpha theta range was reported by six studies and it occurred in 6 23% of patients. 5,8,15,20 22 Epileptiform activity in the alpha theta range was reported by seven studies and ranged from 6 85%. 5,8,11,17 20 Excluded patterns Two studies 5,19 excluded seizures that started as electrodecremental events, brief bursts of spikes, and periodic spikes at <2 Hz. Most studies did not consider a brief burst of spikes (preictal spiking) preceding the sustained IOP as part of the IOP. Three studies demonstrated this phenomenon, 1,8,19 of which one reported this pattern in almost 75% of cases, 19 with no relation to outcome. IOP and surgical outcome Fourteen studies (66.6%) explored the association between seizure outcome and IOP. Generally, IOPs with faster frequencies (>15 Hz) were associated with better outcomes as compared to those with slower frequencies (<15 Hz). Eight studies found an association between LVFA and good surgical outcome. Of these, six focused on various neocortical topographies 5,9,12,13,15,19 21 and two focused on TLE. 10,19 Among 10 studies in which the unit of analysis was patients (as opposed to seizures), the pooled proportion of seizure-free patients who had an LVFA pattern was 59% (95% confidence interval [CI], 47 70%) (Fig. 3). Two studies found that LFRS was more frequently associated with good outcomes in TLE, one focusing on neocortical TLE 12 and one on mesial TLE. 17 The median proportion of seizure-free patients exhibiting LFRS was 74% (interquartile range [IQR] 72 84%). Repetitive spikes (unspecified frequency) were associated with good outcome in 74% cases in a study on TLE. 13 Two studies assessing outcome in relation to an ED, found it to be associated with a poor outcome (ranging from 92% to 100% of patients). 11,19 Jimenez-Jimenez et al. 11 also showed that if an ED was part of the IOP in conjunction with LVFA, the seizure freedom rate was 45% as compared to 70% with LVFA alone. However, two other studies reported better outcomes with ED. 7,10 Although these authors did not include ED in formal analyses of surgical outcomes, 57% of all their patients became seizure-free. 10 Alarcon et al. 7 also showed that 66% of patients with ED who underwent surgery had a good outcome. A few studies found relevant associations with less frequently reported IOPs. Jung et al. 12 reported 5 20 Hz repetitive sharp waves in two patients, both of whom were seizure-free. Good outcomes were reported by two studies with rhythmic sinusoidal delta-theta activity, 5,22 two studies with rhythmic alpha activity, 11,23 and one study with spikes in the theta range. 11 Slow 2 8 Hz sinusoidal activity was universally associated with a poor outcome in three studies. 11,19,23 A high proportion of patients underwent surgery after ieeg (median proportion = 100%, IQR = %). The median proportion of patients with good outcome was 60% (IQR = 42 66%). To assess the presence of bias in the selection of patients for surgery, we performed a correlation analysis between each study s proportion of patients operated and proportion with good outcome. A positive, nonsignificant correlation was found (Spearman Figure 3. Forest plot and meta-analysis of good surgical outcome (Engel class I) among patients exhibiting LVFA as the IOP. Epilepsia ILAE

7 1635 Intracranial Ictal Onset Patterns rank = 0.39, p = 0.13), for example, studies in which almost all patients underwent surgery tended to report better outcomes, suggesting some selection bias. Surgical outcome was associated with lobe of surgery. The median proportion seizure-free was 56% each for temporal lobe, frontal lobe, and mixed topographies; and it was 34% for bitemporal and 33% for purely extratemporal surgeries (one study only). Extent of IOPs Seven studies evaluated the associations among specific IOPs, and their extent and surgical outcomes. Five studies providing quantitative data had significant heterogeneity precluding meta-analysis. However, the median difference in absolute proportions of seizure-free patients was 43% (95% CI 23 53%) in favor of focal IOPs, as compared to nonfocal IOPs. Two studies 6,18 with mesial TLE patients and one study on patients with mixed topographies 4 found that focal onsets were associated with LFRS and regional onsets with LVFA. Although one study found that an ED pattern was associated with regional or hemispheric onsets (75%), 19 Faught et al. 10 found this pattern in a localized or regional distribution in a patients with mtle. Lee et al. 5 found that beta frequencies were associated with focal onsets, whereas gamma frequencies were associated with regional onsets. Four studies 10,12,16,20 agreed that focal IOPs are associated with better outcomes, as compared to regional and multifocal IOPs. Four studies, 5,18,21,22 however, did not find that the extent of IOP was related to surgical outcome. Weinand et al. 18 found that unilateral onset fared significantly better (69% Engel class I) as compared to patients with generalized onsets (29% Engel class I), but there was no difference if the unilateral onsets were focal or regional. IOP and histopathology Of 11 studies exploring the association between histopathology and IOPs, 6 found an association between mesial temporal sclerosis and LFRS, 4,6,13,16,17,24 and 3 found that LFRS was significantly associated with more severe mesial temporal sclerosis as compared to LVFA. 4,6,17 Another study found that LFRS was associated with mesial temporal atrophy (on MRI), whereas LVFA was associated with normal temporal lobe. 13 One study showed that rhythmic IOPs (delta, theta, alpha, and beta) in mesial TLE were associated with milder forms of gliosis, whereas repetitive epileptiform IOPs occurred in more severe gliosis. 16 Only one study found LVFA in association with mesial temporal lobe pathology, 3 and another study noted that 91% of patients with normal histopathology exhibited LVFA as an IOP. 19 Two studies reported on cortical dysplasias and IOPs 4,24 and found an association with LVFA (one with a delta brush component). In addition, Lee et al. 5 described alpha theta spike waves in developmental pathologies and rhythmic sinusoidal waves in acquired pathology. Most studies did not find a correlation between histopathology and surgical outcome. 19,21,23 Multiple types of IOPs Five studies reported on multiple types of IOPs in individual patients 4,5,10,11,19 ; these studies chose the most prominent pattern as the IOP of interest. Park reported multiple IOPs in 51% of recorded seizures, and 82% of patients with this pattern achieved a good outcome. 16 Faught reported 43% patients with multiple IOPs, and 58% of these patients achieved a good outcome. The remaining three studies described varying proportions (16 30%) of patients with multiple IOPs, but outcomes were not described. 4,5,19 IOP and seizure propagation In three studies providing quantitative data on time to seizure spread, the pooled analyses revealed an odds ratio (OR) for good outcome of 3.9 (95% CI ) with slower (>1 s) as compared to faster (<1 s) propagation times (p = 0.001), with no evidence of heterogeneity (Fig. 4). 12,21,22 Schiller et al. 8 showed that remote seizure propagation (>4 s after onset, different lobe) could be consistently identified by rhythmic theta or delta patterns. Perucca et al. noted that LFRS was seen as a propagation rhythm in mesial temporal sclerosis. By comparison, LVFA and <13 Hz sharp activity were seen in regions of seizure spread across all regions. 4 Velasco et al. found that LFRS was associated with slower seizure propagation to the contralateral hemisphere (54 s) as compared to LVFA (30 s) in mesial TLE. Similarly, Dolezalova et al. found that in TLE, slower frequency IOPs (<8 Hz) were associated with slower propagation as compared to LVFA. One study reported that Taylor type dysplastic lesions were associated with more rapid propagation (10 s) than nondysplastic lesions (15 s) regardless of IOP type. 24 Wennberg et al. showed that seizures originating in the hippocampus were less likely to propagate (28%) and to produce clinical manifestations (23%). By contrast, 72% of seizures arising from the amygdala propagated, and 77% manifested clinically. Similarly 87% of parahippocampal seizures propagated and 100% manifested clinically. 15 All studies assessing seizure propagation agreed that slower propagation times were associated with better outcomes, 12,18,20 22 except one that found similar propagation times regardless of seizure outcome. 13 Four 12,20 22 of these studied neocortical epilepsy (temporal and extratemporal) and one 18 studied TLE, although no comparison was provided regarding propagation times in mesial versus neocortical TLE.

8 1636 S. Singh et al. Figure 4. Good surgical outcome (Engel class I) was four times higher among patients with propagation time of IOP longer than 1 s (OR 3.9, 95% CI ) (random effects, p < 0.001). Epilepsia ILAE Methodologic quality of studies Of the four a priori criteria to assess the validity of the various IOPs described, 14 studies assessed the association of IOP with seizure outcomes, 9 assessed concordance with lesions (usually by histopathology), and none reported the association of IOPs with ictal semiology. Of a maximum quality score of 5 (higher = better), the median was 2.5 (range 1 3.5). Only two studies 4,17 fulfilled all criteria (score = 5), and 17 (81%) studies met three or fewer criteria. Only two studies 4,16 mentioned interobserver agreement for the type of IOP observed, which was 47.7% and 86%. Thirty-three percent of studies reported blinded EEG interpretation, 43% reported accounting for all included patients, and 52% described each of the following three criteria: including consecutive patients, EEG interpretation by >1 reader, and concordance among EEG readers. Discussion This review highlights the various IOPs and their associations with the probable epileptogenic zone, seizure spread, and pathology. The description of IOPs varied widely from study to study, and therefore interpretation was limited by what the authors focused on. Because surgical success would imply that the epileptogenic zone had been removed, we used surgical outcome as the proxy measure for determining whether an ictal pattern was an IOP. LVFA was the IOP that occurred most consistently across all topographies, especially in neocortical epilepsy. LFRS was associated with mesial temporal epilepsy. Of interest, all studies evaluating mesial temporal epilepsy that used depth electrodes found LFRS as a prominent IOP, except one that used epidural electrodes. 10 This may suggest that LFRS is an IOP better detected by depth electrodes. However, most of the studies using a combination of electrodes did not analyze IOPs based on electrode type. The most common criterion for determining what an IOP is has been postoperative seizure freedom. Good outcome has been described with LFRS in mesial TLE and LVFA in neocortical epilepsy. In addition, focal IOPs have often been associated with better outcomes. The few studies that found poorer outcomes with focal IOPs had a high proportion of regional and multifocal onsets, 5,22 suggesting that the actual IOPs might be located outside the monitored region. Slower propagation times (>1 s) were also associated with better surgical outcome. 12,21,22 There is a notion that surgical outcomes are poorer when mesial temporal structures are resected in patients with rapid propagation times, but this is not necessarily the case for neocortical resections, where rapid propagation occurs often and surgical outcomes are favorable. The evidence does not systematically explore the relationship between IOP type, propagation time, topography, and outcomes. The distinction between ED and LVFA was not clear because the latter, at very low voltage and high frequency, may be superimposed on the former. At low sensitivities, the low-voltage fast activity is not recognized. Yet, LVFA has been shown to portend a good prognosis and ED a poor prognosis in most studies. The fact remains that standards are needed to distinguish between ED and LVFA or the combination. It was thought that when the IOP is a generalized ED, seizure onset was either diffuse or occurs at a remote site that has not been implanted, consequently the surgical outcome was poor. On the other hand, two studies 7,10 found that patients with ED at onset had a good surgical outcome even if it was a not a focal rhythm. It could be argued that ED in these cases was not part of the IOP itself, but reflects diffuse cortical changes associated with ictal activity in susceptible regions On assessing outcomes based on topography, it was found that irrespective of IOP, the topography likely determines how completely the ictal onset zone was removed by surgery and in turn the surgical success. This was logical, as

9 1637 Intracranial Ictal Onset Patterns the extent of resection is limited in areas housing eloquent cortices. Only one study investigated the relationship between the extent of the resection, including the area with ictal onset and spreading ictal rhythm, and the surgical outcome 12 ; authors found that wider resections were associated with better outcomes in extratemporal topographies. Some studies excluded specific ieeg patterns as potential IOPs, such as ED, brief bursts of spikes, and <2 Hz repetitive spikes. However, this seemed unjustified because LFRS is an IOP consistently described in many studies in relation to mesial temporal sclerosis. Similarly ED was commonly reported as a more widespread rhythm that could evolve into another, more focal IOP. On the other hand, it seemed prudent to exclude brief bursts of spikes at onset, as this pattern did not correlate with IOP topography, extent of IOP, or surgical outcome, 19 suggesting that this may be an interictal phenomenon. It is clear that, at least in some studies, preexisting notions of what constitutes an IOP resulted in selection bias. Histopathologic analyses demonstrated a strong association between mesial temporal sclerosis and LFRS, especially in severe cases. 4,6,14,16,17,22 On the other hand, LVFA was associated with normal histopathology in TLE 9 and with focal cortical dysplasia. 4,14 Opportunities to address important questions have been missed in the existing evidence. The problem of multiple IOPs in different electrodes within a single seizure was not addressed by any study. Authors either excluded these seizures or focused on a single IOP, thereby missing the opportunity to address this vexing problem encountered in clinical practice. The role of electrographic seizures was not interrogated by any study. Eight studies included electrographic seizures in their analyses, but their frequency, localization, and relation to surgical outcome were not described. No light was shed on their localizing or prognostic value. The known limitations of ieeg, that is, small recording zone, limited sampling, and uncertainty of targets, create barriers to the full assessment of what constitute IOPs. The clinical application of HFOs to localize the ictal onset zone is still limited to a few centers where the technology and expertise are available. One of the studies 4 explored the relationship of the IOPs to HFO density and found no significant differences in HFO density between the seizure-onset and seizure evolution section. In periodic spikes and spikeand-wave activity, ripple and fast ripple densities continued to increase after seizure onset. In LVFA, however, ripple density was found to decrease after seizure onset. Several remediable methodologic deficits contribute to the uncertainty and heterogeneity of the results. First, the evidence suffered from selection bias affecting types of patients, IOPs included and excluded, and the focus of the analyses. There is a need for studies that systematically include a broad representative sample of patients, assess initial ictal changes inclusively rather than exclusively, and encompass relevant analyses, such as lesion, topography, ictal spread, and surgical outcomes. Second, when assessing the methodologic quality pertaining to elements required to measure associations, most studies were found wanting, increasing the opportunity for bias and limiting the robustness of their conclusions. Conclusion Despite these limitations, the following clinically relevant conclusions about IOPs can be offered with a moderate level of certainty. IOPs consisting of LVFA are associated with neocortical, and LFRS with mesial temporal epilepsy, where they are associated with better surgical outcomes. Slower LFRS correlate with more severe forms of MTS. No single IOP is associated consistently with a focal onset. In general, faster and sharper waveforms seem to be stronger indicators of an IOP. ED patterns are usually regional or widespread (not focal), their distinction from LVFA is unclear, and their association with surgical outcome is controversial. Sinusoidal rhythms in the theta-delta range are propagation rhythms rather than IOPs. Focal IOPs and slower propagation times are associated with better outcomes. The relationship between type of IOP and surgical outcome is confounded by topography (i.e., better outcomes in temporal and frontal surgery in general). The interpretation of multiple IOPs within a seizure, and that of subclinical electrographic seizures, remains unknown. Acknowledgments Samuel Wiebe is supported by the Hopewell Professorship in Clinical Neurosciences Research, Hotchkiss Brain Institute, University of Calgary. Disclosure None of the authors has any conflict of interest to disclose. We confirm that we have read the Journal s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines. References 1. Spencer SS. Surgical options for uncontrolled epilepsy. Neurol Clin 1986;4: Spencer SS, Williamson PD, Bridgers SL, et al. Reliability and accuracy of localization by s scalp ictal EEG. Neurology 1985;35: Spencer SS, Guimaraes P, Katz A, et al. Morphological patterns of seizures recorded intracranially. Epilepsia 1992;33: Perucca P, Dubeau F, Gotman J. Intracranial electroencephalographic seizure-onset patterns: effect of underlying pathology. Brain 2014;137: Lee SA, Spencer DD, Spencer SS. Intracranial EEG seizure-onset patterns in neocortical epilepsy. Epilepsia 2000;41: Velasco AL, Wilson CL, Babb TL, et al. Functional and anatomic correlates of two frequently observed temporal lobe seizure-onset patterns. Neural Plast 2000;7: Alarcon G, Binnie CD, Elwes RD, et al. Power spectrum and intracranial EEG patterns at seizure onset in partial epilepsy. Electroencephalogr Clin Neurophysiol 1995;94:

10 1638 S. Singh et al. 8. Schiller Y, Cascino GD, Busacker NE, et al. Characterization and comparison of local onset and remote propagated electrographic seizures recorded with intracranial electrodes. Epilepsia 1998;39: Engel J Jr, Van Ness PC, Rasmussen TB, et al. Outcome with respect to epileptic seizures. In Engel J Jr (Ed) Surgical treatment of epilepsies. 2nd Ed. New York, NY: Raven; 1993: Faught E, Kuzniecky RI, Hurst DC. Ictal EEG wave forms from epidural electrodes predictive of seizure control after temporal lobectomy. Electroencephalogr Clin Neurophysiol 1992;83: Jimenez-Jimenez D, Nekkare R, Flores L, et al. Prognostic value of intracranial seizure onset patterns for surgical outcome of the treatment of epilepsy. Clin Neurophysiol 2015;126: Jung WY, Pacia SV, Devinsky R. Neocortical temporal lobe epilepsy: intracranial EEG features and surgical outcome. J Clin Neurophysiol 1999;16: Spanedda F, Cendes F, Gotman J. Relations between EEG seizure morphology, interhemispheric spread, and mesial temporal atrophy in bitemporal epilepsy. Epilepsia 1997;38: Toczek MT, Morrell MJ, Risinger MW, et al. Intracranial ictal recordings in mesial frontal lobe epilepsy. J Clin Neurophysiol 1997;14: Wennberg R, Arruda F, Quesney LF, et al. Preeminence of extrahippocampal structures in the generation of mesial temporal seizures: evidence from human depth electrode recordings. Epilepsia 2002;43: Park YD, Murro AM, King DW, et al. The significance of ictal depth EEG patterns in patients with temporal lobe epilepsy. Electroencephalogr Clinical Neurophysiol 1996;99: Schuh LA, Henry TR, Ross DA, et al. Ictal spiking patterns recorded from temporal depth electrodes predict good outcome after anterior temporal lobectomy. Epilepsia 2000;41: Weinand ME, Wyler AR, Richey ET, et al. Long-term ictal monitoring with subdural strip electrodes: prognostic factors for selecting temporal lobectomy candidates. J Neurosurg 1992;77: Dolezalova I, Brazdil M, Hermanova M, et al. Intracranial EEG seizure onset patterns in unilateral temporal lobe epilepsy and their relationship to other variables. Clin Neurophysiol 2013;124: Holtkamp M, Sharan A, Sperling MR. Intracranial EEG in predicting surgical outcome in frontal lobe epilepsy. Epilepsia 2012;53: Kim DW, Kim HK, Lee SK, et al. Extent of neocortical resection and surgical outcome of epilepsy: intracranial EEG analysis. Epilepsia 2010;51: Kutsy RL, Farrell DF, Ojemann GA. Ictal patterns of neocortical seizures monitored with intracranial electrodes: correlation with surgical outcome. Epilepsia 1999;40: Wetjen NM, Marsh WR, Meyer FB, et al. Intracranial electroencephalography seizure onset patterns and surgical outcomes in nonlesional extratemporal epilepsy. J Neurosurg 2009;110: Turkdogan D, Duchowny M, Resnick T, et al. Subdural EEG patterns in children with taylor-type cortical dysplasia: comparison with nondysplastic lesions. J Clin Neurophysiol 2005;22: Bertashius KM. Propagation of human complex-partial seizures: a correlation analysis. Electroencephalogr Clin Neurophysiol 1991;78: Gotman J. Interhemispheric interactions in seizures of focal onset: data from human intracranial recordings. Electroencephalogr Clin Neurophysiol 1987;67: Lieb JP, Hoque K, Skomer CE, et al. Inter-hemispheric propagation of human mesial temporal lobe seizures: a coherence/phase analysis. Electroencephalogr Clin Neurophysiol 1987;67: Supporting Information Additional Supporting Information may be found in the online version of this article: Appendix S1. Search strategy.