Anatomo-electro-clinical Correlations





Acknowledgments


The authors thank all staff and patients from their Epilepsy Monitoring Units.


Introduction


Jean Bancaud, who greatly contributed to the study of seizure semiology through his observations from the inception of the stereoelectroencephalographic (SEEG) era, commented that semiological expression of an epileptic seizure is similar to the arrangement of words in forming a meaningful sentence. This statement highlights the difference between studying clinical signs occurring in the context of epileptic seizures and those arising in most other neurological conditions: in seizures (which by definition occur paroxysmally), there is an inherent temporal relation between the evolving signs, the nature of which is essential in order to make sense of the clinical picture. The metaphor of a meaningful sentence also conveys the notion that a certain “sense” exists in terms of the grouping and the evolution of clinical signs; semiological expression does not occur haphazardly but necessarily follows a sequence that is structured according to the effects of the epileptic discharge on functioning of brain networks.


Attempts to understand correlations between clinical symptoms/signs and brain electrical activity (i.e., using behavioral data to infer possible alterations in brain networks produced by epileptic discharges) must consider both the anatomical spread of seizure and the timescale over which changes occur. Interpreting seizure semiology therefore implies not only studying discrete elements (individual symptoms and signs) but how they are assembled together, and over which time course. In the framework of epilepsy presurgical evaluation, the clinician’s main goal is to formulate, and then prove or disprove, hypotheses of likely anatomical zone of seizure onset and early spread.


In this chapter, we will discuss (1) the relevance of semiologic analysis in conjunction with simultaneous EEG during the initial non-invasive phase of presurgical evaluation ( electroclinical correlations ), how this helps to formulate hypotheses of likely epileptogenic zone (EZ) and how it guides the clinician to decide upon an SEEG implantation strategy to confirm or refute these; (2) how to approach the key methodology of SEEG interpretation ( anatomo-electroclinical correlations ). We will highlight some common pitfalls and methodological challenges and will illustrate anatomo-electroclinical correlations using 2 case vignettes.


From phase 1 to SEEG: From electroclinical correlations to anatomo-electroclinical correlations


The “Phase 1” non-invasive part of presurgical evaluation uses electroclinical correlation , i.e., careful study of seizure semiology and its surface EEG correlates, via review of all available recorded seizures. This important step, which is performed in conjunction with study of the ensemble of non-invasive data (patient history, neuroimaging, neuropsychology, etc.) is key to formulating hypotheses of likely cerebral organization of a specific patient’s epilepsy. The surface-EEG based electroclinical correlation can be contrasted with SEEG-based anatomo-electroclinical correlation . The “ anatomo- ” part of this terminology becomes possible with SEEG due to the precise anatomic descriptors of the source of the electrical signal, based on intracerebral electrode contract positions, confirmed (nowadays) on post-implantation imaging, usually computerized tomography of electrodes in situ co-registered with the patient’s pre-implantation magnetic resonance imaging. Anatomo-electroclinical correlation is the core methodology of SEEG interpretation. Its aim is to try to make clinical sense of the observed signal changes on SEEG in conjunction with concomitant semiologic features, in order to extrapolate the likely seizure onset and propagation dynamics underlying the clinical expression. SEEG interpretation cannot meaningfully be done without this careful reference to semiologic output, and no amount of quantitative tools for SEEG signal analysis (interictal or ictal) can be a substitute for this core methodology that is grounded in expert study of the patient’s habitual seizure-related symptoms and signs. Apart from other aspects, semiologic analysis can help to check that a major SEEG electrode sampling error has not occurred. Going back to the Phase 1 and comparing the original electroclinical correlations with SEEG data can be a helpful last step to make sure that the conclusion following SEEG is compatible with the original non-invasive data.


Importance of Phase I Semiological Data Prior to SEEG


The quality of the interpretation of the ensemble of Phase 1 data, including seizure semiology, will help in (1) optimally choosing candidates for SEEG (i.e., patients with focal, drug-resistant epilepsy in whom surgical decision making will be significantly altered by SEEG results, in a way that could benefit outcome, such that the risk-benefit ratio for that specific patient is clearly in favor of proceeding to invasive recording); and (2) optimally designing an implantation strategy to confirm/refute the main hypotheses. Decision making in these 2 steps largely determines the success or otherwise of the definition of the epileptogenic zone using SEEG via anatomo-electroclinical correlation. The quality of the electroclinical correlations derived from the Phase 1 thus directly shapes the robustness of the anatomo-electroclinical correlations of the SEEG.


Prior to getting to the point of SEEG and tackling anatomo-electroclinical correlations, detailed background knowledge of the particular patient’s seizure semiology is very important. Starting with the Phase 1, interpretation of the video-EEG will ideally have been done by a clinician who knows the full case. In particular, it is important to be as familiar as possible with the semiologic history since onset of epilepsy (since patterns can evolve over time and with treatment changes), especially whether subjective symptoms have been reported at any time in the epilepsy history (see Clinical Vignette 1), and their precise description as provided by the patient. It is also important to know whether a single seizure type is described by the patient and their family, or whether different types occur (either variations of severity or clearly distinct semiologic patterns). These aspects will be important in correctly interpreting what is eventually recorded at the time of SEEG, both for spontaneous and stimulation-induced seizures.


Semiologic analysis: Methodological considerations


Some basic methodological considerations can be discussed in terms of optimal conditions for semiological analysis as follows: (1) technical conditions of recording; (2) adequate ictal examination; (3) specialist expertise. These apply to all seizure recordings captured in the Epilepsy Monitoring Unit (EMU); since this chapter is on SEEG, we emphasize that all opportunities for capturing maximally useful seizure data should be prioritized, especially adequate ictal examination.


Technical Conditions of Recording


Since semiological analysis is largely based on expert visual inspection of recorded video data, the information retrieved is dependent on both physical recording conditions (placement of patient in the camera field, quality of lighting, video resolution, etc.) and the observer’s visual perception and ability to recognize and correctly interpret the signs. Video quality can be optimized through adequate technical parameters (such as including a close-up image of the patient’s face, automated tracking of the head, and use of infra-red cameras for night-time filming). Placement of 2 different cameras recording simultaneously may allow better visualization depending on patient position, but in most cases a single high density camera is sufficient if the seizure occurs while the patient is within the camera frame. With a wide-angle lens, the field is large enough to get the full body image, and HD camera allows for replay zooming with excellent resolution, for instance on the face.


Ictal Examination


The quality of the clinical information obtained during the seizure can be greatly augmented by appropriate ictal examination, comprising evaluation of conscious level, language and if required motor examination. This should be meticulously done during SEEG, with the healthcare team on stand-by to examine the patient at the very first sign of any seizure activity on the trace or clinically. Indeed, for some seizure types, it is essentially the ictal examination that reveals all relevant signs, such as a patient presenting a non-signaled seizure with loss of awareness and/or language dysfunction without motor features. Appropriate training in ictal examination for personnel who will be attending to the patient is an important component for optimal seizure recording. The key aspect of the ictal examination is the interaction between the examiner and the patient, to allow evaluation of awareness and language function, and the quality of this (depending on the individual questions to be answered for each patient’s case) is influenced by the experience of the examiner in performing ictal testing. A standardized battery for ictal clinical testing has been proposed following a European study performed under the auspices of the ILAE (see also chapters on stimulation in this book). Attaining optimal ictal examination requires specific training and ongoing practice of nurses, EEG technicians and medical staff. Lack of such training has indeed been identified as a major obstacle to obtaining good quality semiological data in many EMU .


The first step is to make that the patient is visible on camera (e.g., in camera field and if possible, facing the camera; remove bedsheets while protecting privacy), and then immediately test whether the patient is responsive by calling their name and asking simple commands. If the patient can respond, the examiner asks the patient what they can feel; simple naming tasks are given and orientation is checked. Audible commentary by the examining practitioner can be very helpful, for example mentioning out loud the presence of subtle signs that may be poorly visualized on video recording, such as skin color or pupillary change. Verbal memory can be tested by asking the patient to remember a word, or using recall of object naming as an episode that can be recalled by the patient afterward. If a trained person is not present to perform ictal testing, as a minimum any verbal interaction with a patient during a seizure (e.g., by a family member) can bring useful information. Depending on the clinical features of each case, other examination may be indicated, such as motor function, visual fields, proprioception or specific cognitive functions such as face recognition.


Post-ictal examination is also important (language recovery, memory of seizure, presence of focal deficit, etc.), and it may be during this period that the patient can give some description of early subjective symptoms experienced at seizure onset. Any symptoms mentioned by the patient should be carefully explored to achieve maximum descriptive detail, since the specific nature of a hallucination or visceral sensation could be an important clue as to localization of seizure onset. Stimulation studies require their own examination protocol (described in separate chapter on stimulation-induced seizures).


The Importance of Specialist Expertise


Semiologic analysis in conjunction with concomitant electrophysiologic data very largely depends on clinician expertise and experience, at the individual and at the team level, because of the importance of pattern recognition. As experience progresses, an important aspect is refined recognition of which semiologic patterns are robust with high specificity for certain brain regions, lobes or structures; and in contrast, which patterns have low specificity and relatively low predictive value (on their own) for localization. This last point is very important when using semiologic data to support hypotheses for subsequent SEEG implantation (discussed below in the section “Localizing value” of semiologic features: importance for SEEG hypotheses”). Equally, the correct interpretation of SEEG using anatomo-electroclinical correlations requires sufficient understanding of the different electrical seizure patterns, and pitfalls of SEEG including sampling limitations. The diagnostic framework of clinical acumen and electroclinical pattern recognition based on experience, and the “apprenticeship” process, are not yet replaceable by any other means of analysis of semiology or seizures, and this point is even more crucial for the more complex practice of SEEG compared to video-EEG.


General tips for analyzing seizure recordings


Seizures are viewed as video recordings (time-locked with electroencephalographic and other electrophysiologic data) that should allow sufficient time before the onset of the seizure, to determine, as far as possible, the precise moment at which the first clinical sign occurs. This may be obvious, such as a signaled aura or onset of motor signs, or may be very subtle, such as a slight change of facial expression. The clinical onset may be visible on the video, may be highlighted by an observer’s recognition of the seizure (e.g., family member in the room), or can be another recorded change such as increased heart rate. This time-point can be annotated on the trace to facilitate comparison with the signal. Next, the sequence of signs should be carefully reviewed, often by more than one visualization of the video data to make sure that nothing has been missed, and again annotating key moments of the semiologic evolution is helpful for comparing with the signal. How best to describe semiologic signs can be challenging for rarer or subtle patterns; a general rule it is preferable to use standardized vocabulary despite the limitations of this in some cases.


The full seizure period including the post-ictal phase should be reviewed; post-ictal signs may also provide important information (e.g., confusion, aphasia, visual field deficit, motor deficit, etc.). Visualizing seizure videos more than once, especially the early part of the seizure, can help in assessing the important features. Using video close-up on the face or specific body segments can be useful for looking at subtle changes (e.g., slight facial expression change), and it can sometimes be helpful to slow down the video playback speed (e.g., to study hyperkinetic movements) or speed it up (e.g., to review semiologic evolution in a prolonged seizure).


Since semiologic analysis is largely based on clinicians’ pattern recognition, a seizure may immediately have recognizable features that we are drawn to. This however should not preclude detailed searching for other specific features that may be more or less obvious. It is useful to assess both isolated signs or groups of signs, and the overall “gestalt” of the semiology, and degree of reproducibility across recorded seizures from that patient. The order and time course of signs are very important, with more emphasis placed on early compared to late seizure signs for cerebral localization purposes, since the early signs are produced by earliest propagation (i.e., closest to the seizure onset zone). Attention should be paid to asymmetry of signs, which in some cases are very important for cerebral lateralization, although it is important to be aware of which asymmetric signs do not have consistent lateralizing value (e.g., head turning, eye blinking) versus those that do (contralateral dystonia, tonic contraction or clonic jerks).


If more than one seizure has been recorded, this allows assessment of degree of reproducibility in both semiology and electrical pattern. The degree of inter-seizure clinical similarity should be checked against the electrical pattern, and may again provide helpful information about propagation, since different semiologic evolution may reflect differences in seizure propagation even if onset of seizure discharge is similar. The most obvious and common example of this would be a patient who has both focal seizures and focal to bilateral tonic-clonic seizures, or focal seizures with and without impaired consciousness. Anti-seizure medication reduction should be done judiciously to try to record the most focal version of the patient’s usual seizures, since rapid seizure propagation due to low ASM levels may make it harder to assess relevance of both semiologic and electrical features (e.g., seizures in clusters and/or focal to bilateral tonic clonic seizures).


“Localizing value” of semiologic features: Importance for SEEG hypotheses


Knowledge of the relative predictive power (specificity) of different semiologic features in understanding likely cerebral localization of seizures has to be built up over time. This occurs not just by reading literature and attending courses but through looking at as many videos as possible, of a large variety of seizures, in conjunction with discussion with specialist practitioners with large experience as mentioned earlier. When formulating hypotheses from Phase I data with a view to planning SEEG implantation, the specificity (or otherwise) of semiological features should be considered. The specificity of the semiologic features for a given patient is one component amongst the ensemble of non-invasive data that is used to judge whether a unifocal, spatially restrained and potentially surgical treatable zone of seizure organization is likely. To investigate predictors of focality of seizure onset on SEEG, a tool for clinicians called the 5-SENSE score, has been proposed, based on a composite of non-invasive data including “strongly localizing semiology”. Of course for this to be useful and correctly applied, the team needs to understand which semiologic features are “strongly localizing” and which are not; this is quite easy for some signs/patterns (and teams), much more difficult or even unknown for others. It should be remembered that this can be somewhat of a moving target since knowledge in this domain is always advancing. In addition, semiology (like EEG interpretation) is not an exact science and experts will not always agree on the significance or even the appearance and nomenclature of patterns, adding to the complexity of this area. Inter-observer agreement between experts can however be high even for complex semiologic patterns. As research advances on seizure semiologic expressions and their neural correlates, clinical knowledge in this field can potentially benefit from more evidence-based data and be less prone to the risks of dogma.


As well as keeping in mind some individual clinical correlations that are usually fairly robust (e.g., déjà vu linked to peri-hippocampal structures; auditory hallucinations linked to lateral temporal neocortex; visual hallucinations linked to occipital cortex, clonic jerks linked to contralateral motor cortex), it can be helpful to think in terms of systems based on clusters of semiologic features being correlated to clusters of connected brain structures (i.e., networks). As such, one approach is to think about the patient’s semiology in broad terms of “motor system,” “sensory system,” “memory system,” “autonomic system,” “emotional system,” “consciousness system,” and so on, and consider the groups of connected anatomic structures that underlie these general brain functions. This can allow the clinician to look at the bigger picture of possible anatomic correlates and avoid being too rapidly confined to a narrow set of choices, which is very important when trying to decide on optimal SEEG electrode implantation strategy. For example, a rising epigastric sensation (within the larger group of sensations termed abdominal aura ), is considered to be an autonomic sign and may occur with focal seizures involving any structure within or projecting to the central autonomic network (which is usually considered to comprise amygdala, anterior insula, anterior cingulate and their closely connected structures, i.e., structures related to the limbic system). When epigastric sensation is reported at seizure onset, especially with a rising character, the balance of probabilities of localization is most often in favor of a role for mesial temporal structures because of their particularly frequent involvement in focal epileptic seizures , , but all other semiologic features of the seizure should be considered in their sequence (as well as imaging data etc.), bringing additional clues to primarily temporal or extra-temporal organization. This is why it is often appropriate in SEEG exploration to sample some structures that are anatomically connected to the main hypothesized region of seizure onset, as alternative hypotheses that could potentially produce a similar clinical picture. As such, in a case undergoing SEEG of suspected mesial temporal seizures with epigastric rising sensation as first clinical feature, it would be typical to decide to implant not only the mesial temporal structures but the anterior insula and often the orbitofrontal cortex, which could be alternate seizure routes potentially capable of producing similar semiology. Confirming or refuting the role of these different structures in seizure organization lends robustness to surgical decision making and can help predict likelihood of seizure freedom after surgical therapy.


How does electrical seizure onset and spread influence semiologic output?


Not only the presence of different semiologic features, but the timing of their appearance relative to each other and relative to electrophysiologic data are very important. In general, most weight is placed on symptoms/signs occurring at seizure onset or in the early part of the seizure, as these will likely reflect brain activity changes most directly linked to electrical seizure onset. Some seizures evolve gradually with a stepwise appearance of semiologic features, which if present tend to be intuitively and technically more straightforward for the stereo-electroencephalographer to compare with the equivalent timepoint on the electrical trace. Many focal seizures can show progressive semiologic evolution over seconds or even minutes, for example, an insulo-opercular seizure beginning with feeling of buccal sensation progressing to facial contraction then facial clonus evolving over 30–60 s. If SEEG sampling is optimal or at least sufficient, it will often be possible to see a relation between different signs emerging and changes in SEEG activity, which may be spatial (spreading to other structures), temporal (frequency, amplitude of discharge), and most often both (e.g., oro-alimentary automatisms associated with opercular seizure discharge propagation in a temporal lobe seizure as shown in Clinical Vignette 1).


However, some seizures on SEEG do not show such an evident stepwise progression either semiologically or electrographically, or both. These are always more challenging to interpret, in which the repertoire of possible semiologic features can also be wide and the discharge may appear synchronized over various distributed structures from onset. This could be for example a frontal lobe seizure with a semiologic picture of hyperkinetic behavior, complex tonic posturing and vocalization displaying simultaneous SEEG onset across widespread prefrontal and premotor structures; or a parietal lobe seizure presenting with sensory symptoms, tonic limb signs and head version, showing synchronous SEEG seizure activity in parietal, precentral and premotor structures of the same hemisphere. In addition, the main seizure dynamics driving semiologic output in some cases (such as repetitive motor behaviors, altered awareness, emotional expression) may critically depend on activities in non-sampled structures included subcortical ones, albeit tightly linked to the cortical component of the seizure. We must therefore always keep in mind when interpreting SEEG that much of the brain participating in the seizure dynamic is inevitably not sampled-we have to try to “see” the seizure organization by extrapolating the information that we can obtain from the electrodes that are there, hopefully well-enough placed to allow this, helped by our knowledge of structural and functional anatomy (brain connectivity).


As the timing of appearance of different semiologic features may give essential information as to seizure propagation, it is usually very helpful to annotate these different time points onto the SEEG trace, to facilitate comparison with electrical activity. In particular, it is important to pay attention to the time difference (latency) between the first electrical change and the first semiologic feature, and whether this is subjective (signaled by the patient), objective (visible on video) or objective from another source (e.g., observed by witness, tachycardia observed on electrocardiogram). The time-lag between onset of electrical discharge and onset of semiologic features is important in any recording but becomes essential in SEEG, because appearance of clinical features of a spontaneous seizure that precedes obvious discharge usually indicates a problem of inadequate electrode sampling.


Analyzing semiology of stimulated seizures triggered during SEEG of course brings very useful complementary information to spontaneous seizures, using the same principles of anatomo-electroclinical correlation, and this topic is discussed in a separate chapter in this book. It is worth mentioning here that some stimulation-induced seizures on SEEG may have somewhat different time relations, with clinical signs appearing relatively earlier with respect to electrical discharge compared to spontaneous seizures. Ideally stimulated seizures will always be compared to spontaneous SEEG seizures in the same patient.


“Simple” anatomo-electroclinical correlations—Example of a mesial temporal seizure


To illustrate the simpler and relatively straightforward end of the spectrum of anatomo-electroclinical correlations, the example will be taken here of a prototypical mesial temporal seizure of predominantly hippocampal onset. However, many seizures studied on SEEG will be much more complex than the recognizable pattern described below. In addition, temporal seizures as a group are not necessarily simple in their organization and many different patterns and sub-groups have been characterized using SEEG group-level data.


Seizure onset: In this hypothetical seizure, the discharge starts in the hippocampus with repetitive high amplitude spikes followed by low amplitude discharge. This type of ictal discharge is often completely asymptomatic for the patient with mesial temporal lobe epilepsy, and the seizure onset, which will be readily seen on SEEG, would most often not be visualizable at this same point on scalp EEG. This is known from clinical experience in comparing SEEG and video-EEG seizures in the same patients, and also demonstrated in studies using simultaneous SEEG/EEG.


Early spread network: Perhaps after 10 s or so, the hippocampal discharge starts to spread spatially and temporally (i.e., in its frequency and amplitude) to involve other closely connected structures (e.g., amygdala, anterior insula, entorhinal cortex). This is often the point at which the patient may signal their aura, that is, the first subjective semiologic manifestation, which if reported indicates that the patient feels something but remains aware and able to communicate, a clinical observation that itself yields additional semiologic information. Thus, the semiologic production related to seizure onset depends on the “seizure onset zone” (SOZ) (in this case, hippocampus, to which seizure onset was limited according to this SEEG example) but also the “early spread network” (the connected structures in which seizure discharge first spreads to, and the temporal features of the electrophysiologic relationship to the SOZ). The original sense of the epileptogenic zone (EZ) as conceived by Bancaud takes into account both SOZ and early spread network underlying the initial semiologic expression, in both its spatial and temporal aspects.


The clinical manifestations of this “early spread network” activity for a given patient’s seizure will tend to reflect the specific structures involved and how their activity is linked to that of the seizure onset. For example, a seizure that spreads predominantly from hippocampus to amygdala and/or anterior insula might manifest as feeling of fear ± autonomic signs, whereas hippocampal seizure onset that is synchronous with or rapidly spreads to involve entorhinal cortex might be more likely to manifest as a feeling of déjà vu. The same seizure onset beginning in hippocampus and rapidly engaging lateral temporal neocortex would be more likely to be associated with impairment of consciousness, critically dependent on discharge frequency. However, a direct one-to-one relationship between structure and sign usually cannot be inferred outside of primary cortical areas, even in the relatively “simple” case of the mesial temporal seizure described here, and as a general rule it is more useful to think in terms of mapping clusters of signs onto clusters of connected brain structures.


Later propagation: Next in our prototypical example, let’s say the seizure discharge then further spreads spatially and temporally (at this point likely with a lower frequency, higher amplitude recruiting spike discharge, synchronized across the various involved anatomic structures). This spread may be from mesial temporal structures into temporal pole, lateral temporal cortex, basal temporal cortex, opercular regions, insula, posterior cingulate, orbitofrontal cortex, as well subcortical structures such as hypothalamus and basal ganglia, and/or contralateral homologous anatomical structures, i.e., any of the usual connected structures related to hippocampus, which may be more or less co-engaged in different patients’ seizure propagation patterns.


During this propagation phase (which might be taking place over a 30–90 s time window in this prototypical mesial temporal example), other signs will emerge that constitute further objective semiology for that patient’s seizure. Again the clinical manifestations of the seizure will depend on the specific seizure dynamics and spatiotemporal propagation pattern, which will be prone to recur in a stable pattern across seizures for each patient , and may reflect changes in connectivity underlying those particular propagation patterns. This helps to explain why patients tend to have a reproducible core semiologic signature, even if semiologic variations can occur according to various other influences including environmental conditions, antiseizure medications, psycho-physical state of the patient at the time of the seizure, and so on.


The semiologic expression of this later phase of seizure propagation will depend on both the specific anatomic structures involved (e.g., predominantly basal or lateral temporal spread, vs. operculo-insular spread, vs. frontal spread, vs. contralateral spread) and the timescale relationship between these structures’ involvement and the original seizure onset (e.g., low voltage fast discharge vs. rhythmic spike discharge; degree of synchronization between structures; delay (latency) after which each new structure becomes involved; etc.). For example, mesial temporal seizures predominantly spreading to temporal and extra temporal cortex and thalamus, with abnormally increased synchronization of the rhythmic discharge across these structures, may be predisposed to manifest altered consciousness. , (Initial/early ictal loss of consciousness is uncommon in typical mesial temporal seizures and is more usually a feature of lateral or mesio-lateral temporal lobe seizures ). Mesial temporal seizures spreading to opercular cortex with a rhythmic theta band discharge are more likely to manifest chewing automatisms, if theta band coherence is present between temporal and opercular regions (see also Fig. 8.6 ). On the other hand, mesial temporal seizures spreading to temporal pole and/or orbitofrontal cortex with a low voltage fast discharge are more likely to manifest complex motor behaviors that may display hyperkinetic features. Mesial temporal seizures with prominent basal ganglia spread would probably be more prone to display contralateral dystonia, which has not so far been explicitly documented on SEEG (since limited electrophysiologic data on basal ganglia correlations with semiology have been documented) but would be in keeping with data from functional neuroimaging of dystonia in temporal lobe seizures. , Mesial temporal seizures that progress to secondary generalization with head version and facial contraction most likely do so via perisylvian spread to precentral and premotor cortex, although specific propagation patterns of focal to bilateral generalized tonic clonic seizures remain incompletely known.


Thus a common focal seizure onset pattern, in this case from hippocampus, may display a variety of propagation patterns and thus a corresponding variety of anatomo-electroclinical correlations. This example can be seen as “simple” since the semiologic repertoire of mesial temporal seizures is both fairly limited and well known and the SEEG dynamics of mesial temporal seizure production are also well characterized, , including their propensity to be reproducible for a given patient. , See also Clinical Vignette 1 for an example of anatomo-electrical correlations of a different mesial temporal seizure type, characterized by amygdalar onset.


What About More Complex Anatomo-electroclinical Correlations?


The above scenario of a prototypical mesial temporal seizure seems fairly intuitive when trying to understand links between spatiotemporal seizure discharge and semiologic expression. However, a misleading aspect of SEEG work is to assume that seizures “should” be very focal (as in, with a seizure onset limited to one or possibly two electrodes) and that anatomo-electroclinical correlations “should” be explicit. In fact, fairly often in SEEG practice, patients may not have such a clear-cut seizure organization, even when an appropriate implantation has been performed based on expert interpretation of non-invasive data. Complex seizure types are very likely to be encountered by stereoelectroencephalographers, so it is best to be prepared for intrinsic differences between simple and more complex anatomo-electroclinical patterns. The difficulties can arise from semiologic challenges, complex SEEG signal patterns, sub-optimal electrode sampling, or any combination of these.


In terms of semiology, the interpretation of complex semiologies should be somewhat anticipated on SEEG following the patient’s Phase I video-EEG recording of habitual seizures. Hyperkinetic or other complex motor patterns are always more difficult to interpret than seizures with simpler motor signs. (In addition hyperkinetic seizures during SEEG can create management issues if movements are very agitated or violent, as there can be a risk of electrode damage and/or harm to the patient). Difficulties can also present when clinical features are subtle and difficult to assess (e.g., some seizures with no motor or other clear objective features, perhaps also inadequately interrogated for consciousness/language features), or when initial symptoms are non-specific and poorly defined, or fluctuating. The ictal examination is key here; stimulation studies can also be helpful in more clearly revealing semiologic features and their electrophysiologic correlates.


In terms of electrophysiologic pattern and relation to semiology, some very complex semiologic patterns can still have a clearly defined and focal seizure onset pattern and spread (e.g., a prefrontal seizure starting with a “silent” orbitofrontal or frontopolar discharge that expresses with complex motor behavior only during the early propagation phase involving cingulate and dorsolateral frontal cortex; see also Case Vignette 2 for an example of a frontal lobe seizure). As a general rule, propagation patterns of seizures with complex motor semiology (as opposed to elementary motor signs) will tend to be more widespread, as they arise within more densely connected regions, thus facilitating rapid, multi-directional propagation, because of the connectivity pattern of heteromodal cortex.


A main challenge arises when seizure patterns on SEEG look very widespread in a synchronized (simultaneous) fashion from the outset across many brain regions, which can potentially occur for any lobar localization, perhaps especially for frontal and posterior cortex seizures. In its most extreme from this can appear like a “generalized” epileptic seizure pattern at seizure onset, for example, as described in some frontal seizures. This can raise the question for stereoelectroencephalographers of whether the seizure onset is genuinely characterized by a widespread synchronization, or whether the true seizure onset has been missed through electrode sampling error. Here, the relation of signal to semiology is very important in resolving this question, since semiologic features preceding the earliest SEEG changes will indicate inadequate electrode sampling; in addition the semiologic expression may provide clues as to the likely degree of focality (or not) of the recorded seizure. (The use of SEEG signal analysis tools in conjunction with semiologic data can also help to make more sense of widespread fast activity, see other chapters for more discussion of this). It is worth reiterating here that a crucial component of SEEG methodology is patient selection, as mentioned earlier: some widespread ictal patterns on SEEG are the end result of sub-optimal patient selection (e.g., patient with pre-existing “red flag” signs of a potentially widespread epilepsy organization, such as bilateral surface EEG abnormalities, normal MRI and PET, non-localizing semiology).


Framework for Thinking About Semiology: Cortical Hierarchy


As a general rule, seizures involving associative cortical areas rather than primary cortical areas are prone to display a much greater repertoire of clinical features across patients and tend to involve more variable and widespread network patterns, both spatially and temporally. , This observation is most likely explained by the more complex cytoarchitecture and connectivity of associative cortical areas. We can think of seizure discharge within primary cortex as primarily expressing elementary clinical signs with a more linear, “one-to-one” mapping of sign to brain region. This spatial specificity can still be modulated by temporal features of the electrical discharge, for example different seizure discharge frequencies determining the occurrence of either tonic or atonic signs from the same region of motor cortex. On the other hand, seizures involving cytoarchitecturally higher level heteromodal cortex, because of its multi-level connectivity and integrative role, will tend to arise within co-involved connected structures from the early spread phase. As such these seizures will be clinically expressed as more elaborate behaviors or symptoms, made up of complex combinations of signs, determined by the specific anatomic networks involved (i.e., constrained by connectivity) and influenced by many temporal aspects of electrical discharge (e.g., latency between structures, frequency, synchrony, coherence, phase lag) ( Fig. 8.17 ). This explains why we may not be able to visually see obvious correlations between individual semiologic features with SEEG activity in individual anatomic structures in such seizures, even if such correlations nevertheless exist at the sub-visual level of analysis. We know however that patient group level correlates can be demonstrated by mapping clusters of signs to clusters of connected brain structures, , and as such even complex ictal behaviors and symptoms show relation to cortical seizure organization as assessed by standard visual analysis of SEEG, for many sublobar localizations of seizures. This last point is important because pathophysiologic mechanisms of semiologic production, especially complex ictal behaviors, are poorly understood for the most part, and almost certainly involve inhibitory as well as activating effects of cortical and subcortical circuitry in different seizure types at the epileptogenic network level. As such, we should not really expect to fully understand all of the semiological correlates from individual SEEG electrode signal in such cases, at least not in a step by step way throughout the duration of the seizure discharge. Nevertheless, cortical signal-semiology correlations can be demonstrated even for complex patterns in which a main subcortical driving dynamic is present yet tightly linked to cortical seizure discharge, such as ictal rhythmic body rocking. This example serves to highlight the intricate relations between the epileptogenic zone and the emergence of semiology, in which the cortical seizure onset activity is perhaps inevitably and reproducibly “hooked” to the network underlying semiologic outflow, whether this is simple or complex, focal or widespread. This serves as a working framework in which we can aim to better stratify the specificity of relation of behavior (semiology) to brain activity, that is, better define the predictive power of different semiologic patterns for cerebral localization of seizures.


Conclusion


Electroclinical correlations of the Phase 1 non-invasive data are key to correct diagnosis, classification and localization of epilepsy. These pave the way for optimal anatomo-electroclinical correlations during SEEG. We recall that semiology and EEG have been considered as “inseparable” sources of data with regards to seizure classification. , This is because seizure semiologic expression and its corresponding brain electrical activity (constrained by structural and functional connectivity) can be seen as two modalities expressing the same underlying dynamic system operating within a network that alters reproducibly with each seizure. The electrical activity is the causal mechanism of the semiologic expression, but not necessarily in a linear and transparent way. In the context of presurgical evaluation, the degree of specificity of semiological patterns in relation to certain brain systems/regions/structures that may be involved in initial seizure organization carries great weight when deciding whether clear hypotheses of unifocal, spatially constrained EZ exist, and thus largely influences decision to proceed (or not) to SEEG, and implantation strategy. In other words, as clinician-electroencephalographers, we need to be able to recognize when we can predict likely cerebral correlates from other data with some certainty, and we also need to “know when we don’t know.” This important question requires ongoing clinical research to better understand the stratification of specificity of different semiologic signs and patterns.


In patients proceeding to SEEG, the process of anatomo-electroclinical correlation is the core methodology for data interpretation. Much progress has been made on elucidating granular spatial (sublobar) correlations at group level as well as putative mechanisms underlying some semiologic expression, but many unknowns remain. SEEG is the only available method allowing distributed, multi-lobar sampling and millisecond signal capture that is time-locked with clinical seizure expression, and as such is currently the best available tool for improving knowledge of seizure patterns, dynamics and anatomic correlates. In the future, artificial intelligence methods might optimize data obtained from video-recorded seizures, for example by allowing quantification of ictal movements or eventually detecting subtle features that are hard for humans to see , which will be a very valuable tool for ongoing research and in the future could potentially help increase clinicians’ accuracy in decision making. This is where further harnessing signal analysis of SEEG, linked to video analysis techniques applied to seizure semiology, , ideally studying large datasets reflecting the large repertoire of semiologic expression, will be very important for making future progress in our recognition and understanding of complex seizure patterns, with important implications for clinical practice.


Clinical Vignette 1


31-year-old right-handed female, with no significant past medical history, presents with panic attack like symptoms for the last 5 years. Patient was diagnosed with focal epilepsy at age 26. Patient did not respond to two different antiseizure medications. Her seizures occur on daily basis, lasting up to 1 min.


Semiology


Her seizures start with feeling rushed and anxious, they are characterized by a feeling of déjà vu; feeling as if she was in the same place before. Moreover, she tries not to talk during the seizure to avoid making her seizures worse. On examination, the patient has a flushed face then 10–15 s later she has rhythmic oro-alimentary movements. She also has tachycardia at the beginning of her seizures. Patient is able to follow commands during the seizures. There are no speech arrest or language deficit during or after seizures.


Phase 1 Video-EEG


Patient had multiple typical seizures characterized by déjà vu and feeling anxious, lasting up to 60–90 s. There were no clear electroencephalogram (EEG) changes during these episodes. Two years later, patient was re-admitted to the epilepsy monitoring unit (EMU) and 13 habitual clinical seizures were detected. Interictal scalp EEG showed occasional spike and slow wave discharges in the right temporal region (sphenoidal) SP2>F8>T8 ( Fig. 8.1 ). Ictally, EEG change was seen 9 s after the clinical onset. Patient was able to push seizure button for all of her seizures. The earliest EEG changes were seen in the right sphenoidal electrode Sp2 with low amplitude 8–12 Hz followed by rhythmic 2–4 Hz slow waves were seen in the right temporal chain (Sp2>F8>T8). Her first oro-alimentary movements co-occurred with high amplitude rhythmic 2–4 Hz 60–100 uV in Sp2>F8>T8 ( Fig. 8.2 ). Postictally, the patient was able to communicate, repeat, and remember.


Mar 2, 2025 | Posted by in NEUROSURGERY | Comments Off on Anatomo-electro-clinical Correlations

Full access? Get Clinical Tree

Get Clinical Tree app for offline access